Implementation research in health: a practical guide / edited by David Peters … [et al]

A PRACTICAL GUIDE IMPLEMENTATION RESEARCH IN HEALTH ©WHODavid H. Peters, Nhan T. Tran, Taghreed Adam David H. Peters, Nhan T. Tran, Taghreed Adam Implementation Research in Health: A Practical Guide Implementation Research in Health A Practical Guide2 © World Health Organization 2013 All rights reserved. Publications of the World Health Organization are available on the WHO web site (www.who.int) or can be pur- chased from WHO Press, World Health Organization, 20 Avenue Appia, 1211 Geneva 27, Switzerland (tel.: +41 22 791 3264; fax: +41 22 791 4857; e-mail: bookorders@who.int). Requests for permission to reproduce or translate WHO publications –whether for sale or for non-commercial distribution– should be addressed to WHO Press through the WHO web site (www.who.int/about/licensing/copyright_form/en/index.html). The designations employed and the presentation of the material in this publication do not imply the expression of any opinion what- soever on the part of the World Health Organization concerning the legal status of any country, territory, city or area or of its authori- ties, or concerning the delimitation of its frontiers or boundaries. Dotted lines on maps represent approximate border lines for which there may not yet be full agreement. The mention of specific companies or of certain manufacturers’ products does not imply that they are endorsed or recommended by the World Health Organization in preference to others of a similar nature that are not mentioned. Errors and omissions excepted, the names of proprietary products are distinguished by initial capital letters. All reasonable precautions have been taken by the World Health Organization to verify the information contained in this publication. However, the published material is being distributed without warranty of any kind, either expressed or implied. The responsibility for the interpretation and use of the material lies with the reader. In no event shall the World Health Organization be liable for damages arising from its use. The named authors alone are responsible for the views expressed in this publication. Printed by the WHO Document Production Services, Geneva, Switzerland Designed by Corrales Creative Suggested citation: David H. Peters, Nhan T. Tran, Taghreed Adam. Implementation research in health: a practical guide. Alliance for Health Policy and Systems Research, World Health Organization, 2013. WHO Library Cataloguing-in-Publication Data Implementation research in health: a practical guide / David Peters … [et al]. 1.Health services research - standards. 2.Health policy. 3.Delivery of health care. 4.Research design. 5.Health plan imple- mentation. I.Peters, David. II.Tran, Nhan. III.Adam, Taghreed. IV.Alliance for Health Policy and Systems Research. V.World Health Organization. ISBN 978 92 4 150621 2 (NLM classification: W 84.3) 3Implementation Research in Health A Practical Guide contents Acknowledgements ........................................................................................................................................................4 Preface ...........................................................................................................................................................................5 Foreword ........................................................................................................................................................................6 Executive Summary .........................................................................................................................................................8 Chapter 1: Why is research on implementation needed? ...................................................................................13 Chapter 2: How is implementation research used? .............................................................................................19 Chapter 3: What is implementation research? .....................................................................................................27 Chapter 4: Who should be involved in implementation research? ...................................................................35 Chapter 5: What approaches and methods are appropriate for implementation research? ........................45 Chapter 6: How should implementation research be conducted? ....................................................................57 Chapter 7: How can the potential of implementation research be realized? .................................................61 References ....................................................................................................................................................................64 List of Tables Table 1: Effects of quality improvement on screening and follow-up for cervical cancer in El Salvador ........................23 Table 2: Types of strategies used to improve implementation in health ......................................................................29 Table 3: Implementation outcome variables ..............................................................................................................30 Table 4: A comparison of participatory action research and conventional research .....................................................50 Table 5: Types of implementation research objectives, implementation questions, and research methods ...................55 List of Figures Figure 1: Reported smallpox cases, by month, from 1960-1967, and in 1968-1969, in 20 West and Central African countries ...................................................................................................................14 Figure 2: The Prevention of Mother-To-Child-Transmission (PMTCT) cascade in Zambia (2007-2008) .........................15 Figure 3: The continuum of implementation research.................................................................................................31 Figure 4: Plan-Do-Study-Act cycle and research tools that can be used at each stage ................................................48 List of Boxes Box 1: Ghana’s Dangme West health insurance scheme ............................................................................................17 Box 2: Context-specific research becomes context-specific implementation ...............................................................20 Box 3: The role of implementation research in assessing and improving performance ................................................21 Box 4: The importance of implementers in implementation research ..........................................................................36 Box 5: Policy-makers and researchers come together on road traffic injuries in Malaysia ............................................40 Box 6: District health teams use implementation research to build human resource capacity in Africa ........................41 Box 7: A pragmatic trial in South Africa .....................................................................................................................46 Box 8: Effectiveness-implementation research applied in Bangladesh newborn care study .........................................47 Box 9: Participatory action to improve neonatal health care ......................................................................................50 Box 10: Implementation theory .................................................................................................................................53 Implementation Research in Health A Practical Guide4 Acknowledgements This Guide was developed by the Alliance for Health Policy and Systems Research (AHPSR) with support from the Implementation Research Platform (IRP). The IRP was formed in rec- ognition of the need for reliable, relevant research to inform the implementation of health policies and programmes, and to ensure that needed interventions are made widely avail- able within health systems through effective scale-up. The AHPSR is the host of the IRP Secretariat and leads its work. The AHPSR’s main goal is to promote the generation and use of health policy and systems research as a means to improve health and health systems in low- and middle-income countries. Although the Guide has benefited from the contributions and advice of many individuals, responsibility for the views expressed and for any errors of fact or judgment rests with the authors. In particular, the authors would like to acknowledge Irene Agyepong for her contri- butions to the planning, case-studies, and review of the Guide; George Pariyo for his contri- butions to the planning and review of the Guide; and Sally Theobald for her contributions to the case-studies and review of the Guide. The authors would also like to thank the following individuals for their review and feedback on the Guide: Garry Aslanyan, Rajiv Bahl, Neal Brandes, Somsak Chunharas, Soraya Elloker, Abdul Ghaffar, Lucy Gilson, Margaret Gyapong, Luis Huicho, Jose Martines, Karstein Maseide, Garrett Mehl, Olumide Ogundahunsi, Kelechi Ohiri, Enola Proctor, Krishna Rao, Suzanne Reier, Abha Saxena, Jim Sherry, Rajendra Shukla, and Jeanette Vega. Finally, we would like to acknowledge Gary Humphreys for his help in bringing this project together on the page. 5Implementation Research in Health A Practical Guide PReFAce One of the greatest challenges facing the global health community is how to take proven interventions and implement them in the real world. Research on health systems, such as implementation research, is crucial to meeting that challenge, providing a basis for the context-specific, evidence-informed decision-making needed to make what is possible in theory a reality in practice. The World Health Organization (WHO) has long played a leading role in promoting health policy and systems research (HPSR)—including implementation research, with notable recent initiatives including the 2011 report: Implementation Research for the Control of Infectious Diseases of Poverty and the 2012 publication of its first strategy on HPSR called ‘Changing Mindsets’, which advocated for greater embedding of research into decision- making and called for more demand-driven research. With this Guide, WHO continues its support for this area, offering an introduction to the often challenging field of implementa- tion research. Because implementation research is a relatively new and somewhat neglected field, there is a need to bring it into sharper focus, defining exactly what it is and what it can offer. As such, this Guide presents an introduction to basic concepts used in implementation re- search and describes the range of approaches and applications that it can be used for. The main aim of the Guide is to support the development of and demand for implementation research that is problem-focused, action-oriented and above all aligned with health system needs. Research on implementation requires the engagement of a wide range of stakeholders and draws on multiple disciplines in order to address complex implementation challenges. As this Guide points out, at its best, implementation research is often a collective and collab- orative endeavor and in many cases it is people working on the front line of health care, whether running specific programmes, or working in health systems, who ask the questions around which it is built. It is therefore essential that all stakeholders understand the im- portance of collaboration in the implementation research endeavor. It is our hope that this Guide will encourage that collaboration, and facilitate the coming together of stakeholders across the broad spectrum of health systems, all of which, on a daily basis, wrestle with the challenge of implementation. Marie-Paule Kieny Assistant Director-General Health Systems and Innovation Cluster World Health Organization Implementation Research in Health A Practical Guide6 FoRewoRd Interest in implementation research is growing, largely in recognition of the contribution it can make to maximising the beneficial impact of health interventions. As a relatively new and, until recently, rather neglected field within the health sector, implementation research is something of an unknown quantity for many. There is therefore a need for greater clarity about what exactly implementation research is, and what it can offer. This Guide is designed to provide that clarity. Implementation research engages a wide range of interventions in a health system, and for the purposes of the Guide we consider a wide variety of policies, programmes, as well as individual practices and services intended to improve people’s health. Regardless of the way these interventions work – be it through the prevention of disease, promotion of good health, or treatment and/or palliation of illness – research on their implementation is crucial to understanding how they work in the real world. Intended to support those conducting implementation research, those with responsibility for implementing programmes, and those who have an interest in both, the Guide provides an introduction to basic implementation research concepts and language, briefly outlines what it involves, and describes the many opportunities that it presents. The main aim of the Guide is to boost implementation research capacity as well as demand for implementation research that is aligned with need, and that is of particular relevance to health systems in low- and middle-income countries (LMICs). Research on implementation requires the engagement of diverse stakeholders and multiple disciplines in order to address the complex implementation challenges they face. For this reason, the Guide is intended for a variety of actors who contribute to and/or are impacted by implementation research. This includes the decision-makers responsible for designing policies and managing programmes whose decisions shape implementation and scale-up processes, as well as the practitioners and front-line workers who ultimately implement these decisions along with researchers from different disciplines who bring expertise in systematically collecting and analysing information to inform implementation questions. It is also our hope that the Guide may appeal to educators who teach implementation research, to funders of health research and health programmes who may be interested in supporting this type of research, and to civil society groups interested in health programmes and research who may want to use this evidence to promote good clinical and public health practice. 7Implementation Research in Health A Practical Guide While for the sake of simplicity we refer in the Guide to implementers and researchers, we recognize that these are notional groupings, and that many people play both roles. Indeed it is one of the core contention of the Guide that the interests of implementation research are often best served where there is active participation by people working in the field, since it is those people who see where implementation is going wrong and who are therefore likely to ask the most pertinent questions. We encourage implementers to ask those questions, and to take a lead in demanding better research. We also encourage researchers to be more practical in their work, focusing on the issues that matter to implementers. If research is to improve implementation, research design needs to reflect the specific implementation problems implementers are addressing as well as the contexts in which they occur. The opening chapters make the case for why implementation research is important to decision-making. They offer a workable definition of implementation research and illustrate the relevance of research to problems that are often considered to be simply adminis- trative and provide examples of how such problems can be framed as implementation research questions. The early chapters also deal with the conduct of implementation research, emphasizing the importance of collaboration and discussing the role of implementers in the planning and designing of studies, the collection and analysis of data, as well as in the dissemination and use of results. The second half of the Guide detail the various methods and study designs that can be used to carry out implementation research, and, using examples, illustrates the application of quantitative, qualitative, and mixed-method designs to answer complex questions related to implementation and scale-up. It offers guidance on conceptualizing an implementation research study from the identification of the problem, development of research questions, identification of implementation outcomes and variables, as well as the selection of the study design and methods while also addressing important questions of rigor. Implementation Research in Health A Practical Guide8 A key challenge faced by the global health community is how to take proven interventions and implement them in the real world. Affordable, life-saving interventions exist to confront many of the health challenges we face, but there is little understanding of how best to deliver those interventions across the full range of existing health systems and in the wide diversity of possible settings. Our failure to effectively implement interventions carries a price. Each year more than 287,000 women die from complications related to pregnancy and child birth, for example, while approximately 7.6 million children, including 3.1 million newborns, die from diseases that are preventable or treatable with existing interventions. UndeRstAnding imPlementAtion in the ReAl woRld Implementation issues arise as a result of a range of factors including ‘real world’ contex- tual factors that are either overlooked or not captured by other research disciplines. Imple- mentation research shines a light on those factors, providing the basis for the kind of context-specific and evidence-informed decision-making that is crucial to making what is possible in theory a reality in practice. Because implementation research is embedded in re- ality, people working in the real world (practitioners as opposed to people ‘doing research’) often ask the questions that are the starting point for new thinking. Making sure that those questions are heard, and that the research undertaken is directed at finding answers to the questions asked rather than at the topics researchers themselves may find interesting is one of the key challenges implementation researchers face. A PRActicAl tool Embedded in the real world, implementation research is also a powerful tool for capturing and analysing information in real time, allowing for the assessment of performance, for example, and facilitating health systems strengthening. Implementation research is particu- larly important in supporting the scale-up of interventions and their integration into health systems at the national level. Too often interventions that work in small-scale pilot studies fail to live up to expectations when rolled out in national strategies, or fail to transfer from one country to another as a result of contextual differences. Implementation research not only helps to clarify why that happens, but can be used to support the process of re-iterative refinement needed for successful adaptation. The same capacities make implementation re- search a useful tool for helping organizations develop the capacity to learn, enabling them to assimilate and put into effect knowledge developed on an iterative basis. execUtive sUmmARy 9Implementation Research in Health A Practical Guide A collAboRAtive endeAvoR Implementation research is often at its most useful where implementers have played a part in the identification, design and conduct phases of the research undertaken. It is for this reason that the fostering of collaborative ties between key stakeholders involved in policy generation, programme management, and research is so important. One way to support collaboration between researchers and implementers is to integrate implementation re- search into policy and programmatic decision-making processes right from the beginning, rather than an endeavor that is carried out separately from the implementation process. In this way, scientific inquiry can also be integrated into the implementation problem-solving process in an iterative, and continuous manner. Implementation research can also play an important role in acting as a vehicle for grass roots communities by identifying neglected issues, exposing performance shortcomings and increasing the accountability of health or- ganizations. In all of these collaborative endeavors, implementation researchers are called upon to be receptive and flexible in the work they do. Indeed the understanding of context and sys- tems, and the flexibility to identify appropriate methodological approaches, can be as im- portant as or even more important than adherence to a fixed research design. A bRoAd ReseARch sPectRUm Broadly speaking, the term implementation research describes the scientific study of the processes used in the implementation of initiatives as well as the contextual factors that af- fect these processes. It can address or explore any aspect of implementation, including the factors affecting implementation (such as poverty, geographical remoteness, or traditional beliefs), the processes of implementation themselves (such as distribution of fully-subsi- dised insecticide-treated bednets (ITNs) through maternal health clinics, or the use of mass vaccination versus surveillance-containment), and the outcomes, or end-products of the implementation under study. Implementation research is applicable and relevant to many different research domains, and to different degrees, depending on the subject under study. For example, basic research into new medicines typically involves no implementation is- sues, while ensuring that those medicines are available to the people who need them does. Implementation research often focuses on the strategies needed to deliver or implement new interventions here called ‘implementation strategies’, a term used to distinguish them from clinical and public health interventions. In order to study implementation process- Implementation Research in Health A Practical Guide10 es, a framework for conceptualizing and measuring implementation outcomes based on variables such as ‘acceptability’, ‘appropriateness’ and ‘feasibility’, can be used to understand how well a given implementation process is actually working. A wide RAnge oF APPRoAches Because it draws on a wide variety of research approaches and disciplines, it makes little sense to talk in terms of a set of ‘implementation research methods.’ However, certain research approaches and designs – including pragmatic trials, effectiveness–implementa- tion hybrid trials, quality improvement studies and participatory action research – are particularly useful because they generate actionable intelligence, are good at capturing the subtleties of context over time, and offer the iterative flexibility needed to respond to change. While such tools are vital to the implementation researcher, it is important to bear in mind that in implementation research, the “question is king” which is to say that it is the question that determines the method used, rather than the method that determines the kinds of questions asked. Implementation research questions are often complex, reflecting the wide array of contextual factors that can influence implementation, producing unpre- dictable effects, and requiring continuous adaptation by implementers. Embracing that complexity requires considerable flexibility on the part of researchers, particularly in regard to the complex and dynamic nature of the subject under study. Aligning ReseARch with need And ensURing QUAlity Ideally, implementation research should be aligned with need, both in the sense that it ad- dresses the concerns of the intended audience, and is also responsive to the particularities of the subject under study. A key consideration in this regard is the level of certainty required regarding results or projections. A policy-maker working with his or her own constraints, for example, may be looking for strong indications that a given intervention will work, but may not necessarily have the time required for multi-year studies that could generate a higher level of certainty. Responding to the needs of different audiences, may have important implications for the basic design of research, and for budget and scheduling. In order to ensure that implementation research is aligned with need and of high quality, it is helpful to ask the following key questions: k Does the research clearly address a question concerning implementation? k Is there a clear description of what is being implemented (e.g. details of the practice, programme, or policy)? 11Implementation Research in Health A Practical Guide k Does the research involve an implementation strategy? If so, is it described and examined appropriately? k Is the research conducted in a real-world setting? If so, are these conditions de- scribed in sufficient detail? k Does the research appropriately consider implementation outcome variables? k Does the research appropriately consider context and other factors that influence implementation? k Does the research appropriately consider changes over time, and the level of com- plexity of the system? k Does the research clearly identify the target audience for the research and how it can be used? getting moRe oUt oF imPlementAtion ReseARch Despite the importance of implementation research, it continues to be a neglected field of study, partly because of a lack of understanding regarding what it is and what it offers, and partly because of a lack of investment in implementation research activities. We spend billions on health innovations, but very little on how best to use them. This problem affects everyone, but in particular populations in low- and middle-income countries where the implementation challenges are greatest. This guide is an attempt to redress the deficit in understanding of implementation research and to encourage programme personnel and implementers to take a greater interest in the subject, recognizing that implementation research is in fact an integral part of programme planning and execution, rather than some- thing that happens once programmes are up and running, and conducted largely for the benefit of other researchers. For their part implementation researchers can do much more to engage with implementers and programme personnel in research process. Only by coming together can implementers, with their intimate understanding of context, and research- ers, with their understanding of the methods and science of inquiry, hope to advance our understanding of the implementation issues that compromise so many of our public health endeavors. why is ReseARch on imPlementAtion needed?1 k Despite abundant evidence of the efficacy of afford- able, life-saving interventions, there is little under- standing of how to deliver those interventions effec- tively in diverse settings and within the wide range of existing health systems. k Implementation issues often arise as a result of con- textual factors that policy-makers and health system managers may not even have considered. k Implementation research is crucial to improving our understanding of the challenges we face in confront- ing the real world by broadening and deepening our understanding of these real-world factors and how they impact implementation. k Implementation research is of immense value in shin- ing a light on the often bumpy interface between what can be achieved in theory and what happens in practice. key Points ©WHO 13Implementation Research in Health A Practical Guide why is ReseARch on imPlementAtion needed? “ Neglecting implementation challenges costs lives and money.” In December 1966 Doctor William Foege drove into the Eastern Nigerian bush to investigate a reported smallpox outbreak. A great deal was already known about how to deal with the disease at that time; indeed a global campaign to eradicate smallpox had been running since 1959, and there had been several breakthroughs in vac- cine production and delivery, notable among which the development of freeze-dried smallpox vaccine, and the use of a simple bi- furcated needle to deliver an effective dose. But at the end of 1966 smallpox was still circulating in 31 countries and territo- ries, and there were concerns that the goal upon which the whole campaign was based – mass vaccination – was unattainable. In reality mass vaccination meant vaccinating 80% of the population concerned, a level at which so-called herd immunity is achieved. But even with good quality vaccine and the simple magic of the bifurcated needle, achieving that lev- el of coverage was proving extremely difficult [1]. There was a problem of implementation. Then came Nigeria. Having confirmed that the reported outbreak was indeed smallpox, Foege set about mount- ing a response, and realized that he did not have suf- ficient vaccine to achieve the mass vaccination of the population, only 35% of which was currently vaccinated against the disease. Meanwhile, fresh supplies of vaccine and the extra vehicles needed to transport it were several weeks away. He needed a backup plan. In the end, that plan involved going to the affected vil- lages and vaccinating the people who had not yet caught the disease, effectively ring-fencing the hot spots with a barrier of immunity. Realizing that some people, many of whom might be unaware that they were infected, were going to be moving around, notably to the markets where they bought and sold food and goods, Foege also mapped out local transportation routes and the markets they served. Then, using the remainder of the vaccine available to him, he built rings of resistance in the areas he had identified as being most likely to see new cases. This surveillance-containment strategy – as it came to be known – shut down smallpox outbreaks in Eastern Nige- ria in just five months, and was achieved by vaccinating just 750 000 of a population of around 12 million [2]. Foege was by no means the first to use the method, indeed a similar approach had been used to control outbreaks in England in the 19th century, but he was the first to use it in the demanding conditions of sub-Saharan Africa, and the achievement resonated. The problem of achieving herd immunity had been shown to be irrelevant. The power of the approach was soon con- firmed in other African countries (see Figure 1), and even more impressive results were achieved elsewhere, nota- bly in the Indian state of Tamil Nadu, where, the following year, D. A. Ramachandra Rao led a single team on a sur- veillance-containment campaign which halted smallpox transmission among the 41 million inhabitants in just six months [2]. Global scale-up of surveillance-containment followed, and smallpox was officially declared eradicated in 1979. Of course the implementation of the surveillance-con- tainment strategy was not, strictly speaking, implementa- tion research or what was then known as ‘field research’ – in many ways a more descriptive term. Foege was sim- ply responding to an outbreak. However, the results of his efforts were of immense value for research, and, as chAPteR 1 • Why is research on implementation needed? Despite abundant evidence of the efficacy of life-saving interventions, there is little understanding of how to deliver those interventions effectively. Implementation Research in Health A Practical Guide14 this Guide will show, some of the most significant break- throughs in implementation knowledge have come from people who were not ‘doing research’ at the time of their discovery. One of the central mes- sages of this Guide is that contributions to imple- mentation research can be made by people both inside and outside academia, and that very often it is the per- son in the field – the doctor in the remote rural clinic or the midwife working in the local community – who, facing some particular problem, asks the questions that are the starting point for new thinking. Making sure that those questions are heard, and that the research undertaken is directed at finding answers to the questions asked rather than at the topics researchers themselves may find interesting is a subject that we will return to. Roughly half a century since Foege made that drive into the Eastern Nigerian bush, we are still struggling to make the best use of the vaccines available to us, even if the challenges we face have, generally speaking, changed. At present, LMICs seeking to imple- ment vaccination strategies are more likely to be struggling to in- troduce new vaccines or integrate immunization programmes into the services routinely provided by their health systems. In doing so, they face a range of obstacles to effective implementation that include mana- gerial, systems, socio-behavioural, and financial challenges, any one of which can limit the impact of a vaccination programme and hinder progress towards better health. Just as it underpinned efforts to scale-up Foege’s surveil- lance-containment approach to smallpox immunization, effective implementation research can play a vital role in improving our understanding of these challenges and chAPteR 1 • Why is research on implementation needed? Very often it is the person in the field – the doctor in the remote rural clinic or the midwife work- ing in the local community – who, facing some particular problem, asks the questions that are the starting point for new thinking. Figure 1. Reported smallpox cases, by month, from 1960-1967, and in 1968-1969, in 20 West and Central African countries JAN 200 0 400 600 800 1,000 1,200 1,400 1,600 1960-1967 Average N um be r of C as es 1960-1967 Average Includes 20 Countries 1968 1,800 2,000 FEB FEBMAR MARAPR APRMAY MAYJUNE JUNEJULY JULYAUG AUGSEPT SEPTOCT OCT 1968 1969 NOV NOVDEC DECJAN Source: Foege et al., 1971 [2] 15Implementation Research in Health A Practical Guide support the implementation of routine immunization pro- grammes within health systems. In this way, immuniza- tion, instead of being delivered only through special cam- paigns, can be delivered on a sustainable basis as part of routine new-born infant and child health services. And the same is true of many other interventions, which, like immunization, often fall short of their promise through failures of implementation. Despite the medical and public health advances of the past century, each year more than 287 000 women die from complications related to pregnancy and child birth, and approximately 7.6 million children, including 3.1 mil- lion newborns, die from diseases that are preventable or treatable with existing interventions [3]. More often than not, these deaths are the result of problems with implementation, problems that can be elucidated with carefully designed, planned and executed implementa- tion research. As stated in the introduction, the central concern of this Guide is implementation research as it relates to LMICs, where, despite abundant evidence of the efficacy of af- fordable, life-saving interventions, there is little under- standing of how to deliver those interventions effectively in diverse settings and within the wide range of existing health systems [4]. We know that insecticide-treated bed- nets reduce malaria transmission rates, but need more research into the most effective and financially sustain- able way to deliver those bednets and how to ensure that they are properly used. We know that oral rehydration therapy (ORT) is extremely effective in treating diarrhoea- chAPteR 1 • Why is research on implementation needed? Source: Mandala et al., 2009 [5] Figure 2. The Prevention of Mother-To-Child-Transmission (PMTCT) cascade in Zambia (2007-2008) 4,000 6,000 8,000 10,000 12,000 14,000 16,000 Pregnant women tested HIV+ 14,815 2,528 1,680 796 581 35.5% drop out 82.9% drop out 27.0% drop out Those checked for CD4 A P R I L 2 0 0 7 - M A R C H 2 0 0 8 Those with results available Those found eligible for cART Those initiated on cART 0 2,000 Implementation Research in Health A Practical Guide16 related disease, but struggle to achieve adequate levels of ORT use. Similarly, we know that anti-retroviral treat- ment (ART) programmes can prolong the lives of people living with HIV, but too often fail to ensure that everyone who needs treatment gets it. As shown in Figure 2 ,only a small proportion – less than 4%, of women who test positive for HIV during pregnancy in Zambia – are actu- ally initiated on ART in order to prevent mother-to-child- transmission (PMTCT) of HIV [5]. The reasons for these failures are all too fa- miliar to implementers and programme man- agers: human resource shortages, supply-chain issues, inefficient dis- tribution channels, and barriers to patient ac- cess that include obvious obstacles such as high out-of- pocket costs or lack of conveniently located facilities, to more subtle barriers such as socioeconomic or gender discrimination, or cultural values and preferences that may prevent communities from accessing and/or ben- efiting from the interventions available. In some cases implementation issues arise outside the health system, occurring as a result of contextual factors that policy- makers and health system managers may not even have considered. This was certainly true of the voucher scheme to support the distribution of ITNs that was first piloted in Ghana in 2004, for example [6]. The basic idea of the scheme was to give each pregnant woman a voucher for an ITN at her first visit to an ante- natal clinic. The voucher entitled her to a discount of ap- proximately US$ 4.20 on the purchase of an ITN available through formal sector retail outlets. A similar scheme had been launched in the United Republic of Tanzania with considerable success at the end of the 1990s. In Ghana however, the scheme ultimately failed, the main reason for the difference in outcomes being the nature of ITN production and distribution in the two countries. Both the United Republic of Tanzania and Ghana have polyester textile mills capable of making nets for ITN pur- poses, but only in the United Republic of Tanzania did these mills convert some of their textile production into nets, largely as a response to direct appeals by the gov- ernment and the business opportunity provided by the initial social marketing for the ITN programme. In Ghana, private sector businesses produced locally-sewn nets. Unfortunately, during the roll-out of the voucher scheme in Ghana, the informal sector was overlooked, the scheme’s designers preferring to work through the formal private sec- tor which was not only not be- ing supplied by the domestic polyester textile mills, but had limited capacity to import ITNs into the country. As a conse- quence, the formal private sector was unable to meet the demand generated by the voucher scheme, and mothers found themselves presenting vouchers that could not be honoured. While it has often been remarked that hindsight is a won- derful thing, it is probably fair to say that properly con- ducted implementation research, with its all-important focus on context, would have helped implementers fore- see and anticipate the problems that arose in Ghana, notably by highlighting the role of the informal private sector in supplying ITNs. It is because of its capacity to il- luminate contextual issues that implementation research is such an important tool for implementers at the plan- ning stage and there are numerous examples of imple- mentation research supporting effective policy-making and programme design. The development of Ghana’s Dangme West Health Insurance Scheme is a case in point, the scheme’s designers making full use of different imple- mentation research tools to anticipate likely challenges (Box 1) [7, 8]. chAPteR 1 • Why is research on implementation needed? Properly conducted implementation research, with its all-important focus on context can help implementers foresee and anticipate problems. 17Implementation Research in Health A Practical Guide conclUsion Implementation research is, then, of immense value in shining a light on the often bumpy interface between what can be achieved in theory and what happens in practice. Engaging with the real world, and drawing much of its strength from real-world practitioners and the communities they serve, implementation research generates context-specific insights that are simply not available from narrower research perspectives. How exactly the potential of implementation research can be realized in different applications is the subject of the next chapter. chAPteR 1 • Why is research on implementation needed? In the late 1990s, the needs of the people of Dangme West District, a poor rural district just outside of Accra, Ghana, were served by just four public sector rural health centers, none of which had doctors. To make mat- ters worse, user fees, introduced in 1985, had resulted in many poor people foregoing even basic health services. There was thus increasing pressure to find a viable alternative method of financing health that did not include out-of-pocket fees imposed at point of service use. Interest in the possibility of developing health insurance was qualified by the fact that most people in the district were either very poor or worked outside the formal sector. Research to test the feasibility of a health insurance scheme in Dangme West was conducted using a consul- tative and iterative process of formative research involving the District Health Management Team (DHMT) in collaboration with the District Assembly (local government) and community members of the Dangme West District. Extensive discussions, at both the community and individual level were used to arrive at a feasible design, drawing on examples from other LMICs, as well as a study of the history of health insurance and its evolution in high-income countries. Practical details such as the constituents of a suitable benefit package, payable fees, and likely costs were the subject of broad consultation as was the challenging issue of how ben- eficiaries were to be identified and how access to care was to be ensured, given the limited service availability in the district. After protracted consultation, the decision was taken to introduce a benefit package covering all outpatient care in the health centers in the district. The health centers were also assigned a gatekeeper role, providing referrals for patients needing hospital care to the four hospitals in the surrounding districts. All household members had to register with the scheme and the premium was set per member rather than per household. Many of the findings and experiences of this research eventually informed the Ghana National Health Insurance scheme development, which was passed into law in 2003 as a national scheme. Box 1. Ghana’s Dangme West Health insurance scheme Sources: Waddington et al., 1990 [7]; Agyepong et al., 2007 [8] how is imPlementAtion ReseARch Used?2 k Implementation research is vital to understanding context, assessing performance, informing implemen- tation and facilitating health systems strengthening. k Implementation research is particularly important in supporting the scale-up of interventions and integrat- ing them into health systems at the national level. k Implementation research can also be used to help organizations develop the capacity to learn. key Points ©WHO 19Implementation Research in Health A Practical Guide how is imPlementAtion ReseARch Used? “ Implementation research takes what we know and turns it into what we do.” Having made the broad argument for the importance of implementation research, we now turn to some of the specific areas where it can be of enormous value to a range of stakeholders from ministerial-level decision- makers, who may use implementation research to sup- port health policy formation, to programme managers seeking to understand context-specific issues, and health providers looking to assess performance, make changes, or introduce innovations. For all these stakeholders, im- plementation research offers a window into the practical challenges presented by the provision of health services in the real world. UndeRstAnding context As the Ghana ITN example cited in the previous chapter demonstrates, implementation research has an important role to play in elucidating the contextual factors that can influence the outcomes of interventions. This is im- portant because even when interventions are designed in similar ways, there is evidence to suggest that imple- mentation occurs differently in different contexts, and with many different effects [4]. For example, it has been amply demonstrated that the Integrated Management of Childhood Illness (IMCI) strategy, a systematic approach to child health which puts equal emphasis on curative care and disease prevention, is hugely dependent on lo- cal health system characteristics such as the basic skills of health workers, the supply of drugs, and the effectiveness of supervision and referral [9]. IMCI is also dependent on the extent to which patients avail themselves of the services on offer, and implementation research can be of particular value in identifying and describing barriers to access, as evidenced by research on the “three delays model” (deciding to seek care, getting transport to care, and receiving quality care once at a health facility), which played a central role in overcoming the challenges of re- ducing maternal and newborn mortality [10]. As noted in the previous chapter, implementation re- search is particularly valuable in shining a light on the sometimes subtle cultural barriers which may escape strictly quantitative information gathering. For example, gender roles and household decision-making author- ity can determine the extent to which individuals access needed health services, notably where decisions relat- ing to health seeking are deferred to the male heads of households; women often delay seeking care—even in emergencies and at the detriment of their own health and the health of their children, if the man is not home. Where such barriers do exist, implementation research- ers and programme managers can play an important part in changing the approach used to inform communities about the care available; for example, employing mes- saging about alternate decision-makers for health seek- ing when male heads of households are not home. This sort of initiative ensures that pregnant women do not delay accessing emergency obstetric services—a signifi- cant contributor to high maternal mortality rates in some areas. Thus implementation research can offer crucial insights at a number of levels for implementers who, generally speak- ing, recognize that implementation goes beyond simply reapplying the same template in country after country [11]. The value of such insights is illustrated by work go- ing on in Andhra Pradesh, India, where implementation research based on micro-planning has been used to focus on understanding the local context, including the percep- tions and needs of vulnerable populations (Box 2) [12]. chAPteR 2 • hoW is implementation research used? Even when interventions are designed in similar ways, implementation occurs differently in different contexts, and with many different effects. Implementation Research in Health A Practical Guide20 Assessing PeRFoRmAnce Implementation research can also be used to gauge performance over time and to serve as the basis for projections into the future. Where appropriate, it can also be used to look at the way similar organizations or programmes have per- formed, noting any disparities in outcomes. It is worth noting here that monitoring and evaluation (M&E) – activities that many implementers and managers already engage in on a routine basis – play a significant part in this kind of implementation research, often helping to define important research questions. This was certainly true of the work done in Afghanistan after the collapse of the Taliban regime in 2002 when a simplified ‘scorecard’ was used for M&E on health system performance and to establish annual priorities for improvement (Box 3) [13-15]. The approach revealed a number of constraints in delivering basic services that needed to be addressed through reallocation of funding, contract decisions, training, and further research, and formed a basis to assess several new implementation strategies to finance and deliver health services. chAPteR 2 • hoW is implementation research used? Efforts to collect data on India’s HIV/AIDS epidemic, have greatly sharpened the overall epidemiological pic- ture, but significant challenges remain to generating the evidence needed to allocate resources efficiently and implement an effective AIDS response. In the state of Andhra Pradesh, USAID’s Samastha Project, which is de- signed to support the implementation of India’s HIV strategy, has been working to refine such a response. The project’s strategies were based on findings from a needs assessment carried out in Andhra Pradesh in 2006 which identified a number of gaps in HIV health services that compromised the quality of care, and left many unaware of available services. To address these shortcomings, Samastha mounted a four-pronged response focusing on: improving access to quality clinical care in facilities; instituting a system of self-assessment and monitoring to prevent infection and improve quality; setting up a computerized management information system (CMIS) that tracks individuals to support programme planning and monitoring; and consolidating community outreach services by using micro-planning and support groups. Micro-planning is a context-specific “bottom-up” tool for developing, implementing, and monitoring activi- ties tailored to the needs of local communities. Communities wishing to improve local HIV services can use micro-planning to identify vulnerable populations, analyze the availability and accessibility of services, and prioritize service delivery according to the available services. Micro-planning is also invaluable for tracking cli- ents’ use of services. A key aspect of the micro-planning effort was the use of peer outreach workers (PORWs), people living with aids (PLHIV) who were chosen from the community to conduct outreach, referrals, and follow-up in the target areas. These PORWs played a critical role in that they initiated and maintained contact with the project’s target populations, and also provided essential details about the local context, and notably the needs and characteristics of the community in their coverage area. The PORWs used a range of tools to help them identify PLHIV and vulnerable individuals, develop action plans to address each person’s needs, and make sure that clients used the necessary services. As a result of the micro-planning effort, registration of PLHIV increased from 610 in 2007 to 5 907 in 2011. Utilization of ART services also increased sharply, the number of those receiving ART or pre-ART services nearly quadrupled, while the number of eligible PLHIV who were not enrolled for ART services diminished from 228 to 18. Box 2. Context-specific research becomes context-specific implementation Source: Sankar, 2013 [12] 21Implementation Research in Health A Practical Guide chAPteR 2 • hoW is implementation research used? After decades of war and with the fall of the Taliban regime in 2002, Afghanistan’s health sector was in shambles. The physical infrastructure was destroyed, there were few health workers and managers, no func- tional health information systems. The meagre health services provided were largely managed by nongovern- mental organizations (NGOs). The new government produced a national health policy and strategy for the delivery of a Basic Package of Health Services (BPHS) to be implemented across the country with the help of donor organizations and local and international NGOs. With inputs by government officials, donor organiza- tions, NGOs, and front-line providers, and facilitation by an external evaluation team, a Balanced Scorecard of BPHS was developed to provide a summary of performance of the BPHS in each of the 34 provinces on an annual basis. The scorecard fits on a single sheet of paper, and covers six main domains (1. Perceptions of patients and community; 2. Staff perspectives; 3. Capacity for health services; 4. Service provision (quality and volume); 5. Financial systems; and 6. Overall vision on equity for women and the poor) and 29 indicators. Data were collected from observations made of over 700 randomly selected health facilities, over 7000 ob- servations of randomly selected patient-provider interactions and exit interviews. There were also over 2000 interviews with randomly selected health workers. The Balanced Scorecard has been used to highlight areas for training and allocation of resources, to provide bonuses on contracts or to cancel contracts, to motivate staff and provide accountability to the public, and as a basis for testing new policies. It also led to the removal of user fees at primary clinics and the extension of performance based payment schemes. Overall, trends have been improving in all areas on a national basis. Box 3. The role of implementation research in assessing and improving performance Sources: Peters et al., 2007 [13]; Hansen et al., 2008 [14]; Edward et al., 2011 [15] 40 50 60 70 80 90 100 2004 Sc or e (0 -1 00 ) 2005 2006 2007 2008 0 30 20 10 Domain A: Patients and Community*** Domain B: Staff* Domain C: Capacity and service provision*** Domain D: Service Provision*** Domain E: Financial Systems* Domain F: Overall Vision* *P<0.05, **p<0.001 ***p<0.0001 National Trends in Balanced Scorecard Domains Implementation Research in Health A Practical Guide22 chAPteR 2 • hoW is implementation research used? sUPPoRting And inFoRming scAle-UP The crucial importance of context in the successful imple- mentation of interventions clearly has implications for the way interventions are scaled up. Interventions that work and can be shown to work in small-scale pilot studies, too often fail to live up to expectations when rolled out in national strategies. We know what works, but do not always know how to make it work in the real world. The successful implementation of even simple interventions entails an understanding of the different ways in which the intervention is affected when the real world (commu- nity, health system, economy) interacts with it. Of course many new interventions are not simple at all, and may involve the introduction of a new diagnostic tool or new information/communication technology. Innova- tion may also take the form of an organizational change designed to support, for example, a new outreach ap- proach. Then again, the innovation might involve a change in a process in clinical, or administrative services, or a new project, programme, strategy, or policy. What- ever form the innovation does take, it is clear that simply bolting on a new component in the hope that things will improve is not sufficient. Actions have consequences and it is the job of the implementation researcher to identify and understand them. Nor is scale-up a simple matter of doing the same thing on a bigger scale. Achieving scale-up is generally equated with increasing geographic coverage from a limited study area. In fact this kind of scale-up is more accurately de- scribed as horizontal scale-up, or ‘spreading’, which typi- cally involves replicating an intervention, and stands in contrast to vertical scale-up, which is defined as involving the institutionalization of an innovation through policy, regulatory, budgetary, or other health system changes – in other words, the complex process of embedding an innovation in the institutional structure of a health sys- tem [16]. Scale-up can also concern the expansion of the organizational, financial, and technical capabilities of a health system. Needless to say, each of these forms of scale-up – and there are others, present particular chal- lenges for decision-makers and the researchers who sup- port them. To make that support effective implementation research- ers need to ask a number of key questions, including: k What are the projected effects of introducing and scaling-up the new component (both the intend- ed and unintended consequences)? k What lessons can be learned from other initia- tives to introduce and scale-up (or not scale-up) a similar intervention? k What are the main obstacles to/opportunities for scaling-up the intervention? k How will key stakeholders be affected? k How are different stakeholders disposed towards scaling-up (e.g. as facilitators, blockers, etc.)? k How well do different strategies for dealing with the different stakeholders work over time? k What are the projected costs of scaling-up? Interventions that work in small-scale pilot studies, too often fail to live up to expectations when rolled out in national strategies. 23Implementation Research in Health A Practical Guide sUPPoRting QUAlity imPRovement And heAlth systems stRengthening With regard to supporting quality improvement and health system strengthening, as with all health policy and systems research – of which implementation research is a form – the core con- cern is to ask questions that are relevant to the challenges faced. Imple- mentation research can yield many benefits but those benefits are maxi- mized where research is answering the questions that decision-makers and practitioners are asking, or should be asking. Health policy and systems research is often referred to as the brains of the health system, but it is also the eyes and the ears, the only mechanism that decision-makers can rely on for the constant stream of information they need to be able to adapt to changing circumstances optimally [11]. For this reason, some of the best implementation research is often supported, if not actually conducted, by practitioners in the field – the doctors, nurses, and managers who are confronted with quality issues on a daily basis. Implementation research is also of great value where it allows for an iterative approach to improvement, as ex- emplified by the case of El Salvador, where, in 2002 a Pan American Health Organi- zation and Ministry of Health- sponsored study sought to enhance organizational fea- tures of an early detection, cytology-based programme. El Salvador’s cervical cancer pro- gramme had many problems back in 2002: few women were getting screened with PAP smears, many laboratory samples were unsatisfactory, and follow-up colposcopy for those with positive tests was rare. The use of quality improvement cycles (plan- do-study-act) led to the training of outreach workers to identify women needing screening, to support access to screening, and encourage follow-up visits. After just one year, results improved dramatically (Table 1) [17]. The pro- cess was picked up by the Ministry of Health to expand the programme. chAPteR 2 • hoW is implementation research used? The benefits of implementation research are maximized where research is answering the questions that decisions-makers and practitioners are asking, or should be asking. Table 1. Effects of quality improvement on screening and follow-up for cervical cancer in El Salvador Pre-intervention Post-intervention Number of women aged 30–59 years of age screened for first time in last year Unknown (2446 samples) 3408 Number of unsatisfactory samples 41 (1.7%) 14 (0.4%) Turnaround time from clinic to laboratory (days) 23 9 Turnaround time from laboratory to clinic (days) 27 11 Number of women followed up with colposcopy for positive Pap 22 (24%) 196 (100%) Source: Agurto et al., 2006 [17] Implementation Research in Health A Practical Guide24 chAPteR 2 • hoW is implementation research used? A key driver of quality improvement, and indeed of health system strengthening generally is the learning capacity of organizations. Learning organizations have been defined as those structured in such a way as to facilitate learning as well as the sharing of knowledge among members or employees [18, 19]. Like learning people, learning orga- nizations are better placed to anticipate problems, and develop responses, thus becoming more effective. Key to the learning organisation’s ability to learn is the capac- ity to assimilate and put into effect new knowledge that develops iteratively as knowledge is tested and new les- sons learned [20]. Implementation research has an im- portant role to play in helping decision-makers ensure that the organizations for which they are responsible are able to learn, notably by asking questions such as: what approaches and processes can be used to create/build a learning organization; how can a health organization interact with its various stakeholders (e.g. governing bod- ies, financing bodies, regulators, beneficiaries, internal staff) to improve learning and organizational effective- ness? Needless to say, this is a complex issue and one to which we will return. tAcit knowledge Having discussed the importance of implementation re- search and outlined some of its key applications, it is per- haps worth acknowledging that implementation research cannot provide all the knowledge needed for successful implementation. Implementation know-how is also ac- quired through apprenticeship and experience involving observation and practice. This kind of know-how, some- times referred to as ‘tacit knowledge’, is an important part of learning for individuals and organizations [21]. That said, it is clear that the boundaries between tacit knowledge and the kind of formal knowledge derived from implementation research often overlap. For example, research on the tacit knowledge of health managers may provide important insights about implementation. Those who conduct implementation research or use research for decision-making learn much of their craft through ap- prenticeship and the tacit knowledge they gain in addi- tion to the formal methods they apply. conclUsion This chapter has attempted a brief overview of the broad applications of implementation research, including its value in elucidating the challenges and opportunities that arise when interventions are moved from the pilot study to the real world. The next chapter will discuss what exactly implementation research is, offering a practical definition that is applicable across the different research areas that it covers. It will also consider the implementa- tion strategies that support improved delivery of services, programmes and policies. Finally, it will describe the implementation outcome variables that can be used to characterize the various ways in which implementation occurs. 25Implementation Research in Health A Practical Guide whAt is imPlementAtion ReseARch?3 k Implementation research can address any aspect of implementation, including the factors affecting implementation, the processes of implementation themselves and the outcomes, or end-products of the implementation under study. k Implementation research is applicable and relevant in many different domains and, depending on the subject under study, is applicable and relevant to dif- ferent degrees, with certain research questions being implementation-light and others implementation- heavy. k Implementation research often focuses on the strate- gies needed to deliver or implement new interven- tions, which are referred to as ‘implementation strategies’. k In order to understand implementation processes, it is essential to use a framework for conceptualizing and measuring implementation outcomes. Implementa- tion outcome variables are the building blocks of this framework and serve as indicators of how well a given implementation is actually working. key Points ©WHO 27Implementation Research in Health A Practical Guide chAPteR 3 • What is implementation research? whAt is imPlementAtion ReseARch? “ The basic intent of implementation research is to understand not only what is and isn’t working, but how and why implementation is going right or wrong, and testing approaches to improve it.” Implementation research is a growing field of study with roots in many disciplines and research traditions. It ad- dresses a wide range of implementation problems in di- verse contexts. In many ways this is the great strength of implementation research – its capacity to tap into dif- ferent sources, bring to bear multiple perspectives and offer multisectoral insights – but it also presents some obvious taxonomic challenges of the kind that academ- ics sometimes shy away from. It is perhaps not surprising then that there is some con- fusion regarding nomencla- ture as well as significant debate regarding the scope of implementation research [22, 23]. Broadly speaking, the term implementation research describes the scientific study of the processes used in the implementation of initiatives as well as the contextual factors that affect these processes. One major purpose of implementation research is to support and promote the successful application of interventions that have been demonstrated to be effective – a drug that is known to kill malaria parasites, a diagnostic test that identifies who has tuberculosis, or a strategy to prevent the transmission of HIV from mother to child. It can be used to figure out how to deploy human resources so that geographically remote communities can access care when needed, and identifying how to eliminate financial barriers that pre- vent vulnerable populations from getting needed services. It is also about reducing costs and making organizations more efficient and accountable. Finally, implementation research is about learning how to bring promising strate- gies to scale, and importantly, how to sustain these strat- egies over the long term. Implementation research, as it applies specifically to health, is a type of health policy and systems research concerned with the study of clinical and public health policies, programmes, and practices, with the basic intent being to understand not only what is and isn’t work- ing, but how and why implementation is going right or wrong, and to test approaches to improve implementa- tion. As noted at the outset, very often it is concerned with the problems arising when an initiative is rolled-out or scaled-up. For the purposes of this Guide we propose a very simple and very broad definition of implementation research that can be used across research communities, and that has meaning for health sector practitioners, and policy- makers, as well as for the interested public: implementation research is the scientific inquiry into questions concerning implementation. Under this definition implementa- tion research can address or explore any aspect of implementation, in- cluding the factors affecting imple- mentation (such as poverty, geographical remoteness, or traditional beliefs), the processes of implementation themselves (such as distribution of fully-subsidised ITNs through maternal health clinics, or the use of mass vac- cination versus surveillance-containment), and the out- comes, or end-products of the implementation under study. As described above, implementation research may focus on issues such as: identifying common implementa- tion problems; understanding the factors that hinder or facilitate access to health interventions; developing and testing solutions to tackle implementation barriers, either within a specific context or across a range of environ- ments; and determining the best way to introduce inno- vations into a health system, or to promote their large scale use and sustainability. Implementation research is the scientific inquiry into questions concerning implementation. Implementation Research in Health A Practical Guide28 imPlementAtion stRAtegies While implementation research may not be concerned with discovering new health products, or testing the safety or efficacy of clinical interventions, it often deals with the strategies needed to deliver or implement those products or interventions. These strategies are sometimes referred to as ‘implementation strategies’, a term used to distinguish them from clinical and public health interven- tions [24]. For example, while outreach clinics and super- vision checklists are implementation strategies commonly used to improve the coverage and quality of immuniza- tion programmes, the provision of the vaccine itself is considered the health intervention. Implementation strat- egies may also be designed to improve the sociocultural aspects of implementation, for example by improving the acceptability or adoption of the intervention, or may af- fect things like the quality and cost of the services pro- vided. Implementation research may focus on the imple- mentation strategy itself, or incorporate consideration of the implementation strategy into a broader study of the health intervention. As Table 2 shows, one way of talking about implementa- tion strategies is to group them in terms of the actor or stakeholder using them. Typical implementation strate- gies include: (1) enhancing the capabilities of govern- ment (public policy, oversight and financing agencies); (2) improving the performance of implementing and provider organizations; (3) strengthening the capabilities and per- formance of individual providers and front-line workers; (4) empowering communities and households; and (5) supporting multiple stakeholders engaged in improving health [4]. imPlementAtion oUtcomes In order to advance our understanding of implementation processes, and enhance the efficiency of implementation research, a framework for conceptualizing and measuring implementation outcomes is essential. Such a framework also allows for much-needed studies of the comparative effectiveness of implementation strategies. In order to conceptualize and evaluate the success or failure of im- plementation, it is useful to employ some form of consis- tent taxonomy that allows us, for example, to talk about different aspects of implementation response – the ac- ceptability of an intervention, say, or the extent to which an intervention has been taken up or adopted. These characteristics can be seen as the outcomes of implementation, and are referred to as implementation outcome variables. Implementation outcome variables serve as indicators of how well a given implementation is actually working. The implementation outcome variables – acceptability, adoption, appropriateness, feasibility, fi- delity, implementation cost, coverage and sustainability – can also be seen as intermediate factors that contrib- ute to other important outcomes such as satisfaction with health care or health status [25, 26]. Not all implementation outcome variables are of equal importance in the delivery of an intervention, or for re- search on implementation (see Table 3) [25]. With regard to a novel intervention, for example, the main focus might be on issues of acceptability, adoption, appropriateness, and feasibility. For existing interventions, the degree to which the intervention is implemented as it was originally designed, or is faithful to the original (measured by the fidelity variable) is often very important, as are costs and coverage. Although sustainability issues should be con- sidered from the earliest phases of an intervention, they are frequently neglected in research on health interven- tions [27]. Each variable represents an important aspect of imple- mentation that can be studied through implementation research. chAPteR 3 • What is implementation research? 29Implementation Research in Health A Practical Guide chAPteR 3 • What is implementation research? Table 2. Types of strategies used to improve implementation in health Main Actor and Areas of Intervention Implementation Strategy Examples Government • Policy-making, oversight and regulation • Public financing • Policy reviews • Governance strengthening and corruption reduction strategies • Contracting with performance measures • Decentralize public service provision • Public financing incentives and rules (ways to raise revenues, pool funding, and payment mechanisms) • Public education, behaviour change communication Implementing and Provider Organizations • Organizational improvement and accountability • Quality improvement/quality assurance/performance management strategies: team problem-solving; develop and apply guidelines and standard operating procedures; regular supervision • Provide financial incentives for teams and individuals based on performance • Reorganize and/or integrate services • Human resource management systems • Facility management and logistics systems strengthening • Strengthen financial management • Marketing health services and products Individual Providers and Front-line Workers • Individual practices • Continuing education and training • Peer learning and support • Job aids Communities and Households • Empowerment, participation, education • Individual practices • Community information and education: training community health workers; train- ing of community members such as youth, mothers (in groups, home, mass media); social marketing and demand creation • Strengthen inclusion and participation: community-managed services; community partnerships and co-management; community-owned services • Strengthen local accountability: joint monitoring; provider accountability schemes; community-based information systems • Local organizational capacity building: community mobilization; community boards and structures to oversee and manage • Financial empowerment: community financing; in-kind subsidies and vouchers; community participatory budgeting; incorporation with income generating and micro-financing schemes • Peer support for health services and healthy behaviours Multiple Actors • Assess needs and constraints: constraints reduction plans • Obtain broad-based support of stakeholders: engage powerful interest groups; coordinate with community organizations • Flexible management processes and modification through stakeholder feedback Adapted from: Peters et al., 2009 [4] Implementation Research in Health A Practical Guide30 chAPteR 3 • What is implementation research? Table 3. Implementation outcome variables Implementation Outcome Working Definition* Related terms** Acceptability The perception among stakeholders (e.g. consumers, providers, managers, policy-makers) that an intervention is agreeable Factors related to acceptability: (e.g. comfort, relative advantage, credibility) Adoption The intention, initial decision, or action to try to employ a new intervention Uptake, Utilization, Intention to try Appropriateness The perceived fit or relevance of the intervention in a particu- lar setting or for a particular target audience (e.g. provider or consumer) or issue Relevance, Perceived fit, Compatibil- ity, Trialability, Suitability, Usefulness, Practicability Feasibility The extent to which an intervention can be carried out in a particular setting or organization Practicality, Actual fit, Utility, Suitability for everyday use Fidelity The degree to which an intervention was implemented as it was designed in an original protocol, plan, or policy Adherence, Delivery as intended, Treat- ment integrity, Quality of programme delivery, Intensity or dosage of delivery Implementation cost The incremental cost of the delivery strategy (e.g. how the services are delivered in a particular setting). The total cost of implementation would also include the cost of the intervention itself. Marginal cost*** Coverage The degree to which the population that is eligible to benefit from an intervention actually receives it. Reach, Access, Service Spread or Effec- tive Coverage (focusing on those that need an intervention and its delivery at sufficient quality, thus combining cover- age and fidelity), Penetration (focusing on the degree to which an intervention is integrated in a service setting) Sustainability The extent to which an intervention is maintained or institution- alized in a given setting. Maintenance, Continuation, Durabil- ity, Institutionalization, Routinization, Integration, Incorporation *The original definitions referred to individual “innovations or evidence-based practices”. This table uses the term “intervention” so that the definitions are more broadly applicable to programmes and policies. **Other terms are more commonly found in implementation literature on large-scale programmes and policies (Peters et al 2009; Rogers 2003; Carroll et al. 2007, Victoria et al 2005) ***Provides the numerator for related measures of efficiency and measures of cost-utility, cost-benefit, or cost-effectiveness. Many cost analysis examine the total cost of implementation of an intervention, including the cost of the intervention itself, as well as the costs of implementing a particular delivery strategy in a given setting. Adapted from: Proctor et al., 2011 [25] 31Implementation Research in Health A Practical Guide chAPteR 3 • What is implementation research? the continUUm oF imPlementAtion ReseARch As discussed earlier, one of the criticisms sometimes lev- elled at implementation research is that it lacks definition as a field of study. This is partly because it is applicable and relevant in so many different domains and partly be- cause, depending on the subject under study, it is applica- ble and relevant to different degrees. In order to grasp this idea, it is helpful to think about implementation research in terms of a continuum, with certain research activities being Implementation-light and others Implementation- heavy. In Figure 3 we represent this idea as a flow chart, with questions becoming more implementation intensive or heavy as we move downstream. Figure 3. The continuum of implementation research Implementation studied as contributing factors Research Questions: Co-primary or secondary question, e.g. effective- ness of program in all its variation Context: Real-world setting and population Implementation Strategies: One or more studied Implementation variables: May be used as independent variables Implementation relevant but effects reduced Research Questions: Secondary question, e.g. average effectiveness of a program Context: Real-world setting with partially controlled intervention Implementation strategies: Identified and described, but uses one type only and effects are controlled Implementation variables: Assumed to be equal or unchanging, or effects controlled (e.g. adjusted as confounding factors) Implementation as primary focus Research Questions: Primary question, e.g. How do parts of a program change and why? What are the ef- fects of implementation strategies? Context: Real-world set- ting and population Implementation strategies: May be primary focus Implementation variables: May be primary outcomes or determinants INNOVATION Examples: Basic science; Phase I & II clinical trials; Qualitative studies unre- lated to implementation issues (e.g. perceptions of illness) Examples: Efficacy stud- ies, Phase III randomized controlled clinical trial; Qualitative study on health service use that does con- sider how well the services are provided. Example: Pragmatic trials, Quasi-experimental study with intervention and com- parison areas; Observational studies with implementation as secondary issue Examples: Effectiveness- Implementation trials; Ob- servational studies assessing implementation variables as secondary factors; Participa- tory research Examples: Mixed methods and quasi-experimental studies to determine the changes in delivery or acceptability of a program; Observational studies on adaptation, learning, and scaling-up of a programme Implementation not relevant Research question: Basic sciences, product development, or inquiry unrelated to implemen- tation Context: Controlled or not related to implemen- tation Implementation strategies and variables: not relevant Implementation relevant but not considered Research question: Susceptible to implemen- tation variables, but not considered Context: Largely con- trolled, highly selected population, factors af- fecting implementation fixed or ignored Implementation strategies: None or one type only, not considered in research Implementation variables: Can influence results but assumed to be controlled or not relevant Informing Scale-up: Health systems integration and sustainability Proof of implementation: How does it work in real-world settings? R E S E A R C H Proof of concept: Is it safe and does it work? Implementation Research in Health A Practical Guide32 chAPteR 3 • What is implementation research? Thus, on the left side of the flow chart, we find research that does not involve implementation issues at all (such as basic research into a Zivoudine as a means of preventing mother-to-child-transmission of HIV). While at the right of the chart, we find research that is primarily concerned with question about implementation in real-world set- tings (for example, how to ensure that pregnant women who test positive for HIV in low-income countries get the prophylactic Zivoudine treatment needed to reduce the risk of transmission of HIV to her baby). As described be- low, as it focuses more on implementation strategies and studies implementation variables, the research becomes more implementation- heavy. It is important to note that research that is not concerned with imple- mentation can become highly relevant for im- plementation research. For example research into freeze-dried small- pox vaccine meant that the vaccine no longer had to be kept cold which had major implications for the implemen- tation of the smallpox eradication campaign. Thus basic research may yield new products that can later be tested as an intervention that improve health. Similarly, dis- coveries in the field (oral rehydration salts can be mixed on site and dispensed by non-specialized personnel, for example) can have implications further up the research chain (focus on developing non-caking oral rehydration salts). Research is implementation-heavy when it addresses questions focused on implementation, occurs in real- world settings, and at least considers factors that affect implementation, if not actually testing implementation strategies and implementation outcome variables. It is the focus of implementation research on context and the interaction between the real world and the intervention being studied that sets it apart from, say, routine monitor- ing, which measures progress in a specific area over time, without necessarily seeking to understand what may be influencing that progress (or lack of it). Monitoring is very often the starting point for implementation research when it extends to using research methods to investigate issues beyond the routing monitoring. Research occurring in controlled, non-practice settings, with highly selected samples that do not represent the intended population for the intervention, and where implementation outcome variables and factors affecting implementation are not relevant or are controlled in attempts to eliminate their effects, can be said to be implementation-light. Finally, it is important to note that this flow chart represents a significant simplification of a much more complex real- ity and is used here merely to illustrate the basic idea. In reality, health systems rarely deal with new innovations one at a time, for example, and often have to accom- modate multiple innovations (and other disruptions/changes) simultaneously. Thus the process of absorption and adaptation is much messier and more complex than has been presented here. conclUsion This chapter has presented some of the basic ideas that constitute implementation research, including a practical definition that can be used across research traditions. It has also attempted to offer a typology of implementation barriers and problems, a description of implementation strategies, and a description of implementation outcome variables that can be used in research to describe various aspects of the ways in which implementation occurs. In the next chapter we will be looking at the role of col- laborations and why they are necessary to understanding implementation in real-world settings, reflecting perti- nent contextual factors. Research is implementation-heavy when it addresses questions focused on implementation, occurs in real-world settings, and at least considers factors that affect implementation 33Implementation Research in Health A Practical Guide who shoUld be involved in imPlementAtion ReseARch?4 k Good implementation research is collaborative re- search, and often most useful where implementers have played a part in the identification, design and conduct phases of the research undertaken. k The fostering of collaborative ties between key stake- holders involved in policy generation, programme management, and research is essential. k Implementation research should be integrated into policy and programmatic decision-making so that sci- entific inquiry becomes a part of the implementation problem-solving process. k Implementation research can play an important role in “speaking truth to power”, by identifying neglected issues or by demonstrating performance and increas- ing accountability of health organizations. k Understanding of context and systems, and the flex- ibility to identify appropriate methodological ap- proaches, can be as important as or even more impor- tant than adherence to a fixed-research design. key Points ©WHO 35Implementation Research in Health A Practical Guide chAPteR 4 • Who should be involved in implementation research? who shoUld be involved in imPlementAtion ReseARch? “ Successful implementation research begins and ends with successful collaboration.” As highlighted in the previous chapter, one of the defin- ing aspects of implementation research is that it seeks to understand the reality of implementation in real-world contexts. Unlike other forms of research, it does not seek to filter out the extraneous or accidental; indeed, in many ways it is precisely such factors that are of interest to the implementation researcher. This approach implies a readiness on the part of the implementation researcher to embrace the unpredictable and otherwise problematic on occasion, in a way that other researchers might not. It also implies using study subjects in all their complexity and in their natural environments. This means working with popula- tions that are actually going to be affected by an intervention, for example, rather than selecting populations on the basis of narrow eligibility criteria (for example, choosing volunteers who have only one health condition when it is people with co-morbidities who are the target for the intervention). The considerations of context that are relevant to imple- mentation research include the social, cultural, economic, political, legal, and physical environment, including the prevailing demographic and epidemiological conditions. They can also include the institutional setting, compris- ing the particular institutional structures in place, and the various stakeholders working within them. The way health systems are structured, and the roles played by government, the private sector and non-governmental organizations (NGOs) are also of interest. Another key consideration for implementation research- ers is the audience for whom the research is being under- taken, or for whom it may be relevant. Specifically, imple- mentation researchers need to bear in mind the needs and/or limitations of their intended audience – the people or organizations that are going to be using the results. As noted in the last chapter, implementation research cov- ers a broad range of subjects, and the audience for this research is similarly broad, including, for example, health managers and teams working on quality improvement (QI) strategies, policy-makers looking to effect a change in policy or to introduce an entirely new policy, and prac- titioners who need support in using interventions that are based on the best evidence available. Each of these po- tential audiences has different requirements and differ- ent perspectives, and for implementation research to be of optimal utility it needs to take those differences into consider- ation. More often than not, the audience for this research is not another researcher, or academic, but a non-specialist in need of clear, evidence-based analysis uncluttered by jargon, which can form the basis of future deci- sions. Though there are exceptions, which we will discuss be- low, implementation research is most likely to be useful where implementers have played a part in the identifica- tion, design and conduct of the research undertaken, and are not just a passive recipients of results. As already not- ed, people on the front line of health care, whether run- ning specific programmes, or working in health systems have a great deal to contribute to the information-gath- ering endeavor (Box 4). Routine monitoring, for example, is often the starting point for many implementation re- search questions, and relies on the routine collection and analysis of current administrative and health data. There are a number of ways of encouraging feedback from the field, one of which being so-called Participatory Action Research (PAR), which, as the name suggests, involves participation of the research subjects, and has been de- scribed as research “carried out with and by local people Implementation research is most likely to be useful to its audience where that audience is not just a passive recipient of results. Implementation Research in Health A Practical Guide36 rather than on them” [28]. PAR was not designed spe- cifically to answer questions about implementation, but implementation is an obvious subject for participatory research. PAR also creates opportunities for communities that may not otherwise have a voice to “speak truth to power”, for example by offering an account of exploitive or abusive service provision. We will return to PAR, and other research methods in more detail in Chapter 5. The importance of both researchers and implementers coming together in the conduct of implementation re- search is of course of considerable importance in situ- ations where the core issues relate to quality improve- ment and the scale-up of a programme, both of which impact many stakeholders. As noted in the previous chapter, scale-up takes many forms and may involve not only the expansion of services, but also the development of the organizational, financial, and political capabilities of implementing organizations, and learning processes that engage implementing bodies, beneficiaries, funders and officials [29]. Such endeavors are necessarily multi- faceted, and thus require multifaceted research studies. Such studies are best carried out where there is strong collaboration. Above all, regardless of the subject under study, it is im- portant for both researchers and implementers to recog- nize the value of coming together in what is in fact a sym- biotic relationship – a relationship in which implementers generate feedback from the front lines, while researchers provide expertise in research methods needed for trust- worthy studies. Only by coming together in this way, can stakeholders ensure that the knowledge generated is valid, and aligned with need. the imPoRtAnce oF PARtneRshiPs in imPlementAtion ReseARch Given the importance of collaboration in implementation research, the skills needed to build and maintain part- nerships are a critical consideration. This includes the fostering of collaborative ties between key stakeholders involved in policy generation, programme management, and research. One interesting example of the way col- laboration can work is the above-cited case from Ghana, where researchers and implementers came together in the late 1990s to consider the optimal approach to devel- oping a national health insurance scheme (Box 1) [7]. Be- ginning with what is known as formative research, which involves studying the community for which a given inter- vention is being planned, the researchers and implement- chAPteR 4 • Who should be involved in implementation research? Given the nature of the questions implementation research asks, it is clear that implementers have a crucial part to play in its conduct, and should take an active role rather than being participants or passive partners in the overall research process. Indeed in certain cases – quality improvement (QI) studies, for example – implementers are ideally placed to lead this type of research, taking responsibility for almost all aspects of the research cycle. Implementers can make a number of important contributions, starting with the planning stage where they are best placed to identify implementation barriers and propose implementation research questions, as well as to identify solutions that can be tested. They can also play a role in designing studies, notably by facilitating understanding of context and the contextual factors that impact implementation, and contributing to frame research questions to reflect the existing situation. In terms of data collection, implementers are clearly in a privileged position with regard to accessing data sources, and interviewing respondents, while in terms of analysis (particularly for qualita- tive data), implementers can play an important role in making sense of and interpreting data. Finally implementers have a key part to play in the dissemination of results, notably by incorporating lessons learned into programme practice. Box 4. The importance of implementers in implementation research 37Implementation Research in Health A Practical Guide ers worked in close collaboration, making joint decisions about the design of the research to ensure that it not only had the capacity to provide evidence that was sound, but would do so while respecting the time constraints faced by the District Health Authorities [7]. The experience in Ghana shows not only the value of collaboration, but the importance of an evolving, iterative approach to imple- mentation research. embedding imPlementAtion ReseARch One way to improve collaboration and encourage part- nerships in implementation research is to integrate it into policy and programme decision-making. Because imple- mentation research often flows from well-established programme activities and is of direct benefit to pro- grammes, it makes sense to include it as an integral part of programme processes from the beginning rather than a tangential activity that is then used to provide context/ comparison that may inform programme development at a later date. Had implementation research been an in- tegral part of Ghana’s 2004 ITN pilot from the very be- ginning, the differences between Ghana and the United Republic of Tanzania with respect to ITN production and distribution would have been identified early enough for the pilot’s designers to make the necessary adjustments, thus saving time and vital resources [6]. In order to effectively integrate implementation research into the decision-making processes related to implemen- tation it is not enough to simply open lines of communi- cation with implementation researchers at an early stage; implementation research needs to be embedded in the overall design, planning and decision-making endeavor. This embedding can be achieved in three ways: first, by integrating the funding for research and programmatic activities; second, by systematically applying research and scientific inquiry as part of problem-solving; and, third, by using joint decision-making in the research and imple- mentation process. We consider each of these impera- tives in turn. k Integrating funding into research and programme activities A major driver of the way implementation research is conducted is the way it is funded. Research funding gen- erally flows through separate channels from programme funding. This is true for both international donors and governments that typically have one budget for pro- grammes and one for research. As a result of this separa- tion, research funding cycles are not always aligned with programme needs. Similarly, a good deal of research is awarded on a competitive basis, often on the initiative of individual investigators, whereas programme fund- ing typically is not competitive, and may require teams or programme managers to identify the questions of rel- evance. This too creates mismatches between programme needs and research objectives, and creates a disincentive for researchers to link their work to the actual barriers and challenges that are encountered during the imple- mentation of programmes. Moreover, the process and timeline for issuing calls for research proposals and se- lecting studies often exceeds the time available to those responsible for programmes to make important decisions about implementation. Where funding for IR is integrated within programme budgets, there are greater opportunities to align the re- search with programme needs. This is the approach that has been adopted by Bloomberg Philanthropies in two major global initiatives on tobacco prevention and road safety. In both instances, the foundation provided fund- ing for a consortium of partners, ranging from civil society, chAPteR 4 • Who should be involved in implementation research? Implementers generate feedback from the front lines, while researchers provide expertise in research methods needed for trustworthy studies. Implementation Research in Health A Practical Guide38 academic researchers, and technical agencies to collabo- rate on the implementation of large scale public health interventions. Each consortium partner plays a role and contributes to different aspects of the implementation. In the case of the tobacco initiative, the focus was on legislating tobacco control and research on the costs of tobacco control and policy implications was paired with advocacy efforts [30]. Similarly, for the Road Safety in 10 Countries (RS-10) initiative, researchers are working with implementing partners to support surveillance efforts, monitor and evaluate progress, as well to provide critical insights related to perceptions of stakeholders and com- munities on the interventions being implemented [31]. This collaborative approach not only gives important rec- ognition to the individual roles played by different part- ners in the implementation process, it enables alignment of research questions with programme needs while also facilitating the engagement of multiple stakeholders in the research process—thus ensuring that the evidence generated is used to inform implementation. k The systematic application of research and scientific inquiry in programme activities In addition to integrating research and programme fund- ing, it is crucial that the systematic application of scientific research be institutionalized within programme decision- making so that implementation research becomes a core part of the problem-solving process. This can be achieved in a number of ways. To begin with, establishing protocols and/or processes for decision-making related to imple- mentation and scale-up that explicitly refer to research is a good way of ensuring that problems and questions that need investigation are addressed in a systematic manner. The WHO/ExpandNET framework for scale-up is a useful example of how research and scientific inquiry can be integrated into processes. The framework includes ques- tions that need to be answered – sometimes through implementation research – as part of the nine steps that implementers need to consider when scaling-up a pro- gramme [16]. Integrating implementation research ques- tions into a framework for scale-up effectively embeds research into decision-making related to the scale-up of interventions. The WHO/ExpandNet framework also sug- gests a multidisciplinary approach using different actors to guide the scale-up process so that there is partnership and collaboration at all stages, including during the con- duct of implementation research [16]. Another way in which research and scientific inquiry can be integrated into programme decision-making is through mandatory M&E, the value of which as a basis for imple- mentation research has already been noted [32]. Requir- ing a link between the monitoring and implementation of programmes ensures that problems and challenges, some of which may be addressed through implementation re- search, are identified on a regular basis. Required evalua- tion of programmes also serves to enhance accountability and longer term learning. It also helps to systematize the way in which implementation challenges are understood, and to ensure the relevance of implementation research for the programme. An interesting example of the value of this type of embedding is the legislation requiring the systematic evaluation of all publicly-funded social pro- grammes in Mexico, which has helped to de-politicize policy-making while at the same time increasing the rel- evance of research evidence to implementers and other policy-makers [32, 33]. k Shared responsibility for decision-making The final aspect of embedded research is perhaps the most important, and also the most difficult to achieve. Programme implementation and scale-up both demand a certain level of flexibility and often develop under time pressure. Thus, any research study undertaken in sup- port of these activities needs to be responsive to these realities. This means that decisions about study designs, methods, and outcomes need to be informed not just by the perspectives of researchers, but must also reflect the views of implementers and other stakeholders. For example, researchers alone may desire evidence that is probabilistic and which can be substantiated through statistical analyses, whereas time pressures may dictate a study design which simply generates evidence regarding, chAPteR 4 • Who should be involved in implementation research? 39Implementation Research in Health A Practical Guide for example, whether a strategy has resulted in a certain level of coverage in a population or whether the quality of services meets certain standards (sometimes referred to as an adequacy statement) without necessarily taking into consideration whether outside influences caused the change [4, 34]. Similarly the questions that are the subject of implemen- tation research need, in many cases, to be jointly devel- oped by researchers and decision-makers to reflect their different perspectives. As already noted, implementers and researchers often come at problems from slightly different angles, implementers focusing on the specific barriers and challenges to implementation, and research- ers looking for ways to formulate questions that are suit- able for study and can be answered through research. This difference in agenda was apparent in the roll-out of the study on the use of visibility enhancement materi- als for motorcycle safety in Malaysia (Box 5). Researchers and decision-makers at different levels worked together to come up with research questions and a study design that generated evidence (on the feasibility of scale-up at a district level), which responded to the decision-makers’ needs, while employing quasi-experimental methods which ensured rigor and allowed for the independence and objectivity that the researchers were comfortable with [35]. The Malaysia example is evidence that different agendas can be satisfied, given a readiness to compro- mise, but it should not be seen as typical. The fact is shar- ing responsibility for decision-making is not always easy, nor will every decision be agreed upon by all. However, it does show that decisions can be informed by multiple perspectives and that the expertise and insights of differ- ent actors can be given due consideration. the chAllenges PResented by PARtneRshiPs Even in this brief presentation of embedded implementa- tion research, it should be clear to the reader that while collaborative approaches offer a number of opportunities, they also entail challenges. This is partly a reflection of the complexity of health systems and the way that the multiplicity of actors working within them interact. While it is commonplace to talk about critical actors in terms of their basic roles, breaking them down as decision-makers, programme managers, front-line health workers, and the patients they serve, for example, it is important to recog- nize that each role can be played out at different levels and that the boundaries between participants are some- times blurred. Thus, decision-making that sets the agenda and leads to the crafting of policies and programmes may initially oc- cur at the national level of a given health system, but when central decisions are handed down for implemen- tation, problem-solving or process-enhancing initiatives may lead to modification of the originally planned imple- mentation arrangements and of the central decisions. As such, while front-line workers are often predominantly re- sponsible for implementing already agreed decisions and policies, because of these local effects they can also end up being the setters of the decision-making agenda and the formulators of related policies and programmes. From this perspective, implementers, at all levels of health sys- tem operations, have a crucial part to play—sometimes even the leading role (particularly in quality improvement studies), in the implementation research endeavor. A good example of implementers engaging with the re- search process is provided by the district health manage- ment teams (DHMT) in Ghana, Uganda and the United Republic of Tanzania, who, supported by the PERFORM research group, used the Action Research (plan, act, ob- serve and reflect) approach to develop and then test con- text-specific management strengthening processes fo- cused on improving workforce performance (Box 6) [36]. chAPteR 4 • Who should be involved in implementation research? Collaborative approaches present a number of opportunities for implementation research, but they also present challenges. Implementation Research in Health A Practical Guide40 It is apparent then that the flow of information and the formation of ideas is often fed by feedback loops, and is not a strictly linear process. Moreover, because of the importance of process in implementation research, an understanding of context and systems, and the flexibility and creativity to identify appropriate methodological ap- proaches, can be as important as or even more important than adherence to a fixed research design informed by a particular disciplinary perspective. The clinical random- ized controlled trial, for example, depends on a fixed and reproducible intervention such as the taking of a pill, whereas implementation may involve interventions that change frequently and are not strictly reproducible. A fixed research design like a randomized controlled trial, while appropriate for studies of efficacy and effective- ness, may not be suitable for answering questions related to implementation. Another challenge inherent in collaboration is the some- times competing priorities of participants. For example, researchers may be under pressure to publish in high- impact journals that often favour specific disciplinary ap- proaches, while implementers may be under pressure to resolve the problem in the shortest time possible. Such chAPteR 4 • Who should be involved in implementation research? Despite the clear link between knowledge and action, interactions between those who generate information and those who are expected to use and apply that information are the exception rather than the rule, especially in LMICs. The trial of preventive measures designed to reduce motorcycle crashes and deaths in Malaysia is a good example of how this kind of collaboration can be made to work. In July 2005, a Department of Road Safety (DRS) was established within the Ministry of Transport (MoT) and meet- ings between DRS and other stakeholders were set up to assess policy-makers’ appetite to engage in new research regarding the possible benefits to be derived from using reflective materials to improve visibility of motorcyclists. Crucially, researchers and policy-makers came together very early on to agree on their shared objectives, and how to attain them. Policy-makers initially had a sceptical view of research, believing it took too much time to conduct and that they could resolve most of the challenges related to the district wide implementation of an intervention using visibility enhancement materials (VEMs). At the same time, the research team had initially envisaged a more complex study design of a longer duration in order to obtain a high degree of generalizability. Through the stakeholder consultations, discussions, and compromises, the group agreed upon a study field trial using a quasi- experimental design that would enable the scale-up of the VEM intervention in an entire district, as it was not politically feasible to randomize motorcyclists to receive the intervention within a district. This design would also enable the use of routinely collected police and hospital data for the analysis which reduced the time required for the baseline assessment; it also strengthened the link between research and programme activities. A field trial was particularly appealing to policy-makers because it presented an opportunity to achieve immediate, concrete results. It also offered the opportunity to understand how such an intervention could be scaled-up in the Malaysian context and how it could be sustained once the trial was finished. Of course the potential downside of conducting what would be very public research was that the VEM might turn out to be useless, but research- ers were able to convince the DRS that even negative findings would result in savings given that the cost of the research would be far less than the cost of a national VEM campaign that was ineffective. In the end the trial was given a public launch to raise awareness of the issue and the project was eventually branded the “Be Seen and Be Safe” campaign for motorcycle safety. Box 5. Policy-makers and researchers come together on road traffic injuries in Malaysia Source: Tran et al., 2009 [35] 41Implementation Research in Health A Practical Guide chAPteR 4 • Who should be involved in implementation research? differences, and there are many others, are not insur- mountable, but it is clear that without compromise and in some case sacrifices on both sides, collaboration will be problematic at best. One way for implementation researchers to improve their chances of successful collaboration with partners in the field is by getting out into the field. The best understand- ing of context and others’ perspectives comes from some experience of that context. Ideally, implementation researchers should spend some time living and working in the context (and the organizations) they intend to study, thus gaining insights that can inform their research de- sign, and working methods. This kind of immersion also helps in the development of listening skills, understanding others’ perspectives, engaging in dialogue, negotiation The dearth of competent, motivated health workers is a major impediment to improving health and saving lives in Africa. To address this deficit, a two-pronged strategy is required, ensuring the training of new health personnel and improving the performance of the existing workforce. A number of complex factors affect workforce perfor- mance, including staffing levels and distribution, the organization of work and required resources, working condi- tions and remuneration. Understanding the nature of these factors and developing appropriate responses to the challenges faced offers the opportunity to not only improve the performance of the existing workforce but also to reduce staff losses. PERFORM, a research group currently working in Ghana, Uganda and the United Republic of Tanzania, focuses on the building of local capacity and ownership for processes supporting human resource management and re- lated health systems activities. Specifically, PERFORM supports district health management teams (DHMT) who, through a series of workshops and review meetings use so-called Action Research (plan, act, observe and reflect) to develop and then test context-specific management strengthening processes focused on improving workforce performance. The DHMTs identify areas of health workforce performance to be improved, implement integrated human resource and health systems strategies feasible within the existing context to improve health workforce performance, and monitor the implementation of the strategies, evaluate the processes and impact on health workforce performance and the wider health system. The DHMTs then plan their own ‘bundles’ of human resource and health systems strategies, act to implement these strategies and observe the impact of the strategies they have developed on health workforce performance. They then reflect on how well their plans have been achieved and if necessary, revise the plan or address new challenges thus beginning to embed the process within their districts. The main outputs of this ongoing work will include developed and tested district-level interventions to improve both health workforce performance and the management of decentralised health systems more generally in the three African countries. These interventions can be potentially replicated in other districts in Ghana, Uganda and the United Republic of Tanzania and in other countries with decentralised health systems. The outcomes of the research will contribute to the body of knowledge of how strengthening management in sub-Saharan Africa can improve workforce performance and the wider health system. In addition, the action research approach will con- tribute to the development of skills and abilities of participating managers to resolve other management problems in the future, with the possibility of scaling-up this approach if it proves successful. Box 6. District health teams use implementation research to build human resources capacity in Africa Source: PERFORM Consortium, 2011 [36] Implementation Research in Health A Practical Guide42 chAPteR 4 • Who should be involved in implementation research? and collaborative problem-solving. Unfortunately, while such skills are critical aspects of implementation research, they are not generally recognized as research skills. imPlementAtion ReseARch in the FAce oF oPPosition While the kind of immersion just described is helpful, there are situations where embedding research in pub- lic policy processes is simply not possible. Policy-makers, managers, and funding agencies do not always want to know how their programmes are being implemented, un- less of course they can be shown to be doing well. They may have invested considerable political and financial capital in a policy, and be afraid of not producing the desired results or of poorly managing resources. Funders are frequently resistant to research that might highlight sustainability issues or the negative unintended conse- quences of their programmes, such as the human resource distribution problems arising as a result of hiring people for single purpose projects, an issue often encountered with HIV projects, among others [27]. Similarly, the con- cerns of minority groups may not be of interest to those groups in power, particularly if there are social and politi- cal sensitivities. Areas where this kind of problem arises include issues related to men who have sex with men (MSM), the treatment of aboriginal groups, the provision of abortion services, and pervasive dangerous practices such as female genital mutilation, etc. Implementation researchers who collaborate with disadvantaged groups or civil society organizations may find themselves unable to collaborate with those who oppose them. This can be a particular problem when research is conducted in an area suffering from ongoing civil conflict. In some cases participatory action research may even be considered revolutionary to the existing power structures. In these circumstances, an important aspect of implementation researchers’ work is to find ways to get their research into agenda-setting processes to influence policy. This may also require approaches that rely more on advocacy strategies that can make use of well-designed research. conclUsion This chapter has sought to identify the core issues related to implementation research, focusing in particular on the importance of conducting implementation research in real-world settings, paying attention to context, and be- ing mindful of the needs of the audience for whom the research is intended. It has also emphasized the impor- tance of collaboration and the need for partnership be- tween implementers, researchers and other stakeholders. In the next chapter we will focus on the principles behind common methods used in implementation research, and look at the ways in which research questions provide a basis for the research methods used. We will also be looking at ways in which methods can be designed to address the problem of complexity that is so often a de- fining characteristic of implementation. 43Implementation Research in Health A Practical Guide whAt APPRoAches And methods ARe APPRoPRiAte FoR imPlementAtion ReseARch? 5 k Implementation research, like all research, is governed by two broad principles: that its findings should be warranted, and that its methods should be transpar- ent. k Because it draws on a wide variety of qualitative, quantitative, and mixed-method research approaches, it makes little sense to talk in terms of a narrow set of ‘implementation research-methods’ k In implementation research, the “question is king”, and it is the question that determines the method used, rather than the method that determines the kind of questions asked. k The questions asked are often complex, reflecting the complexity of the real world. A wide array of contex- tual factors influence implementation, producing un- predictable effects that require continuous adaptation by implementers. key Points ©WHO 45Implementation Research in Health A Practical Guide chAPteR 5 • What approaches and methods are appropriate for implementation research? whAt APPRoAches And methods ARe APPRoPRiAte FoR imPlementAtion ReseARch? “ The question is king.” In discussing different approaches to implementation re- search it is helpful to keep in mind its basic goals, which, as we have already discussed, are to understand how, and why clinical and public health policies, programmes, and practices work, or fail to work in real-world settings, and to learn how to make them work better. More spe- cifically implementation research can be used to: assess change in real-world contexts, drawing on past experi- ence, where appropriate; understand complex phenom- ena; generate and/or test new ideas; and predict, or at least help anticipate what may happen in the future as a result of a particular innovation or change. It also plays an important role in informing stakeholders, thereby im- proving understanding, transparency and accountability [37]. Last, but certainly not least, the goal of implemen- tation research is to make a difference, to improve the effectiveness, quality, efficiency and equity of policies, programmes and services. Before discussing some of the possible research ap- proaches that can be used to achieve these goals, it is worth noting that, while implementation research may in some ways be different to other forms of research, like all research, it is governed by two broad principles. The first of these is that findings should be warranted, that is to say backed by sufficient evidence. The second is that its methods should be transparent, that is to say sufficiently explicit for others to be able to judge whether the processes are adequate and justify the conclusions reached, and can be repeated [38]. Whichever approach is used, these principles need to be borne in mind. Because implementation research draws on a wide vari- ety of qualitative, quantitative, and mixed-method re- search approaches, it makes little sense to talk in terms of a narrow set of ‘implementation research-methods’. There are however a number of research approaches that are particularly useful to the implementation researcher because they are inherently practical and generate ac- tionable intelligence; are good at capturing the subtleties of context, and in particular context as it changes over time; and offer the iterative flexibility needed to respond to change and evolve. A short description of a selection of these approaches is presented below. PRAgmAtic tRiAls Tests or trials of health interventions are generally de- scribed as either explanatory or pragmatic. The terms were originally coined to distinguish between trials de- signed to help choose between options for care, and tri- als designed to test underlying causal hypotheses. Thus, explanatory trials generally seek to understand and explain the benefit pro- duced by an intervention under controlled condi- tions, often using care- fully selected subjects in a research clinic, whereas pragmatic trials focus on the effects of the interven- tion in routine practice. In contrast to explanatory tri- als, pragmatic trials seek to maximize the variability in the way the intervention is implemented, (e.g., in terms of settings, providers or types of patients), in order to maximize the generalizability of results to other settings [39]. In this way, pragmatic trials can provide strong evi- dence of the effectiveness of an implementation strategy in ‘real-world’ conditions. Implementation research draws on a wide variety of qualitative, quantitative, and mixed-method research approaches so it makes little sense to talk in terms of a narrow set of ‘implementation research-methods’. Implementation Research in Health A Practical Guide46 Pragmatic trials usually include an extensive formative period involving implementers and policy-makers to design the intervention strategy, which may sometimes generate a false sense of confidence that the design is robust and suitable for the setting in which it is imple- mented. Therefore, unless additional research methods are added to the trial, the kind of changes that happen in the ‘real world’ – changes in implementation strategies, changes in implementation outcome variables, or other non-random changes in contextual factors – may not be captured. Ideally, the design of the intervention in a prag- matic trial, and the outcomes the research is designed to generate, should be developed in collaboration with the participants, funders, and practitioners who are making decisions about the intervention, and who are directly af- fected by the outcome. The value of pragmatic trials in LMIC settings is well documented, one good example be- ing a recent study undertaken by researchers in South Africa (Box 7) [40]. chAPteR 5 • What approaches and methods are appropriate for implementation research? One of the biggest obstacles to improving access to antiretroviral therapy (ART) in LMICs is the lack of trained medical staff needed to administer it. In South Africa shortages of doctors have tended to restrict access to the treatment and researchers at the Knowledge Translation Unit of the University of Cape Town Lung Institute in Cape Town, South Africa used pragmatic trials to demonstrate that health workers other than doctors were ca- pable of meeting the demand for care. Specifically, the trial focused on the Streamlining Tasks and Roles to Expand Treatment and Care for HIV (STRETCH) programme, which provides educational outreach training of nurses to initiate and re-prescribe ART, and to decentralise care. Thirty-one primary care clinics were randomly assigned to either the nurse-run program or the usual, ‘standard’ care. The study followed over 8000 patients in the nurse- run programme and 7000 patients in the standard care group for one and a half years, and found that mortality rates, viral suppression rates, and other measures of quality of care did not differ, or were actually higher in the nurse-run programme. Source: Fairall et al., 2012 [40] eFFectiveness-imPlementAtion hybRid tRiAls Effectiveness-implementation hybrid trials combine ele- ments of effectiveness and implementation research in order to assess both the effectiveness of a health inter- vention, and the implementation strategy used to deliver it. Whereas pragmatic trials do not try to control or en- sure the delivery of services to meet a realistic standard in normal practice settings, effectiveness-implementation hybrid trials also intervene and/or observe the implemen- tation process as it actually occurs, for example by assess- ing implementation outcome variables [24]. One recent paper proposes three basic types of effective- ness-implementation hybrid research designs, based largely on the priority given to the effectiveness or imple- mentation components in the research aims [24]. k Type 1 designs test the effects of a health inter- vention on relevant outcomes while observing and gathering information on implementation. In this kind of research patient functioning or symptoms in response to a health intervention are measured, while at the same time the fea- sibility and acceptability of the implementation approach taken is evaluated through qualitative, process-oriented, or mixed-methods. k Type 2 designs involve the dual testing of health interventions and implementation strategies. Box 7. A pragmatic trial in South Africa 47Implementation Research in Health A Practical Guide k Type 3 designs test an implementation strategy while observing and gathering information on the health intervention’s impact on relevant out- comes. Type 3 designs primarily test the imple- mentation strategy using measures of adoption of and fidelity to health interventions. Effectiveness-implementation hybrid trials offer a number of benefits. Instead of taking a step-by-step approach to problem-solving, starting with a randomized clinical trial to determine if an intervention implemented under con- trolled conditions works, and then moving on to designs such as cluster randomized controlled trials in order to determine the best way to introduce the intervention in real-world settings, effectiveness-implementation hybrid approaches allow researchers to simultaneously evalu- ate the impact of interventions introduced in real-world settings and the implementation strategy used to deliver them. Such designs not only speed up what may other- wise be a very time-consuming process, they also allow researchers to identify important intervention-implemen- tation interactions. These can then be used to inform de- cisions about optimal implementation approaches. The testing of the newborn care packages in Sylhet, Bangla- desh offers a good example of the different purposes for which effectiveness-implementation hybrid trials are ap- propriate (Box 8) [41-44]. chAPteR 5 • What approaches and methods are appropriate for implementation research? In order to capture the full range of qualitative and quantitative data needed to evaluate the cluster randomized controlled trial that was designed to test a homecare and community health worker intervention and a commu- nity-care intervention in comparison to the usual neonatal care provided, researchers used a range of research methods, including: • Quantitative household survey research, which provided estimates of existing neonatal mortality and levels of skilled birth attendance, which motivated the need to intervene and also provided baseline levels; • Formative qualitative research, which was used to explore home practices that put newborns at risk of death, and the barriers for safe delivery and postnatal care. It was then used as the basis for the development of the home-based newborn package of care (“participant enrichment”), and to design quantitative research instruments (“instrument validity”); • Observations of newborn care were conducted to show that community health workers could diagnose new- born illness; • Household surveys and in-depth interviews were used to show that the intervention was being implemented as planned, while surveys, observations, and in-depth interviews were also used to demonstrate whether the new newborn practices were actually being implemented (“implementation fidelity”); • End-line household surveys were used to assess both neonatal mortality and service coverage levels, while qualitative research was used to explain in detail how and why delivery and postnatal practices changed, largely because of the engagement of the local community in the programme, and supportive supervision of the community health workers (“meaning enhancement”). Sources: Baqui et al., 2008 [41]; Baqui et al., 2009 [42]; Choi et al, 2010 [43]; Shah et al. 2010 [44] Effectiveness-implementation hybrid trials offer a number of benefits, including speeding up the translation of knowledge into action. Box 8. Effectiveness-implementation research applied in Bangladesh newborn care study. Implementation Research in Health A Practical Guide48 QUAlity imPRovement stUdies As highlighted by the example of cervical screening in El Salvador (Table 1), the study of quality improvement (QI) in health care presents three main challenges: first, the evaluation of quality is inherently context-dependent; second, quality is something of a moving target – with improvement interventions being repeatedly modified in response to outcome feedback; third, conducting re- search into quality also generally involves taking into ac- count complex, multi-component interventions. All of this has implications for research design [17]. In order to reflect the iterative, ‘moving-target’ nature of QI, studies typically involve a set of structured, cyclical processes, governed by a paradigm referred to as the plan-do-study-act (PDSA) cycle or a variant thereof [45]*. The PDSA cycle allows for the application of scientific methods on a continuous basis to formulate a hypothesis or plan to improve quality, implement the plan, analyze and interpret the results, then generate a plan for what to do next. A detailed description of the range of quality improvement tools that can be used in PDSA studies is beyond the scope of this Guide, though some examples are shown in Figure 4, listed according to the stages of the PDSA cycle [46]. Another instructive example is the PDSA work done in El Salvador in relation to cervical can- cer screening, which is discussed in Chapter 2. chAPteR 5 • What approaches and methods are appropriate for implementation research? *The PDSA cycle has various names, including the plan-do-check-act-cycle, the Shewhart cycle, after Walter Shewhart who first pioneered statistical “control” methods in the 1930s to improve manufacturing processes, and later the Deming cycle, after the W. Edwards Deming, who popularized modern quality control methods and the PDCA and later PDSA cycle. Adapted from: Brassard et al., 1994 [46] Figure 4. Plan-Do-Study-Act cycle and research tools that can be used at each stage PLAN for changes to bring about improvement ACT to get the greatest benefit from changes STUDY changes to see if they are working and investigate selected processes • Small group leadership skills • Experiment design • Conflict resolution • On-Job training DO changes on small scale first to trial them • Data check sheets • Graphical analysis • Control charts • Key performance indicators • Customer/supplier mapping Flowchart- ing • Pareto analysis • Brainstorming • Nominal group technique solution/faults tree • Evaluation matrix • Cause-and-effect diagrams • Process mapping • Process standardisation • Control reference informa- tion formal training for standard processes A Act D Do S Study P Plan 49Implementation Research in Health A Practical Guide Typically, PDSA QI research assesses whether the inter- vention being studied – usually a change in process – produces a significant improvement in an outcome. The results are then used to effect changes in the interven- tion on a cyclical, iterative basis. PDSA interventions thus typically involve repeated testing over time. PDSA studies are often referred to as quasi-experimental because the experimenter lacks complete control of the study, nota- bly with respect to his or her ability to randomly allocate the intervention to specific subjects [47]. There is a wide range of quasi-experimental research designs, some com- mentators identifying as many as sixteen different types, and noting their different strengths and weaknesses with respect to issues of internal validity (whether the inter- vention in the specific study actually made a difference), and external validity (asking to which populations, set- tings, treatment and outcome variables can the observed effect be generalized) [47]. PDSA research designs are often used for QI in health systems, offering as they do the assessment of responses measured repeatedly and regularly over time, either in a single case or with comparison groups [48]. Research de- signs typically include: time series studies, with baseline and post-intervention assessments of levels and trends in results; multiple time series studies, where the interven- tion and baseline are repeated at different times: time series studies where the interventions are lagged across different groups at different times: and factorial design studies where the intervention is randomized to groups to compare time series. The data for these quasi-experi- mental designs may come from routine health manage- ment information, or from special surveys to specifically measure the outcomes of interest. For standardized guidance on using good quality health information systems and health facility surveys, readers can go to the Lindelow and Wagstaff report on data and measurement issues arising in the assessment of health facility performance [49]. For general advice on how to design and report on research involving QI interventions, readers should consult the Standards for Quality Improve- ment Reporting Excellence (SQUIRE) guidelines [45]. PARticiPAtoRy Action ReseARch All research on human subjects involves human participa- tion, but the particularity of participatory action research (PAR) is that it assigns power and control over the re- search process to the subjects themselves. Thus, PAR re- fers to a range of research methods that typically involve iterative processes of reflection and action “carried out with and by local people rather than on them” [28]. The kind of “bottom-up” approaches that involve locally defined priorities and perspectives are set out in Table 4 [28]. Although most of the PAR methods involve quali- tative techniques, increasingly quantitative and mixed- methods techniques are being used, as, for example, in participatory rural appraisal or participatory statistics [50] [51]. Guidelines for conducting and reporting participa- tory action research are available, though the emphasis tends to be on local action by participants rather than ways in which external researchers can work with them [52, 53]. chAPteR 5 • What approaches and methods are appropriate for implementation research? Implementation Research in Health A Practical Guide50 A number of PAR-based initiatives have been undertaken in LMICs in recent decades, one notable example being the work of the Indian nongovernmental organization, Ekjut, which helps women’s groups to improve maternal and neonatal health in tribal areas of the Indian states of Jharkhand and Odisha (Box 9) [54]. chAPteR 5 • What approaches and methods are appropriate for implementation research? Source: Cornwall and Jewkes 1995 [28] Table 4. A comparison of participatory action research and conventional research Participatory Action Research Conventional Research What is the research for? Action Understanding with possible later action Who is the research for? Local people Institutional, personal, and professional interests Whose knowledge counts the most? Local people’s Scientist’s Who chooses the topic? Local priorities Funding agency, institutional agendas, professional interests Methodology is chosen for what reasons? Empowerment and learning Disciplinary convention, “objectivity”, “truth” Who takes part in the stages of research? Problem identification Local people Researcher Data collection Local people Researcher, data collector Interpretation Local concepts and frameworks Disciplinary theories and frameworks Analysis Local people Researcher Presentation of findings Locally accessible and useful By researcher to academics and funding agency Action on findings Integral to process Usually separate or may not happen Who takes action? Local people, with or without external support External agencies Who owns the results? Shared The researcher or funder Emphasis of process or outcomes? Process Outcomes The success and sustainability of community-based programmes for improving maternal and neonatal health require the active involvement of women, families and community health-care workers, yet the strategies used to engage these groups are often externally driven and top-down in character. Since 2005, the Indian nongov- ernmental organization known as Ekjut has sought to reverse this trend by helping women’s groups to improve maternal and neonatal health in tribal areas of the Indian states of Jharkhand and Odisha. Local female facilitators guide women’s groups through a cycle of activities involving participatory learning and action, during which women identify, prioritize and analyse local maternal and neonatal health problems and subsequently devise and implement strategies to address them. The Ekjut intervention was initially evaluated in a cluster randomized controlled trial carried out between 2005 and 2008 in 36 largely tribal clusters of three con- tiguous districts of Jharkhand and Odisha. A recent study reported significant falls in neonatal mortality in those districts as a result of the interventions and concluded that community mobilization through women’s groups can produce a sustainable and reproducible improvement in neonatal survival in rural areas of India. Source: Roy et al., 2013 [54] Box 9. Participatory action to improve neonatal health care 51Implementation Research in Health A Practical Guide ReAlist Review The aim of realist review is to enable decision-makers to reach a deeper understanding of the intervention and how its potential can be maximised in different settings. The approach can be extremely helpful in dealing with policy and programme interventions, where complexity and variation in implementation are significant factors [55]. Used to address complex social interventions or programmes, realist review is anchored in particularity, providing explanatory analysis fo- cused on what works for whom, in what circum- stances, in what respects, and how [55]. That said, it is clear that synthesis requires some form of generalisa- tion and in realist review that generalisation comes from the identification of underlying assumptions. Thus, the first step in realist review is to discover and make explicit the underlying assumptions of the intervention in ques- tion, revealing how it is meant to work in theory. This is an iterative process that involves sharpening the focus of questions around the nature of the intervention by as- sessing the integrity of the underlying theory, comparing rival theories, and assessing the same theory in different settings. Realist review then seeks empirical evidence in the literature that supports, contradicts or modifies the underlying programme assumptions, combining theoreti- cal understanding and empirical evidence, while focusing on the relationship between the context in which the in- tervention is applied, the mechanisms by which it works and the outcomes which are produced. The search and review of publications eventually leads to a formal audit trail of the literature reviewed, and a final search is then conducted once the review is nearly com- plete. Individual articles are assessed for their relevance and methodological rigour, and synthesized according to the key questions of the review. This might include ques- tions such as: Which parts of the programme are working and which are not? For whom? Under what circumstanc- es? Why? Finally, the results of the review are shared in ways that provide a stronger link to those who commis- sioned the review or who can use the review, preferably as part of a policy dialogue [56]. The 2011 WHO report, Realist Review and Synthesis of Reten- tion Studies for Health Work- ers in Rural and Remote Areas, by Dieleman and colleagues, is an example of how this type of review can be applied to under- stand contextual factors and the main mechanisms that underpin many retention strategies [57]. This report adopts an analyti- cal framework that is centered around a ‘context-mechanism-outcome’ which describes how an intervention—in this case retention strategies for workers, interacts with a specific context—in this case rural and remote areas in LMICs, and how it results in certain outcomes. Through this approach, the underly- ing theory supporting the intervention are clarified and tested using scenarios where the intervention has been implemented [57]. mixed-methods ReseARch Mixed-methods research, as the name suggests, uses both qualitative and quantitative methods of data collec- tion and analysis in the same study. While not designed specifically for implementation research, mixed-methods research is particularly suitable for these research activi- ties because it provides a practical way to understand multiple perspectives, different types of causal pathways, and multiple types of outcomes – all of which are com- mon in implementation settings. Mixed-methods research approaches can be extremely useful and are applicable for a range of purposes – as many as 65 according to chAPteR 5 • What approaches and methods are appropriate for implementation research? Realist review provides explanatory analysis focused on what works for whom, in what circumstances, in what respects, and how. Implementation Research in Health A Practical Guide52 one study [58]. These can be boiled down to four main rationales [59]*. k Participant enrichment: to gain the most informa- tion from a sample of participants (e.g. by admin- istering a standard survey questionnaire and then asking for in-depth explanations). k Instrument validity: to make sure that the instru- ments used are appropriate and useful (e.g. using focus groups to identify items for a questionnaire or testing its validity). k Implementation fidelity (treatment integrity): to assess whether the intervention or programme is being administered as intended. k Meaning enhancement: to maximize the interpre- tation of the findings, such as by using qualitative measures to explain the statistical analysis or vice versa. A number of different schemes exist that describe differ- ent types of mixed-methods research, based on the emphases used in different approaches, the sampling schemes used for the different parts of the study, the tim- ing and sequencing of the qualitative and quantitative methods, and the level of mixing between the qualitative and quantitative methods [60, 61]. Fully mixed-designs use qualitative and quantitative methods in each stage of the research, including sampling and data collection, analysis, and drawing of inferences. Interested readers can consult Tashakkori and Teddlie’s handbook which outlines 35 different designs, or Onwuegbuzie and Col- lins’ article describing 24 different sampling schemes [55, 61]. Broad guidance on the design, conduct, and reporting of mixed-methods designs are available by sev- eral authors [59, 60, 62, 63]. A simple scheme for good reporting of a mixed-methods study (GRAMMS) is listed below [64]. k Describe the justification for using a mixed methods approach to address the research ques- tion. k Describe the design in terms of the purpose, priority and sequence of methods. k Describe each method in terms of sampling, data collection and analysis. k Describe where the integration has occurred, how it has occurred, and who has participated in it. k Describe any limitation derived from associating one method with another method. k Describe any insights gained from mixing or integrating methods. the imPoRtAnce oF the ReseARch QUestion While the different research approaches described above may be said to constitute a kind of basic toolbox for implementation researchers, it is important to remember that in implementation research it is the question posed that determines the tool to be used and not the tool that determines the shape of question. Put simply, in imple- mentation research the question is king. This does not mean that it is an entirely adhoc endeavor and that there is no room for overarching concepts; indeed certain theo- retical constructs are of value in helping to think about implementation processes, notable among them are theories of change that try to explain what is needed to achieve a long-term goal such as improved health. A theory of change should describe a logical sequence or pathway with sets of outcomes along the way to the goal, and should reflect the assumptions about the changes made. Often the steps along the way and the assump- tions about why they occur are the subject of research for chAPteR 5 • What approaches and methods are appropriate for implementation research? * The authors use slightly different terms than used here. We’ve changed the terms to avoid confusion with other terms used in implementation research. The authors use the term “instrument fidelity”, which we label “instrument validity” to avoid confusion with “implementation fidelity”. The authors use the term “treatment integrity” as a term for “implementation fidelity”. The authors also use the term “significance enhancement”, which can be confused with statisti- cal significance testing in quantitative research, whereas we use the term “meaning enhancement”. 53Implementation Research in Health A Practical Guide verification, testing, or further explanation. It is also worth noting the valuable insights that have come from some of the work done on implementation theory, notable among which the Consolidated Framework for Implementation Research (CFIR), a useful template for organizing key concepts in implementation research (Box 10) [65-67]. chAPteR 5 • What approaches and methods are appropriate for implementation research? A number of theories have been developed to promote the effective implementation of health interventions. Many try to explain individual or group behaviours around implementation issues. For example the RE-AIM (reach, efficacy, adoption, implementation, and maintenance) framework is commonly used in health promo- tion interventions, and provides a practical approach to evaluating the effects of health interventions through changes in individuals, organizations and communities. The Diffusion of Innovations Theory seeks to explain how innovations spread, highlighting the importance of the perceived attributes of the innovation (relative advantage, compatibility with current approaches, ability to observe results, ability to test the innovation, and its complexity), innovativeness of the adopter, the social system, individual processes for adoption, and the system for diffusion. The Consolidated Framework for Implementation Research (CFIR) was developed as a way to consolidate the vari- ous theories and terms used to support further development of theory and testing on how to translate health interventions that have been shown to be effective. The CFIR comprises five elements: 1. Intervention charac- teristics; 2. Outer setting; 3. Inner setting; 4. Characteristics of the individuals involved; and 5. The process of implementation. Constructs like the strength and quality of the evidence are related to the intervention domain, whereas issues like patient needs and resources are part of the outer setting. Constructs related to inner setting of the organization include its culture and leadership engagement. Individual attitudes, beliefs and capabilities also play a prominent role, whereas other influential factors are related to the process of implementation itself (e.g., plan, evaluate, and reflect). Box 10. Implementation theory Sources: Glasgow et al., 2009 [65]; Rogers et al., 2003 [66]; Damschroder et al., 2009 [67] Implementation Research in Health A Practical Guide54 mAtching methods with ReseARch QUestions Given the importance of the research question, it is worth taking a moment to consider the kinds of questions that are likely to arise, and the research methods that may be appropriate to answering them. One way of going about this is to break down research questions into a limited number of categories based on the core objective of the research to be undertaken, as we have done in Table 5 [4, 34, 58]. Thus, in the second example, one of the key questions asked with a view to describing implementation phe- nomena, is: ‘What describes the main factors influencing implementation in a given context?’ This is followed by the kinds of research methods likely to generate the in- formation required, a list that includes both quantitative (cross-sectional or descriptive surveys, network analysis) and qualitative methods. What is noticeable about the methods proposed is the importance given to qualitative research, and in particular to research designed to absorb a richness of detail through, for example, case-studies, key informant interviews, and focus groups. In the predictions example, the research questions in- tended to generate predictions draw on prior knowledge or theories to forecast future events. The research may rely on a variety of quantitative research methods, includ- ing agent-based modeling, which relies on computational models to simulate the actions and interactions of au- tonomous agents such as organizations or groups with a view to assessing their effects on the system as a whole, and data extrapolation and sensitivity analysis. These methods may be supplemented by qualitative research such as scenario building exercises and so-called ‘delphi’ techniques which rely on a panel of experts making pre- dictions, usually based on a questionnaire, and conducted over several rounds between which a facilitator provides an anonymous summary of forecasts. Such forecasts are obviously of particular value where a significant innova- tion is being introduced, as was the case with the ITN voucher programme in the United Republic of Tanzania or the insurance scheme in Ghana. As noted in the previous chapters, implementation re- search is of greatest use where it is conducted in real- world settings and assesses the context and other factors influencing implementation, with some research specifi- cally testing implementation strategies and/or measuring implementation outcome variables [68]. conclUsion This chapter has sought to identify the defining character- istics of research approaches selected on the basis of their usefulness to implementation researchers, and to suggest ways in which those methods can be applied to specific implementation research questions. In the next chapter the focus will be on aligning implementation research with need, both in terms of the response imposed by the material under study and the needs of the intended audi- ence. chAPteR 5 • What approaches and methods are appropriate for implementation research? 55Implementation Research in Health A Practical Guide chAPteR 5 • What approaches and methods are appropriate for implementation research? Adapted from: Peters et al., 2009 [4], Habicht et al 1999 [34] Table 5. Types of implementation research objectives, implementation questions, and research methods Objective Description Implementation Question Research methods and data collection approaches * Explore Explore an idea or phenomenon to make hypotheses or generalizations from specific examples What are the possible factors and agents responsible for good implementation of a health intervention? For en- hancing or expanding a health intervention? Qualitative methods: Grounded theory, ethnography, phenom- enology, case-studies and narrative approaches; key informant interviews, focus groups, historical reviews Quantitative: Network analysis, Cross-sectional surveys Mixed methods: Combining qualitative and quantitative methods. Describe Identify and describe the phenomenon and its correlates or possible causes What describes the context in which implementation occurs? What describes the main fac- tors influencing implementa- tion in a given context? Quantitative: Cross-sectional (descriptive) surveys, network analysis Qualitative methods: Grounded theory, ethnography, phenom- enology, case-studies and narrative approaches; key informant interviews, focus groups, historical reviews Mixed methods: Both qualitative and quantitative inquiry with convergence of data and analyses Influence Test whether an intervention pro- duces an expected outcome. With Adequacy With sufficient confidence that the intervention and outcomes are occurring Is coverage of a health intervention changing among beneficiaries of the interven- tion? Before-after or time-series in intervention recipients only; Partici- patory action research With Plausibility With greater confidence that the outcome is due to the intervention Is a health outcome plausibly due to the implemented intervention rather than other causes? Concurrent, non-randomized cluster trials: health intervention implemented in some areas and not in others; before-after or cross-sectional study in programme recipients and non-recipients; Typical quality improvement studies With Probability With a high (calcu- lated) probability that the outcome is due to the interven- tion Is a health outcome due to implementation of the intervention? Partially controlled trials: Pragmatic and cluster randomized trials; Health intervention implemented in some areas and not in others; Effectiveness-Implementation Hybrids Explain Develop or expand a theory to explain the relationship between concepts and the reasons for the occurrence of events, and how they occurred? How and why does implemen- tation of the intervention lead to effects on health behaviour, services or status in all its variations? Mixed methods: Both qualitative and quantitative inquiry with convergence of data and analyses Quantitative: Repeated measures of context, actors, depth and breadth of implementation across subunits; network identifica- tion; can use designs for confirmatory inferences; Effectiveness- implementation hybrids; Qualitative methods: Case-studies, phenomenological and ethno- graphic approaches with key informant interviews, focus groups, historical reviews Participatory action research Predict Use prior knowledge or theories to fore- cast future events What is the likely course of future implementation? Quantitative: Agent-based modeling; Simulation and forecast- ing modeling; Data extrapolation and sensitivity analysis (trend analysis, econometric modeling) Qualitative: Scenario building exercises; Delphi techniques from opinion leaders how shoUld imPlementAtion ReseARch be condUcted? 6 k Implementation research should be aligned with need, both in the sense that it meets the requirements of the intended audience and is also responsive to the particularities of the subject under study. k Research designs need to be responsive and capable of capturing changing elements at multiple points in time. k There are no fixed rules for justifying the selection of a particular research method for implementation research, other than the fact that the methods used should reflect the questions asked. key Points ©WHO 57Implementation Research in Health A Practical Guide chAPteR 6 • hoW should implementation research be conducted? how shoUld imPlementAtion ReseARch be condUcted? “ Be responsive to the demands of your subject and your audience.” Having looked at ways in which the question in imple- mentation research can be seen to be ‘king’, in this chap- ter we look at the importance of doing research that is aligned with need, both in the sense that it meets the requirements of the intended audience and is also re- sponsive to the particularities of the subject under study. Although there is a role for implementation research in generating theory, most often the evidence generated by implementation research is produced to be used in the real world, rather than consumed by other researchers. This means that the implementation researcher needs to be aware of the use to which his or her work will be put. A key considerations in this regard is the level of certainty required regarding results or projections. A policy-maker, for example, may be looking for strong indications that a given intervention will work, but may not have the time required for protracted studies that could generate a higher level of certainty. The quasi-experimental field trial of the kind described in the previous chapter about reflective materials and motorcycle injuries may be ad- equate and more importantly, more appropriate for an- swering the research questions being asked. Such differences have important implications for the ba- sic design of research, in terms of sample size, the need for concurrent comparison groups and randomization of participants within these groups, as well as the cost and time requirements for the research to be undertaken [4, 34]. Different research approaches also have implications for budget and scheduling. Simple exploratory or descrip- tive studies seeking to establish what is happening may not require long-term observation, for example. However, studies assessing the influence of an intervention usually need observations from at least two points in time, while more frequent measurement may be needed if changing implementation factors are to be assessed. If the main research question calls for an explanation of how or why implementation is developing in a particular way, the demands on research design become that much greater, requiring multiple observations to capture chang- es from different types of respondents. Thus, in a situation where lack of compliance with guidelines on, say, antimi- crobial prescription is the main concern, a quantitative study of antimicrobial prescription levels would need to be accompanied, at the very least, by research captur- ing levels of awareness regarding rational drug use, and doctors and patients attitudes and interactions. But even there, the implementation researcher would probably miss essential contextual factors without digging a little deeper, or throwing the net a little wider. For example, where public sector generalists are paid per consulta- tion, they may find it quicker and thus more profitable to prescribe the antimicrobial that the patient is demanding rather than spend time informing the patient about the dangers of antimicrobial resistance [69]. Simply looking at questions of awareness, would miss the essential part played by health system financing. the imPoRtAnce oF Flexible, ResPonsive ReseARch Health systems are complex. A complexity that is aggra- vated by the complexity of its main actors including decision-makers, implementers and people who are the ultimate beneficiaries of health services. Both health sys- tems and its actors are constantly changing and adapt- ing to new actions, often reacting in unpredictable ways [70]. In addition, a wide array of contextual factors typi- cally influence implementation and these factors often change over time, producing unpredictable effects that require continuous adaptation by implementers. This has profound implications for the research methods used in studying implementation, particularly in regard to the need for flexibility and creativity in response to changes in the subject under study. In addition, since implementa- tion of policies, programs, and practices is rarely a static or linear process, research designs need to be responsive and capable of capturing changing elements at multiple Implementation Research in Health A Practical Guide58 points in time. Research designs that depend on having a single and unchangeable intervention, such as a typical randomized controlled clinical trial, are poorly adapted to study phenomena that change, especially when that change is unpredictable. The challenge posed by com- plexity is all the greater where the research design itself is complex, calling for multiple methods and different sources of information. JUstiFying the selection oF ReseARch methods As in other areas of research, there are no fixed rules for justifying the selection of a particular research method for implementation research, other than the fact that the methods used should reflect the questions asked. An understanding of the research objective and specific research question is a good starting point, and should be informed by the proposed theory of change. Figure 3 offers a useful template for thinking about the range of research questions and methods along the continuum of implementation research, with implementation becoming more important with the testing of proof of implementa- tion and the consideration of implementation outcome variables. Generally speaking however, proof of implementation usually concerns a clinical or public health intervention that has already been shown to work. As the issues move further downstream to questions of scaling-up cover- age and integration and sustainability of interventions in health systems, the focus of research turns to implemen- tation. Needless to say, this is solely a theoretical model and in reality the process is neither linear nor complet- ed in one iteration, since health systems are constantly adapting and incorporating multiple innovations with dif- ferent entry points. While it is clear that a degree of subjective judgment is necessary to assess any particular study, as a general rule, good implementation research should be able to address each of the following questions: k Will the research answer a relevant and impor- tant implementation problem? k Is the new knowledge potentially worth the cost of the research? k Are there clear research objectives and questions related to implementation, and does the pro- posed research design match them (see Table 5) k Does the research fit with a theory of change or causal chain in a coherent way? If not, what is the potential for generating new theories or questions? k Will the research produce results that can be acted on in a timely way by the intended audi- ences? k Does the research design reflect an understand- ing about whether the intervention is stable and simply replicable, or whether the intervention is expected to change? k Does the research adequately capture changes that occur over time and place in both the inter- vention, the context, and the effects? k In complex environments, can the research iden- tify the main components of the health system and their relationships, as well as the unintended consequences that are likely to occur from an in- tervention? chAPteR 6 • hoW should implementation research be conducted? Contextual factors can influence implementation and these factors can also change over time, producing unpredictable effects that require continuous adaptation by implementers. 59Implementation Research in Health A Practical Guide Assessing the QUAlity oF imPlementAtion ReseARch Standards for assessing the quality of conventional quan- titative and qualitative methods of research are largely the same when the research involves implementation. Guidelines already exist to help with the design and re- porting of health research, many of which are catalogued by the EQUATOR “Library for Health Research Reporting” (at: http://www.equator-network.org/). These include such guidelines as the CONSORT statement on random- ized controlled trials, STROBE guidelines for observational studies, PRISMA for systematic reviews and meta-analy- sis, and COREQ for qualitative research. Because none of these guidelines focuses on the particu- lar issues related to implementation research, we propose the following set of questions that can be used as a sum- mary guide to assessing the issues particularly related to implementation research, which can be used in addition to the conventional guidelines. It should be noted that these guidelines are designed to help with the reporting and assessment of research reports, but they can also be adapted for purposes of assessing research proposals or the design of implementation research. The key ques- tions include: k Does the research clearly address a question con- cerning implementation? k Is there a clear description of what is being imple- mented (e.g. details of the practice, programme, or policy)? k Does the research involve an implementation strategy? If so, is it described and examined ap- propriately? k Is the research conducted in a real-world setting? If so, are these conditions described in sufficient detail? k Does the research appropriately consider imple- mentation outcome variables? k Does the research appropriately consider context and other factors that influence implementation? k Does the research appropriately consider chang- es over time, and the level of complexity of the system? k Does the research clearly identify the target audi- ence for the research and how it can be used? It is our hope that, using this guidance, practitioners, policy-makers, researchers and users of research will begin to have the confidence to assess whether imple- mentation research is conducted and reported in a trans- parent way and warrants the conclusions drawn through valid and reliable methods. conclUsion This chapter has sought to identify key considerations for implementation researchers seeking to optimise the im- pact of the work they do. In the next chapter we will look at the challenges faced by the implementation research field itself, and discuss ways in which both implementers and researchers can do more to support this crucial area of study and make better use of the potential it offers. chAPteR 6 • hoW should implementation research be conducted? how cAn the PotentiAl oF imPlementAtion ReseARch be ReAlized?7 k Without implementation research we are at best com- mitting valuable resources to implementation in the hope that things will work out. k Despite the importance of implementation research, it continues to be a neglected field of study for two reasons: a lack of understanding regarding what it is and what it has to offer; and a lack of funding for implementation research activities. k Implementation research should be seen as a core function of programme implementation, which means embedding it into the programme cycle. k More funding is needed for implementation research, and this funding should be aligned with funding for programmes. k More opportunities for researchers and implementers in LMICs who want to undertake implementation research are needed. key Points ©WHO 61Implementation Research in Health A Practical Guide chAPteR 7 • hoW can the potential of implementation research be realized? how cAn the PotentiAl oF imPlementAtion ReseARch be ReAlized? “ Implementation research is not on the whole an expensive pursuit so investments in it go a long way.” Implementation research is essential to ensuring that the benefits accruing from the effective implementation of policies, programmes, and services are realized. Without it, and the knowledge it generates, we are at best com- mitting valuable resources in the naive hope that things will work out rather than applying the kind of evidence- and experience-informed decision-making that imple- mentation research makes possible. This approach comes with a cost. Every time a programme fails be- cause insufficient attention was paid to ‘real-world’ or context-specific factors not foreseen or anticipated by programme designers, there is a cost both in terms of wasted resources and of un- necessary human suffering. Sometimes this cost can be enormous. But despite the importance of implementation research, it continues to be a neglected field of study, a fact borne out by numerous studies, including, for example the 2006 report produced by Sir David Cooksey on United Kingdom (UK) health-care research, which noted “perverse incen- tives that value basic science more highly than applied research” and reported that two thirds of public and charitable research funding went into basic as applied research, including research into implementation [71]. Needless to say private sector research funding was even more heavily skewed to developing new products for new markets, rather than to making the most of products al- ready on the shelf. And what is true of the UK is true of most developed countries. Billions of dollars are spent on health innovations, yet only a fraction of this is spent on how best to use those innovation [72]. The problem affects everyone, but where implementation challenges are greatest, notably in LMICs, the adverse consequences of neglecting implementation research are most keenly felt. Even when products programmes and services are specifically designed for LMICs, too often they never get to the end user because of failures in implementation. Given the need for implementation research, why isn’t more of it done? And why does the implementation re- search that is undertaken not always have the impact we might expect? There are several reasons for the lack of implementation research, among the most important be- ing the lack of understanding of what it is and the role it can play in maximising the benefit of the interventions at our disposal. Another obvious constraint is the lack of funding to sup- port this type of research. As it is clear from the various approaches de- scribed in this Guide, implementation research is not on the whole an expensive pursuit, and certainly not expensive compared to investment in cutting edge biomedical engineering. So investment in implementa- tion research goes a long way. Moreover it is possible to argue that we now have many of the interventions and technologies needed to reduce morbidity and mortality, and should focus more on making better use of them. That is not to say that investment in basic science should stop, and it is worth noting that the above-cited Cooksey report stressed the importance for the UK of maintain- ing the level of investment in so-called ‘blue sky’ proj- ects. Nor is it simply a question of raising new funds for implementation research, but of making better use of the funds available, channelling them towards research that is aligned with need, by, for example, increasing oppor- tunities for implementers and programme personnel to access this funding. Without implementation research we are at best committing valuable resources in the hope that things will work out Implementation Research in Health A Practical Guide62 To address these different challenges, it is necessary to support and promote implementation research on a num- ber of different fronts, which we here break down into an agenda for action. Action #1: Implementation research should be seen as a core part of programme implementation. This means embedding research into the programme cycle in an it- erative progression that allows for continuous learning and, where necessary, adaptation. One way of supporting this change would be to encourage ownership or primary responsibility by implementers or programme personnel with support and guidance from specialist academics. Bottom line: implementers need to take a more active role in implementation research. Action #2: To ensure that implementation research be- comes more accessible, researchers should be encour- aged to engage in programme activities. This will include entering into dialogue with implementers and pursuing initiatives that embed implementation research in the real world such as living in the field in order to better ap- preciate the complexities of implementation. Action #3: Implementers need to make programmes more accessible to researchers and invite researchers to participate in programmes. There are times when it is the implementers who resist collaboration. Knowledge gen- eration activities should be made a part of programme implementation and those with the expertise to carry out research should be engaged to support this process. Action #4: Make more funding available for implemen- tation research and align this funding with funding for programmes. To bring this realignment about it is essen- tial that funding for implementation research be made available within programme budgets or explicitly tied to programme activities through structured collaborations and partnerships. Action #5: More training opportunities for implementa- tion research need to be made available to programme personnel or implementers. Implementation research should also be built into training programmes such as Masters in Public Health so that it is recognized as a core function of public health practice, a cycle in which knowl- edge is generated through research and used to inform programme implementation. Action #6: Provide more guidance and opportunities for mentorship to researchers and implementers in LMICs who want to undertake implementation research. No one is better placed to provide the kind of context-specif- ic research needed to support implementation research in LMICs, and thus researchers in those countries represent a huge untapped resource. Action #7: Incentives for researchers should be linked to engaging in making changes in policies and programmes, in addition to incentives related to academic publication and teaching. So some reorientation of researchers is needed, with opportunities for researchers to experience and understand field work and programmes. This is pos- sibly something that donors can support. conclUsion This Guide is an attempt to begin to redress the deficit in understanding of implementation research and it is the writers’ hope that readers will pursue the subject in greater depth in the source material cited in these pages. In particular, it is hoped that programme personnel and implementers will take a greater interest in the subject, recognizing that implementation research is in fact a programme issue, that is to say an integral part of pro- gramme planning and execution, rather than something that happens once the programme is up and running, something that is practiced in the research ‘bubble’, and conducted largely for the benefit of other researchers. For their part implementation researchers can do much more to engage with implementers and programme personnel in research process. For things to change, it is essential that programme personnel and implementers who under- stand the importance of context and researchers who un- derstand methods and science of inquiry come together. Only in this way can we hope to advance our understand- ing of implementation issues. chAPteR 7 • hoW can the potential of implementation research be realized? 63Implementation Research in Health A Practical Guide Implementation Research in Health A Practical Guide64 ReFeRences • implementation research in health: a practical Guide ReFeRences 1. Fenner, F., Smallpox and its eradication. 1988, Geneva, Switzerland: World Health Organization. 2. Foege, W.H., J.D. Millar, and J.M. Lane, Selective epidemio- logic control in smallpox eradication. American Journal of Epidemiology, 1971. 94(4): p. 311-315. 3. NORAD, Accelerating progress in saving the lives of women and children, in The Global Campaign for the Health Mille- nium Development Goals. 2013, Ministry of Foreign Affairs, Norway: Oslo, Norway. 4. Peters, D.H., et al., Improving Health Services in Developing Countries: From Evidence to Action. 2009, Washington, DC, USA: World Bank. 5. Mandala, J., et al., Prevention of mother-to-child transmis- sion of HIV in Zambia: implementing efficacious ARV regimens in primary health centers. BMC Public Health, 2009. 9(314). 6. de Savigny, D., et al., Introducing vouchers for malaria prevention in Ghana and Tanzania: context and adoption of innovation in health systems. Health Policy and Planning, 2012. 27(4): p. iv32-iv43. 7. Waddington, C.J., A price to pay, part 2: The impact of user charges in the Volta region of Ghana. International Journal of Health Planning and Management, 1990. 5(4): p. 287- 312. 8. Agyepong, I.A. and S. Adjer, Public social policy develop- ment and implementation: a case study of the Ghana National Health Insurance scheme. Health Policy and Plan- ning, 2007. 23(2): p. 150-160. 9. Victoria, C.G., et al., Context matters: Interpreting impact findings in child survival evaluations. Healthy Policy & Plan- ning, 2005. 20(1): p. i18-i31. 10. Thaddeus, S. and D. Maine, Too far to walk: maternal mortality in context. Social Science Medicine, 1994. 38(8): p. 1091-110. 11. de Savigny, D. and T. Adam, Systems thinking for health systems strengthening. 2009, Alliance for Health Policy and Systems Research. World Health Organization. 12. Sankar, B., Micro-planning in Andhra Pradesh: Ensuring Quality HIV Care for Individuals and Communities, in AIDSTAR-One. 2013, USAID: Arlington, VA, USA. 13. Peters, D.H., et al., A balanced scorecard for health services in Afghanistan. Bulletin of the World Health Organization, 2007. 85: p. 146-151. 14. Hansen, P.M., et al., Measuring and managing progress in the establishment of basic health services: the Afghanistan health sector balanced scorecard. International Journal of Health Planning and Management, 2008: p. 107-117. 15. Edward, A., et al., Configuring balanced scorecards for measuring health system performance: evidence from 5 years evaluation in Afghanistan. PLoS Med, 2011: p. e1001066. 16. WHO and ExpandNET, Nine steps for developing a scaling- up strategy. 2010, Department of Reproductive Health and Research, WHO: Geneva, Switzerland. 17. Agurto, I., et al., Improving cervical cancer prevention in a developing country. International Journal for Quality in Health Care, 2006: p. 81-86. 18. Faden, R.R., et al., An ethics framework for a learning health care system: A departure from traditional research ethics and clinical ethics, in Hastings Centre Report. 2013. p. S16-27. 19. Smith, M., et al., Best Care at Lower Cost: The Path to Continuously Learning Health Care in America. 2013, Insitute of Medicine of the National Academies: Washing- ton, DC, USA. 20. IOM, Crossing the Quality Chasm: A New Health System for the 21st Century. 2001, Institute of Medicine: Washington, DC, USA. 21. Lam, A., Tacit knowledge, organizational learning and societal institutions: An integrated framework. Organiza- tion Studies, 2000. 21(3): p. 487-513. 22. Ciliska, D., et al., Diffusion and dissemination of evidence- based dietary srategies for the prevention of cancer. Nutri- tion Journal, 2005. 4(13). 23. Remme, J.H.F., et al., Defining Research to Improve Health Systems. PLoS Medicine, 2010. 7(11). 65Implementation Research in Health A Practical Guide ReFeRences • implementation research in health: a practical Guide 24. Curran, G.M., et al., Effectiveness-implementation hybrid designs: Combining elements of clinical effectiveness and implementation research to enhance public health impact. Medical Care, 2012. 50(3): p. 217-226. 25. Proctor, E., et al., Outcomes for implementation research: conceptual distinctions, measurement challenges, and re- search agenda. Administration and Policy in Mental Health, 2011. 38(2): p. 65-76. 26. Brownson, R.C., G.A. Colditz, and E.K. Proctor, Dissemina- tion and Implementation Research in Health: Translating Science to Practice. 2012: OUP USA. 27. Peters, D.H., L. Paina, and S. Bennett, Expecting the unex- pected: Applying the Develop-Distort Dilemma to maximize positive market impacts in health. Health Policy Plan, 2012. 27(4): p. iv44-iv53. 28. Cornwall, A. and R. Jewkes, What is participatory research? Social Science & Medicine, 1995. 41(12): p. 1667-1676. 29. Subramanian, S., et al., Do we have the right models for scaling up health services to achieve the Millennium Devel- opment Goals? BMC Health Services Research, 2011. 11. 30. WHO. Bloomberg initiative to reduce tobacco use. Tobacco Free Initiative (TFI) 2013 [cited 2013; Available from: http://www.who.int/tobacco/about/partners/bloomberg/ en/. 31. Peden, M., G. diPietro, and A. Hyder, Two years into the road safety in 10 countries project: How are countries do- ing? Injury Prevention, 2012. 18(4): p. 279. 32. WHO, Changing Mindsets: Strategy on Health Policy and Systems Research. 2012, Alliance for Health Policy & Systems Research 33. Hanney, S. and M. Gonzalez-Block, Evidence-informed health policy: Are we beginning to get there at last? Health Research and Policy Systems, 2009. 7(30). 34. Habicht, J.P., C.G. Victora, and J.P. Vaughan, Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology, 1999. 28: p. 10-18. 35. Tran, N., et al., Engaging policy makers in road safety research in Malaysia: A theoretical and contextual analysis. Health Policy, 2009. 90(1): p. 58-65. 36. Consortium, P. PERFORM: Improving Health Workforce Performance. 2011 [cited 2013; Available from: http:// www.performconsortium.com/. 37. Newman, I., et al., A typology of research purposes and its relationship to mixed methods, in Handbook of mixed methods in social and behavioral research, A. Tashakkori and C. Teddlie, Editors. 2003, Sage: Thousand Oaks, CA. p. 167-188. 38. American Educational Research, A., Standards for reporting on empirical social science research in AERA publications. Educational Researcher, 2006. 35: p. 33-40. 39. Zwarenstein, M., et al., Improving the reporting of prag- matic trials: An extension of the CONSORT statement. BMJ: British Medical Journal, 2008. 40. Fairall, L., et al., Task shifting of antiretroviral treatment from doctors to primary-care nurses in South Africa (STRETCH): A pragmatic, parallel, cluster-randomised trial. Lancet, 2012: p. 889-898. 41. Baqui, A.H., et al., Effect of community-based newborn- care intervention package implemented through two service-delivery strategies in Sylhet district, Bangladesh: A cluster-randomised controlled trial. Lancet, 2008: p. 1936-1944. 42. Baqui, A.H., et al., Community-based validation of assess- ment of newborn illnesses by trained community health workers in Sylhet district of Bangladesh. Tropical Medicine & International Health, 2009. 14(12): p. 1448-1456. 43. Choi, Y., et al., Can mothers recognize neonatal illness cor- rectly? Comparison of maternal report and assessment by community health workers in rural Bangladesh. Tropical Medicine & International Health, 2010. 15(6): p. 743-753. 44. Shah, R., et al., Community-based health workers achieve high coverage in neonatal intervention trials: A case study from Sylhet, Bangladesh. Journal of Health, Population, & Nutrition, 2010. 28: p. 610. ReFeRences Implementation Research in Health A Practical Guide66 45. Davidoff, F., et al., Publication guidelines for quality im- provement in health care: Evolution of the SQUIRE project. Quality & Safety in Health Care, 2008. 17: p. i3-i9. 46. Brassard, M., The Memory Jogger II: A Pocket Guide of Tools for Continuous Improvement and Effective Planning. Vol. First Edition. 1994, Methuen, MA: Goal/QPC. 47. Campbell, D.T., J.C. Stanley, and N.L. Gage, Experimental and quasi-experimental designs for research. 1963, Boston, USA: Houghton Mifflin. 48. Speroff, T. and G.T. O’Connor, Study designs for PDSA qual- ity improvement research. Quality Management in Health- care, 2004. 13: p. 17-32. 49. Lindelow, M. and A. Wagstaff, Assessment of Health Facility Performance: An Introduction to Data and Measurement Issues, in Are You Being Served? New Tools for Measuring Service Delivery, S. Amin, J. Das, and M. Goldsteim, Editors. 2008, The World Bank: Washington, DC, USA. p. 19-66. 50. Mergler, D., Worker participation in occupational health research: theory and practice. International Journal of Health Services, 1987. 17: p. 151-167. 51. Chambers, R., Revolutions in development inquiry. 2008, UK: Earthscan. 52. Smith, L., L. Rosenzweig, and M. Schmidt, Best Practices in the Reporting of Participatory Action Research: Embracing Both the Forest and the Trees The Counseling Psychologist, 2010. 38: p. 1115-1138. 53. Israel, B.A., et al., Critical issues in developing and follow- ing CBPR principles. Community-based participatory research for health: From process to outcomes, 2008: p. 47-66. 54. Roy, S.S., et al., Improved neonatal survival after participa- tory learning and action with women’s groups: A prospec- tive study in rural eastern India. Bulletin of the World Health Organization, 2013. 91(6): p. 426-433B. 55. Onwuegbuzie, A.J. and K.M.T. Collins, A typology of mixed methods sampling designs in social science research. The Qualitative Report, 2007. 12: p. 281-316. 56. Pawson, R., et al., Realist review: “A new method of systematic review designed for complex policy interven- tions. Journal of Health Services Research & Policy, 2005. 10: p. 21-34. 57. Dieleman, M., et al., Realist Review and Synthesis of Reten- tion Studies for Helath Workers in Rural and Remote Areas, in Increasing access to health workers in rural and remote areas. 2011, World Health Organization: Geneva, Switzerland. 58. Collins, K.M.T., A.J. Onwuegbuzie, and I.L. Sutton, A model incorporating the rationale and purpose for conducting mixed methods research in special education and beyond. Learning Disabilities: A Contemporary Journal, 2006. 4: p. 67-100. 59. Leech, N.L. and A.J. Onwuegbuzie, Guidelines for Conduct- ing and Reporting Mixed Research in the Field of Counsel- ing and Beyond. Journal of Counseling & Development, 2010. 88: p. 61-69. 60. Creswell, J.W., et al., Best practices for mixed methods research in the health sciences. Maryland, USA. 2011: National Institutes of Health. 61. Tashakkori, A. and C. Teddlie, Mixed Methodology: Combin- ing Qualitative and Quantitative Approaches. 1998: SAGE Publications. 62. Creswell, J.W., Research design: Qualitative, quantitative, and mixed methods approaches. 2008: Sage Publications, Incorporated. 63. Creswell, J.W. and V.L.P. Clark, Designing and Conducting Mixed Methods Research. 2010: SAGE Publications. 64. O’Cathain, A., E. Murphy, and J. Nicholl, The quality of mixed methods studies in health services research. Journal of Health Services Research & Policy, 2008. 13: p. 92-98. 65. Glasgow, R.E., T.M. Vogt, and S.M. Boles, Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health, 1999. 89(9): p. 1322-1327. 66. Rogers, E.M., Diffusion of Innovations, 5th Edition. 2003: Free Press. ReFeRences ReFeRences • implementation research in health: a practical Guide 67Implementation Research in Health A Practical Guide ReFeRences ReFeRences • implementation research in health: a practical Guide 67. Damschroder, L.J., et al., Fostering implementation of health services research findings into practice: A consoli- dated framework for advancing implementation science. Implementation Science, 2009. 4(50): p. 50. 68. Epstein, J., Generative Social Science: Studies in Agent- Based Computational Modeling. 2006, Princeton, NJ, USA: Princeton University Press. 69. Humphreys, G., Are antibiotics still “automatic” in France? Bulletin of the World Health Organization, 2011. 89(1): p. 8-9. 70. Agyepong, I.A., et al., When solutions of yesterday become problems of today: Crisis-ridden decision making in a complex adaptive system (CAS) - the Additional Duty Hours Allowance in Ghana. Health Policy and Planning, 2012. 27: p. iv20-iv31. 71. Cooksey, D., A review of UK health research funding. 2006, London: HM Treasury. 72. Sanders, D. and A. Haines, Implementation research is needed to achieve international health goals. PLoS Med, 2006. 3: p. 719-722. World Health Organization Avenue Appia 20 CH-1211 Geneva 27 Switzerland Tel: +41 22 791 29 73 Fax: +41 22 791 41 69 alliancehpsr@who.int http://www.who.int/alliance-hpsr http://www.implementationresearchplatform.org ISBN 978 92 4 150621 2 We spend billions on health innovations, but very little on how best to apply them in real-world settings. Without implementation research, we are at best committing valuable resources in the hope that things will work out. This guide is an attempt to redress the defi- cit in understanding of implementation research and to encourage programme personnel and implementers to take a greater interest in the subject, recognizing that implementa- tion research is in fact an integral part of programme planning and execution, rather than something that happens once programmes are up and running. Intended for newcomers to the field, those already conducting implementation research, and those with responsibility for implementing programmes, the guide provides an introduction to basic implementation research concepts and language, briefly outlines what it involves, and describes the many exciting opportunities that it presents. IMPLEMENTATION RESEARCH IN HEALTH A PRACTICAL GUIDE

Published by the World Health Organization in 2013 Implementation Research in Health: A Practical Guide ©World Health Organization 2013 The World Health Organization has granted translation and publication rights for an edition in Chinese to Guangdong Provincial Academy of Chinese Medical Sciences (Guangdong Provincial Hospital of Chinese Medicine; The Second Affiliated Hospital of Guangzhou University of Chinese Medicine), which is solely responsible for the quality and faithfulness of Chinese version. In the event of any inconsistency between the English and the Chinese version, the original English version shall be the binding and authentic version. 健康领域的实施研究：实用指南 ©广东省中医药科学院（广东省中医院、广州中医药大学第二附属医院）中医药 标准化重点研究室 2020年 顾问 Anne E. Sales, Bryan J. Weiner, 徐 东 主译 李 慧 谢润生 陈文嘉 译者 王慧敏 任泳燕 沈 敏 郭蔚驰 黄金科 著者 David H. Peters, Nhan T. Tran, Taghreed Adam 世界卫生组织授权广东省中医药科学院（广东省中医院、广州中医药大学第二附 属医院）翻译和出版本手册的中文版，并由后者对本手册的中文版负责。 目录 致谢.......................................................................................................................................... 1 序言.......................................................................................................................................... 2 前言.......................................................................................................................................... 3 内容概要.................................................................................................................................. 5 第一章 为什么需要进行实施研究？.................................................................................. 9 第二章 如何使用实施研究？............................................................................................ 16 第三章 实施研究是什么？................................................................................................ 24 第四章 哪些人群应该参与实施科学研究？.................................................................... 32 第五章 哪些方式和方法适用于实施研究？.................................................................... 42 第六章 如何进行实施研究？............................................................................................ 56 第七章 如何让实施科学的潜能变为现实？.................................................................... 60 参考文献................................................................................................................................ 63 表格 表 1 质量改进对萨尔瓦多宫颈癌筛查及随访的影响................................................ 23 表 2 卫生领域中用于实施促进策略的类型................................................................ 27 表 3 实施结局变量........................................................................................................ 28 表 4 参与式行动研究与传统研究的比较.................................................................... 49 表 5 实施研究目标、实施问题和研究方法的类型.................................................... 55 插图 图 1 1960-1967 年和 1968-1969 年按月报告 20 个西非和中非国家的天花病例 .... 10 图 2 2007-2008 年赞比亚预防母婴传播的级联关系 ................................................. 12 图 3 实施研究的连续过程............................................................................................ 31 图 4 可用于各阶段的计划-执行-研究-改进循环和研究工具.................................... 47 补充资料 补充资料 1 加纳 Dangme West 地区医疗保险计划 ................................................... 14 补充资料 2 特定背景的研究转化为特定背景的实施................................................ 19 补充资料 3 实施研究在效果评估与促进中的作用.................................................... 20 补充资料 4 实施研究中实施者的重要性.................................................................... 34 补充资料 5 政策制定者和研究者共同关注在马来西亚的交通意外伤害................ 38 补充资料 6 在非洲的区域保健小组利用实施研究开展人力资源能力建设............ 41 补充资料 7 南非的一项实用性试验............................................................................ 44 补充资料 8 孟加拉国新生儿护理研究中应用的效果-实施混合试验研究 .............. 46 补充资料 9 改善新生儿健康的参与式行动................................................................ 50 补充资料 10 实施理论.................................................................................................. 53 健康领域的实施研究：实用指南 - 1 - 致谢 本指南由卫生政策与系统研究联盟（Alliance for Health Policy and Systems Research，AHPSR）在实施研究平台（Implementation Research Platform，IRP） 的支持下制定。卫生政策和项目的实施需要可靠的研究提供信息作为参考。所 以，实施研究平台被组建起来以通过有效的推广措施，确保在卫生系统内广泛 地提供所需的干预措施。卫生政策与系统研究联盟是实施研究平台秘书处的主 要承担单位，并领导后者的工作。卫生政策与系统研究联盟的主要目标是促进 卫生政策和系统研究的产生和使用，以期改善低中收入国家卫生与卫生系统。 尽管本指南已经得到许多人的帮助和建议，但由作者对本书的全部内容负 责。作者特别要感谢 Irene Agyepong对本指南的策划、案例研究和审查做出的 贡献；感谢 George Pariyo 对本指南的策划和审查做出的贡献；以及 Sally Theobald对本指南的案例研究和审查做出的贡献。作者还想感谢以下学者对本 指南的审查和反馈：Garry Aslanyan, Rajiv Bahl, Neal Brandes, Somsak Chunharas, Soraya Elloker, Abdul Ghaffar, Lucy Gilson, Margaret Gyapong, Luis Huicho, Jose Martines, KarsteinMaseide, Garrett Mehl, Olumide Ogundahunsi, Kelechi Ohiri, Enola Proctor, Krishna Rao, Suzanne Reier, Abha Saxena, Jim Sherry, Rajendra Shukla, and Jeanette Vega。 最后，我们要感谢 Gary Humphreys 将本项目内容进行整合。 健康领域的实施研究：实用指南 - 2 - 序言 如何采取经过验证的干预措施并在真实世界中实施是全球卫生领域面临的 最大挑战之一。开展实施研究等卫生系统的研究对于应对这一挑战至关重要。 它为符合特定背景的知证决策提供了基础，从而帮助将理论中的可能性转化为 现实中的实践。 世界卫生组织（World Health Organization，以下简称 WHO）长期以来在促 进卫生政策和系统研究（health policy and systems research，以下简称 HPSR） 方面发挥了主导作用——包括实施研究。其中包括了 2011年报告中的一些举措 （控制贫困传染病的实施研究）和 2012年出版的第一个关于 HPSR的战略（该 战略被称为“改变思维模式”，主张将研究更多地嵌入决策中，并呼吁进行更 多以需求为导向的研究）。WHO为继续支持这一领域，组织编写了本指南来介 绍实施研究这一富有挑战性的领域。 实施研究是一个相对较新且稍被忽视的领域。因而我们需要给予它更密切 的关注，确切地定义它的概念以及明确它能提供什么。因此，本指南介绍了实 施研究中使用的基本概念，并描述了可以使用的方法及应用的范围。该指南的 主要目的是支持实施研究的发展和需求。实施研究是以问题为中心，以行动为 导向，最重要的是与卫生系统的需求相一致。 实施研究需要广泛的利益相关者参与，并利用多个学科来应对复杂的实施 挑战。正如本指南所指出的那样，实施研究往往是一项集体协作的努力成果， 并且在许多情况下，提出问题的都是从事卫生保健第一线工作的人员。无论他 们是在运行特定的项目，还是在卫生系统中工作，他们会围绕实施提出问题。 因此，所有利益相关者必须了解合作在实施研究工作中的重要性。我们希望本 指南能够鼓励这种合作，并促进利益相关者在广泛的卫生系统中的聚集，以应 对这些系统每天面临的实施挑战。 Marie-PauleKieny 世界卫生组织卫生系统与创新部门助理总干事 健康领域的实施研究：实用指南 - 3 - 前言 人们对实施研究的兴趣正在增长，很大程度上是因为实施研究可以最大限 度地发挥卫生干预措施的益处。实施研究作为一个在卫生领域中相对较新的， 而且至今仍受忽视的领域，对许多人来说仍然比较陌生。因此，有必要更清楚 地了解实施研究究竟是什么，以及它能提供什么。本指南旨在明确这些方面。 实施研究涉及卫生系统中的各种干预措施。基于本指南的目的，我们考虑 了各种旨在改善人们健康的政策、方案以及个人实践和服务。无论这些干预措 施（例如：预防疾病，促进身体健康，或者是治疗和/或缓解疾病）如何运作， 研究干预措施的实施对于理解它们在真实世界中的如何发挥作用至关重要。 为了帮助实施研究执行者、实施计划负责者以及对两者都有兴趣的研究者， 本指南介绍了实施研究基本的概念和词汇，简要概述其涉及的内容，并描述实 施研究所提供的许多机会。本指南的主要目的是提高人们实施研究的能力，使 得与现实需求相一致的实施研究的需求能够得到激发。这与中低等收入国家的 卫生系统密切相关。 实施研究需要不同利益相关者和多个学科的共同参与，以解决他们所面对 的复杂实施挑战。因此，本指南适用于参与和/或受实施研究影响的各种行动者。 这包括负责制定政策和管理方案（这些方案决策决定了实施和推广的进程）的 决策者，最终执行这些决策的从业者和一线工作人员，以及来自不同学科的研 究人员（他们能专业系统地收集和分析信息以回答实施问题）。 我们还希望本指南可以吸引实施研究的教育工作者，有兴趣支持这类研究 的健康研究和卫生健康项目的资助者，以及对卫生健康项目和卫生研究感兴趣、 希望利用这些证据促进良好临床和公共卫生实践的民间社会团体。 虽然为了简单起见我们在指南中提到了实施者和研究者，但我们认识到这 些是理论上的分组，许多人同时扮演这两个角色。事实上，指南的核心论点之 一是在实施研究领域的工作人员积极参与的情况下，实施研究的获益往往能够 最大化。正是这些人看到了实施过程中哪里出错，因此他们更可能会提出最相 关的问题。我们鼓励实施者提出这些问题，并牵头开展更好的研究。我们还鼓 励研究人员在工作中更加务实，关注于对实施者至关重要的问题。如果研究是 为了促进实施，那么研究设计需要反映实施者正在解决的具体实施问题以及它 健康领域的实施研究：实用指南 - 4 - 们发生的背景。 本指南开篇的章节说明了为什么实施研究对决策很重要。其为实施研究提 供了一个实用定义，并说明了实施研究与通常被人们认为是单纯管理问题的相 关性，并举例说明了如何将这些问题界定为实施研究问题。前几章论述了实施 研究的开展，强调了协作的重要性，论述了实施者在研究策划和设计、数据收 集和分析以及结果推广和使用中的作用。 本指南的后半部分详细介绍了可用于实施研究的各种研究设计和方法，并 举例说明了定量、定性和混合方法的设计应用，回答与实施和推广有关的复杂 问题。这部分通过研究问题的识别、研究问题的提出、实施结果和变量的确定 以及研究设计和方法的选择等方面，为实施研究的概念化提供了指导，同时要 求通过严谨的研究设计解决一些重要的问题。 健康领域的实施研究：实用指南 - 5 - 内容概要 全球卫生界面临的一个关键挑战是如何将经证实的干预措施在真实世界中 加以实施。面对我们面临的许多健康方面的挑战，我们拥有负担得起并能够拯 救生命的干预措施，但是对于如何在现有的卫生系统中与各种可能的环境中广 泛地实施这些干预措施，我们仍知之甚少。未能成功有效地实施这些干预措施 使我们付出了代价。例如，每年有超过 28.7万妇女死于与妊娠和分娩有关的并 发症，而大约 760万儿童（包括 310万新生儿）死于可通过现有干预措施预防 或治疗的疾病。 理解真实世界中的实施 实施问题是由包括“真实世界”环境因素等一系列因素引起的，这些因素 要么被忽视，要么在其他研究领域仍未被攻克。实施研究揭示了这些因素，为 特定的环境和知证决策提供了依据，这对于理论上可能的事情在实践中成为现 实至关重要。因为实施研究是嵌入在现实中的，所以在真实世界中工作的人们 （从业者而不是研究人员）经常会提出一些问题，这些问题是新思考的起点。 我们应确保能够认识到这些现实问题，并且为了找到这些问题的答案进行研究， 而不是研究人员自己感兴趣的问题。这些是实施研究人者面临的关键挑战之一。 一种实用的工具 考虑到真实世界中效果评估的问题，实施研究是强有力工具，例如能够实 时捕捉、分析信息，和促进卫生系统的完善。实施研究在支持干预措施的推广， 及将其纳入国家级卫生系统方面尤为重要。在国家级政策推行的时候，小规模 试点研究中工作的干预措施往往达不到预期，或者因为背景差异无法从一个国 家推行到另一个国家。实施研究不仅有助于阐明发生这种情况的原因，而且可 以通过反复改进以支持某项政策在另一个国家的成功推行。因此，实施研究成 为一个帮助组织机构提升学习能力的工具，能够使他们对在迭代基础上发展起 来的知识进行理解、吸收并实施。 健康领域的实施研究：实用指南 - 6 - 协同合作 当实施者参与到一项研究的识别、设计和实施阶段时，实施研究通常是最 有用的。正因为如此，在政策制定、项目管理和研究中，促进关键利益相关者 之间的合作是非常重要的。一种鼓励研究人员和实施者之间进行协作的方法是 从一开始就将实施研究整合到政策和方案决策过程中，而不是将其与实施过程 分开进行。通过这种方式，科学探索可以以一种迭代、连续的方式整合到实施 问题解决过程中。实施研究可以作为一种载体，帮助基层社区识别被忽视的实 施问题、暴露实施效果的不足和增加卫生组织的可问责性。 在所有这些协同合作中，实施研究人员应该在他们的工作中善于聆听且能 够灵活变通。事实上，对背景、环境和体制的理解，以及灵活地采用适当的研 究方法，可能与坚持某种既定的研究设计同等重要，甚至更重要。 广泛的研究范围 一般来说，“实施研究”一词既描述实施过程中所用的科学研究，也描述 影响这个过程的背景因素。它可以探索或解决实施中遇到的任何方面的问题， 包括：影响实施的因素（例如贫穷、地理偏远或传统信仰）、实施过程本身 （例如通过分发经杀虫剂处理的防虫蚊帐[insecticide-treated bednet，ITN，下文 简称“防虫蚊帐”]，或大规模疫苗接种与重点人群监控策略相比），以及实施 结局或最终产物。 取决于所研究的主题，实施研究或多或少地适用于许多不同的研究领域， 并与之关联。例如，新药物研发的基础研究通常不涉及实施问题；然而，当确 保新药物能被所需的患者获取时则涉及实施问题。实施研究通常侧重于提供或 实施新干预所需的策略，这里称为“实施策略”，这是一个用来区分临床和公 共卫生干预措施的术语。为了研究实施过程，我们需要制定一个概念框架，基 于“可接受度”“适应性”“可行性”等指标来评价实施过程是否有效运行。 方法的多样性 因为实施研究涉及到各种各样的研究方法和学科，所以用一套“实施研究 方法”来讨论是没有意义的。然而，某些研究方法和设计特别有用，包括实用 健康领域的实施研究：实用指南 - 7 - 性试验（pragmatic trials）、效果-实施混合试验（effectiveness-implementation hybrid trials）、质量改进研究（quality improvement studies）和参与式行动研究 （participatory action research）。因为它们能够产生可用于实践的知识，善于捕 获背景中的细微差别，并提供响应变化所需的迭代灵活性——即研究方法能够 随着研究背景、影响因素的变化而进行反复多次的调整，从而使实施研究更加 适用、贴合于所研究的背景环境。虽然这些方法对于实施研究人员至关重要， 但应记住，在实施研究中“问题为王”——即问题决定所使用的方法，而不是 由方法决定所提问题的类型。实施研究问题通常很复杂，反映了可能影响实施 的多种背景因素，产生不可预测的影响，这需要实施者不断进行调整。特别是 在研究对象复杂而多变的情况下，研究人员需要具有相当大的灵活性来应对。 协调研究与需求，确保质量 理想情况下，实施研究应该与需求保持一致，既要从某种意义上解决目标 受众的关切问题，也要回应研究对象的特殊性。在这方面的一个关键考虑因素 是结果或预测所需的确切度。例如，一个政策制定者在自己的工作范围内，可 能正在寻找一个强有力的证据表明某项干预措施是可行的，但未必有时间开展 一项为期多年的研究来有力证明其可行性。响应不同目标受众的需求，可能对 研究的基本设计、预算和安排具有重要意义。为了确保实施研究符合需求和具 有高质量，有必要提出以下关键问题： ► 研究是否清楚地解决了有关实施的问题？ ► 是否清楚地描述了正在实施的工作（例如，实践、项目或政策的详细 信息）？ ► 研究是否涉及实施策略？如果是，是否进行了适当的描述和验证？ ► 研究是否在真实世界中开展？如果是，是否足够详细地描述开展条 件？ ► 研究是否适当考虑了实施结局变量？ ► 研究是否适当考虑了影响实施的背景因素和其他因素？ ► 研究是否适当考虑了系统随时间的变化及其复杂程度？ ► 研究是否清楚地确定了研究的目标受众和如何应用研究结果？ 健康领域的实施研究：实用指南 - 8 - 从实施研究中收获更多 尽管实施研究很重要，但其仍然是一个被忽视的研究领域，一部分原因是 缺乏了解实施研究是什么和它能提供什么，另外一部分原因是缺乏对实施研究 活动的人财物投入。我们花费了数十亿美元用于健康创新，但却很少关注如何 更好地使用这些创新成果。这个问题影响到每个人，尤其是实施挑战最大的中 低收入国家的人口。本指南旨在纠正对实施研究理解上的不足，并鼓励项目人 员和实施者更加关注该主题。让他们认识到实施研究实际上是项目规划和实施 的一个组成部分，而不是在项目启动和运行后才做的事情；且也要认识到实施 研究是为了其他研究人员的利益而进行的研究。对实施研究人员来说，他们可 以在研究过程中与实施者和项目人员进行更多的交流。只有通过联合深入了解 背景的实施者与熟悉科研方法的研究人员，才能促进我们对许多影响公共卫生 工作实施问题的理解。 第一章 为什么需要进行实施研究？ - 9 - 第一章 为什么需要进行实施研究？ 本章要点  尽管有大量证据表明，能挽救生命的干预措施是有效的，但在不同的环境和 现有广泛的卫生系统范围内，如何有效地实施这些干预措施却鲜有人关注。  实施问题往往是由于政策制定者和卫生系统管理者可能未曾考虑到的背景因 素造成的。  实施研究对于提高我们在真实世界所面临的问题的认识至关重要，其能扩大 和深化我们对真实世界因素以及其如何影响实施的认识。  实施研究对于揭示理论上可行但实践中难以实现的不契合点，具有巨大的价 值。 为什么需要进行实施研究？ “忽视实施问题会耗费生命和金钱。” 1966年 12 月，威廉·佛吉 （William Foege）医生驱车进 入尼日利亚东部的丛林，调查 已报告的天花暴发疫情。当时 人们已经知道了很多关于如何 处理这种疾病的知识；实际上， 自 1959年以来一场根除天花的 全球运动一直在进行，且当时疫苗生产和投放取得了几项突破，其中冻干天花 疫苗的研制以及使用简单的分叉针来提供有效剂量取得了显著进展。但到 1966 年底，天花仍在 31个国家和地区传播，人们担心大规模疫苗接种战略目标是不 可能实现的。实际上，大规模疫苗接种意味着要给 80%的相关人群接种疫苗， 达到所谓的群体免疫水平。但是，即使有了高质量的疫苗和简单神奇的分叉针， 要达到群体免疫水平的覆盖率也非常困难 [1]。这存在实施问题。 第一章 为什么需要进行实施研究？ - 10 - 回到尼日利亚，在证实报告的疫情确实是天花之后，佛吉医生开始采取应 对措施，并意识到他没有足够的疫苗在人群中实现大规模接种，并且目前只有 35%的人接种了该疾病的疫苗。与此同时，新的疫苗供应和运输所需的额外车 辆还需几周的时间才能到达。因此，他需要一个后备计划。 最后，这项计划设计在受天花影响的村庄实施，给尚未感染疾病的人接种 疫苗，并有效地用免疫屏障隔离疾病高发区域。佛吉意识到有许多感染者可能 不知道自身已感染了病毒，因而他们仍会四处走动，特别是到买卖食品和货物 的市场。所以，佛吉绘制了当地的运输路线和选定了他们常去的市场。然后， 他利用剩余可用疫苗，在最有可能发现新病例的区域内建立了免疫屏障圈。 正如人们所知，这项监控策略（surveillance-containment strategy）仅用 5个 月就停止了尼日里亚东部的天花暴发，并通过在 1200万左右的人口中仅为 75 万人接种疫苗来实现[2]。佛吉并不是第一个使用这种方法的人。事实上 19世 纪英国也曾采用类似的方法来控制疫情。他在非洲撒哈拉沙漠以南地区这样苛 刻条件下首次使用这种方法，所取得的成效广为称道。实现群体免疫的问题已 被证明无关紧要。监控策略方法的力量很快在其他非洲国家得到了证实（见图 1），在其他地方甚至取得了更令人印象深刻的成果，特别是在印度的泰米尔纳 德邦。次年，罗摩占陀罗·拉奥（D.A. Ramachandra Rao）领导了一个小组，开 展了一项遏制天花的运动，在短短六个月内就阻止了天花在 4100万居民中的传 图 1 1960-1967 年和 1968-1969 年按月报告 20 个西非和中非国家的天花病例 资料来源：Foege等，1971 [2] 第一章 为什么需要进行实施研究？ - 11 - 播[2]。随后，这种监控策略在全球范围内得到推广，最终天花于 1979年被正 式宣布根除。 当然严格地说，监控策略的实施并不是一项实施研究，而只是一种当时人 们所称的描述性的“实地研究”。虽然佛吉只是在应对疫情暴发，但是他的努 力成果对研究具有巨大价值。正如本指南所示，对实施认知的一些重要突破并 非由“从事研究的人员”所发现。 本指南的核心信息之一是，学术界内外的人都可以对实施研究做出贡献。 而往往一线工作者（如在偏远农村诊所工作的医生或当地社区工作的助产士） 面临着一些特殊的困难，能提出激发新思考的问题。所以，我们应确保所开展 的研究是为了回答这些来自实施的问题，而不是针对研究者自己感兴趣的问题。 下文将继续讨论这个话题。 虽然自佛吉进入尼日利亚 东部丛林大约已过半个世纪， 我们目前面临的问题总体上已 经发生改变，但我们仍在努力 充分利用现有的疫苗。目前， 有意实施疫苗接种策略的中低 收入国家更有可能努力引进新 疫苗，或将免疫接种计划纳入 提供常规服务的卫生系统中。 在这个过程中，这些国家面临着一系列阻碍有效实施的因素，包括管理、制度、 社会行为和财政挑战，其中任何一个方面都有可能影响疫苗接种计划，并阻碍 健康卫生水平的提高。 正如上文所提到，实施研究支持了佛吉推广对天花免疫的监控方法。这些 有效的实施研究可以在提高我们对上述挑战的认识，同时在支持卫生系统内实 施常规免疫接种计划方面发挥至关重要的作用。通过这种方式，免疫接种可作 为新生儿和儿童保健服务的常规部分持续开展，而不仅是通过特殊活动进行。 许多其他干预措施也应如此，否则他们往往会由于实施失败而达不到预想效果。 第一章 为什么需要进行实施研究？ - 12 - 尽管医疗和公共卫生在上个世纪取得了进步，但每年仍有 28.7 万妇女死于 与妊娠和分娩有关的并发症，约 760 万儿童（包括 310 万新生儿）死于现有干 预措施本可有效防治的疾病[3]。这些死亡往往是实施过程出现的问题所导致的。 而这些问题是可以通过精心设计、策划和执行的实施研究加以阐明的。 如前文所述，本指南的主要关注点是实施研究，其与中低收入国家密切相 关。尽管有大量证据表明中低收入国家能负担得起有效挽救生命的干预措施， 但人们却不甚了解如何在不同环境和卫生系统内有效地实施这些干预措施[4]。 我们知道，使用防虫蚊帐可以降低疟疾传播率，但需要更多的研究探索出最有 效以及经济上可持续的方式派发这些防虫蚊帐，同时确保这些蚊帐得到正确使 用。我们知道口服补液疗法（oral rehydration therapy）在治疗腹泻相关疾病方 面极为有效，但仍需努力使口服补液疗法的采用率达到足够水平。 同样，我们知道抗逆转录病毒药物治疗（anti-retroviral treatment，下文简称 图 2 2007-2008 年赞比亚预防母婴传播的级联关系 译者注：cART（combined anti-retroviral treatment）为联合抗逆转录病毒治疗。 资料来源：Mandala等，2009 [5] 第一章 为什么需要进行实施研究？ - 13 - ART）项目可以延长艾滋病毒（human immunodeficiency virus，下文简称 HIV） 感染者的寿命，但往往无法确保每个需要治疗的患者都能获得治疗。如图 2所 示，赞比亚实际上只有小部分孕妇（不到 4%的 HIV阳性感染者）接受了为防 止 HIV母婴传播（prevent mother-to-child transmission of HIV，PMTCT）而进行 的抗逆转录病毒药物治疗[5]。 这些失败的原因对于实施者和项目负责人来说都太熟悉了：人力资源短缺、 供应链问题、低效率的分销渠道和患者可获得性障碍，其中包括一些明显的障 碍，如自付费用高，或缺乏便利的设施；以及一些更深层次的障碍，如社会经 济或性别歧视，或文化偏好与价值观。上述因素可能妨碍人们从社区获得和/或 从现有干预措施中受益。在某些情况下，政策制定者和卫生系统管理者没有考 虑到相关的背景因素，导致出现了一些卫生系统以外的实施问题。2004年首次 在加纳开展的用于支持发放防虫蚊帐的代金券试点项目正是一个很好的例子[6]。 这个项目的基本思想是： 在每位孕妇第一次去产前诊所 时，给她发放一张防虫蚊帐的 代金券。这样，当她通过正规 零售店购买防虫蚊帐时，可获 得约 4.20美元的折扣。坦桑尼 亚联合共和国启动了类似的项 目，并在 20世纪 90年代末取得了相当大的成功。然而，该项目在加纳却最终 失败了。导致两者结果差异的主要原因是这两个国家的防虫蚊帐生产和分销模 式的不同。 坦桑尼亚联合共和国和加纳的聚酯纺织厂均具备生产防虫蚊帐的能力。但 是只有在坦桑尼亚联合共和国的聚酯纺织厂才能将部分纺织产品制作成网，这 更多是为了响应政府政策的号召，抓住由防虫蚊帐项目的首次社会营销带来的 商业机会。在加纳，是由私营企业生产本地的蚊帐聚酯网。遗憾的是，在加纳 推行代金券项目时，由于该项目的设计者更偏好与正规私营企业合作，而忽略 了非正规私营企业。这些正规私营企业不是由国内聚酯纺织厂供货，而且他们 第一章 为什么需要进行实施研究？ - 14 - 的防虫蚊帐的进口能力十分有限。因此，正规私营企业无法满足防虫蚊帐代金 券项目的需求，当母亲们出示代金券时才发现自己无法兑现。 人们经常说事后经验总结、亡羊补牢是件好事。但应该说，如果我们适当 地开展实施研究，将重点放在事件发生的背景上，则有可能帮助实施者提前预 见这些在加纳出现的问题，特别是认识并重视非正规私营企业在提供防虫蚊帐 方面的作用。正是因为实施研究能够阐明背景问题、遇见一些实施中可能出现 补充资料 1 加纳 Dangme West 地区医疗保险计划 20 世纪 90 年代末，加纳的 Dangme West 地区（阿克拉郊外一个贫穷的农村地 区），只有四个连医生都没有的公立农村卫生中心为当地人民提供服务。更糟的是， 于 1985年开始加收服务费导致许多穷人甚至无法享受基本的医疗保健服务。因此，寻 求一种可行的、不含自付费用的健康卫生筹资替代方法的压力越来越大。发展医疗保 险的想法得到认可，因为该地区的大多数人非常贫穷，或者并不在正规企业工作。 为了测试在阿克拉地区开展医疗保险计划的可行性，当地开展了一项形成性咨询 研究，邀请了该地区卫生管理团队（District Health Management Team）、区议会（地 方政府）以及当地居民进行了多轮磋商。经过社区层面和个人层面的广泛讨论，借鉴 其他中低收入国家的例子，并研究高收入国家的医疗保险历史及其演变，最终得出了 可行的设计方案。在该地区服务有限的情况下，广泛协商的主要内容包含符合当地情 况的福利方案内容、应付费用和可能出现的成本费用等实施层面的细节。同时，如何 确定受益人以及如何确保卫生保健服务的可获得性这一个具有挑战性的问题，也是重 点讨论内容之一。在经过长时间的磋商之后，最终引入一个涵盖全区卫生院所有门诊 服务的福利方案。 卫生中心被授予“守门员”的角色，为需要进一步接受医疗服务的患者提供前往 周边地区四家医院的转诊服务。所有家庭成员都必须在该计划中登记，保费额度是以 个人为单位计算，而不是以家庭为单位。这项研究的许多发现和经验对加纳国家医疗 保险计划的制定产生了积极影响，最终该计划于 2003年通过法律正式在国家层面推 行。 资料来源: Waddington等，1990 [7]；Agyepong等，2007 [8] 第一章 为什么需要进行实施研究？ - 15 - 的问题，因而它是实施者在设计阶段的重要工具。而且有许多实施研究案例可 以有效辅助政策制定和方案设计。加纳 Dangme West医疗保险计划的发展就是 一个很好的案例。该计划的设计者充分利用不同实施研究工具成功预测了可能 出现的挑战（补充资料 1） [7, 8]。 结论 因此，实施研究对于揭示理论上可行但实践中往往难以实现的不契合点， 具有巨大的价值。实施研究融入到真实世界，并从真实世界的实施者和他们所 服务的社区中汲取力量，产生了特定于事件发生背景的、在研究者有限视角中 难以获得的经验。下一章将阐述如何在不同的应用场景中准确地实现实施研究 的所有潜力。 第二章 如何使用实施研究？ - 16 - 第二章 如何使用实施研究？ 本章要点  实施研究对于理解背景、评估成效、有依据地实施和促进卫生系统的加强至 关重要。  实施研究在支持推广干预措施并将其纳入国家级卫生系统方面尤为重要。  实施研究也可以用于帮助组织机构提升学习能力 如何使用实施研究？ “实施研究将我们所知道的转化为我们所做的。” 在广泛论证了实施研究的重要性之后，接下来我们将聚焦于一些对各种利 益相关者均具有巨大价值的特定领域。这些利益相关者可能包括能够利用实施 研究支持卫生政策的部长级决策者、希望理解特定背景问题的项目管理者，以 及寻求评估绩效、做出改进或引进创新的医疗保健提供者。实施研究为他们打 开了一扇窗户，让他们能了解到在真实世界中提供卫生服务所带来的实际挑战。 了解背景 正如前一章引用的加纳经杀 虫剂处理的蚊帐实例所表明的， 在阐明可能影响干预结果的背 景因素方面，实施研究发挥着 重要作用。这一点很重要。因 为证据表明，即使干预措施是 以相似的方式进行设计，但在 不同的环境中可能会发生各样 的变化，同时会带来不同的效果[4]。例如，有充分的证据表明，儿童疾病综合 管理（Integrated Management of Childhood Illness）是一种同时强调治疗护理与 第二章 如何使用实施研究？ - 17 - 疾病预防的、系统的儿童健康管理方法。儿童疾病综合管理很大程度上依赖于 当地卫生系统的特点，如卫生保健工作人员的基本技能、药品供应、监督与转 诊的有效性[9]。儿童疾病综合管理的成效也与患者是否充分利用现有服务有关。 正如有关“三种延误模式”（决定寻求医疗保健、前往医疗机构、到达医疗机 构后接受优质医疗保健）的研究所证明的，该模式在应对降低产妇和新生儿死 亡率的挑战中发挥了核心作用，并体现了实施研究在识别和描述关于医疗服务 可及性方面的障碍有独特的价值[10]。 如前一章所述，有些关于文化障碍的微妙的、细微的信息，即便是通过严 谨的定量信息收集方法也可能遗漏，而实施研究对于挖掘这类信息有着独特的 价值。 例如，性别角色和家庭决策权可以决定个人获得所需医疗服务的程度。特 别是当就医有关的决策权落在男性家庭负责人身上时，如果男性不在场，女性 往往会拖延寻求治疗（即使在紧急情况下），并且可能损害自己和子女的健康。 在存在这种文化障碍的地方，实施研究人员和项目管理人员能在转变向社区居 民告知可用医疗保健的方式上发挥重要作用。例如，当男性户主不在家时，将 这些女性和子女的就医需求信息转达给替代决策者。这类举措确保孕妇不会耽 误获得紧急产科医疗服务的时机——在某些地区，延误治疗是造成产妇高死亡 率的重要因素。 由此可见，实施研究可以在多个层面上为实施者提供重要的见解。实施者 一般都知道所谓的实施并不仅仅是简单地将相同的模板套用到一个又一个国家 [11]。这些见解的价值体现在印度安得拉邦正在进行的工作中。在那里，基于 微观规划的实施研究被用来重点了解当地背景，包括弱势群体的看法和需求 （补充资料 2）[12]。 评估成效 实施研究还可以用来衡量一段时间内干预的成效，并作为预测未来的基础。 适当的情况下，它还可以用来观察相似组织或方案的执行方式及其结果是否存 在差异。监测和评估是许多实施者和项目管理者长期开展的工作。值得注意的 第二章 如何使用实施研究？ - 18 - 是，监测和评估在这类实施研究中发挥着重要作用，通常有助于定义重要的研 究问题。2002年塔利班政权垮台后，阿富汗所做的工作便是如此，当时一个简 化的“计分卡”被用于医疗系统成效的监测与评估，以及确定年度优先改进事 项（补充资料 3）[13-15]。这种方法揭示了在提供基本服务方面存在的一些限 制因素，这些因素需要通过重新分配资金、决策、培训和进一步研究加以解决， 从而为评估一些新的实施策略奠定了基础，进而更好地提供资金和开展卫生服 务。 影响和促进干预措施推广的因素 在成功实施干预措施的过 程中，背景因素显然对干预措 施的推广方式有至关重要的影 响。在小规模试点研究中有效 的干预措施，在国家层面实施 时往往达不到预期。我们知道 什么有效，但不一定知道如何使它在真实世界中发挥作用。即使是某一简单的 干预措施，其成功实施也需要了解真实世界（社区、卫生系统、经济）与干预 措施是如何相互作用并相互影响。 当然，许多新的干预措施实施起来谈何容易，当中可能涉及引入新的诊断 工具或新的信息通信技术，也可能采取组织形式的变革支持创新，例如支持一 种新的外展方式。此外，创新可能是临床或行政服务中某环节的改变，或是一 个新的项目、方案、战略或政策。无论创新采取什么形式，很明显，如果希望 事情会有所改善，仅仅依靠某个新的措施是不够的。采取措施后所产生的结果 需要实施者去识别和了解。 第二章 如何使用实施研究？ - 19 - 补充资料 2 特定背景的研究转化为特定背景的实施 收集印度 HIV感染者和艾滋病病人数据的工作极大地提高了对 HIV流行病学整体 状况的认识水平，但是要产生指导高效分配资源和实施有效的艾滋病应对措施所需的 证据，仍然面临重大挑战。在安得拉邦，美国国际开发署（United States Agency for International Development，USAID）为支持印度实施 HIV 政策而签署的萨曼沙 （Samastha）项目一直在努力完善这种应对措施。该项目的策略是基于 2006年在安得 拉邦进行的需求评估的结果。该评估结果发现了艾滋病医疗服务存在的较多不足，既 降低了医疗保健质量，也使许多人不了解现有服务。为了弥补存在的不足，萨曼沙项 目重点采取了四个方面的对策：①提高患者在医疗机构中获得优质医疗资源的机会； ②建立自我评估和监测制度，以预防感染，提高医疗质量；③建立计算机管理信息系 统，以追踪患者信息，支持项目规划和监测；④利用微型规划和支持小组巩固社区外 展服务。 微型规划是一种基于特定背景的“自下而上”的工具，用于开发、实施和监测为 当地社区需求而量身定做的活动。希望改善当地艾滋病服务的社区可以利用微型规划 来识别弱势群体，分析服务的可用性和可及性，并根据现有可用服务进行优选排序。 微型规划对于追踪客户的服务使用情况也是非常有价值的。微型规划工作的一个关键 方面是利用了同行外展工作人员（peer outreach worker，下文简称 PORW）、从社区 中选择的 HIV感染者（people living with HIV，PLHIV）在目标地区进行外展服务、转 诊和随访。PORW在与项目目标人群建立和保持联系方面发挥了关键作用，并能够提 供当地背景的重要细节，尤其是在覆盖范围内社区的需求和特征。 PORW使用一系列工具来帮助他们识别 HIV感染者和易感人群，制定满足个体需 求的计划，并确保客户使用了必要的服务。由于微型规划的实施，PLHIV 的登记从 2007 年的 610 人增加到 2011 年的 5907 人。抗逆转录病毒药物治疗（anti-retroviral treatment，下文简称 ART）服务的利用率也急剧上升，接受 ART或前 ART服务的人 数几乎翻了两番，而未登记参加 ART服务的合格 HIV感染者人数从 228人锐减到 18 人。 资料来源: Sankar, 2013 [12] 第二章 如何使用实施研究？ - 20 - 补充资料 3 实施研究在效果评估与促进中的作用 经过几十年的战争和 2002年塔利班政权的倒台，阿富汗的卫生部门处于混乱 无序的状态。基础设施遭到破坏，卫生工作者和管理人员很少，没有功能健全的 卫生信息系统。主要由非政府组织（NGO）提供极为有限的卫生服务管理。新政 府制定了一项国家卫生政策和战略，以便在捐赠组织、国际及当地非政府组织的 帮助下，在全国范围内实施基本卫生服务包（BPHS）。在政府官员、捐助组织、 非政府组织和一线供应商的投入和外部评估小组的协助下，新政府制定了关于基 本卫生服务包的平衡计分卡，每年对 34个省的基本卫生服务包的成效进行总结。 这张一页纸大小的记分卡共有 29个条目，涵盖了六个主要领域：1、患者和社区 的看法；2、员工的看法；3、卫生服务供应能力；4、服务提供（质量和数量）； 5、金融系统；6、妇女和穷人平等的总体愿景。 研究者随机选择了 700多个卫生机构、7000多个患者-卫生服务提供者，在他 们的医疗活动和面谈中收集观察数据。此外，还对随机纳入的卫生工作者进行了 2000多次访谈。平衡计分卡用于发现需要培训和资源分配的领域、提供合同奖金 或取消合同、激励员工并让公众进行问责，并作为试验新政策的基准。此外，它 推动了初级诊所患者费用的取消和基于成效的支付方案的推广。总的来说，全国 所有相关领域的趋势都在不断改善。 资料来源：Peters等，2007 [13]；Hansen等，2008 [14]；Edward等，2011 [15] 第二章 如何使用实施研究？ - 21 - 此外，干预措施的推广，不是简单地在更广范围内重复做同样的事情。推 广通常等同于从有限的研究区域增大地理覆盖范围。事实上，这种推广更准确 的定义是横向规模扩展，或“传播”，通常包括复制干预措施。这与纵向规模 扩展形成鲜明对比——纵向规模扩展被定义为通过政策、监管、预算或其他卫 生系统变化的涉及创新的制度化。换句话说，纵向规模扩展是在卫生系统体制 结构中进行创新的复杂过程[16]。干预措施的推广还可能涉及卫生系统的组织、 财务和技术能力的扩展。毋容置疑，每一种形式的推广，对决策者和支持他们 的研究者来说都是特别的挑战。 为了获得有效的实施支持，研究人员需要解决一些关键问题，包括： ► 引入和增加某个新的措施的预期效果是什么？（包括期望的和不期望 的结果） ► 对同种干预措施，有无其他引入和推广（或不推广）方案的相关经验 可供参考？ ► 推广干预的主要阻碍/机遇是什么？ ► 关键利益相关者将受到什么影响？ ► 不同的利益相关者在干预措施的推广过程中扮演何种角色？（例如， 倡导者、反对者等角色） ► 随着时间的推移，协调不同利益相关者策略的效果如何？ ► 推广干预措施的预计成本是多少？ 支持质量改进，完善卫生系统 与所有卫生政策和卫生系统研究（包括实施研究）一样，在支持质量改进 和卫生系统的完善方面，核心是提出与所面临挑战相关的问题。当研究能够解 决决策者和实践者提出的问题时，实施研究优势可得到充分的发挥。卫生政策 和卫生系统研究通常被称为卫生系统的大脑、眼睛和耳朵。决策者能够依赖这 些研究作为用于持续获得信息的唯一机制，从而能以最佳方式适应不断变化的 第二章 如何使用实施研究？ - 22 - 环境[11]。因此，一些优秀的实施研究通常能得到该领域从业人员的支持。这 些从业人员包括每天都面临着质量问题的医生、护士和项目管理人员。 由于实施研究使得循序渐进 的迭代改进变为可能，所以它被 赋予巨大的价值。一个典型例子 是 2002 年一项由泛美卫生组织 和卫生部（Pan American Health Organization and Ministry of Health）资助的、在萨尔瓦多发 起的基于细胞学方案的研究。该研究试图增强早期检测的机构特点。萨尔瓦多 的宫颈癌防治项目早在 2002年就出现了许多问题：很少有妇女接受巴氏涂片 （PAP涂片）检查，许多实验室样本不尽人意，且很少对阳性者进行阴道镜检 查随访。值得高兴的是，质量改进循环（Plan-Do-Study-Act）的使用推动了外 展工作者的培训工作，为需要筛查的妇女更好地提供筛查的机会，并鼓励随访。 仅仅一年后，结果显著改善（表 1）[17]。这一举措随后被卫生部采纳并推广。 质量改进和加强卫生系统的关键驱动力通常是组织的学习能力。学习型组 织被定义为以促进成员或员工之间学习和知识共享而构建的组织[18, 19]。像学 习型人才一样，学习型组织能更好地预测问题，并制定对策，从而提高效率。 学习型组织学习能力的关键在于能够吸收和实施新知识，而这些新知识是由旧 知识经过不断的检验，并吸收新的经验教训，迭代发展而成的[20]。 决策者需要确保其负责的组织机构能持续学习，而实施研究在这方面为决 策者提供了重要的帮助，特别是通过提出以下问题：可以用哪些方法和步骤来 创建学习型组织？卫生组织如何与各利益相关者（例如管理机构、融资机构、 监管机构、受益人、内部员工）互动，从而提高学习效果和组织效能？毋庸置 疑，这是一个复杂的话题，我们后面还会对其进行探讨。 第二章 如何使用实施研究？ - 23 - 资料来源：Agurto等，2006 [17] 隐性知识 经上文讨论实施研究的重要性并概述了它的一些关键应用之后，我们应该 认识到实施研究无法提供成功实施所需的全部知识。通过师承学习和实操经验 （包括观察和实践）也可获得实施经验。这种专业知识有时被称为“隐性知 识”，是个人和组织学习的重要组成部分[21]。很明显，隐性知识和从实施研 究中获得的正式知识之间的界限往往存在重叠。例如，对卫生管理人员的隐性 知识的研究可能可以提供有关实施的重要见解。那些开展实施研究或使用研究 结果来做决策的人，可以通过师承学习和从正式途径以外的方式学习到很多实 践经验和的隐性知识。 结论 本章试图简要概述实施研究的广泛用途，包括其在阐明干预措施从试验研 究转移到真实世界时所面临的挑战和机遇方面的价值。下一章将具体讨论什么 是实施研究，并提供一个适用于涵盖各种研究领域的实际定义。它还将涉及帮 助服务、方案和政策交付改进的实施策略。最后，它将介绍实施结局变量—— 可以用于描述和评价各种实施开展的情况的变量。 表 1 质量改进对萨尔瓦多宫颈癌筛查及随访的影响 干预前 干预后 上一年首次筛查的 30-59岁女性人数 未知 （样本量 2446） 3408 不满意样本数量 41（1.7%） 14（0.4%） 标本从门诊到实验室的转运时间（天） 23 9 标本从实验室到门诊的转运时间（天） 27 11 PAP涂片阳性妇女接受阴道镜检查随访的人数 22（24%） 196（100%） 第三章 实施研究是什么？ - 24 - 第三章 实施研究是什么？ 本章要点  实施研究可以解决实施中各方面的问题，包括影响实施的因素，实施过程和 结果，或实施的最终产品。  实施研究或多或少地相关并适用于许多不同领域。取决于所研究的对象，有 的研究问题涉及大量的实施相关内容（重实施），而有的则涉及较少（轻实 施）。  实施研究通常侧重于提供或实施新干预措施所需的策略，即“实施策略”。  为了理解实施过程，有必要使用一个框架来概念化和衡量实施结局。实施结 局变量是该框架的基础，可作为实际实施工作情况的指标。 实施研究是什么？ “实施研究的基本目的不仅是要了解哪些措施是有效的或无效的，还要了 解实施成败的原因和方式，以及检验改进实施的方法。” 实施研究是一个立足于许 多学科和研究传统，而在不断 发展的研究领域。它解决了在 各种背景下的一系列实施问题。 在许多方面，这是实施研究的 巨大优势——它能够利用不同的资源，从多个角度思考，并提供多领域见解。 但它也面临着一些明显的、学者们往往会选择回避的分类学挑战。术语、名词 命名中仍存在着一些混乱，同时关于实施研究的范畴也存在不少争议[22,23]。 一般而言，“实施研究”一词描述的是关于实施某些倡议活动或项目所采 用的过程，以及影响这些过程的背景因素的科学研究。实施研究的一个主要目 的是支持和促进已证明有效的干预措施的成功应用——如一种已知可杀死疟原 虫的药物，一种结核病患者的诊断试验，或一种防止艾滋病病毒母婴传播的策 略。它可用于确定如何部署人力资源，以便地理位置偏远的社区在需要的时候 获得医疗服务，并确定如何消除阻碍弱势群体获得所需服务的财务障碍。它还 第三章 实施研究是什么？ - 25 - 涉及降低成本和提高组织的效率和可问责性。最后，实施研究的目的是学习如 何将具有发展前景的实施策略规模化，更重要的是，如何长期维持这些策略。 卫生领域的实施研究，是一种涉及临床和公共卫生政策、规划和实践研究 有关的卫生政策与系统研究，其基本目的不仅是要了解哪些措施是有效的或无 效的，还要了解实施成败的原因和方式，以及检验改进实施的方法。正如前文 所言，实施研究往往关注的是一项计划或项目推出或推广时出现的实施问题。 基于本指南的目的，我们提出了一个非常简单和广义的实施研究定义：实 施研究是对实施问题的科学探究。它可以在不同研究团体间使用，并且对于卫 生部门从业者、政策制定者以及感兴趣的公众都有意义。 根据这一定义，实施研究可以解决或探讨实施的任何方面，包括影响实施 的因素（如贫困，地理位置偏远或传统信仰），实施过程本身（如通过防虫蚊 帐的分发，或大规模疫苗接种与重点人群监控策略的比较），以及实施的结局 或最终产品。如上所述，实施研究可能侧重于以下主题：识别常见的实施问题； 了解促进或阻碍获得卫生干预措施的因素；无论是在特定背景中还是在一定范 围的环境中，开发和测试解决方案以解决实施障碍；确定将创新引入卫生系统 或促进其大规模使用和可持续发展的最佳方式。 实施策略 实施研究通常涉及提供或实施健康产品或临床干预所需的策略，可能不关 注这些产品或干预的安全性或有效性。这种策略有时被称为“实施策略”，这 一术语用于将其与临床和公共卫生干预区分开[24]。例如，虽然外展诊所和监 督检查表通常用于提高免疫规划覆盖面和质量的实施策略，但疫苗提供被认为 是一种健康干预。实施策略可用于改善实施中存在的社会文化问题，例如通过 提高干预的接受度或采用率。实施策略也可影响所提供服务的质量和成本等。 实施研究可能侧重于实施策略本身，或考虑把实施策略纳入到更广泛的卫生干 预研究中。 如表 2所示，讨论实施策略的一种方式是根据使用它们的参与者或利益相 关者对它们进行分组。典型的实施策略包括：（1）提高政府能力（公共政策， 监督和金融机构）；（2）提升实施和卫生服务提供方的成效；（3）提升个体 服务者和一线工作者的能力和表现；（4）赋予社区和家庭更多权力；（5）支 第三章 实施研究是什么？ - 26 - 持多个利益相关者参与健康促进 [4]。 实施结局 为了促进我们对实施过程的理解，并提高实施研究的效率，有必要建立一 个框架来概念化和衡量实施结局。该框架使得开展实施策略的有效性比较研究 变为可能——此类研究往往有着大量而迫切的需求。为了概念化和衡量实施的 成败，采用某种惯用统一的分类方法是有用的。例如，在我们讨论实施响应的 不同方面（如干预的接受度，或者是干预措施已被使用或采用的程度）时。 这些特征可以看作是实施的结局，被称为实施结局变量。实施结局变量可 作为某个实施项目实际开展情况的指标。实施结局变量（包括接受度、采用率、 适当性、可行性、保真度、实施成本、覆盖范围和可持续性）也可被视为促成 其他重要结局（如对医疗保健或健康状况的满意度）的中间因素[25,26]。 并非所有实施结局变量在实施干预或实施研究中都同等重要（见表 3）[25]。 例如，一项新的干预措施，主要关注的可能是接受度、采用率、适当性和可行 性问题。而对于现有的干预措施，干预措施的实施程度与最初设计的干预程度 相同，或忠实于原始干预措施的程度（通过保真度进行衡量）通常非常重要， 成本和覆盖范围也是如此。虽然可持续性问题应该从干预的最初阶段就加以考 虑，但在卫生干预研究中这些问题往往被忽视[27]。 每个变量代表了实施的一个重要方面，均可通过实施研究进行探讨。 第三章 实施研究是什么？ - 27 - 改编自：Peters等，2009 [4] 表 2 卫生领域中用于实施促进策略的类型 主要参与者和 干预领域 实施策略实例 政府 •决策，监督和管理 •公共筹资 •政策述评 •强化治理和减少腐败的策略 •签订成效指标合同 •下放公共服务供应 •公共筹资激励机制和规则（提高收入、筹集资金和支付机 制） •公共教育，行为改变交流模式 实施和卫生服务提供组织 •组织改进和问责制 •质量促进/质量保证/成效管理策略：解决团队问题；制定、 应用指南和标准操作程序；定期监督 •根据绩效为团队和个人提供财务激励 •重组和/或整合服务 •人力资源管理制度 •加强设施管理和物流制度 •加强财务管理 •健康服务和产品的营销 个体服务者和一线工作者 •个体实践 •继续教育和培训 •同伴学习和支持 •工作辅助工具 社区和家庭 •赋权（ empowerment）， 参与，教育 •个体实践 •社区信息和教育：培训社区医务人员；培训社区成员，例如 青年、母亲（在团体、家庭、大众媒介中）等；社会营销和 需求创造 •加强包容和参与：社区管理服务；社区伙伴关系和共同管 理；社区自有服务 •加强地方可问责性：联合监督；服务提供者可问责性；基于 社区的信息系统 •地方组织能力建设：社区动员；社区委员会监督与管理 •财政赋权：社区筹资；实物补贴和代金券；社区参与式预 算；采用创收和小额融资计划 •卫生服务和健康行为的同伴支持 多方参与者 •评估需求和限制：规划并减少限制 •获得利益相关者的广泛支持：吸引有实力的利益团体；与社 区组织协调 •根据利益相关者的反馈对管理流程进行灵活的调整 第三章 实施研究是什么？ - 28 - *最初的定义是指某一种“创新或循证实践”。本表使用“干预”一词，以便其含义能 更广泛地适用于各个项目和政策。 **其他术语在大型项目和政策的实施文献中更为常见。（Peters等，2009；Rogers等， 2003； Carroll等，2007；Victoria等，2005） ***为计算效率、成本效用、成本效益或成本效果的相关测量指标提供分子。许多成本 分析考察的是干预措施的实施总成本，包括干预本身的成本，以及在特定环境中实施特定 策略的成本。 改编自： Proctor等，2011 [25] 实施研究的连续过程 正如前面所讨论的，针对实施研究的批评集中在缺乏关于实施研究范畴的 定义。其中一部分原因是实施研究不同程度地适用于许多不同领域、并与之相 关。为了理解这一观点，从连续过程的角度看待实施研究是有帮助的。有的研 表 3 实施结局变量 实施结局 实用性定义* 相关术语** 接受度 （acceptability） 利益相关者（例如消费者，服务提供 者，管理者，决策者）认为干预是可以 接受的程度 与接受度相关的因素（例如舒适 度，相对优势，可信度） 采用率 （adoption） 尝试采纳新的干预的意愿、初步决策或 行动 吸收度，利用率，尝试的意愿 适当性 （appropriateness） 一种干预在特定环境中，或对于特定目 标受众（例如提供者或消费者），或与 特定主题的契合度或相关性 相关性，契合度，兼容性，可试 性，适应性，有效性，实用性 可行性 （feasibility） 干预可以在特定环境或组织中进行的程 度 实用性，实际适合度，效用，适合 日常使用 保真度（fidelity） 一项干预措施按原始方案、计划或政策 所设计的实施程度 依从性，按期交付，治疗完整度， 按计划履行的质量、强度或剂量 实施成本 （implementation cost） 履行策略的增量成本（例如，如何在特 定环境中完成服务）。实施的总成本还 包括干预本身的成本。 边际成本*** 覆盖范围 （coverage） 有资格从干预中受益的人群实际接受干 预的程度。 影响范围，可及性，服务传播或有 效覆盖（关注于需要干预的人群并 以足够的质量交付，从而将覆盖面 和保真度相结合），渗透率（一种 干预整合到服务场景中的程度） 可持续性 （sustainability） 干预在特定环境中得以维持或体制化的 程度。 维持，延续，持久性，体制化，常 规化，整合，融入 第三章 实施研究是什么？ - 29 - 究问题涉及大量的实施相关内容（可称为“重实施”），而有的则涉及较少 （可称为“轻实施”）。在图 3中，我们将此想法用流程图的形式进行表示： 越靠近下游的研究问题越是“重实施”的。 因此，在流程图的左侧， 我们发现研究根本不涉及实 施问题（例如齐多夫定作为 预防 HIV母婴传播手段的基 础研究）。在流程图的右侧， 我们发现研究主要关注真实 世界中实施问题（例如，如 何确保低收入国家 HIV检测 阳性的孕妇获得必要的预防性齐多夫定治疗以降低艾滋病病毒传染给婴儿的风 险）。由于这些靠近流程图右侧的研究更关注实施策略和研究实施变量，研究 变得更重实施。 值得注意的是，某些研究虽然不直接包含实施相关内容，但也可能与实施 研究有密切联系。例如，研发冻干天花疫苗意味着疫苗不再需要冷藏保存，这 对根除天花运动的实施具有重大意义。可见，基础研究开发的新产品后续可以 作为改善健康的干预措施进行测试。同样，该领域的发现（例如，口服补液盐 可以现场混合调配，并由非专业人员分发）可能会拓展研究思路（例如，后续 可进一步专注于开发不结块的口服补液盐）。 如果在真实世界中开展的某一研究关注实施，并至少考虑到影响实施因素， 则该研究是重实施的，无论它是否实际测试某一实施策略，或是否评估实施结 局变量。实施研究的焦点关注在背景与环境，以及某一干预措施与真实世界之 间相互作用。这使得实施研究有别于常规监测。相比之下，常规的监测主要测 量评估一段时间内特定领域的进展，而未必会去了解这一进展受到什么因素的 影响。当把监测范围扩展，并使用各种研究方法用于调查常规监测之外的问题 时，这往往是实施研究的起点。如某研究发生在条件变量受控的非真实环境中， 研究样本经过严格筛选、不能代表干预所针对的目标人群，并且实施结局变量 和影响实施的因素与该研究无关、或被控制以消除其影响，则这样的研究可以 说是轻实施的。 第三章 实施研究是什么？ - 30 - 最后，需要注意的是，此流程图是复杂现实的简化，仅用于说明基本概念。 实际上，卫生系统很少一次只单单采用一项新的创新，而经常需要同时考虑多 项创新（以及其他突发事件/改变）。因此，现实中的吸收和适应性改造过程往 往比文中提出的情况更加混乱和复杂。 结论 本章介绍了构成实施研究的一些基本概念，包括可以在传统研究中使用的 实用性定义。本章还介绍了关于实施障碍和问题的类型，实施策略，以及可用 于描述实施方式的各个方面的实施结果变量。在下一章中，我们将探讨实施研 究中协作的作用，以及为什么协作对于理解真实世界中的实施十分重要，并反 映与之密切相关的背景因素。 第三章 实施研究是什么？ - 31 - 图 3 实施研究的连续过程 创 新 实施验证： 在真实世界中如何运作？ 宣传推广： 融入卫生系统和可持续性 与实施无关 研究问题：与实 施无关的基础科 学、产品开发或 探究 背景：受控制或 与实施无关 实施策略和变 量：不相关 与实施有关 但未被考虑 研究问题：易受 实施变量影响， 但未考虑实施 背景：受到严格 变量控制、严格 纳入的人群，影 响实施的因素被 忽略或控制 实施策略：在研 究中未考虑，无 或仅一种类型， 实施变量：可影 响结果，但受控 制或与研究无关 与实施相关 但效果降低 研究问题：次要 问题（例如项目的 平均有效性） 背景：真实世界 中部分控制的干预 实施策略：已被 识别和描述，但仅 使用一种类型且效 果受控 实施变量：假设 相同或不变，或效 果受控（例如调整 为混杂因素） 示例：基础科 学； I 期和 II 期临床试验； 与实施问题无 关的定性研究 （例如对疾病 的理解） 实施研究作为 影响因素 研究问题：同为主 要问题或作为次要 问题（例如：项目 在其所有变化中的 有效性） 背景：真实世界环 境和人群 实施策略：一项或 多项研究 实施变量：可用作 自变量 以实施为重点 研究问题：主要问 题（例如，项目的 各个部分为何和如 何变化？实施策略 有什么影响？） 背景：真实世界环 境和人群 实施策略：可能是 研究重点 实施变量：可能是 主要结局或决定因 素 概念验证： 是否安全并且有效？ 示例：疗效研 究，III 期随机 对 照 临 床 试 验；为了解卫 生服务的使用 程度而开展的 定性研究 示例：实用性 临床试验，干 预和比较领域 的 准 实 验 研 究；以实施为 次要问题的观 察性研究 示例：实施有 效性试验；将 实施变量评估 为次要因素的 观察性研究； 参与式研究 示例：混合方法 和准实验研究， 以确定项目的履 行程度或接受度 的变化；关于某 项目的适应性改 造、学习和推广 的观察性研究 研 究 第四章 哪些人群应该参与实施科学研究？ - 32 - 第四章 哪些人群应该参与实施科学研究？ 本章要点  良好的实施研究是协作性的。当实施者参与研究的识别，设计和实施阶段时， 实施研究通常最有用。  构建政策制定管理者与研究人员间的协作关系至关重要。  实施研究应成为项目的决策过程的一个环节，从而使科学探究能够进到实施 的问题解决过程当中。  实施研究有助于发现一些被忽视的问题，证明卫生保健组织的成效、提高责 任可追溯性。所以实施研究可以在“向权力说真话”方面发挥重要作用。  熟悉实施的环境与体制，认识到适宜的实施研究方案往往是是灵活可变的， 其重要性并不亚于坚持固定的研究方案设计。 哪些人群应该参与实施科学研究？ “成功的实施研究始于成功的协作。” 正如前一章所强调的那样， 实施研究定义之一是，它试图 探讨真实世界情景下的实施情 况。与其他形式的研究不同， 它并不尝试过滤掉外来的或偶 然的因素。事实上，在许多方 面，正是这些被忽视因素才是 实施研究者的兴趣所在。实施研究人员与一般研究者不同，他们愿意接受、拥 抱这些存疑的、不可预测的因素。而其他研究人员可能不会这样做。它还意味 着在所有复杂性情况下和自然环境中观察研究对象。这意味着与实际受干预影 响的人群合作，而不是根据苛刻的纳入标准选择研究对象。（例如，当患有合 并症的人也是干预的目标时，研究者不应只选择了患单一疾病的志愿受试者）。 实施研究关注的背景因素包括社会、文化、经济、政治、法律和自然环境， 包括一般人口统计和流行病学状况。它还可以包括社会制度环境，现有的特定 第四章 哪些人群应该参与实施科学研究？ - 33 - 体制结构，以及在其中工作的各种利益相关者。卫生系统的结构以及政府，私 营机构和非政府组织的作用也是应该关注的。 实施研究的另一个关键考虑因素是研究所针对的受众，或与之相关的研究 对象。具体而言，实施研究人员需要牢记目标受众（那些将要使用实施研究结 果的人或组织）的需求或局限性。正如前一章所讨论的，实施研究涉及广泛的 主题，涉及的人员也同样广泛（例如，致力于质量改进策略的卫生管理者和团 队，期望改进政策或引入新政策的政策制定者，以及需要使用基于现有最佳证 据的干预措施的卫生从业者）。这些潜在受众中的每一个都有不同的要求和不 同的观点。为了使实施研究发挥最佳效用，我们需要考虑到这些差异。通常情 况下，本研究的受众不是另一位研究者或学者，而是非专业人士。他们需要清 晰的、基于证据的分析，且不受专业术语的干扰，用以构成未来决策的基础。 尽管有一些例外情况，我们将在下文中讨论，但如果实施者在确定、设计 和实施所进行的研究中发挥了作用，而不仅仅是被动接受结果，那么实施研究 很可能是有用的。如上所述，无论是在卫生保健工作第一线实施具体方案的人， 还是在卫生系统工作的人，都对信息收集工作作出了巨大贡献（补充资料 4）。 例如，常规监测往往是许多实施研究问题的起点，并依赖于对当前行政和卫生 数据的常规收集和分析。鼓励来自实地的反馈有多种方式，其中之一就是所谓 的参与式行动研究（Participatory Action Research，PAR），顾名思义，这涉及 到研究对象的参与，并被描述为“由被研究者发起和参与的研究”[28]。尽管 参与式行动研究并非专门设计来回答实施问题，但实施是参与式研究的一个主 题。参与式行动研究还为那些本没有发言权的社区创造了“向权力说真话”的 机会，例如，针对剥削性或粗暴的服务提供1说明。我们将在第五章还将回到参 与式行动研究上，进一步详细探讨它和其他研究方法。 研究者与实施者的配合，在质量改进和推广某一项目的过程中，起到重要 的作用。前面的章节中，我们提到过，推广某一项目的实施规模有多种形式。 它可能不仅涉及服务规模的扩大，而且还涉及到负责实施的单位或组织需要提 升的组织协调能力，加大财政投入，和政治能力。项目的推广同时也是实施单 位或组织、受益人群、资助方和政府官员的一个共同参与的学习过程 [29]。 1 原文为“exploitive or abusive service provision”，结合上下文，译者理解原文作者旨在表达在某些地方，患 者可能遭遇到医疗卫生工作者不合理的粗暴对待，而参与式行动研究有助于为这些群体提供发声的机会。 第四章 哪些人群应该参与实施科学研究？ - 34 - 上述的过程必然需要多方面的努力和研究。因此，这种研究最好通过强有 力的合作方式进行。最重要的是，无论研究对象是谁，研究人员和实施者都必 须认识到，协作方实际上是互利共生的——在这种关系中，实施者从第一线产 生反馈，而研究人员提供专业的研究方法，使得研究更为可靠。只有这样聚集 在一起，利益相关者才能确保所产生的知识是有效的，并与需求保持一致。 协作关系在实施研究中的重要性 鉴于协作在实施研究中的重要性，建立和维持合作关系所需的技能是一个 关键的考虑因素。这包括促进参与政策制定、项目管理和研究的主要利益相关 者之间的合作关系。上文中加纳的案例是协作可发挥作用的一个有趣例子。在 20世纪 90年代后期，研究人员和实施者共同商讨制定国家健康保险计划的最 佳方法（补充资料 1）[7]。从形成性研究（针对某项特定干预措施的目标社区 而进行设计的研究）开始进行商讨，研究人员和实施者密切合作，共同确定研 究设计，以确保研究有能力在紧迫的时间内，提供可靠的证据。加纳的经验不 仅彰显了协作的价值，也显示了循序渐进、逐步迭代改进的方法对实施研究十 分重要。 补充资料 4 实施研究中实施者的重要性 鉴于实施研究问题的性质，显然，实施研究人员在实施研究中扮演着至关重 要的角色。他们应该在整个研究过程中发挥积极作用，而不仅是参与者或被动合 作者。事实上，在一些具体研究中——如质量改进（quality improvement，QI）研 究——实施者是主导此类研究最适合的人选，负责研究周期的几乎各个方面。研 究策划阶段，实施者最适合明确实施所遇到的障碍并提出实施研究的问题，以及 确定可供测试的解决方案。研究设计阶段，研究者通过促进对影响实施的背景因 素的理解，可使得研究问题框架能够反映现有情况。在数据收集阶段，实施者在 获得数据源、开展调查访问方面显然具有优势。在数据分析阶段（特别是对于定 性数据），实施者可在理解和解释数据方面发挥重要作用。最后，在传播、扩散 研究结果方面，实施者发挥着关键作用，尤其是将研究总结出的经验教训整合到 项目的实践中。 第四章 哪些人群应该参与实施科学研究？ - 35 - 实施研究的嵌入 改善协作、鼓励在实施 研究方面建立伙伴关系的方 法是将实施研究嵌入到政策 项目的决策过程中。实施研 究往往产生于既定的项目， 并对项目有直接的帮助。所 以，应当从项目设计之初就将实施研究嵌入进来，成为整个设计与实施进程的 一个部分，而不是将实施研究当做一个后续的额外工作（为项目的开展提供实 施背景信息）。如果实施研究从一开始就纳入到 2004年加纳防虫蚊帐试点项目， 那么加纳和坦桑尼亚联合共和国在防虫蚊帐生产和分发方面将不会有如此大的 差异——试点项目的设计者可以更早地做出必要调整，从而节省时间和重要资 源[6]。为了有效地将实施研究融入于实施相关的决策过程，仅在初期与实施研 究人员进行沟通是不够的。实施研究还需要嵌入整体设计、规划和决策工作中。 我们可以通过以下三种方式来嵌入实施研究：第一，整合研究和方案活动的资 金；第二，系统地将研究和科学调查作为问题解决过程的一部分；第三，在研 究和实施过程中采用联合决策的机制。 ► 将资金投入到研究和项目过程中 开展实施研究的主要驱动力来自各种研究资助。实施研究经费来源一般与 项目经费的来源渠道不同。国际资助者和各国政府都将预算分为项目预算和科 研预算。这种割离导致研究经费周期并不总是与项目需求相一致。同时，大量 的研究项目资金是竞争性的，且往往是在个别研究者的倡议下进行的。相比之 下，项目经费通常不具有竞争性，可能需要成立小组或项目管理人员去识别有 意义的问题。这也造成项目需求与研究目标之间的不匹配。研究人员难以将研 究工作与项目实施过程中遇到的实际障碍和挑战联系起来。此外，发出研究建 议、选择研究方案的程序和时间表，往往超出了项目负责人就实施问题做出重 要决定的时间。 如果将实施研究的资金需要纳入项目预算，就有更多的机会促使实施研究 合乎项目的需要。下面是布隆伯格慈善基金（Bloomberg Philanthropies）在两个 第四章 哪些人群应该参与实施科学研究？ - 36 - 关于预防烟草使用和道路安全的大型全球倡议项目中采取的做法。在这两个倡 议项目中，布隆伯格慈善基金为一个合作联盟提供资金。合作联盟包括民间团 体、学术研究人员和技术机构，以合作实施大规模公共卫生干预。每个联盟合 作伙伴都发挥作用，并对实施的不同方面做出贡献。就烟草倡议而言，重点是 将控烟立法和控烟成本，以及将政策内涵和宣传工作相结合[30]。 同样，对于道路安全的十国倡议，研究人员正在与实施伙伴合作，支持监 控工作，监测和评估进展，深入分析利益关相方和当地社区对于将要实施的干 预措施的看法[31]。这种协作方式不仅认可了不同合作伙伴在实施过程中发挥 的个人作用，还使研究问题与项目需求相一致，同时促进多个利益相关方参与 研究过程，从而确保所产生的证据用于指导实施。 ► 科学调查与研究在项目过程中的系统应用 除了将研究和项目基金结合起来之外，至关重要的是，在项目决策中把系 统应用科学研究制度化，使实施研究成为解决问题过程的核心部分。这可以通 过多种方式实现。首先，制定明确涉及研究的实施和推广的决策方案和/或程序， 是确保以系统的方式解决需要调查的问题和疑问的好方法。WHO/扩大网络 （ExpandNet）框架是如何将科学调查与研究纳入进程的一个有用例子。该框架 包括需要回答的问题（有时通过实施研究），作为实施者在推广方案时需要考 虑的九个步骤的一部分 [16]。将实施研究问题整合到扩大框架中，有效地将研 究嵌入到与干预措施推广普及相关的决策中。WHO/扩大网络框架还建议采用 多学科方法，使用不同的参与者来指导推广普及过程，以便在所有阶段（包括 实施研究期间）建立伙伴关系和开展合作 [16]。研究和科学调查可以纳入方案 决策的另一种方式是通过强制性监测和评估，其价值作为实施研究的基础已经 被注意到[32]。我们也需要在监测和实施计划之间建立联系，以确保定期识别 问题和挑战，其中一些问题和挑战可通过实施研究加以解决。对方案进行必要 的评价也有助于长期学习和加强可问责性。它还有助于系统化理解实施挑战， 并确保实施研究与方案的相关性。一个体现这种嵌入式研究价值的有趣例子是， 墨西哥要求对所有公共资助的社会项目进行系统评估与研究的立法。这有助于 使政策制定去政治化，同时增加这些研究证据对实施者、政策制定者的影响 [32,33]。 第四章 哪些人群应该参与实施科学研究？ - 37 - ► 决策中的责任分担 嵌入式实施研究的最后一个方面也 许是最重要的，也是最难以实现的。项 目的执行和普及推广都需要一定程度的 灵活性，而且往往是在时间压力下制定 的。因此，为支持这些活动而进行的任 何研究都需要面对这些现实层面的要求。 这意味着关于研究设计、方法和结果的决策不仅需要从研究人员的角度进行告 知，还必须反映实施者和其他利益相关者的观点。例如，研究人员可能希望证 据是基于概率的，并且可以通过统计分析得到证实。而时间方面的压力可能迫 使研究设计主要为了生成效果证据。例如，某一实施策略是否使得政策在人群 中的覆盖水平达到某一水平，或者服务质量是否达到了一定的标准（这些有时 被称为充分性说明），而不一定会考虑这样的变化是否由其他外部影响导致 [4,34]。同样，在许多情况下，作为实施研究主题的问题需要由研究人员和决策 者共同确定，以反映其不同的观点。如前所述，实施者和研究人员往往从略有 不同的角度提出问题。实施者专注于实施面临的具体障碍和挑战，研究人员寻 找方法来制定适合研究并能通过研究得到解答的问题。例如，在马来西亚使用 可见性增强材料提升摩托车安全的研究中，双方的研究关注点有着明显差异 （补充资料 5）。 研究者和决策者在多个层面上合作，提出研究问题和研究设计，可以收集 决策者所需要的证据（关于在地区层面上推广的可行性）。采用准实验方法， 可确保在研究者看来，研究是严谨、独立而客观的 [35]。马来西亚的例子证明， 只要愿意互相让步，不同的研究关注点是可以得到满足的。但这样的让步并达 成共识的案例不应被视为典型案例。实际上，共同承担决策的责任并不总是容 易的，每个决定也未必会得到各方的同意。然而，它确实表明，可以从多个角 度为决定提供信息，并且可以适当考虑不同行为者的专门知识和见解。 第四章 哪些人群应该参与实施科学研究？ - 38 - 补充资料 5 政策制定者和研究者共同关注在马来西亚的交通意外伤害 尽管知识和行动之间有明确的联系，但信息产生方与预期的信息使用方之间 的相互作用往往让人无法估计，特别是在中低收入国家。一项针对为减少马来西 亚摩托车交通意外伤亡的预防措施的试验就是这种合作如何发挥作用的一个很好 的例子。 2005年 7月，交通部内正式设立道路安全部。道路安全部和其他利益相关者 之间开会讨论，评估政策制定者是否有兴趣从事关于反光材料的新研究。摩托车 驾驶员如使用反光材料，可能可提高其能见度，从而有助于避免交通事故的发 生。 重要的是，研究人员和政策制定者联合起来，就双方共同的目标以及如何实 现这些目标达成了一致。政策制定者最初对反光材料的研究持怀疑态度，认为研 究的开展太费时间，而且他们有能力解决可见性增强材料（visibility enhancement materials，VEMs）在整个地区广泛推广使用时所带来的主要困难。与此同时，研 究小组则设计了一种复杂的、持续时间更长的研究设计，以便获得高度的可推广 性。 通过利益相关者磋商、讨论和妥协，最终同意使用类实验的现场试验，以使 得可见性增强材料在整个地区得到推广（在地区内将骑摩托车驾驶员分为接受与 不接受干预，显然不合理也难以操作）。这一设计还将能够利用日常收集的来自 警察和医院的数据进行分析，从而减少基线评估所需的时间。此外，它还加强了 研究和方案活动之间的联系。 现场试验对政策制定者特别有吸引力，因为它快速、直接地提供可靠的研究 结果。另一方面，现场试验既了解了在马来西亚当地条件下如何推广这种干预， 同时也了解了在试验研究结束后如何可持续地推行这项干预。此外，可能出现一 种不尽人意的情形——这项公开研究可能证明 VEM材料是无用的。但即使是这 样，这笔研究开支至少让研究者有依据去说服道路安全部不继续推广这项干预， 从而避免了在国家层面上推广 VEM材料的浪费。毕竟现场试验的花费比在全国 开展一项没有实际效果的干预要少得多。最终，为了提高人们对这一问题的认 识，该试验得以公开启动。该项目被冠以“看得见的摩托车安全”作为推广活动 口号。 资料来源：Tran等，2009 [35] 第四章 哪些人群应该参与实施科学研究？ - 39 - 伙伴关系带来的挑战 在这个嵌入式实施研究的简短介绍中，读者也应该清楚地看到，尽管协作 方法提供了许多机会，但它也带来了挑战。这在一定程度上反映了卫生系统的 复杂性，其中各方面参与者的多样性，以及他们相互作用的方式。谈论关键参 与者的基本作用很常见。例如，将他们区分为决策者、项目管理者、一线卫生 工作者和被服务的患者。但是务必认识到，每一个参与者的作用都可以在不同 的层面上得到发挥。而且，参与者之间的界限有时是模糊的。 因此，议程制定、政策和项目起草，最初可能由国家层面的卫生部门发起。 但当核心决定被下达、执行时，解决问题或加强进程的举措可能导致改变原计 划的实施安排和原本的核心决定。因此，虽然一线研究人员往往主要负责执行 已经商定的决定和政策，但一些本地的影响因素可导致他们最终也可能成为决 策议程和相关项目的制定者。从这个角度来看，在各级卫生系统运作中，实施 者在实施研究努力中发挥着至关重要的作用——有时甚至起主导作用（特别是 在质量改进研究中）。加纳、乌干达和坦桑尼亚联合共和国的地区卫生管理团 队提供了实施者参与研究过程的一个很好的例子，他们在 PERFORM研究组的 支持下，使用“行动研究”（Action Research，包括：计划、行动、观察和反 思）方法开发并测试了在特定环境下以提高工作成效为重点的人力管理强化过 程（补充资料 6）[36]。 显然，信息流和思想形成往往是由反馈环路促成的，而不是一个严格的线 性过程。此外，由于实施过程在实施研究中的重要性，对实施研究背景和系统 的理解，以及创造性地、灵活地确定适当的研究方法，可能与坚持特定学科视 角所告知的固定研究设计一样重要，甚至更重要。例如，临床随机对照试验依 赖于一种固定的、可重复的干预措施（如服用一种药丸）。而实施可能涉及的 干预措施变化频繁且不严格可重复。像随机对照试验这样的固定研究设计，虽 然适用于疗效和有效性的研究，但可能不适合回答与实施相关的问题。合作中 固有的另一个挑战是，不同参与者所关注的问题往往是不同的。例如，研究人 员可能面临在高影响力期刊上发表论文的压力（这些期刊通常倾向于特定的学 科方法），而实施者可能面临在尽可能短的时间内解决问题的压力。 第四章 哪些人群应该参与实施科学研究？ - 40 - 实施研究存在的反对声音 虽然刚才描述的嵌入式实施研究是有益的，但在某些情况下，根本不可能 在公共政策进程中嵌入实施研究。决策者、管理人员和供资机构并不总是想知 道他们的方案是如何执行的，当然，除非能够证明他们做得很好。 他们可能在一项政策上投入了相当多的政治和财政资本，害怕没有产生预 期的结果或缺乏管理的资源。资助者往往不愿资助一些关注可持续性的、或可 能出现非预期阴性结果的研究。如因雇用人员从事单一用途的项目而引起的人 力资源分配问题——这是艾滋病项目经常遇到的一个问题[27]。同样，少数群 体所关注的问题可能不会受到当权者的重视，特别是一些关于社会或政治方面 的敏感问题。出现这类问题的领域包括与男男性行为者有关问题、土著群体的 待遇、提供堕胎服务，以及其他普遍的危险做法，如阉割女性生殖器官等。与 处境不利的群体或民间社会组织合作的实施研究人员可能发现自己无法与反对 他们的人合作。尤其是当在一个正在遭受内部冲突的地区进行研究时。在某些 情况下，参与式行动研究甚至被认为是对现有权力结构的革命性研究。在这种 情况下，实施研究者的一个重要工作是，设法使他们的研究尽快进入影响政策 制定的过程中。这也可能需要更多基于精心研究的宣传策略。 结论 本章试图找出与实施研究相关的核心问题，并着重介绍了在真实世界环境 中进行实施研究、关注背景和目标受众需求的重要性。它还强调了协作的重要 性，以及实施者、研究人员和其他利益相关者之间建立伙伴关系的必要性。在 下一章中，我们将关注实施研究中常用方法背后的原则，并关注研究问题如何 为研究方法的选定提供指引。同时，复杂性往往是实施研究的一个典型特征， 所以我们还将研究如何设计方法来解决复杂性问题。 第四章 哪些人群应该参与实施科学研究？ - 41 - 补充资料 6 在非洲的区域保健小组利用实施研究开展人力资源能力建设 缺乏有能力、有积极性的卫生工作者是非洲改善健康和拯救生命的主要障 碍。为了解决这一缺陷，需要采取双管齐下的策略，确保培训新的卫生人员，改 善现有医疗人员的业绩表现。一些复杂的因素影响着工作人员的业绩，包括医疗 工作人员的水平和分布、工作安排和所需资源、工作条件和报酬。了解这些因素 的性质，并对所面临的挑战作出适当的反应，不仅可以提高现有医疗人员劳动力 的业绩，而且可以减少工作人员的损失。PERFORM是目前在加纳、乌干达和坦 桑尼亚联合共和国开展工作的一个研究小组，其工作重点是建立地方能力和自主 权，以支持人力资源管理和相关卫生系统活动。具体来说，PERFORM支持地区 卫生管理团队（district health management teams，下文简称 DHMT），他们通过 一系列研讨会和审查会议，利用所谓的行动研究（计划、行动、观察和反思）开 发并测试针对具体情况的管理强化过程，重点是提高员工绩效。DHMT确定了需 要改进的卫生人力绩效领域，实施现有背景下可行的综合人力资源和卫生系统策 略，以改善卫生人力绩效，并监测策略的实施，评价流程及其对卫生人力绩效和 更广泛卫生系统的影响。然后，DHMT计划自己的人力资源和卫生系统策略“捆 绑”，采取行动实施这些策略，并观察其制定的策略对卫生人力绩效的影响。然 后，他们反思他们的计划是如何实现的，如有必要，修订计划或应对新的挑战， 从而开始将这一进程纳入他们的地区。这项正在进行的工作的主要产出将包括制 定和测试地区一级的干预措施，以改善三个非洲国家卫生工作人员的业绩和更普 遍地管理分散的卫生系统。这些干预措施可在加纳、乌干达和坦桑尼亚联合共和 国的其他地区以及卫生系统分散的其他国家推广。这项研究的结果将有助于了解 如何加强撒哈拉以南非洲的管理，以及如何提升员工绩效、如何在更广的范围内 完善卫生系统。此外，行动研究方法将有助于提升参与管理人员的技能和能力， 以解决未来的其他管理问题，如果证明这种方法是成功的，则有可能推广这种方 法的规模。 资料来源：PERFORM研究组，2011 [36] 第五章 哪些方式和方法适用于实施研究？ - 42 - 第五章 哪些方式和方法适用于实施研究？ 本章要点  实施研究与所有研究一样，也应遵循两大原则：研究结果应当是有根据的， 研究方法应当是透明的。  由于实施研究借鉴了各种各样的定性、定量和混合方法的研究方法，因此狭 义地讨论“实施研究方法”是没有意义的。  在实施研究中，“问题为王”。由问题决定所用的研究方法，而不是由方法 决定可以提什么样的问题。  因为真实世界具有复杂性，所以提出的问题往往是复杂的。由于各种背景因 素影响着科学研究的开展，并产生不可预测的效果，因此实施者需要不断进 行调整。 哪些方式和方法适用于实施研究？ “问题为王。” 在讨论实施研究的不同方法 时，应牢记研究的基本目的。正 如我们已经讨论过的，实施研究 的目标是了解临床和公共卫生政 策、项目和实践为何在真实世界 环境中没有产生预期中的效果， 并了解如何使其更好地发挥应有 的作用。具体而言，实施研究可 用于：在合适的地方借鉴过去的 经验，进而在真实世界中评估变化；理解复杂的现象；产生或测试新的想法； 帮助预测未来可能发生的事情，并将其作为特定创新或变化的结果。它还在告 知利益相关者方面发挥重要作用，从而增进理解、提高透明度和加强可问责性 [37]。最后，实施研究的目标是要有所作为，提高政策、方案和服务的效果、 质量、效率和公平性。 第五章 哪些方式和方法适用于实施研究？ - 43 - 在讨论可用于实现这些目标的一些可能的研究方法之前，需要注意的是， 尽管实施研究在某些方面可能与其他研究的形式不同，但如所有研究一样，它 同样遵循两大原则。首先，研究结果应该是有根据的，也就是说有足够的证据 支持。二是其研究方法要透明，也就是要足够明确地让其他人能够判断过程是 否充分，进而证明所得出的结论是正确的，并且可以重复[38]。无论采用哪种 方法，都必须铭记这些原则。 由于实施研究借鉴了各种各样的定性、定量和混合方法的研究方法，因此 狭义地讨论“实施研究方法”是没有意义的。但有一些研究方法对实施研究人 员特别有用，因为它们天然就是实用的，而且可以为研究人员的后续工作提供 可参考的信息。同时，这些方法能够很好地适应实施背景、环境中的微妙之处， 尤其是随着时间变化的背景因素，并为了适应这些变化提供灵活可变、可供反 复多次改进的研究方案。下面简要描述这些研究方法的选择。 实用性试验 卫生干预措施的试验通常被 描述为解释性的或实用性的。该 术语最初是为了将以帮助选择治 疗方案为目的的试验与以检验潜 在因果关系假设为目的的试验区 分开。因此，解释性试验通常寻 求认识并解释在可控条件下进行 干预所产生的益处，通常在临床 研究中心中纳入经严格筛选的受 试者。而实用性试验则侧重于常 规实践中干预的效果。与解释性试验相比，实用性试验力求最大限度地扩大干 预实施方式的变异性（如，场景、环境、服务、患者类型），并希望把研究结 果最大化地应用到其他场景中[39]。通过这种方式，实用性试验可以提供有力 的证据，证明实施策略在“真实世界”条件下的效果。 实用性试验通常包括一个广泛的设计期，由实施者和决策者参与设计干预 措施。这个阶段，有时可能会产生一种错误的自信心，认为设计是稳健的，是 第五章 哪些方式和方法适用于实施研究？ - 44 - 适合实施环境的。除非在试验中加入额外的研究方法，否则在“真实世界”中 发生的多种变化，如实施策略的变化，实施结果变量的变化，或背景因素中的 其他非随机变化等，可能不会被认识到。理想情况下，在实用性试验中，干预 措施的设计和研究结果的设计都应该由参与者、资助者和决定干预措施的从业 者以及直接受结果影响的从业者共同合作制定。在中低收入国家环境下实用主 义试验的价值是有据可查的，一个很好的例子是南非研究人员最近进行的一项 研究（补充资料 7）[40]。 效果-实施混合试验 效果-实施混合试验将有效性和实施研究结合在一起，以评估健康干预措施 的效果和实施策略。虽然实用性试验并不试图控制或确保服务的提供以达到正 常实践环境中的现实标准，但效果-实施混合试验也会在实际发生时进行干预和 观察实施过程，例如通过评估实施结果变量[24]。 最近的一篇论文提出了效果-实施混合研究设计的三种基本类型，主要基于 研究目标中的效果或实施部分的优先地位[24]。 Ⅰ型实施研究设计：设计测试健康干预措施对相关结果的影响，同时观察 补充资料 7 南非的一项实用性试验 改善中低收入国家获得抗逆转录病毒疗法的最大障碍之一是缺乏管理该疗法所 需的训练有素的医务人员。在南非，医生的短缺往往会限制患者接受治疗的机会， 南非开普敦大学肺研究所知识转化部门的研究人员利用实用性试验证明，非医生的 卫生工作者能够满足护理需求。具体而言，该实用性试验的重点是“理顺角色与工 作的关系，以推广艾滋病毒的治疗和护理方案”（Streamlining Tasks and Roles to Expand Treatment and Care for HIV，缩写为“STRETCH项目”）。该方案为护士提 供外部教育培训，以启动和重新开展抗逆转录病毒疗法，并下放护理权力。31家 初级保健诊所被随机分配到护士管理组或常规的“标准”护理组。该研究对护士管 理组项目中的 8000多名患者和常规标准护理组中的 7000名患者进行了一年半的跟 踪随访，发现两组患者在死亡率、病毒抑制率和其他护理质量指标上没有差异，甚 至在某些项目中护士管理组的护理质量更高。 资料来源：Fairall等，2012 [40] 第五章 哪些方式和方法适用于实施研究？ - 45 - 和收集关于实施情况的信息。在此类研究中，测量了患者接受健康干预措施后 的功能或症状，同时通过定性、过程导向或混合方法评价了实施方法的可行性 和接受度。 Ⅱ型实施研究设计：设计涉及对健康干预措施和实施策略的双重测试。 Ⅲ型实施研究设计：设计测试实施策略，同时观察和收集关于健康干预措 施对相关结果的影响信息。Ⅲ型实施研究设计主要使用健康干预措施的采用率 和保真度来测试实施策略。 效果-实施混合试验带来了 许多益处。它不是采取循序渐进 的方法来解决问题，而是从随机 临床试验开始，以确定在可控制 条件下实施的干预是否有效，然 后再继续进行如整群随机对照试验等设计，以确定在真实世界环境中引入干预 的最佳方式。因此效果-实施混合试验方法可以使研究人员能够同时评估在真实 世界环境中引入的干预措施的影响以及用于实施这些干预措施的实施策略。这 样的设计不仅加快了原本可能非常耗时的过程，而且也使研究人员能够识别重 要的干预-实施交互作用，然后可以利用这些信息来为有关最佳实施方法的决策 提供信息。孟加拉国锡尔赫特新生儿护理包试验便提供了一个很好的例子，说 明效果-实施混合试验适用于不同的目的（详见补充资料 8）[41-44]。 质量改进研究 正如萨尔瓦多共和国子宫颈癌筛查的例子（表 1）所强调的，卫生保健质 量改进研究存在个主要挑战：首先，质量评估的本质上是依赖于其所处的背景、 环境；其次，质量是一个永无止境的目标（有的专家将质量比喻为“移动 靶”）——质量改进措施需要根据反馈结果不断调整；第三，质量研究通常涉 及到复杂、多组成部分的干预。这些都会对研究设计产生影响 [17]。 第五章 哪些方式和方法适用于实施研究？ - 46 - 为了反映质量研究的“移动靶”特性，研究通常涉及一组结构化的周期性 过程，由一种被称为计划-执行-研究-改进循环（又称“戴明环”，下文简称 PDSA循环）或它的变体[45]。PDSA循环允许持续应用科学方法制定假设或计 划，以提高质量、实施计划、分析和结果解释，然后生成下一步计划。对可用 于 PDSA循环研究的质量改进工具的范围的详细描述超出了本指南的范围，尽 管图 4 展示了根据该循环的各个阶段所列出的示例[46]。另一个有启发性的例 子是在萨尔瓦多共和国进行的与宫颈癌筛查相关的 PDSA循环研究工作，将在 第 2章中讨论。 补充资料 8 孟加拉国新生儿护理研究中应用的效果-实施混合试验研究 孟加拉国的一项整群随机对照试验旨在检验家庭护理和社区卫生工作者及 社区护理干预，并与常规新生儿护理作比较，以获取这个试验所需的全部定性 和定量数据，研究人员使用了一系列的研究方法，包括： •定量家庭调查研究，提供对现有新生儿死亡率和接生熟练程度的评估，从 而激发干预的需要，并提供基线水平。 •形成性定性研究，过去用于探索使新生儿面临死亡风险的家庭实践以及安 全分娩和产后护理的障碍，然后将其作为开发家庭新生儿护理包（“充分参 与”）的基础，并设计定量研究工具（“工具有效性”）； •对新生儿护理的观察表明，社区保健工作者能够诊断新生儿疾病； •采用家庭调查和深入访谈的形式展示干预措施正在按计划进行实施，同时 还采用调查、观察和深入访谈的形式证明新的新生儿做法是否正在实际实施 （“实施保真度”）； •使用终期家庭调查来评估新生儿死亡率和服务覆盖水平，而使用定性研究 来详细解释分娩和产后的做法为何和如何发生变化，这主要是因为当地社区参 与了该方案及社区卫生工作者的支持性监督（“含义增强”）。 资料来源：Baqui等，2008 [41]；Baqui等，2009 [42]；Choi等，2010 [43]；Shah等， 2010 [44] 第五章 哪些方式和方法适用于实施研究？ - 47 - 通常情况下，PDSA质量改进研究评估某种正在研究的干预措施（通常是 流程的某个改变）是否会显著改善结局，然后周期性、迭代地利用这些结果改 进干预措施。因此 PDSA干预通常涉及随时间推移的重复检测。PDSA研究通 常被称为准实验，因为实验者缺乏对研究的完全控制，特别是将干预随机分配 给特定受试者的能力[47]。对照组在大范围的准实验研究设计中多达 16 种不同 类型，并在内部效度（具体研究中的干预是否确实产生了差异）和外部效度 （询问哪些人群、背景、治疗和结局变量可以概括观察到的效应）方面注意到 它们的不同优势和劣势[47]。 PDSA研究设计通常用于卫生系统中的质量研究，在单个病例或对照组中， 图 4 计划-执行-研究-改进（PDSA）循环和其中各阶段可使用的研究工具 注：PDSA循环有许多不同的名称，包括“计划-执行-检查 -改进循环（PDCA循环）”，即 “休哈特循环”。休哈特博士是 20世纪 30年代在工业流程改进中开创性使用统计学“质量 控制”的人。此后，爱德华·戴明将现代质量管理方法、PDCA循环和改良而成的 PDSA循 环（即以其名字命名的“戴明环”）进行推广。 改编自: Brassard等，1994 [46] 执行 先试验性地开展小规模改进 •小组领导技能 •试验设计 •矛盾解决 •在职培训 研究 分析观察 改进措施是否取得 预期效果，同时分析 一些特定的流程 •数据检查表 •图解分析 •控制图 •KPI 绩效考核 计划为改进措施制定计划 •回执客户/供应商流程图 •Pareto 优选排序分析 •头脑风暴 •名义群体法 /故障树分析法 •因果图 改进 从改变 中获得最大利益 •流程图 •流程标准化 •依据参照信息进行 标准流程的正式培训 第五章 哪些方式和方法适用于实施研究？ - 48 - 提供对随时间推移重复和定期测量的反应的评估 [48]。 研究设计通常包括：①时间序列研究（对基线与干预后结果的水平和趋势 进行分析）；②多时间序列研究（其中基线和干预在不同时间重复进行）。时 间序列研究里，干预在不同分组与时间中是滞后的，且析因分析研究中干预被 随机分组以比较时间序列。 这些实验设计的数据可能来自常规健康管理信息，也可能来自专门测量关 注结果的专项调查。 对于使用高质量卫生信息系统和卫生设施调查的标准化指南，读者可以访 问林德罗（Lindelow）和瓦格斯塔夫（Wagstaff）关于卫生设施性能评估中出现 的数据和测量问题的报告[49]。关于如何设计和报告涉及质量研究干预的研究 的一般建议，读者应参考《质量改进卓越报告标准》（SQUIRE）指南 [45]。 参与式行动研究 所有以人作为受试者的研究均有人类参与，但参与式行动研究的特殊性在 于，它将研究过程的权力和控制权赋予了受试者。因此参与式行动研究指的是 当地居民不仅充当受试者，也可成为研究的发起者或参与者的一系列研究方法 [28]。 表 4通过“由点到面”的方式，列出了由当地条件决定的优先关注事项和 观点[28]。尽管大部分参与式行动研究方法涉及定性技术，但越来越多地使用 定量和混合方法技术，如在参与式农村评估或参与式统计方面的例子[50，51]。 有开展和报告参与式行动研究的准则，虽然重点往往是参与者的地方行动，而 不是外部研究人员与他们合作的方式[52，53]。 近几十年来，中低收入国家采取了一些基于参与式行动研究的举措，一个 值得注意的例子是印度非政府组织的一个有助于妇女团体改善印度贾坎德邦和 奥里萨邦部落地区的产妇和新生儿健康的工作（补充资料 9）[54]。 现实性综述 现实性综述（Realist Review）的目的是使决策者能够更深入地了解干预措 施，以及如何在不同背景下最大限度地发挥其潜力。这种方法对处理政策和方 案干预措施方面非常有帮助，因为在这些干预措施中，实施的复杂性和差异性 往往是重要影响因素[55]。 第五章 哪些方式和方法适用于实施研究？ - 49 - 对于处理复杂的社会干预或项目，现实性综述是以特殊性为基础的，提供 了对谁工作、在什么情况下、在什么方面、以及如何工作的解释性分析 [55]。 换句话说，很明显，这种解释性分析的整合需要进行某种形式的归纳。而在现 实性综述中，这种归纳来自对基本假设的识别（这是一个迭代过程）。因此， 现实性综述的第一步是围绕干预性质凝练问题，发现并明确所研究干预的基本 假设，并揭示它在理论上能起什么作用。 进行这种关于干预性质的分析，我们需要评估基本理论的完整性，比较对 立的理论，以及在不同的环境中评估相同的理论。然后，现实性综述在文献中 寻求支持、反驳或修正基本假设的经验证据，将理论理解和经验证据结合起来， 同时将重点放在干预的应用背景、作用机制和产生的结果之间的关系上。 表 4 参与式行动研究与传统研究的比较 参与式行动研究 传统研究 研究目的是什么？ 行动 了解可能的后续行动 研究对象是谁？ 当地居民 机构、个人和专业兴趣者 谁的知识最重要？ 当地居民的 科学家的 谁来选择主题？ 当地优先 资助机构、行政机构日程安排、 专业兴趣 方法学的选择原因是什么？ 增强能力和学习 遵循惯例、“客观性”、“真实 性” 谁参与各研究阶段？ 问题识别 当地居民 研究者 数据收集 当地居民 研究者，数据收集者 干预措施 当地的概念和框架 学科理论和框架 分析 当地居民 研究者 结果展示 本地可访问和有用的 由研究者转给学术和资助机构 针对调查结果采取的行动 过程的一部分 通常是分离的或可能不会发生 谁采取行动？ 当地居民，有或没有外部支持 外部机构 结果归谁所有？ 共享的 研究者或投资者 强调过程或结果？ 过程 结果 资料来源：Cornwall and Jewkes，1995 [28] 第五章 哪些方式和方法适用于实施研究？ - 50 - 检索和回顾公开发表文献后，对所收集文献进行审查跟踪。当文献回顾接 近完成，则进行最终检索。然后对每篇文献的相关性和方法学严谨性进行评估， 并根据文献综述的关键问题对纳入文献进行归纳。这可能包括以下问题，例如： 项目哪些部分正在运作？哪些部分没有运作？为谁开展？为何开展？在什么情 况下开展？最后，通过更好的方式分享综述结果，以便综述作者和读者之间建 立更紧密的联系，并促使其成为政策对话的一部分[56]。 蒂耶勒曼（Dieleman）和他的同事们在 2011年的 WHO报告发表了以《农 村和偏远地区卫生工作者保留研究的综合与现实性综述》为主题的报告，展示 了如何将这种类型的综述应用于理解背景因素和支撑众多保留策略的主要机制 [57]。 这个报告采用了以“背景-机制-结局”为中心的分析框架，描述了干预措 施（在此为工人的保留策略）如何与特定情境（在此为中低收入国家的农村和 偏远地区）相互作用，以及如何产生特定结局。通过这种方式，支持干预的基 础理论得到了阐明，并且这些理论在已实施干预的场景中得到了检验[57]。 补充资料 9 改善新生儿健康的参与式行动 以社区为基础的改善孕产妇和新生儿健康方案是否能够成功及可持续性实施需 要妇女、家庭和社区保健工作者的积极参与，然而，与这些群体接触的策略往往是 外部驱动和自上而下的。自 2005年以来，一家名为埃朱特（Ekjut）的印度非政府 组织一直试图通过帮助妇女团体改善印度贾坎德邦和奥里萨邦部落地区的产妇和新 生儿健康来扭转这一趋势。 当地妇女协调者通过涉及参与式学习和行动的一系列活动指导妇女团体，在这 一过程中，妇女识别、确定优先事项并分析当地产妇和新生儿的健康问题，并随后 制定和执行解决这些问题的战略。埃朱特干预最初是 2005年至 2008年期间在贾坎 德邦和奥里萨邦三个彼此接壤区域的 36个主要部落群体中进行的一项整群随机对 照试验中进行评估的。最近的一项研究报告指出，由于采取了干预措施，这些地区 的新生儿死亡率显著下降。由此可见，通过妇女团体进行社区动员提高印度农村地 区的新生儿存活率的模式是可持续的、可复制的。 资料来源：Roy等，2013 [54] 第五章 哪些方式和方法适用于实施研究？ - 51 - 混合方法研究 混合方法研究，顾名思义，就是在同一研究中收集和分析数据时既使用定 性的方法，也使用定量的方法。虽然混合方法研究并不是专门为实施研究设计 的，但它特别适合于这些研究活动，因为它提供了一种实用的方式来理解多种 观点、不同类型的因果链，和多种类型的结果——这些在实施环境中都是通用 的。混合方法研究有着非常广泛的用途——一项研究报告显示混合方法研究有 多达 65种用途[58]。这些用途可以归纳为以下四大类[59]2： ①充分参与：从参与者样本中获得最多信息（例如，通过管理标准调查问 卷，然后要求深入解释）。 ②工具有效性：确保所使用的工具合适且有效（例如，使用焦点小组法确 定问卷项目或测试其有效性）。 ③实施保真度（治疗方案的完整性）：评估干预措施或方案是否按设计进 行。 ④含义增强：最大限度地解释调查结果，如通过定性测试方法来解释统计 分析，反之亦然。 许多不同的方案从不同方法使用的侧重点、研究不同部分使用的抽样方案、 定性和定量方法的时间和顺序，以及定性和定量方法之间的混合水平等方面来 描述不同类型的混合方法研究[60,61]。完全混合设计在研究的每个阶段（包括 抽样和数据收集、分析和得出推论）均同时使用定性和定量的方法。有兴趣的 读者可以参考塔萨科里（Tashakkori）和特德利（Teddlie）的手册，其中概述了 35种不同的设计，或者参考乌布泽（Wuegbuzie）和柯林斯（Collins）的文章， 文中描述了 24种不同的抽样方案[55,61]。7位作者提供了关于混合方法设计的 设计、实施和报告的广泛指南[59,60,62,63]。下面列出了一个简单的混合方法研 究（GRAMMS）报告方案[64]： 2 引文作者使用的术语与本文此处使用的术语有所不同，我们调整这些术语是为了避免和实施研究中的相 似术语产生混淆。作者们原本使用“工具保真度”（instrument fidelity），而我们则使用“工具有效性” （instrument validity）以避免和实施保真度（implementation fidelity）产生混淆。原作者使用“治疗方案的 完整性”（treatment integrity）来描述“实施保真度”（implementation fidelity），他们还使用“显著性增 强”（significance enhancement）——这可能和定量研究中的统计学显著性检验产生混淆，因而我们改用 “含义增强”（meaning enhancement）。 第五章 哪些方式和方法适用于实施研究？ - 52 - ► 描述使用混合方法解决研究问题的依据。 ► 按照方法的目的、优先级和顺序描述设计。 ► 从抽样、数据收集和分析方面描述每种方法。 ► 描述整合发生的地点、发生的方式以及参与整合的人员。 ► 描述将一种方法与另一种方法相关联所产生的任何局限性。 ► 描述从混合或整合方法中是否获得新的见解。 研究问题的重要性 上述不同的研究方法可以说是构成了实施研究人员的基本工具箱。但请谨 记，在实施研究中，应由所提出的问题决定使用的工具，而不是由工具来决定 问题的形成。简言之，在实施研究中，问题为王。这并不意味着它是一项完全 点对点的散在模式，也不意味着它完全不可能构建一个包罗万有的总体概念。 事实上，某些理论框架在帮助思考实施过程方面是有价值的，尤其是实现长期 目标（如改善健康）所需的变革理论。 一个变革理论应该描述在实现目标的过程中产生结果的逻辑顺序或途径， 并应反映关于变革的假设。上述过程中，每个环节发生的原因及其背后的假设， 往往是需要验证、测试或进一步解释的研究主题。同样值得注意的是，在实施 理论方面所做的一些工作产生了宝贵的见解，如《实施研究整合框架》 （Consolidated Framework for Implementation Research，CFIR）应该引起重视， 这是组织实施研究关键概念的有用模板（补充资料 10） [65-67]。 匹配研究问题与方法 鉴于研究问题的重要性，因此有必要花点时间考虑可能出现的各种问题， 以及选用可能适合回答这些问题的研究方法。解决这些问题的其中一种方法是， 根据将要进行的研究的核心目标，将研究问题分成有限的几个类别，如表 5所 示[4，34，58]。 因此，在第二个例子中，为了描述实施现象而提出的关键问题之一是： “特定的背景下，什么描述了影响实施的主要因素？”在其次是可能产生所需 第五章 哪些方式和方法适用于实施研究？ - 53 - 信息的研究方法，包括定量（横断面调查或描述性调查、网络分析）和定性方 法。值得注意的是所提出方法是对定性研究的重视，需要重视的是对旨在通过 例如个案研究、关键信息提供者访谈和焦点小组访谈挖掘丰富细节的研究。 在预测示例中，研究问题趋向于产生预测，利用先验知识或理论来预测未 来事件。该研究可能依赖于多种定量研究方法，包括基于主体模型（agent- based modeling，ABM），其依靠计算模型模拟组织或团体等经济主体的行为和 相互作用，以评估其对整个系统的影响，以及数据外推和敏感性分析。这些方 法可以通过定性研究来补充，如情景构建练习和所谓的“德尔菲法”，该技术 依靠专家小组进行预测，通常基于调查问卷，并在几轮之间进行，主持人提供 匿名预测摘要。这种预测显然具有特殊价值，因为正在进行重大创新，坦桑尼 补充资料 10 实施理论 为促进健康干预措施的有效实施，许多理论得到了发展。许多人试图围绕实 施问题解释个人或群体行为。例如，RE-AIM 框架（Reach，Effectiveness， Adopt，Implementation，Maintenance 五个单词的首字母；即覆盖面、效果、采 用、实施和维持）通常用于健康促进干预措施，并提供了一种实用的方法，通过 个人、组织和社区的变化来评价健康干预的效果。创新扩散理论试图解释创新如 何扩散、强调创新的感知属性（相对优势、与当前方法的兼容性、观察结果的能 力、检验创新的能力及其复杂性）、采用者的创新性、社会系统、采用的个体过程 和扩散系统的重要性。 制定《实施研究综合框架》的目的是为了巩固用于支持进一步发展理论和试 验的各种理论和术语，以便对已证明有效的卫生干预措施进行转化。《实施研究综 合框架》由五个部分组成：1.干预特征；2.外部环境；3.内部环境；4.个体特征；5. 实施过程。证据强度和质量等结构与干预领域相关，而患者需求和资源等问题是 外部环境的一部分，与组织内部环境相关的结构包括其文化和领导参与，个人的 态度、信念和能力也发挥着重要作用，而其他影响因素与实施过程本身（例如计 划、评价和反映）有关。 资料来源：Glasgow等，2009 [65]；Rogers等，2003 [66]；Damschroder等，2009 [67] 第五章 哪些方式和方法适用于实施研究？ - 54 - 亚联合共和国的防虫蚊帐代金券项目，或加纳的医疗保险项目就是如此。 如前几章所述，在真实世界环境中进行并评估影响实施的背景因素等的实 施研究是最有用的。当中有一些研究是专门测试实施策略和/或测量实施结果变 量的[68]。 结论 本章基于对实施研究人员的有用性选择了若干研究方法进行介绍，并尝试 描述了这些方法的决定性特征。同时提出如何将这些方法应用于具体的实施研 究问题。在下一章中，重点将是使实施研究与需要保持一致，既包括对研究对 象的应答，也包括目标受众的需求。 第五章 哪些方式和方法适用于实施研究？ - 55 - 改编自：Peters等，2009 [4]；Habicht等，1999 [34] 表 5 实施研究目标、实施问题和研究方法的类型 目的 说明 实施问题 研究方法和数据收集方法* 探索 从具体的例子 中探索一种想 法或现象，以 作出假设或描 述概况 哪些可能的因素可 以带来卫生干预的 良好实施？或帮助 巩固、推广健康干 预措施？ 定性方法：基础理论，民族志，现象学， 案例研究和描述性研究；关键信息访谈， 焦点小组，历史回顾 定量方法：网络分析，横断面调查 混合方法：定性和定量方法相结合 描述 识别和描述该 现象及其关联 或可能的原因 什么描述了实施发 生的背景？ 什么描述了在给定 的背景下影响实施 的主要因素？ 定量：横断面（描述性）调查，网络分 析。 定性方法：基础理论，民族志，现象学， 案例研究和叙事方法；关键信息访谈，焦 点小组，历史回顾。 混合方法：定性和定量研究，数据和分析 合并 影响 测试干预是否 产生预期结 果。 - - 充分性 有信心确保干 预的执行和结 局的发生。 卫生干预的覆盖范围 在干预的受益者中是 否发生了变化？ 针对受试者的开展前后观察，或时间序列 研究；参与式行动研究 合理性 有较充分的信 心认为结果是 卫生干预带来 的 健康结局是否合理归 因于实施的干预而非 其他原因？ 同期、非随机分组试验：在一些地区实施 健康干预，而在其他地区未实施；在项目 接受者和非接受者中开展前后或横断面研 究；典型质量改进研究 概率 经计算，健康 结局大概率是 由于干预所致 健康结局是否归因于 干预措施的实施？ 部分对照试验：实用性和整群随机试验； 在某些地区实施健康干预，而在其他地区 未实施；效果-实施混合 解释 发展或扩展一 种理论来解释 概念和事件发 生的原因之间 的关系，以及 它们是如何发 生的？ 干预措施的实施如 何以及为什么会对 健康行为、服务或 各种状态产生影 响？ 混合方法：定性和定量研究，数据和分析 合并 定量：重复测量的背景，行动者，在子单 位中实施的深度和广度；网络的识别；可 以使用设计的确证推论；有效的简化混合 定性方法：案例研究、现象学和人种学方 法，包括关键信息提供者访谈、焦点小 组、历史回顾 参与式行动研究 预测 使用先验知识 或理论预测未 来事件 未来实施的可能过程 是什么？ 定量：基于主体模型（ABM）；模拟和预 测建模；数据外推和敏感性分析（趋势分 析、经济计量建模） 定性：情景构建练习；来自意见领袖的德 尔菲法 第六章 如何进行实施研究？ - 56 - 第六章 如何进行实施研究？ 本章要点  实施研究应该既要满足目标受众的要求，又要考虑到所研究对象的特殊性。  研究设计需具有应变能力，能够识别多个时间点上的变化情况  实施研究方法并没有固定不变的选择规则。但要注意的是，使用的研究方法 应能够回答所提出的研究问题。 如何进行实施研究？ “响应主题和目标受众的需求。” 前文中，我们讨论了实施研究的“关键”是研究问题。在本章中，我们将 讨论开展与需求相符合的研究的重要性——既是为了满足目标受众的要求，也 是为了响应所研究主题的特殊性。 尽管实施研究在理论形成中起到了一定的作用，但实施研究所产生的证据 主要是为了在真实世界中应用，而不仅仅是为了给研究人员使用。这意味着， 实施研究人员需要认识到他们研究成果的用途是什么。其中要重点考虑的是结 果或预测需要多确切。例如，政策制定者在自己的工作范围内，可能正在寻找 一个强有力的证据表明某项干预措施是可行的，但未必有时间开展一项为期多 年的研究来有力证明其可行性。前一章所述的关于可见性增强材料与摩托车驾 驶员交通安全的准现场试验就是一个很好的例子。这一试验很好地回答了所提 出的实施研究问题。 这些差异对研究的基本设计（包括研究样本量、是否需要设置对照组、是 否需要随机分组、以及研究成本和时间要求）具有重要的提示意义[4, 34]。不 同的研究方法对预算和时间安排也有不同要求。例如，描述现状的简单探索性 或描述性研究可能不需要长期观察。然而，评估干预影响的研究通常至少需要 两个时间点的观察。而如果要评估实施因素的变化，则可能需要更频密的测量。 如果主要的研究问题要求解释如何或为何实施正在以某种方式发展，那么 对研究设计的要求将更多，需要反复多次地观察不同类型受访者的变化。所以， 在对指南（以抗菌药物指南为例）依从性不高的情况下，需要对抗菌药物处方 第六章 如何进行实施研究？ - 57 - 水平进行定量研究，至少需要对合理用药、医生和患者态度和相互作用的认知 水平进行研究。但是，即使这样，实施研究人员也可能因为没有深入挖掘或者 扩大研究范围，而忽视关键的背景因素。 例如，当公立医院按全科医生就诊次数支付薪酬时，这些医生可能会更快 地开具患者所需的抗菌药物处方，而不是花费时间告知患者有关抗菌素耐药性 的危险，因为这样更加有利可图[69]。仅仅关注医生的合理用药意识问题，将 忽视卫生系统财务制度所带来的重要影响。 灵活应变的研究设计很重要 卫生系统很复杂。由于其主要参 与者的复杂性而加剧了复杂性，包括 决策者，实施者和健康服务的最终受 益者。卫生系统及其参与者都在不断 变化和适应新行动，经常以不可预测 的方式作出反应[70]。此外，各种各 样的背景因素通常会影响实施，这些 因素往往会随着时间而变化，产生不 可预测的影响。这需要实施者不断适 应。这对研究实施过程中使用的研究方法具有深远意义，特别是对于响应研究 对象变化的灵活性和创造性的需要。此外，由于政策、程序和实践的实施很少 是静态或线性过程，因此研究设计需要具有响应能力，能够在多个时间点捕获 不断变化的元素。 依赖于单一且不可改变的干预的研究设计（如典型的随机对照临床试验） 很难适应不断变化的研究现象，尤其是当变化不可预测时。在研究设计本身很 复杂的情况下，其复杂性所带来的挑战也越大，这需要多种方法和不同来源的 信息。 论证研究方法的选择 与其他研究领域一样，除了所使用的方法应反映所提出的问题之外，没有 固定的规则来证明实施研究所选择的研究方法的合理性。对研究目标和具体研 第六章 如何进行实施研究？ - 58 - 究问题的理解是一个很好的起点，应该借鉴提出的变革理论。图 3 提供了一个 有用的模板，用于思考在实施研究的连续过程中的一系列研究问题和方法，随 着对实施证明的测试和对实施结果变量的考虑，实施变得更加重要。 然而，一般而言，实施证据通常涉及已经证明有效的临床或公共卫生干预。 当这些问题进一步转向到扩大卫生系统覆盖面、整合和干预措施的可持续性问 题时，研究的重点转向实施。毋庸置疑，这仅仅是一个理论模型，实际上这一 过程既不是线性的，也不是在一次迭代中完成的，因为卫生系统会不断地适应 和整合具有不同切入点的多种创新。 显而易见，评估任何特定的研究需要一定程度的主观判断，但作为一般规 则，良好的实施研究应该能够解决以下每个问题： ► 研究是否会回答相关且重要的实施问题？ ► 新知识是否值得投入研究成本？ ► 是否有明确的研究目标和与实施相关的问题？提出的研究设计是否与 它们相匹配？（见表 5） ► 该研究是否符合改变理论或因果链？如果不符合，那么有什么产生新 理论、提出新问题的潜力？ ► 研究产出的成果是否能被目标受众及时采用并实施？ ► 研究设计中是否关注到干预是否稳定、简单且可复制？或者干预是否 还需进行修改？ ► 研究是否考虑到干预、背景和效应随着时间变化和地点变更而可能发 生的变化？ ► 在复杂的环境中，研究能否识别卫生系统的主要组成部分及其关系？ 是否能发现干预有可能带来的意外效果？ 评估实施研究的质量 当研究涉及实施时，评估常规定量和定性研究方法质量的标准大致相同。 已经存在帮助设计和报告卫生研究的指南，其中许多指南已由 EQUATOR“卫 生研究报告图书馆”进行分类报道（见：http://www.equator-network.org/）。包 第六章 如何进行实施研究？ - 59 - 括 CONSORT随机对照试验声明、STROBE观察性研究指南、PRISMA系统综 述和荟萃分析、COREQ定性研究等指南。 由于这些指南都没有关注与实施研究相关的特定问题，所以我们提出以下 一组问题，可以作为评估与实施研究特别相关问题的总结准则。除了常规指南 之外，还可以使用这些问题作为参考。值得注意的是，这些准则的目的是帮助 报告和评估研究报告，但它们也可用于评估研究建议或设计实施研究。当中的 关键问题包括： ► 研究是否明确解决了实施问题？ ► 是否有明确描述正在实施的内容（例如：实践、项目或政策的详 情）？ ► 研究是否涉及实施策略？如果是，它是否被恰当地描述和检查？ ► 研究是在真实世界中进行的吗？如果是，这些条件是否足够详细？ ► 研究是否恰当地考虑了实施结局变量？ ► 研究是否适当考虑了影响实施的背景和其他因素？ ► 研究是否适当考虑了随着时间推移而发生的变化？是否考虑了制度或 系统的复杂程度？ ► 研究是否清楚地确定了研究的目标受众？是否考虑到目标受众如何使 用研究结果？ 我们希望这些准则能使实施研究的从业者、决策者、研究人员和使用者将 开始有信心评估是否以透明的方式进行和报告实施研究，并保证通过有效和可 靠的方法得出的结论。 结论 本章力求识别实施研究人员优化其工作影响的主要考虑因素。在下一章中， 我们将看到实施研究领域本身面临的挑战，并讨论实施者和研究人员如何能够 做更多的工作来支持这一关键研究领域，以更好地实现它的潜力。 第七章 如何让实施科学的潜能变为现实？ - 60 - 第七章 如何让实施科学的潜能变为现实？ 本章要点  如果没有实施研究，我们充其量只能将宝贵的资源投入到实施中，以期事情 能够成功。  尽管实施研究很重要，但它仍然是一个被人们忽视的领域。一方面因为人们 对实施研究是什么、能提供什么缺乏理解；另一方面因为缺乏开展实施研究 的资金。  实施研究应被视为项目实施的核心部分，从而将其整合到项目周期中。  实施研究需要更多资金，并且这笔资金应与项目资金匹配。  中低收入国家中，有意从事实施研究的研究人员和实施者需要更多的机会。 如何实现实施科学的潜能？ “实施科学研究的总体花费并不多。投入一定的研究经费就可产生很大获 益。” 实施研究对于确保实现有 效实施政策、项目和服务所产 生的效益至关重要。如果没有 它，以及它产生的知识，我们 充其量只能投入宝贵的资源， 而不是应用实施研究所能实现 的证据和经验知识决策。这种 方法往往需要付出代价。每当 对“真实世界”关注不足、或项目设计者未预料到的背景环境因素，从而导致 一个项目失败时，都将导致资源浪费，抑或导致某些人群遭受不必要的痛苦。 有时这样的代价是巨大的。但是，尽管实施研究十分重要，但它仍然是一个被 忽视的研究领域。许多研究证实了这一事实，例如，2006 年大卫·库克西 （David Cooksey）爵士在英国医疗保健研究中所作的报告。该报告指出“存在 一种重视基础科学、轻视应用研究的不合理的奖励机制”，并报告说，三分之 第七章 如何让实施科学的潜能变为现实？ - 61 - 二的公共和慈善研究经费用于基础研究，剩余的三分之一则用于应用研究（当 中包括实施研究）[71]。毋庸置疑，私营机构的研究经费显著倾向于为新市场 开发新产品，而不是帮助现有产品发挥出最大价值。大多数发达国家的情况都 是和英国类似，数十亿美元投入到了医疗创新中，然而当中只有很少一部分被 用于如何用好这些创新[72]。这个问题影响到每个人。特别是在中低收入国家， 忽视实施研究的不利后果最为严重。即使产品程序和服务是专门为中低收入国 家设计的，但由于实施失败，它们往往难以服务到最终用户。 既然实施研究如此必要，为什么不开展更多这方面的研究呢？为什么进行 的实施研究有时并没产生我们预期的影响？实施研究开展较少的原因有多种， 其中最重要的原因是人们不了解什么是实施研究，也不够了解实施研究在最大 限度地发挥干预措施效益方面能够发挥的作用。另一个明显的限制因素是缺乏 支持此类研究的资金支持。 从本指南中描述的各种方法中，可以清楚地看到实施研究总的来说并不是 一种花费巨大的研究。与尖端生物医学工程研究的投资相比，肯定并不昂贵。 因此，对实施研究的投入有很长的路要走。此外，有可能认为我们现在有许多 降低发病率和死亡率所需的干预措施和技术，并应更加注重更好地利用它们。 这并不是说应该停止对基础科学的投资，值得注意的是，上问提到的库克西报 告中，强调了英国在一项名为“蓝天（blue sky）”的项目中保持投资的重要性。 这也不仅仅是为实施研究筹集新资金的问题，而是更好地利用现有资金，将其 引导到符合需求的研究，例如，增加实施者和项目人员获得这笔资金的机会。 为了应对这些不同的挑战，有必要在许多不同的方面支持和促进实施研究， 我们将其分解为一系列行动计划： 行动 1：实施研究应被视为项目实施的核心部分。这意味着将研究以迭代 的方式嵌入到项目周期中，以允许持续学习，并在必要时进行调整。支持这一 变革的一种方法是在专业学者的支持和指导下，鼓励实施者或项目人员承担所 有权或主要责任。底线是：实施者需要在实施研究中发挥更积极的作用。 行动 2：为确保实施研究变得更容易获得，应鼓励研究人员深度参与项目 活动。这将包括与实施者进行对话，并寻求将实施研究嵌入真实世界的举措。 例如研究人员可住在项目现场，以便更好地理解实施的复杂性。 第七章 如何让实施科学的潜能变为现实？ - 62 - 行动 3：实施者需要让研究人员更容易了解各样的项目，并邀请研究人员 参与到项目中。有时候实施者会拒绝合作。知识创造活动应作为项目实施的一 部分，并应聘请具有开展研究专业知识的人员来支持这一过程。 行动 4：为实施研究提供更多资金，并将这笔资金与项目资金合为一体。 实现这一调整至关重要的是，实施研究的资金应在项目预算内提供，或通过结 构化合作和伙伴关系明确地与项目活动建立联系。 行动 5：需要为项目人员或实施者提供更多的实施研究培训机会。实施研 究也应纳入公共卫生硕士等培训计划，以便将实施研究视为公共卫生实践的核 心职能。通过这样的良性循环，实施研究能产生知识，并可为项目实施提供有 益的信息。 行动 6：为希望进行实施研究的中低收入国家的研究人员和实施者提供更 多的指导和机会。没有人能够更好地提供支持中低收入国家实施研究所需的那 种特定背景研究。因此这些国家的研究人员是一股巨大的有待开发发掘的力量。 行动 7：除了与学术出版和教学相关的激励措施外，研究人员的激励措施 应与政策和计划的变革联系起来。因此需要对研究人员进行一些重新定位，让 研究人员有机会体验和理解实地工作和计划。这可能是资助方可以支持的事情。 结论 本指南试图纠正对实施研究的理解上的不足。作者希望读者能够在这本书 中引用的源材料中更深入地探讨这一主题。特别是，希望项目人员和实施者对 该主题更感兴趣，认识到实施研究实际上是一个与项目相关的问题。换句话说， 实施研究应该是项目规划和实施的一个必要组成部分，而不是等到项目落地实 施之后才开始做的事情。不然，这种以其他研究人员的需求而开展的研究将产 生研究“泡沫”。研究人员在其负责的部分中可以做更多的工作，与研究过程 中的实施者和项目人员进行交流。我们应当为此做出改变，理解环境重要性的 程序人员和实施者以及理解探究方法和科学的研究人员必须通力合作。只有这 样，我们才能希望提高对实施问题的理解。 健康领域的实施研究：实用指南 - 63 - 参考文献 1. Fenner, F., Smallpox and its eradication. 1988, Geneva, Switzerland: World Health Organization. 2. Foege, W.H., J.D. Millar, and J.M. Lane, Selective epidemiologic control in smallpox eradication. American Journal of Epidemiology, 1971. 94(4): p. 311-315. 3. NORAD, Accelerating progress in saving the lives of women and children, in The Global Campaign for the Health Millenium Development Goals. 2013, Ministry of Foreign Affairs, Norway: Oslo, Norway. 4. Peters, D.H., et al., Improving Health Services in Developing Countries: From Evidence to Action. 2009, Washington, DC, USA: World Bank. 5. Mandala, J., et al., Prevention of mother-to-child transmission of HIV in Zambia: implementing efficacious ARV regimens in primary health centers. BMC Public Health, 2009. 9(314). 6. de Savigny, D., et al., Introducing vouchers for malaria prevention in Ghana and Tanzania: context and adoption of innovation in health systems. Health Policy and Planning, 2012. 27(4): p. iv32-iv43. 7. Waddington, C.J., A price to pay, part 2: The impact of user charges in the Volta region of Ghana. International Journal of Health Planning and Management, 1990. 5(4): p. 287-312. 8. Agyepong, I.A. and S. Adjer, Public social policy develop-ment and implementation: a case study of the Ghana National Health Insurance scheme. Health Policy and Planning, 2007. 23(2): p. 150-160. 9. Victoria, C.G., et al., Context matters: Interpreting impact findings in child survival evaluations. Healthy Policy & Plan-ning, 2005. 20(1): p. i18-i31. 10. Thaddeus, S. and D. Maine, Too far to walk: maternal mortality in context. Social Science Medicine, 1994. 38(8): p. 1091-110. 11. de Savigny, D. and T. Adam, Systems thinking for health systems strengthening. 2009, Alliance for Health Policy and Systems Research. World Health Organization. 12. Sankar, B., Micro-planning in Andhra Pradesh: Ensuring Quality HIV Care for Individuals and Communities, in AIDSTAR-One. 2013, USAID: Arlington, VA, USA. 13. Peters, D.H., et al., A balanced scorecard for health services in Afghanistan. Bulletin of the World Health Organization, 2007. 85: p. 146-151. 14. Hansen, P.M., et al., Measuring and managing progress in the establishment of basic health services: the Afghanistan health sector balanced scorecard. International Journal of Health Planning and Management, 2008: p. 107-117. 15. Edward, A., et al., Configuring balanced scorecards for measuring health system performance: evidence from 5 years evaluation in Afghanistan. PLoS Med, 2011: p.e1001066. 16. WHO and ExpandNET, Nine steps for developing a scaling-up strategy. 2010, Department of Reproductive Health and Research, WHO: Geneva, Switzerland. 17. Agurto, I., et al., Improving cervical cancer prevention in a developing country. International Journal for Quality in Health Care, 2006: p. 81-86. 健康领域的实施研究：实用指南 - 64 - 18. Faden, R.R., et al., An ethics framework for a learning health care system: A departure from traditional research ethics and clinical ethics, in Hastings Centre Report. 2013. p. S16-27. 19. Smith, M., et al., Best Care at Lower Cost: The Path to Continuously Learning Health Care in America. 2013, Insitute of Medicine of the National Academies: Washing-ton, DC, USA. 20. IOM, Crossing the Quality Chasm: A New Health System for the 21st Century. 2001, Institute of Medicine: Washington, DC, USA. 21. Lam, A., Tacit knowledge, organizational learning and societal institutions: An integrated framework. Organization Studies, 2000. 21(3): p. 487-513. 22. Ciliska, D., et al., Diffusion and dissemination of evidence-based dietary srategies for the prevention of cancer. Nutrition Journal, 2005. 4(13). 23. Remme, J.H.F., et al., Defining Research to Improve Health Systems. PLoS Medicine, 2010. 7(11). 24. Curran, G.M., et al., Effectiveness-implementation hybrid designs: Combining elements of clinical effectiveness and implementation research to enhance public health impact. Medical Care, 2012. 50(3): p. 217-226. 25. Proctor, E., et al., Outcomes for implementation research: conceptual distinctions, measurement challenges, and re-search agenda. Administration and Policy in Mental Health, 2011. 38(2): p. 65- 76. 26. Brownson, R.C., G.A. Colditz, and E.K. Proctor, Dissemination and Implementation Research in Health: Translating Science to Practice. 2012: OUP USA. 27. Peters, D.H., L. Paina, and S. Bennett, Expecting the unexpected: Applying the Develop-Distort Dilemma to maximize positive market impacts in health. Health Policy Plan, 2012. 27(4): p. iv44- iv53. 28. Cornwall, A. and R. Jewkes, What is participatory research? Social Science & Medicine, 1995. 41(12): p. 1667-1676. 29. Subramanian, S., et al., Do we have the right models for scaling up health services to achieve the Millennium Development Goals? BMC Health Services Research, 2011. 11. 30. WHO. Bloomberg initiative to reduce tobacco use. Tobacco Free Initiative (TFI) 2013 [cited 2013； Available from: http://www.who.int/tobacco/about/partners/bloomberg/ en/. 31. Peden, M., G. diPietro, and A. Hyder, Two years into the road safety in 10 countries project: How are countries doing? Injury Prevention, 2012. 18(4): p. 279. 32. WHO, Changing Mindsets: Strategy on Health Policy and Systems Research. 2012, Alliance for Health Policy & Systems Research 33. Hanney, S. and M. Gonzalez-Block, Evidence-informed health policy: Are we beginning to get there at last? Health Research and Policy Systems, 2009. 7(30). 34. Habicht, J.P., C.G. Victora, and J.P. Vaughan, Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology, 1999. 28: p. 10-18. 35. Tran, N., et al., Engaging policy makers in road safety research in Malaysia: A theoretical and 健康领域的实施研究：实用指南 - 65 - contextual analysis. Health Policy, 2009. 90(1): p. 58-65. 36. Consortium, P. PERFORM: Improving Health Workforce Performance. 2011 [cited 2013；Available from: http:// www.performconsortium.com/. 37. Newman, I., et al., A typology of research purposes and its relationship to mixed methods, in Handbook of mixed methods in social and behavioral research, A. Tashakkori and C. Teddlie, Editors. 2003, Sage: Thousand Oaks, CA. p. 167-188. 38. American Educational Research, A., Standards for reporting on empirical social science research in AERA publications. Educational Researcher, 2006. 35: p. 33-40. 39. Zwarenstein, M., et al., Improving the reporting of pragmatic trials: An extension of the CONSORT statement. BMJ: British Medical Journal, 2008. 40. Fairall, L., et al., Task shifting of antiretroviral treatment from doctors to primary-care nurses in South Africa (STRETCH): A pragmatic, parallel, cluster-randomised trial. Lancet, 2012: p. 889-898. 41. Baqui, A.H., et al., Effect of community-based newborn-care intervention package implemented through two service-delivery strategies in Sylhet district, Bangladesh: A cluster-randomised controlled trial. Lancet, 2008: p. 1936-1944. 42. Baqui, A.H., et al., Community-based validation of assessment of newborn illnesses by trained community health workers in Sylhet district of Bangladesh. Tropical Medicine & International Health, 2009. 14(12): p. 1448-1456. 43. Choi, Y., et al., Can mothers recognize neonatal illness correctly? Comparison of maternal report and assessment by community health workers in rural Bangladesh. Tropical Medicine & International Health, 2010. 15(6): p. 743-753. 44. Shah, R., et al., Community-based health workers achieve high coverage in neonatal intervention trials: A case study from Sylhet, Bangladesh. Journal of Health, Population, & Nutrition, 2010. 28: p. 610. 45. Davidoff, F., et al., Publication guidelines for quality improvement in health care: Evolution of the SQUIRE project. Quality & Safety in Health Care, 2008. 17: p. i3-i9. 46. Brassard, M., The Memory Jogger II: A Pocket Guide of Tools for Continuous Improvement and Effective Planning. Vol. First Edition. 1994, Methuen, MA: Goal/QPC. 47. Campbell, D.T., J.C. Stanley, and N.L. Gage, Experimental and quasi-experimental designs for research. 1963, Boston, USA: Houghton Mifflin. 48. Speroff, T. and G.T. O’Connor, Study designs for PDSA qual-ity improvement research. Quality Management in Health-care, 2004. 13: p. 17-32. 49. Lindelow, M. and A. Wagstaff, Assessment of Health Facility Performance: An Introduction to Data and Measurement Issues, in Are You Being Served? New Tools for Measuring Service Delivery, S. Amin, J. Das, and M. Goldsteim, Editors. 2008, The World Bank: Washington, DC, USA. p. 19-66. 50. Mergler, D., Worker participation in occupational health research: theory and practice. International Journal of Health Services, 1987. 17: p. 151-167. 51. Chambers, R., Revolutions in development inquiry. 2008, UK: Earthscan. 健康领域的实施研究：实用指南 - 66 - 52. Smith, L., L. Rosenzweig, and M. Schmidt, Best Practices in the Reporting of Participatory Action Research: Embracing Both the Forest and the Trees The Counseling Psychologist, 2010. 38: p. 1115-1138. 53. Israel, B.A., et al., Critical issues in developing and follow-ing CBPR principles. Community-based participatory research for health: From process to outcomes, 2008: p. 47-66. 54. Roy, S.S., et al., Improved neonatal survival after participatory learning and action with women’s groups: A prospec-tive study in rural eastern India. Bulletin of the World Health Organization, 2013. 91(6): p. 426-433B. 55. Onwuegbuzie, A.J. and K.M.T. Collins, A typology of mixed methods sampling designs in social science research. The Qualitative Report, 2007. 12: p. 281-316. 56. Pawson, R., et al., Realist review: “A new method of systematic review designed for complex policy interven-tions. Journal of Health Services Research & Policy, 2005. 10: p. 21-34. 57. Dieleman, M., et al., Realist Review and Synthesis of Retention Studies for Helath Workers in Rural and Remote Areas, in Increasing access to health workers in rural and remote areas. 2011, World Health Organization: Geneva, Switzerland. 58. Collins, K.M.T., A.J. Onwuegbuzie, and I.L. Sutton, A model incorporating the rationale and purpose for conducting mixed methods research in special education and beyond. Learning Disabilities: A Contemporary Journal, 2006. 4: p. 67-100. 59. Leech, N.L. and A.J. Onwuegbuzie, Guidelines for Conducting and Reporting Mixed Research in the Field of Counseling and Beyond. Journal of Counseling & Development, 2010. 88: p. 61-69. 60. Creswell, J.W., et al., Best practices for mixed methods research in the health sciences. Maryland, USA. 2011: National Institutes of Health. 61. Tashakkori, A. and C. Teddlie, Mixed Methodology: Combining Qualitative and Quantitative Approaches. 1998: SAGE Publications. 62. Creswell, J.W., Research design: Qualitative, quantitative, and mixed methods approaches. 2008: Sage Publications, Incorporated. 63. Creswell, J.W. and V.L.P. Clark, Designing and Conducting Mixed Methods Research. 2010: SAGE Publications. 64. O’Cathain, A., E. Murphy, and J. Nicholl, The quality of mixed methods studies in health services research. Journal of Health Services Research & Policy, 2008. 13: p. 92-98. 65. Glasgow, R.E., T.M. Vogt, and S.M. Boles, Evaluating the public health impact of health promotion interventions: The RE-AIM framework. American Journal of Public Health, 1999. 89(9): p. 1322- 1327. 66. Rogers, E.M., Diffusion of Innovations, 5th Edition. 2003: Free Press. 67. Damschroder, L.J., et al., Fostering implementation of health services research findings into practice: A consolidated framework for advancing implementation science. Implementation Science, 2009. 4(50): p. 50. 68. Epstein, J., Generative Social Science: Studies in Agent-Based Computational Modeling. 2006, Princeton, NJ, USA: Princeton University Press. 健康领域的实施研究：实用指南 - 67 - 69. Humphreys, G., Are antibiotics still “automatic” in France? Bulletin of the World Health Organization, 2011. 89(1): p.8-9. 70. Agyepong, I.A., et al., When solutions of yesterday become problems of today: Crisis-ridden decision making in a complex adaptive system (CAS) - the Additional Duty Hours Allowance in Ghana. Health Policy and Planning, 2012. 27: p. iv20-iv31. 71. Cooksey, D., A review of UK health research funding. 2006, London: HM Treasury. 72. Sanders, D. and A. Haines, Implementation research is needed to achieve international health goals. PLoS Med, 2006. 3: p. 719-722.