Compliance to and perceived benefits of annual ivermectin treatment a multi-country effort

t 1 @) @ A Feasibility Study on Compliance to and Perceived Benefits of Annual Ivermectin Treatment A Multi-Country Effort African Program for Onchocerciasis Control World Health Organization Draft December 2005 Proiect Team Project Coordinator and Principal Investigator Data Analysis and Report Writing Principal Investigador for Imo, Abia, Cross River, Osun and Taraba, States, Nigeria Principal Investigator for Kaduna, Kogi and Zamfara States, Nigeria Principal Investigator for Cameroon and Chad Principal Investigator for Tanzania Principal Investigator for Uganda Adenike Abiose William Brieger Joseph Okeibunor Joseph Okeibunor Elizabeth Elhassan Samuel Wanji William Kisoka Richard Ndyomugyenyi I \'.Ia INTRODUCTION A multi-country team assembled at the African Program for Onchocerciasis Control headquarters to develop a protocol for studying factors that influenced long-term compliance u-ong villagers with annual ivermectin distribution. At this point, APOC had been supporting community directed treatment programs since 1997 and hoped that participating countries would continue ivermectin distribution into the future. A clearer understanding of why people do or do not take ivermectin was needed in order to sustain community mobilization for the long period of time (at least fifteen years) needed to control onchocerciasis. A protocol to achieve this was developed for implementation in five countries. The research group later realized that a feasibility study would be needed to determine if quality data were actually available over an adequate period of time to provide valid answers to Compliance questions. A pilot study on data availability and quality was conducted in Oyo State, Nigeria in mid-April, and from that experience, simple tools were developed to determine the feaiibitity of conducting a larger study. This report describes the series of feasibility studies that were conducted in potential sites for a larger compliance study' Background A large, multi-country study carried out in 1994-95 showed that community-directed treatment with ivermectin (CDTI) was a feasible and effective way to ensure that Mectizan@ reached endemic communities. The study found that significantly higher treatment coverage figures of 68.6%owere achieved when CDTI was used compared to 62.2%o, where the health ryit.rn planned and organized distribution. The ability of CDTI to achieve higher coverage is important because of the need to maintain a minimum annual population treatment coverage level of 65% for at least 15 years in order to effectively control onchocerciasis (Plaisier et al., lggT).This finding informed the decision of the African Program for Onchocerciasis Control (APOC) to adopt the CDTI as its main control strategy. Though community-directed treatment (ComDT) approach has been shown to be effective in reaching hundreds of previously neglected communities or those with limited access to health services, the rate of compliance of individuals who have been on ivermectin treatment since the inception of the APOC sponsored community-directed treatment with ivermectin programs has not been documented. To date published reports of CDTI interventions have focused on coverage since it has been found that the program requires a minimum of 650/o annual population coverage in endemic communities over a minimum of l5 years for effective control of onchocerciasis. Most studies, including independent project monitoring sponsored by APOC, have reported on CDTI coverage in a given distribution period. Only two CDTI studies, one in Nigeria and another in Uganda recorded coverage in a handful of villages over at least 5 years (Emukah et aI.,2004; Ndyomungyenyi et at.,2004).It may be possible to estimate individual compliance from coverage, but unless longitudinal data are available, such estimates will only apply to individual years when coverage dataarc reported. 2 IOf interest once adequate and accurate means are found to measure individual compliance is the need to establish reasons for compliance or non-compliance. Few studies and unpublished reports have shown that in a good number of communities, women are excluded from annual ivermectin treatment for various reasons. Studies showing gender difference in coverage imply possible longer term problems with compliance (Brieger et a|.,2002; Maduka er aI.,2004), but this question needs to be tested through properly designed research. The implication of this is that there currently is ambiguity in the rate of annual compliance to ivermectin treatment by eligible persons and communities. This ambiguity needs to be examined in order for APOC to take corrective measures that would enhance the likelihood of eliminating onchocerciasis as a public health problem in Africa. Issues of Comnliance Studies on medication compliance usually focus on regimes for long term or chronic conditions ranging from high blood pressure to HIV/AIDS. Such regimens usually involve daily medication consumption, multiple drugs and complimentary dietary and behavioral changes. Although control on onchocerciasis requires only an annual dose of ivermectin, this must continue for at least l5 years, and thus, onchocerciasis is a chronic as well as a transmissible disease. Some lessons from the compliance with treatment for other chronic conditions may be relevant. Issues of drug tolerance, adverse reactions and side effects are commonly documented deterrents to adherence (Weiss et a|.,2005; Pulliam et a|,2003; Trotta et a1.,2002). Though expected to decrease over time with ivermectin use, side effects still may deter initial acceptance and set the stage for later non-compliance. Another set of compliance issues is medicine perceptions including preferences for alternatives and a questioning of the need for medicines (Pound et a|.,2005). This is especially important when many of those having onchocerciasis may not perceive themselves as sick or are satisfied with traditional therapies. Studies have also documented minority ethnic status as a factor influencing compliance (Cramer and Pugh, 2005). Involvement of family members has shown a positive effect on compliance (Elliot, 2003). Although long-term compliance studies have not been undertaken concerning CDTI, cross-sectional studies of recent compliance exist. In Uganda it was reported that a variety of factors were associated with taking ivermectin including trust in the Community Directed Distributor, family support and perceptions that ivermectin was available at no cost (Nuwaha er a|.,2005). While pregnancy is an exclusion criteria for ivermectin treatment, coverage, and hence current compliance, for women is often lower than for men even when they are not pregnant, Brieger et aI.,2002; Maduka et aI.,2004). Minority ethnic status is a double edged factor. When persons are of minority ethnic status within a village, their consumption of ivermectin may be low, but when they live together in a homogeneous village even if they are minorities within a district, their coverage may be greater than the indigenous population (Brieger et o1.,2002). Community level variables may influence compliance wherein larger settlements/villagers have lower coverage (Vanamail et aI.,2005; Brieger et a1.,2002). J Research Ouestions and Obiectives The study (of which Phase I meanwhile is proposed herein) is designed to determine the rate of compliance of eligible individuals and communities where ivermectin distribution has been undertaken for upwards of 7 years, with particular attention to the effect of gender, youth and membership of a minority group. Seven years is the maximum duration of CDTI projects at the time of this writing. The need to focus on this longer duration arises from study aims to learn whether perceived benefits and compliance may be linked. Perception of immediate benefits such as worn expulsion, have been documented (Akogun et a|.,2000), but there is need to learn how perception of benefits related directly to onchocerciasis might influence sustained compliance over time. Research on the dermatological effects of ivermectin treatment have shown that while improvements in severe itching may result after about two years, intervention durations of 2-5 years have been inconclusive concerning improvements in onchocercal skin disease (Brieger et a1.,1998; 2001). Prior to undertaking a full study mentioned-above it is necessary to determine whether adequate and accurate data exist in the field from which compliance calculations can be made. Such data sources potentially include district level registers and summary forms, Community Directed Distributor (CDD) registers and recall by individual community members who took ivermectin. It is assumed that a combination of CDD record and individual recall will be needed to validate each other. Therefore, a feasibility study was undertaken in project sites that have run for seven years or whose year of first CDTI distribution was 1998. The feasibility study was designed to determine whether adequate, continuous and quality data are available to form the basis of a detailed study on compliance rates and factors that influence those rates. This information can be used to improve compliance as distribution continues into the foreseeable future and to enhance the likelihood of onchocerciasis control using ivermectin. Field work in the Ibarapa Central Local Govemment Area of Oyo State Nigeria, which has been running CDTI since 2000, was undertaken during the period of revising this protocol and provided insights into the kinds of issues researchers will likely encounter in finding feasible study sites (Appendix l). A workshop held in Ouagadougou between 28 February and 5 March 2005 raised research questions related to the study of compliance with annual ivermectin treatment over a period of approximately seven years since the inception of the first 25 projects. It was soon realized that these could not be adequately addressed until the first question was answered. It is the purpose of this newly designed Phase I or feasibility study to answer the first research question. This feasibility study should also provide preliminary answers to be used to design a Phase II study. Research Ouestion for Phase I Pilot Studv What is the feasibility of measuring compliance to long-term treatment with ivermectin? a. Can we measure compliance to long-term treatment with ivermectin? b. What are the available sources of information and how reliable are these? 4 Studv Obiectives The General Objective of the study was To determine the feasibility of measuring compliance of individuals and communities to annual treatment with ivermectin in onchocerciasis endemic communities. Specific Objectives included the following: l. To review available records at district and community levels to ascertain the existence of accurate information on community and individual compliance to treatment with ivermectin 2. To determine whether an adequate number of accurate records exist at community level to measure compliance to treatment with ivermectin 3. To calculate levels of community compliance in project sites among communities enrolled since inception of the CDTI project (1997/98) 4. To assess the accuracy of community records by age, gender and minority populations Key Definitions Village Compliance with CDTI occurs when the village members have arranged an annual distribution of ivermectin consistently on an annual basis since the inception of the project. o If a village does not comply in a given project year, it would be difficult for individual residents to comply o Villages would also not be in a position to comply if the district level organizers did not provide the ivermectin in a given year o Villages that miss one or more years of distribution for whatever reason would be excluded from the individual compliance study o Village compliance rate would consist of the proportion of villages that were enrolled at the inception of the project (1998) and who have consistently arranged CDTI annually since that time o Availability of ivermectin annually on a continuous, year-by-year basis is an important enabler for regular individual compliance. Individual Compliance is normally defined as consistent following of medical recommendations including taking medications as prescribed and maintaining a dietary and activity regimen as recommended by a qualified health worker. o In the case of annual ivermectin distribution, the process of compliance is simply defined as taking ivermectin at each annual distribution since the inception of CDTI in the village/settlement, unless the person does not meet eligibility requirements in a given year. o For the purpose of calculation, it will be necessary to work from a base of persons who were eligible by age and height to take ivermectin on the date of the first distribution. 5 o Non-compliance involves missing an annual distribution for reasons of refusal or absence, but not because of ineligibility (e.g. being pregnant or very sick at that distribution). o Note that if people in a village did not take ivermectin in a particular year because the medicine was not made available in the village, such a village would be excluded from the study as its residents were denied the chance to try to comply. o Residents in qualified villages would therefore have had an opportunity to take seven doses of ivermectin. o Taking all seven doses on an annual basis would be considered as full compliance. o Individual compliance rate could be calculated as the number of times out of seven that the person took the annual dose. . Based on the desire to achieve 65% minimum annual coverage in a project site, an individual would be expected to have taken at least five of the seven doses. o For categorical analysis, one could define full compliance as 7 doses, minimal compliance as 5-6 doses, and non-compliance as 0-4 doses. METHODS Ponulation Since compliance is best measured over time instead of at one or two individual distributions, it was decided to base the study in project sites that afforded the maximum potential opportunities for individuals and communities to take ivermectin. Altogether 25 CDTI projects started in late 1997 and 1998 (Appendix l) giving a potential of seven annual distributions by the start of this study. Interestingly, this is almost half the number of years foreseen in the original goal of maintaining annual treatment for l5 years. The period also allows adequate time for positive dermatological and clinical benefits to be observed and perceived, thus enabling researchers to address the important issue of how perceived benefits might influence compliance. The chart below addresses sampling and instrumentation issues. Procedures and Materials Study instruments were developed for the various phases in the selection process chart and are found in Appendix l. The first step is represented in Table I and includes a list of 18 projects inaugurated in 1997-98 and approved for the study by APOC. The Step 2 instrument was used at the project level to obtain information from staff on districts/local governments/divisions that had actually undertaken ivermectin distribution from the inception of the project. Divisions on this list that had the required number of annual distributions were used as a frame for sampling five Divisions. The Step 3 instrument repeated the 'screening' process for communities and villages within the Division. Ten communities where records indicated that distribution had taken place over the requisite number of years were chosen from each Division, yielding 50 communities per project. 6 Step 4 had two instruments, one to review the CDD records and the second to obtain information from villagers on number of annual treatments received that was compared with CDD records. Step-Down Process of Site Selection STEPS Instruments 1. List CDTI projects that have offered ivermectin through CDTI consistently q an a!4!lal basis since 1998. List from review of APOC records Checklist of division names and years distributed 3. For each division list villages or settlements that have Checklist of collected ivermectin consistently on an annual basis village names, since 1998. If the number of subdivisions exceeds six, population and sample five for further stud 4. Within a sample of 50 villages determine the Checklist of proportion that has accurate records for five, six overall quality and seven continuous years. Accuracy should be issues, based on whether all years have been entered and Recording form whether sampled villagers concur with what is for interviews entered against their names with a sample of villagers Calculate potential population in villages with accurate records based on findings in steps 3and 4 6. Project sites that successfully pass through the above screening may be considered for inclusion in the main study Field work was undertaken by a site principal investigator (PI) who hired one or two research assistants. In the field it became apparent that there was also need to bring on board members of the local health team as guides. The site PI was ultimately responsible for all aspects field coordination and data management. Each PI with the site team also compiled the the daily field experiences into a diary/report from which logistical and programmatic lessons were extracted to guide further study. 2.List districts, local governments or other appropriate divisions within these projects that have offered ivermectin through CDTI consistently on an annual basis since 1998. 5. Ascertain the number of villages in the project site with accurate records for a minimum of five years 7 Table l: APOC Projects from 1997198 (Step 1) - Approved Projects For The Study S/N Country CDTI Project Launched NGDO Comment CameroonI 2 J 4 Adamaoua II North Province South West I Centre 3 Nov-98 Nov-98 Aug-98 Dec-98 IEF GRBP SSI HKI Included Included Included Included 5 Chad Chad Jan-98 OPC AFRICARE Included 6 Sudan Northern Sector May-97 GRBP Not Used 7 8 9 l0 ll t2 t3 t4 Nigeria Kaduna Taraba Cross River Kogi Osun Imo Abia (originally w/lmo) Zamfara SSI CBM, MITOSATH UNICEF SSI I.JNICEF GRBP GRBP SSI Jan-98 Nov-97 Feb-98 Feb-98 Aug-98 Sep-98 Sep-98 Oct-98 Included Included Included Included Included Included Included Included 15 Tanzania Ruvuma Sep-e8 SlI Included t6 t7 l8 Uganda Uganda Phase I Uganda Phase II Uganda Phase III May-97 Sep-98 Dec-98 SSI, GRBP SSI, GRBP, GTZ GRBP, CBM Included Included Included 8 Figure I Total and Consecutive Distributions in Divisions (N=121) 80 70 P60 .9 :3 50 o 840 Sso 220 10 0 <6 b Tires Distributed 7 I Total 6 Consecutive 7'.t 71 ,ae RESULTS Seventeen project sites were actually examined in five countries between August and November 2005. These included Ruvuma inTanzania and the first three phases of districts in Uganda. Cameroonian sites included Adamaoua, Centre, North and South West, while Nigerian sites were based in Abia, Cross River, Kaduna, Kogi, Imo, Osun, Taraba and Zamfara States. Data were completed with results from Chad. The presentation of results begins with examination of adequacy of division and community level distributions over the period of study. Within the 17 project sites, researchers visited l2l divisions (districts/local governments) and enumerated 984 communities. The first aspect of analysis focuses on Steps 2and3 ofthe study design. Emphasis was placed on consecutive distributions occurring from the most recent (2004) backwards as a key factor to facilitate calculating compliance. Figure I shows that in the period under study 58.6% of divisions undertook distribution in all seven possible years, while 34.7% of divisions had at least 6 consecutive distributions including 2004. None had 5 consecutive d istributions. Table 2 examines consecutive distribution by among the 17 projects. Here a minimum of 6 consecutive distributions was considered acceptable (there were none with 5), and l5 projects scaled through step 2, leaving Ruvuma inTanzania and Adamaoua [I in Cameroon behind. It was decided to be strict in setting the cutoff at 6 consecutive distributions at Step 2 to give adequate opportunity for villagers to comply. Among the 984 communities on record, only 29.0% had seven consecutive distributions, followed by 21.7%o with at least six. It was decided at Step 3 to be a bit more liberal and include those with 5 consecutive distributions (14.7o/o) because by the time the full compliance study takes place, these areas would have had opportunity to hold another one or two distributions. Therefore, 65.4% of communities would qualify for inclusion in further study as seen in Figure 2. 9 Table 2: Annual Consecutive Distributions by Project and Division (Step 2) Project Annual Consecutive Distribution in Divisions Include Step 2<6 6 7 Total Eligible Abia 0 8 0 8 8 Yes Adamaoua II J 1 0 4 I No Centre III I 5 0 6 5 Yes Cross River 0 0 t4 t4 t4 Yes Imo 0 t6 0 l6 t6 Yes Kaduna 0 0 7 7 7 Yes Kogi 0 0 5 5 5 Yes North 0 6 0 6 6 Yes Osun 0 I l3 t4 t4 Yes Phase I 0 0 4 4 4 Yes Phase 2 0 0 aJ 3 J Yes Phase 3 0 0 J J J Yes Ruvuma 4 0 0 4 0 No South West I 0 5 0 5 5 Yes Taraba 0 0 t2 t2 t2 Yes Chad 0 0 5 5 5 Yes Zamfara 0 0 5 5 5 Yes Total 8 42 7l t2r ll3 (e3.4%) Figure 2 Total Consecutive Distributions in Communities (N=984) al,o Etr3 E E oo 450 400 350 300 250 200 150 100 50 0 <5 56 Tirnes Distributed 7 I Total u Consecutive 139 145 Table 3 examines consecutive distribution by project site. In this, three of the 17 projects lack adequate communities to be included in further study of compliance. Joining those that dropped out in Step 2 is Osun State in Nigeria. A note of caution should be observed. It was only by reducing consecutive distributions to five and above that the three Ugandan and the Kaduna projects could scale through Step 3. It is essential therefore, to ensure that subsequent sequential distributions did occur in 2005 and possible 2006 before including these sites in the upcoming full compliance study l0 Table 3: Annual Consecutive Distributions by Proiect and Community (Step 3) Project Annual Consecutive Distribution in Communities Percent Elieible Include Step 3<5 5 6 7 Total Elieible Abia 0 0 20 5 25 25 100.0 yes Adamaoua II 45 0 4 0 49 4 8.2 no Centre III 9 I 4l 0 5l 42 82.4 yes Cross River 0 0 0 80 80 80 100.0 yes Imo 0 0 2t 40 6t 6l 100.0 YES Kaduna l3 t7 l0 l3 53 40 75.5 yes Kogi l0 6 l3 l8 47 3t 78.7 yes North 0 0 50 0 50 50 100.0 yes Osun 68 0 0 I 69 I 1.4 no Phase I 7 4l 0 0 48 4t 8s.4 yes Phase 2 aJ 46 0 0 49 46 93.9 yes Phase 3 t6 34 0 0 50 34 68.0 YES Ruvuma 160 0 0 0 160 0 0.0 no South West I 0 0 50 0 50 50 100.0 yes Taraba 0 0 0 39 39 39 100.0 yes Chad 0 0 0 53 53 53 100.0 yes Zamfara 9 0 5 36 50 4t 82.0 yes Total 343 r45 214 289 984 648 65.4 These two steps so far demonstrate that while it is necessary to confirm distribution took place in various divisions, districts and local governments, this is not sufficient to determine whether adequate work had been done in terms of community distribution. The next aspect of the results focuses on what happened at the community level. In Step Four 852 communities/villages were visited and two procedures were completed. First was a review of CDD records. Second was interview with villagers to determine their treatment status and experience and compare these with the CDD records. Table 4 shows that CDD records were in basically devoid of problems such as not listing age, sex and dosage received. Some problems were found concerning missing persons (12.4% of records) and inclusion of persons not resident in the settlement (10.9%). Unavailability of records made it difficult to determine these items in a few cases. Table 4: Problems Seen in CDD Records Problem Number of Records Percent of Records Missing Persons 106 12.4 Non-Residents Included 93 10.9 Age Recorded 845 99.2 Sex Recorded 847 99.4 Dosage Recorded 841 98.7 N: 852 villages ll Figure 3 looks at CDD records on an annual basis. At the beginning of the study period only 22.3% of the 852 villages indicated that distribution had taken place,50.90 clearly showed that there was no distribution, while 25.2% of records for 1998 could not be obtained. A steady improvement in these values is seen over time with 9l .2Yo recording distribution in 2004, 6.804 showed that there was no distribution, and only 0.3% could not be obtained for review. Figure 4 and Table 5 revisit the issue of consecutive and total distributions from CDD records as previous information was based on State and Local OCP information. In this case only 50.5% of villages had record of performing at least 5 consecutive distributions. This contrasts with the views in Figures I and 2 where nearly all divisions reported distribution at least six times and two-thirds of communities showed at least five distributions. Based on sampled village level data, only l0 project sites would be eligible for inclusion in a further study on compliance. Abia is borderline with 48.1% of villages sampled and visited being eligible, but the logistics of tracing eligible villages may not warrant inclusion. Figure 3 Annual Distribution lnformation from 852 CDD Records 120.0 100.0 80.0 tr o e o(L 60.0 40.0 20.0 0.0 1998 1999 2000 2001 Year 2002 2003 2004 r Distributed z Not Distrib M NotAvailable ffi WWMffi :50.9 62.0 32.1 22.3 91.285.381.6 74.4 t2 Figure 4 Annual Distributions Based on CDD Records o!, oIod oo o tho E" s 5 450 400 350 300 250 200 150 100 50 0 <5 56 Distributions in 852 Villages 7 I Total a Consecutirc WZ356m ruMA 249 232W*mruru 1)9 mWrzq tro tro -777) Table 5: Annual Consecutive Distribution from CDD Records in Annual Consecutive Distribution in VillagesProject <5 5 6 7 Total Elieible Percent Eligible Include Step 4 Abia 27 8 l4 a) 52 25 48. I no 0 0 0 52 0 0.0 noAdamaoua II 52 Centre III 50 I 0 0 5l I 2.0 no 5 20 t3 50 36 72.0 yesCross River t4 Imo 1l 10 26 I 48 37 77.1 yes Kaduna 14 t7 7 14 52 38 73.1 yes yesKogi 4 3l 2 8 45 4t 9l.l North 49 0 I 0 50 I 2.0 no noOsun 37 5 0 7 49 t2 24.5 Phase I t4 34 0 0 48 34 70.8 yes 4t 83.7 yesPhase 2 8 4t 0 0 49 17 5J 0 0 50 33 66.0 yesPhase 3 4 50 4 8.0 noRuvuma 46 0 0 South West I 2 43 8 0 53 5l 96.2 yes l4 0 0 36 50 36 72.0 yesTaraba 0.0 noChad 53 0 0 0 53 0 Zamfara ll 6 I 32 50 39 78.0 yes 852 429 50.4Total 423 232 79 118 l3 Figure 5 Distribution of 16613 Villagers lnterviewed The instrument for Step 48 compared recallof the number of tablets most recently taken by 20-25 residents in each village with that recorded by the CDD. Villagers recalled receiving between0andStablets, although the maximum usually given is 4. Some CDD records showed up to 7 tablets having been given. Figure 6 shows that there was agreement on number of tablets given/received in 68.5% of cases. CDDs recorded more tablets than villagers recalled in 6.6%o The quantitative information concludes with a comparison of villager reports and CDD records concerning the number of tablets received and the number of times a person was treated. Figure 5 shows that a total of 16,613 people were interviewed for this second part ofStep 4. They were fairly evenly divided among adult males (25.2%), adult females (25.4%), young (i.e. 12-19 years) males (24.6%) and young females (24.2%). Only 96 (0.6%) of people interviewed were ethnic minorities in the village visited. These were found at only seven sites, Ruvuma in Tanzania and Cross River, Kaduna, Kogi, Osun, Taraba and Zamfara in Nigeria. Figure 6 *Not recalled VI and/or not recorded CDD Tablets Recalled by Villagers Compared to CDD Record 80.0 70.0 (t @-(o Iz o e o o- 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Not* Less Same CDD Record Minus Recall More 68.5 %m%mru 4E 1 mm9.2 7%Z 66 of cases, and less for 15.7%o. It was not possible to compare for 9.2%o since either the CDD did not record or the villager did not recall. l4 Actually it is useful to note that for the most recent distribution only l2 of the 16,613 entries did not have number of tablets recorded in the CDD record, compared to 25 villagers who could not recall a number of tablets. Also, in only 72 cases did CDD records not have full information on the number of times a person received ivermectin. This contrasts with 1,139 villagers who could not recall the number of times they received ivermectin. This shows a high level of completeness in CDD records, even if those are less than fully accurate. Further analysis was restricted to the l0 project sites where most villagers had an opportunity to receive ivermectin at least 5 sequential times. Compared to CDD records older villagers were more likely to overestimate the number of tablets they received than younger ones as well as overestimate the number of distributions/treatments they had received over time. Compared to CDD records males tended to overestimate the number of times they were treated, while females tended to underestimate. No sex differences appeared in number of tablets recalled compared to CDD records. Table 6: Compliance Variation by Project Site Project Number Percent Comply 5 or More Times Cross River, Nigeria 997 42.3 lmo, Nigeria 1000 48.3 Kaduna, Nigeria 953 66.9 Kogi, Nigeria 849 73.7 Uganda I 886 88.0 Uganda II 973 72.4 Uganda III 923 57.3 South West, Cameroun 1029 51.9 Taraba, Nigeria 1005 58.9 Zamfara, Nigeria 969 68.3 ALL 9s84 62.3 Table 7 indicates that males (64.6%) were more likely to have taken ivermectin at least 5 times than did females (60.6%) in the l0 project sites. Likewise older villagers had better compliance (65.5%) than those aged l2-19 years (59.0%). These findings show a similar pattern to many previous coverage surveys. These recall issues combined with the good level of completeness of CDD records led to the use of CDD records to calculate compliance. Table 6 shows the variation in compliance among the l0 sites. Note that unlike analysis of previous records, it was not possible to judge whether people received treatment in consecutive years. Therefore, reports in CDD records of having received ivermectin 5 or more times is a proxy Table 7: Factors Associated with Compliance 5 Times Factor Number Percent Comply trlA Yates, p value Age Adult 4910 65.s 42.31 t2-19 4642 59.0 0.000000 Sex Male 4791 64.6 21.28 Female 4740 60.6 0.000004 measure for compliance. Five-time compliance varies from a low of 42.3%o in Cross River, Nigeria to a high of 88.0% in Uganda Phase L The overall average among sites was 62.3%. Overall it is expected that continuous coverage of 650/o of the total village population is needed to eliminate onchocerciasis. Although coverage is not compliance, it would have been good if over 75Yo of persons interviewed had complied at least 5 times. l5 Finally, information was collected through field notes on the actual research experience. These are also considered 'data' due to the nature of the research question which related to the feasibility of a larger scale study. Concerns highlighted in the reports fell into three broad categories as seen in Appendix 2. Some of the most common factors that were seen as influencing the quality of data included the following: o Reluctance of some CDDs to cooperate because of perceived neglect; other health programs give incentives o Older records removed from village so coverage difficult to ascertain o CDDs who had dropped out gave village records over to health workers so no information is available in the village o District/LGA level records lacking or incomplete for certain years . When CDDs not present, difficult to locate village records o In order to achieve sample size, mop-up exercises are crucial as is the need to work in the evening and on weekends Logistics posed a challenge in conducting the study in many locations. The most common factors, seen below, likely have implications for general management of the projects. These factors also have implications for the cost of running the follow-on compliance study. o Need for local staff to accompany team for familiarity and acceptance, but reluctance of official state and LGA staff to accompany team without remuneration even though CDTI part of official duties . Staff transfers make it challenging to find staff who know the program and the location of villages o Difficulty in locating CDDs, hence, need for advance village mobilization to ensure CDDs will be available o Courtesy calls on LGA officials necessary but time consuming The most common programmatic problems encountered in the field are seen below. Other interesting insights and concerns are found in Appendix 2. o Poor maintenance of project vehicle implying lack of field supervision for a long time (Researchers had to make repairs before field work) o Staff transfers and lack of continuity due to lack of orientation to CDTI on resuming in the new District o Motivational problems with CDDs related to lack of remuneration/incentive (Other programs give stipends) o Motivational problems with staff related to lack of remuneration/incentive o CDD records vary widely in quality implying lack of direct supervision, monitoring o Lack of funds from LGA or State to facilitate and sustain program o No drugs for distribution or drug shortage t6 DISCUSSION The step-down process of determining feasibility was quite valuable. A record at Project or State level indicating that distribution occurred in a Division level could only be confirmed at that next level. Likewise at Division level, records may show that village level distribution occurred, but on reaching the village, researchers discovered that anomalies like missing CDDs and missing records could render further investigation impossible. The concerns enumerated in Appendix 2 thus, have resonance not only for compliance research but also for smooth and accurate functioning of CDTI within each project. Fortunately, adequate and complete CDD records were available in enough project sites to encourage further study. The key lesson from this feasibility study is that it is possible to proceed to a full fledged compliance study as defined during the first protocol development workshop in early 2005. The first decision in revising that protocol would involve determining a cutoff point for measuring compliance. Records at all levels, project, division and community are very poor or non-existent for the first two years. A greater return would be achieved by defining compliance as taking ivermectin for five continuous years. Instruments for a detailed study would need to incorporate places to enter actual year-by-year data from CDD records for samples individuals. At this juncture, at most l0 projects could be included in a 5-year compliance study. The study has also shown the importance of CDD records for veriffing compliance. Generally people tended to underestimate or forget whether they were treated. The use of CDD records to calculate compliance is essential for the future study, though issues of recall are still of interest. Finally, the study does give a hint of possible factors that may influence compliance including age and sex. There were too few villagers with a different ethnicity than the main residents to analyze the effect of ethnicity in this feasibility study. A future study may need to make a more concerted effort to identiff and include ethnic and other minorities. As noted, it will also be necessary to design the research instrument for the next study to provide year-by-year information copied out from the CDD record for each person being interviewed in more detail about his or her compliance. In that way all persons will be measured using the same time span as standard. Since distribution for 2005 would have taken place by the time the full compliance study could take off, and since the year 2000 was seen as the time when most projects had come around to having good records, it may in fact be possible to look at six years of compliance for that study. The field reports from each team were quite helpful in interpreting the results and portraying the challenges in undertaking a broader and more detailed compliance study. Appendix 2 outlines a number of programmatic problems, especially in terms of record keeping that need to be overcome for studying compliance over five or more consecutive years. The survey shows clearly that CDD records must be used as a standard and that recall, while an interesting study subject, cannot be the sole basis for judging compliance. Among the record keeping problems are the removal of records from the village, the use loose treatment sheets instead of registers/notebooks, and the loss of records when CDDs and staff change. t7 In closing it is worth noting that even where it was possible for villagers to comply at least five times, only 62%o did so. Logistical problems like drug availability, poor rup..uirion, CDD reluctance without incentives and accessibility of villages offer-some explanation for this poor level of coverage. Researchers came across examples of people who were still reluctant to take the drug, indicating that CDTI staff and CDDs were nor living up to expectations for health education. Therefore, while this feasibility study concludes that a more deiailed compliance study is possible and desirable, the broader implications for the CDTI program is that greater effort is needed if onchocerciasis is to be eliminated from the African continent. REFERENCES Akogun, O.B., M. K. Akogun, et al. Community-perceived benefits of ivermectin treatment in northeastern Nigeria. Soc Sci Med 2000; 50(10): 145l-6. Brieger WR, Awedoba AK, Eneanya, CI, Hagan M, Ogbuagu, KF, Okello DO, Ososanya OO, Ovuga BL, Noma M, Kale OO, Burnham GM, Remme JHF. The effects of ivermectin on onchocercal skin disease and severe itching: results of a multi-centre trial. Tropical Medicine and Internationol Health 1 998; 3(1 2) : 1020-1027 . Brieger WR, Kale OO, Ososanya OO. Development of reactive onchocercal skin lesions during a placebo controlled trial with ivermectin among persons without lesions at baseline. Tropical Doctor200l;31: l-3. Brieger WR, Otusanya SA, Oke GA, Oshiname FO, Adeniyi, JD. Factors associated with coverage in community directed treatment with ivermectin for onchocerciasis control in Oyo State, Nigeria. Tropical Medicine and Internationol Health2002;7(l): I l-18. Cramer JA, Pugh MJ. The influence of insulin use on glycemic control: How well do adults follow prescriptions for insulin? Diabetes Care. 2005;28(l):78-83. Elliott WJ. Optimizing medication adherence in older persons with hypenension. Int Urol Nephrol. 2003 ) 5 @) :5 57 -62. Emukah EC, Osuoha E, Miri ES, OnyenamaJ, Amazigo U, Obijuru C, Osuji N, Ekeanyanwu J, Amadiegwu S, Korve K, Richards FO. "A longitudinal study of impact of repeated mass ivermectin treatment on clinical manifestations of onchocerciasis in Imo State, Nigeria", Am J Trop Med Hyg 2004; 70(5): 556-61. Maduka CU, Nweke LN, Miri ES, Amazigo U, Emukah EC, Richards FO. Missed treatment opportunities, for pregnant and breast-feeding women, in onchocerciasis mass-treatment programmes in south-eastern Nigeria. Ann Trop Med Parasitol.2004;98(7): 697-702. l8 Ndyomugyenyi R, Tukesiga E, Buttner DWQ, Garms R., "The impact of ivermectin treatment alone and when in parallelwith Simulium naevei elimination on onchocerciasis in Uganda", Trop i c al me di c ine and Inte rnat i onal He alth 2004 ; 9(8) : 8 82-886. Nuwaha F, Okware J, Ndyomugyenyi R. Predictors of compliance with community-directed ivermectin treatment in Uganda: quantitative results. Trop Med Int Health. 2005: 10(7):659-67. Plaisier AP, Alley ES, van Oortmarssen GJ, Boatin BA, Habbema JDF. "Required duration of combined annual ivermectin treatment and vector control in the Onchocerciasis Control programme in West Africa", Bulletin of the World Health Organizotion 1997; 7 5(3):237 -245 Pound P, Britten N, Morgan M, Yardley L, Pope C, Daker-White G, Campbell R. Resisting medicines: a synthesis of qualitative studies of medicine taking. Soc Sci Med. 2005 Jul;61(l): 133-55. Epub 2005 lan 26. Pulliam C, Gatchel RJ, Robinson RC. Challenges to early prevention and intervention: personal experiences with adherence. Clin J Pain. 2003; l9(2):ll4-20. Trotta MP, Ammassari A, Melzi S,Zaccarelli M, LadisaN, Sighinolfi L, Mura MS, d'Arminio Monforte A, Antinori A; AdlCoNA Study Group. Treatment-related factors and highly active antiretroviraltherapy adherence. J Acquir Immune Defic Syndr.2002 Dec l5;31 Suppl 3:S128- 31. Vanamail P, Ramaiah KD, Subramanian S, Pani SP, Yuvaraj J, Das PK. Pattern of community compliance with spaced, single-dose, mass administrations of diethylcarbamazine or ivermectin, for the elimination of lymphatic filariasis from rural areas of southern India. Ann Trop Med Parasitol. 2005 Apr;99(3):237 -42. Weiss M, Vered I, Foldes AJ, Cohen YC, Shamir-Elron Y, Ish-Shalom S; Israeli Alendronate OW Study Group. Treatment preference and tolerability with alendronate once weekly over a 3- month period: an Israeli multi-center study. Aging Clin Exp Res. 2005;17(2):143'9. l9 APPENDIX 1: APOC Compliance Feasibility Study Instruments Appendix 1A: STEP2 Instrument SUBDIVISION DISTRIBUTION CHECKLIST Country: Project Name/Site: Subdivision (District, LGA, etc.) 1998 1999 2000 2001 2002 2003 2004 2005 Total Times Distributed 20 Appendix 18: STEP 3Instrument COMMUNITY DISTRIBUTION CHECKLIST Project Name/Site: Countr), Subdivision (District/LGA) Name Community Current Population 1998 1999 2000 2001 2002 2003 2004 2005 Total Times Distributed 2t Appendix lC: STEP 4A Instrument QUALITY OF CDD RECORDS _ OVERALL QUALITY Project Name/Site: LGA/District: Community Name: Village Name (Places covered by CDD): Component Yes No Comments Completeness of Years (State reasons for each year missed) 1998: 1999 2000 2001 2002 2003 2004: Completeness of Persons Listed (Number of missing persons) Non-Residents Listed (Number of non-residents listed) Age Information Included Gender Information Included Full Data on Dosage Listed Minority Residents Included Additional Comments by CDD and Community Members 22 Appendix lD: STEP 48 Instrument - Community Member Validation Coun Project Name/Site: District/LGA: Community Name Village Name (Places covered by CDD): Child = l2-19 Years (i.e. minimum 5 years old at inception of project plus seven years of project duration); Adult = 20+ Years; *For minority, note in the parentheses the Gender: M/F and the Age: A/C HH/ID (A) Category (B) Number of Tablets Recalled Last Time(c) Number Tablets in CDD Record (D) Differ- ence between (c) & (D) (E) Number of Times Recalled (F) Number of Times in CDD Record (G) Differ- ence between (F) & (G) (H) Remark about discrepa ncies (t) Adult Male Adult Male Adult Male Adult Male Adult Male Adult Female Adult Female Adult Female Adult Female Adult Female Child Male Child Male Child Male Child Male Child Male Child Female Child Female Child Female Child Female Child Female Minority (" ) Minority ( ) Minority ( ) Minority ( ) Minority ( ) 23 Appendix lE: Selection Process: l. Determine a systematic sampling interval. If, for example, there are 26 households in the community and one is selecting 5 adult males, the sampling interval would be 2615 or 5 (rounded off). One would then ballot for the first household in the register by picking a number between one and 5. If one chose, for example household #3 then the subsequent ones would be #8, #13,#18 and#23. 2. After selecting the households from the register one finds more than one adult male listed, one would ballot for the one to be interviewed. 3. For subsequent categories (adult female, child female, child male), pick a new household number from among the first five and continue as above. 4. If a selected person is unavailable, chose another person in the same category within the selected household, or go to the next household in sequence. 5. Minority is defined as any group of people resident within a village or community who is disenfranchised or marginalized because of ethnicity, migration, displacement or other special characteristic. Such people are not in the mainstream (not integrated into the) social, cultural and economic activities of the community It may be more feasible to select one route in the community randomly, and following this route include any eligible person along the route in the study. This would reduce the cost of looking for persons selected by the earlier process proposed and described above. With that process, one is forced to constantly revisit the sampling process in the field where the person so selected from the register is not available in the community. It also place some logistic difficulties in the sense that the research assistants have to consult with the researcher at each point to resample. This also has implication for time and transport costs in the field. 24 t aAPPENDIX 2: CDTI Feasibility Study Factors Affecting Data and Programming KEY: AB:Abia (Nigeria), IM=Imo (Nigeria), OS=Osun (Nigeria), KA:Kaduna (Nigeria), KG:Kogi (Nigeria), AD=Adamaoua (Cameroun), SW:South West (Cameroun), NO=North (Cameroun), CB:Central (Cameroun), UG=Uganda (Phases l-3 combined),TZ:Tanzania, CR:Cross River (Nigeria), ZA=Zamfara (Nigeria), TA=Taraba (Nigeria), TC:Tchad A. Data QualiW lssues SITES mentioning Number 1. Need to avoid some LGAs/communities because of social/political volatility - raises questions about validity IM,OS 2 2. Reluctance of some CDDs to cooperate because of perceived neglect; other health programs give incentives AB,OS,KA,KG,AD,SW, NO,CN,CR,TA,ZA 11 3. When CDDs not present, difficult to locate village records AB,IM,OS,KG,CR 5 4. Other villagers unlikely to be able to interpret records in absence of CDD AB,OS,KG 3 5. CDDs unavailable due to farming, work KA,7A 2 6. ln order to achieve sample size, mop-up exercises are crucial as is the need to work in the eveninq and on weekends AB,OS,SW,NO,ZA 5 7. CDDs who had dropped out gave village records over to health workers so no information is available in the village AB, IM,OS,KG,AD,SW,CN,TZ 8 8. Older records removed from village so coverage difficult to ascertain AB, IM,OS, KG,AD,SW,NO, CN,UG,CR,TC,TA,ZA 13 L DistricVLGA level records lacking or incomplete for certain years AB, I M, KA, KG,SW, NO, UG, CR,TC 9 10. Difficult get required number of school age children as school not near AB,IM,UG,TZ 4 11. CDDs coach villagers to recall number of tablets AB,KG,ZA 3 12. People were frightened when strangers came to collect information AB,CR,ZA 3 13. CDDs devolved to lower level (e.9. kindred) affecting coverage and continuity of and access to records AB 1 14. CDD tried to avoid team when did not have the register, had kept cost recovery funds, etc. AB,CR,TC,TA 4 15. Records of some households and minorities missed; young people, recently married women not added, transferred KA,ZA 2 16. District staff transfer, death affects handover and/or distribution schedule KA,TC,TA 3 17. Loss, Careless handling, storage of CDTI forms KA,7A 2 18. Recall may have been affected by previous mass treatment TC 1 19. CDDs told not to keep previous years'forms TC 1 20. Use of treatment sheets instead of booklets, registers makes record review and tracing of villaqers difficult TC 1 21. Difficult to interviewer women in purdah TA,7A 2 25 ( a CDTI Feasibility Study Factors Affecting Data and Programming KEY: AB:Abia (Nigeria), IM:Imo (Nigeria), OS:Osun (Nigeria), KA:Kaduna (Nigeria), KG:Kogi (Nigeria), AD=Adamaoua (Cameroun), SW:South West (Cameroun), NO:North (Cameroun), CB:Central (Cameroun), UG:Uganda (Phases 1-3 combined),TZ:Tanzania, CR:Cross River (Nigeria), ZA:Zamfara (Nigeria), TA:Taraba (Nigeria), TC:Tchad B. Logistical lssues During Data Collection Sites Mentioning Number 1. Communication challenges relating to lack of or non-recognition of e-mail and cell phone communication IM,OS,TX 3 2. Need for local staff to accompany team for familiarity and acceptance, but reluctance of official state and LGA staff to accompany team without remuneration even though CDTI part of officialduties IM,OS,KG,SW,NO,ZA 6 3. Staff transfers make it challenging to find staff who know the program and the location of villages AB,IM,OS,KG,AD,NO,TZ 7 4. Difficulty in locating CDDs, hence, need for advance village mobilization to ensure CDDs will be available AB,IM.OS,KG,SW,CR 6 5. Courtesy calls on LGA officials necessary but time consuming AB,IM,OS,KG,ZA 5 6. Complaint from LOCT and SOCT on need for remuneration to assist IM,OS 2 7. Communities not accessible, rains, roads NO,ZA 2 8. Researchers had to repair CDTI project vehicles before embarking on work OS,TC,TA 3 9. Higher fuel prices farther from base TC 1 26 Ia CDTI Feasibility Study Factors Affecting Data and Programming KEY: AB:Abia (Nigeria), IM:Imo (Nigeria), OS:Osun (Nigeria), KA:Kaduna (Nigeria), KG=Kogi (Nigeria), AD:Adamaoua (Cameroun), SW:South West (Cameroun), NO:North (Cameroun), CB:Central (Cameroun), UG:Uganda (Phases l-3 combined),TZ:Tanzania, CR:Cross River (Nigeria), ZA:Zamfara (Nigeria), TA:Taraba (Nigeria), TC:Tchad C. General CDTI lmplementation lssues Sites Mentioning Number 1. Poor maintenance of project vehicle implying lack of field supervision for a long time (Researchers had to make repairs before field work) OS,SW,NO,TC,TA,ZA 6 2. Staff transfers and lack of continuity due to lack of orientation to CDTI on resuming in the new District OS,TC,TA,ZA 4 3. Replacements not made for'retired' CDDs or those who dropped out OS,AD,CN 3 4. Motivational problems with CDDs related to lack of remuneration/incentive (Other programs give stipends) OS,KA,AD,SW,NO,CN,TA 7 5. Motivational problems with staff related to lack of remu neration/incentive OS,AD,SW,NO,CN 5 6. CDTI record keeping appears to be sole purview of CDD, other villagers not aware, involved, nor do they understand how records are kept when the see them OS,SW,NO,CN 4 7. CDD records vary widely in quality implying lack of direct supervision, monitoring OS,AD,SW,NO,CN,ZA 6 8. The five-year attainment of CDTI project gave staff of MOH and NGDO the impression that the proqram was finishinq OS,TC 2 9. Lack of funds from LGA or State to facilitate and sustain proqram KG,TA,ZA 3 10. CDD sold drug leading to refusals NO ,| 11. CDDs formed 'union' and embarked on strike over allowances AB 1 12. Political tussle over traditional leadership divided the community AB 1 13. Distribution changed locally to kindred level such that not allvillagers had access AB 1 14. State policy makers very involved in CDTI (positive factor) AB 1 15. No drugs for distribution or drug shortage t<A,TZ,7A 3 16. Need for refresher traininq rz 1 17. Need for better sensitization on benefits of ivermectin AD,TZ 2 18. Some supervisors were not mobile IM,KG 2 19. Poor community support KA 1 20. poor supervision, wrong dosages, e.g. 6 tablets for adults 7A 1 27 ?o