Consolidated guidelines on person-centred HIV strategic information: strengthening routine data for impact: policy brief on harnessing the strength of routine data for HIV surveillance

BACKGROUND Programmatic data routinely collected at service delivery sites can play an important role in the measuring and monitoring of HIV incidence and prevalence, helping to gauge the impact of the HIV response on the epidemic. Recent World Health Organization (WHO) guidance on HIV strategic information addresses this use of routine data for the first time. Broadly, the new Consolidated guidelines on person-centred HIV strategic information: strengthening routine data for impact1 aim to help countries improve how routine patient data are collected, analysed and used. They propose a minimum dataset that captures key events in an individual’s interaction with the health system, recommend priority indicators for monitoring a person’s health, and make key recommendations for data systems and use. In addition to using routine surveillance data to assess programme impact, the guidelines address monitoring of HIV prevention, testing and treatment, HIV-related infections, supplementing routine patient data with data from other sources, and digital health data in HIV services. Key recommendations NEW 1. It is recommended that national health information systems include and strengthen individual-level HIV surveillance that: a) routinely links individual data on HIV prevention, diagnosis and treatment over time as people move between facilities and locations b) provides granular, subnational strategic information for public health action. NEW 2. Collection of a minimum dataset of routine clinical health information is recommended for national surveillance to monitor and guide the HIV response and support measurement of incidence: a) Methods using person-centred data, including back-calculation and retesting, should be considered together with data from other sources and modelling to improve incidence measurement. NEW 3. A CD4 test conducted at HIV diagnosis is recommended for use in clinical staging, providing clinical information on entry or re-entry to care and estimating HIV incidence. NEW 4. Mortality and causes of death (AIDS-related and non-AIDS-related) should be reported for all people registered in routine HIV information systems. Vital registration records should be consulted to measure the overall burden of AIDS mortality, including as a proportion of total deaths. NEW 5. Expanding and strengthening HIV case surveillance systems that use simple electronic interfaces and built-in validation mechanisms is recommended in order to better capture new HIV diagnoses and risk factors for HIV acquisition. 1 Consolidated guidelines on person-centred HIV strategic information: strengthening routine data for impact. Geneva: World Health Organization; 2022. CONSOLIDATED GUIDELINES ON PERSON-CENTRED HIV STRATEGIC INFORMATION: STRENGTHENING ROUTINE DATA FOR IMPACT POLICY BRIEF ON HARNESSING THE STRENGTH OF ROUTINE DATA FOR HIV SURVEILLANCE 2ESTIMATING HIV INCIDENCE, PREVALENCE AND MORTALITY Several indicators are important for tracking the course of the HIV epidemic and the progress of the response. With the growth of digital health systems, it has become possible to expand and interlink existing HIV information systems to routinely capture and link individual data over time. This will improve data quality, simplify reporting and provide actionable data at granular subnational levels. Strengthened HIV surveillance systems involve the reporting of HIV diagnoses and AIDS cases through a standardized reporting system, with additional data elements that can include sentinel paediatric or pregnancy- related events, related infections and deaths, as well as more specific data elements, such as the probable route of HIV acquisition. Together these data elements facilitate an understanding of the distribution of new HIV infections. Using regular, reliable routine surveillance data to track HIV incidence, prevalence and mortality will enhance timely decision-making and the overall national response. HIV incidence can be measured directly, for example, through longitudinal follow-up studies (repeated population-based surveys or prospective cohort studies) or indirectly, for example, through mathematical models. Direct and indirect methods have different advantages and challenges. Among key population groups in high burden settings, estimates of incidence, particularly direct estimates, are difficult to obtain. The lack of incidence data on key populations can impede effective budgeting and programming for these groups. Accurate information on mortality among people living with HIV is critical to understanding the impact of HIV programmes and to inform methods of HIV prevalence and incidence estimation. Deaths among people living with HIV may be reported directly by clinics. These data can be supplemented by national death registration, cross-referencing other data sources where deaths are recorded and verbal autopsy methods. Given the difficulty and sometimes the cost of these measurement methods, using routinely collected programme data is an attractive additional way to track programme impact on the epidemic. HIV incidence is a fundamental measure of the current state of the epidemic and is critical to guiding the programmatic response. Photo: © Wits RHI, Project PrEP, South Africa 2 3 Methods for estimating HIV incidence and prevalence using routine data To demonstrate how routine programmatic surveillance data can help to guide the global, national and subnational response, the methods described in the strategic information guidelines focus on approaches to measuring HIV incidence and prevalence (Table 1). These methods utilize individual-level data (whether paper- based or, preferably, electronic) that are routinely collected in the course of clinical procedures in service delivery platforms for HIV prevention, testing and treatment, national registries and/or programme-driven community and client surveys. Table 1 Methods for estimating HIV incidence and prevalence from routine programme data CD4 back- calculation In settings with concentrated HIV epidemics, back-calculation is often the only method available for estimating the number of newly acquired HIV infections over time, or HIV incidence. The principle underlying back-calculation is that the levels of a biomarker at the time of diagnosis can be used to estimate, or back-calculate, the time since acquiring HIV infection. CD4 cell counts at diagnosis are the most important biomarker used for back-calculation. The rate at which CD4 counts decrease over time is known from research. WHO recommends that a CD4 test is conducted at the time of HIV diagnosis for clinical and surveillance use. Retesting Specific groups of people in high HIV burden settings or individuals with HIV-related risks are encouraged to retest at specified intervals. Data from individual longitudinal testing and retesting (possibly from different health services) could be used to identify people who seroconvert (that is, an HIV-positive test after at least one HIV-negative test), thus aiding in the estimation and interpretation of HIV incidence patterns. To support incidence estimation using routine testing data, better insights are needed into the potential biases and limitations of this approach and how data may vary by setting. Also, there is work ongoing to assess differences in seroconversion rates calculated using different ways to assign date of seroconversion between a last HIV-negative and first HIV-positive test results. Recency assays Recency assays distinguish recent from long-standing HIV infection in an individual using one or more biomarkers, typically by measuring the evolution of the immune response following initial infection. HIV incidence can be estimated accurately from recency assays only when data are collected through representative surveys. In routine programmatic settings, to reduce selection bias, recency results should be interpreted in the context of the population being tested. It is good practice to conduct recency testing among people who are tested for HIV for reasons unrelated to perceived high HIV risk (such as women attending antenatal clinics or military recruits). Recent infection testing algorithms (RITAs) should include additional clinical data to identify false recent results. WHO does not recommend recency testing for clinical management of individuals or their partners. Routine antenatal HIV testing In light of evidence for the effectiveness of antiretrovirals to prevent vertical transmission of HIV and WHO recommendations for “treat all”, HIV testing at the first antenatal care (ANC) visit is now nearly universal in most countries highly affected by HIV. Following WHO guidelines, many countries now implement retesting of women at the final ANC visit or during labour and delivery to identify any seroconversions. HIV prevalence among pregnant women can be related to population HIV prevalence, but systematic differences between pregnant women and the general population must be considered. Routine ANC HIV testing can be used to estimate population HIV incidence in settings with generalized HIV epidemics using mathematical models or recency assays. Inclusion of routine data in modelling Mathematical models can provide comparable measures of HIV incidence over time and assess the impact of HIV interventions. For routine data to be of sufficient quality for use in modelling, robust HIV case reporting is needed, as is good-quality data on the numbers of people on treatment and data from HIV testing among pregnant women in ANC. Each of these can be supplemented by survey data from key populations in concentrated epidemics and from the general population in generalized epidemic settings. Minimum dataset required for estimating HIV incidence and prevalence To use the methods described above to estimate HIV incidence and prevalence, routine HIV surveillance systems will need to collect a minimum standardized set of reportable data elements (Table 2). Many sources can be used to track key or sentinel events and other data elements: patient clinical records, surveillance programmes, ANC clinics, HIV testing services, laboratories, as well as vital statistics registries that include cause of death. 4Table 2 Recommended minimum dataset for estimating HIV incidence and prevalence Area Data elements Potential data sourcesb Use in HIV incidence & prevalence estimation Potential derived indicators HIV diagnosisa HIV-positive test: HIV test date HIV test result HIV testing data (clinical and confirming laboratory results) Pivotal date from which to estimate incidence and for inclusion in prevalence estimates Estimated HIV testing coverage (when combined with all other testing data) Reclassification of new diagnoses as long-term infection (with or without ARV exposure) Previous HIV-negative (or HIV-positive) test: HIV test date HIV test result HIV testing data Date from which some estimates will take a midpoint between this and first diagnosis for incidence estimation The period in which a previous HIV-positive test occurred will determine whether an individual is included in HIV incidence or prevalence calculations Annualized rate of tests/ person/year by key population Previous linkage to HIV prevention services Initial pretreatment disease assessment Baseline CD4 test result at HIV diagnosis Date of CD4 sample collection HIV clinical stage HIV testing data and/or ART register Measure of immune function, CD4 back- calculation Late diagnosis Reclassification of new diagnoses as long-term infection Initiation of ART Date started ART (among pregnant women, whether already on ART or ART started during pregnancy) ART register Time to ART initiation Viral suppression First viral load test result (at diagnosis) First viral load test date ART register Measure of immune function Reclassification of new diagnoses as long-term infection Subsequent viral load and CD4 test results and test dates ART register Treatment failure Any viral load test <1000 per mL ART register Loss to follow- up >28 days since last missed appointment ART register Implications for viral rebound Disease progression Date of first AIDS diagnosis ART register Back-calculation of HIV incidence Abbreviations: ART = antiretroviral treatment a Recency testing can be considered if conducted routinely in a setting as part of a RITA on all new HIV diagnoses and the conditions described in the section above to address potential sources of bias are met. b Electronic medical records can be used where any of these data elements are recorded. 4 5 Table 2 (continued) Recommended minimum dataset for estimating HIV incidence and prevalence Area Data elements Potential data sourcesb Use in HIV incidence & prevalence estimation Potential derived indicators HIV diagnoses among pregnant women New HIV diagnoses among pregnant women Date of first ANC visit HIV status at first test during current pregnancy (known positive, tested negative, tested positive, not tested) Dates and results of subsequent HIV tests during pregnancy ART register ANC register/data HIV prevalence, incidence in pregnant women Potential estimation of HIV prevalence in general population New diagnoses in pregnant women not on ART Vertical transmission Pregnancy in women living with HIV ART register ANC register/data HIV-exposed infant or child ART register ANC register/data Paediatric HIV prevalence/ incidence HIV-exposed infant or child test result of HIV assay 1 ART register Paediatric HIV prevalence/ incidence % of exposed infants who were tested Final diagnosis of HIV-exposed infant ART register Paediatric HIV prevalence/ incidence Vertical transmission rate Proportion of children not tested Death Date of death Cause of death ART register Civil registration and vital statistics (CRVS) system Verbal autopsy Modelled estimates HIV prevalence AIDS-related death ART register CRVS system HIV prevalence b Electronic medical records can be used where any of these data elements are recorded. The quality of routine data depends upon how well information is captured by the health information system and health service providers and is cleaned and reviewed by data managers. Assessments of data quality (particularly for completeness and identification of outliers) must be integrated into the steps used to analyse routine facility data. It is particularly important for HIV incidence calculations to distinguish individuals newly diagnosed with HIV from those who may have tested previously. In the long run, routine health facility data provide a sustainable, timely, granular source of data for monitoring HIV incidence and prevalence and the health sector response to HIV. FOR MORE INFORMATION, CONTACT: World Health Organization Department of Global HIV Hepatitis and Sexually Transmitted Infections Programmes 20, avenue Appia 1211 Geneva 27 Switzerland E-mail: hiv-aids@who.int www.who.int/hiv Consolidated guidelines on person-centred HIV strategic information: strengthening routine data for impact. Policy brief on harnessing the strength of routine data for HIV surveillance ISBN 978-92-4-006666-3 (electronic version) ISBN 978-92-4-006667-0 (print version) © World Health Organization 2022. Some rights reserved. This work is available under the CC BY-NC-SA 3.0 IGO licence. ISBN 978-92-4-006666-3