INSPIRE datahub: a pan-African integrated suite of services for harmonising longitudinal population health data using OHDSI tools.

Introduction: Population health data integration remains a critical challenge in low- and middle-income countries (LMIC), hindering the generation of actionable insights to inform policy and decision-making. This paper proposes a pan-African, Findable, Accessible, Interoperable, and Reusable (FAIR) research architecture and infrastructure named the INSPIRE datahub. This cloud-based Platform-as-a-Service (PaaS) and on-premises setup aims to enhance the discovery, integration, and analysis of clinical, population-based surveys, and other health data sources. Methods: The INSPIRE datahub, part of the Implementation Network for Sharing Population Information from Research Entities (INSPIRE), employs the Observational Health Data Sciences and Informatics (OHDSI) open-source stack of tools and the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) to harmonise data from African longitudinal population studies. Operating on Microsoft Azure and Amazon Web Services cloud platforms, and on on-premises servers, the architecture offers adaptability and scalability for other cloud providers and technology infrastructure. The OHDSI-based tools enable a comprehensive suite of services for data pipeline development, profiling, mapping, extraction, transformation, loading, documentation, anonymization, and analysis. Results: The INSPIRE datahub's "On-ramp" services facilitate the integration of data and metadata from diverse sources into the OMOP CDM. The datahub supports the implementation of OMOP CDM across data producers, harmonizing source data semantically with standard vocabularies and structurally conforming to OMOP table structures. Leveraging OHDSI tools, the datahub performs quality assessment and analysis of the transformed data. It ensures FAIR data by establishing metadata flows, capturing provenance throughout the ETL processes, and providing accessible metadata for potential users. The ETL provenance is documented in a machine- and human-readable Implementation Guide (IG), enhancing transparency and usability. Conclusion: The pan-African INSPIRE datahub presents a scalable and systematic solution for integrating health data in LMICs. By adhering to FAIR principles and leveraging established standards like OMOP CDM, this architecture addresses the current gap in generating evidence to support policy and decision-making for improving the well-being of LMIC populations. The federated research network provisions allow data producers to maintain control over their data, fostering collaboration while respecting data privacy and security concerns. A use-case demonstrated the pipeline using OHDSI and other open-source tools.

Enabling data sharing and utilization for African population health data using OHDSI tools with an OMOP-common data model.

The COVID-19 pandemic has spurred the use of AI and DS innovations in data collection and aggregation. Extensive data on many aspects of the COVID-19 has been collected and used to optimize public health response to the pandemic and to manage the recovery of patients in Sub-Saharan Africa. However, there is no standard mechanism for collecting, documenting and disseminating COVID-19 related data or metadata, which makes the use and reuse a challenge. INSPIRE utilizes the Observational Medical Outcomes Partnership (OMOP) as the Common Data Model (CDM) implemented in the cloud as a Platform as a Service (PaaS) for COVID-19 data. The INSPIRE PaaS for COVID-19 data leverages the cloud gateway for both individual research organizations and for data networks. Individual research institutions may choose to use the PaaS to access the FAIR data management, data analysis and data sharing capabilities which come with the OMOP CDM. Network data hubs may be interested in harmonizing data across localities using the CDM conditioned by the data ownership and data sharing agreements available under OMOP's federated model. The INSPIRE platform for evaluation of COVID-19 Harmonized data (PEACH) harmonizes data from Kenya and Malawi. Data sharing platforms must remain trusted digital spaces that protect human rights and foster citizens' participation is vital in an era where information overload from the internet exists. The channel for sharing data between localities is included in the PaaS and is based on data sharing agreements provided by the data producer. This allows the data producers to retain control over how their data are used, which can be further protected through the use of the federated CDM. Federated regional OMOP-CDM are based on the PaaS instances and analysis workbenches in INSPIRE-PEACH with harmonized analysis powered by the AI technologies in OMOP. These AI technologies can be used to discover and evaluate pathways that COVID-19 cohorts take through public health interventions and treatments. By using both the data mapping and terminology mapping, we construct ETLs that populate the data and/or metadata elements of the CDM, making the hub both a central model and a distributed model.

Effect of justification of wife-beating on experiences of intimate partner violence among men and women in Uganda: A propensity-score matched analysis of the 2016 Demographic Health Survey data.

INTRODUCTION: In some communities, rationalization of men's controlling attitudes is associated with the justification of gender norms such as wife-beating as a method of correcting spouse behaviour. In this quasi-experimental study, we investigate the causal effects of the acceptability of gender norms justifying wife-beating on experiences of sexual, emotional, and physical intimate partner violence (IPV) among Ugandan men and women. METHODS AND MATERIALS: We analysed the 2016 Uganda Demographic and Health Survey data using propensity-score matching. The exposure variable is the acceptability of gender norms justifying wife-beating measured on a binary scale and the outcomes are the respondent's lifetime experiences of sexual, physical, and emotional IPV. We matched respondents who accepted gender norms justifying wife-beating with those that never through a 1:1 nearest-neighbour matching with a caliper to achieve comparability on selected covariates. We then estimated the causal effects of acceptability of gender norms justifying wife-beating on the study outcomes using a logistic regression model. RESULTS: Results showed that a total of 4,821 (46.5%) out of 10,394 respondents reported that a husband is justified in beating his wife for specific reasons. Among these, the majority (3,774; 78.3%) were women compared to men (1,047; 21.7%). Overall, we found that men and women who accept gender norms justifying wife-beating are more likely to experience all three forms of IPV. In the sub-group analysis, men who justify wife-beating were more likely to experience emotional and physical IPV but not sexual IPV. However, women who justify wife-beating were more likely to experience all three forms of IPV. CONCLUSIONS: In conclusion, the acceptability of gender norms justifying wife-beating has a positive effect on experiences of different forms of IPV by men and women in Uganda. There is, therefore, a need for more research to study drivers for acceptance of gender norms justifying wife-beating to enable appropriate government agencies to put in place mechanisms to address the acceptability of gender norms justifying wife-beating at the societal level.

A call for citizen science in pandemic preparedness and response: beyond data collection.

The COVID-19 pandemic has underlined the need to partner with the community in pandemic preparedness and response in order to enable trust-building among stakeholders, which is key in pandemic management. Citizen science, defined here as a practice of public participation and collaboration in all aspects of scientific research to increase knowledge and build trust with governments and researchers, is a crucial approach to promoting community engagement. By harnessing the potential of digitally enabled citizen science, one could translate data into accessible, comprehensible and actionable outputs at the population level. The application of citizen science in health has grown over the years, but most of these approaches remain at the level of participatory data collection. This narrative review examines citizen science approaches in participatory data generation, modelling and visualisation, and calls for truly participatory and co-creation approaches across all domains of pandemic preparedness and response. Further research is needed to identify approaches that optimally generate short-term and long-term value for communities participating in population health. Feasible, sustainable and contextualised citizen science approaches that meaningfully engage affected communities for the long-term will need to be inclusive of all populations and their cultures, comprehensive of all domains, digitally enabled and viewed as a key component to allow trust-building among the stakeholders. The impact of COVID-19 on people's lives has created an opportune time to advance people's agency in science, particularly in pandemic preparedness and response.

Sex-disaggregated data matters: tracking the impact of COVID-19 on the health of women and men.

Sex and gender matter to health outcomes, but despite repeated commitments to sex-disaggregate data in health policies and programmes, a persistent and substantial absence of such data remains especially in lower-income countries. This represents a missed opportunity for monitoring and identifying gender-responsive, evidence-informed solutions to address a key driver of the pandemic. In this paper we review the availability of national sex-disaggregated surveillance data on COVID-19 and examine trends on the testing-to-outcome pathway. We further analyse the availability of data according to the economic status of the country and investigate the determinants of sex differences, including the national gender inequality status (according to a global index) in each country. Results are drawn from 18 months of global data collection from over 200 countries. We find differences in COVID-19 prevention behaviours and illness outcomes by sex, with lower uptake of vaccination and testing plus an elevated risk of severe disease and death among men. Supporting and maintaining the collection, collation, interpretation and presentation of sex-disaggregated data requires commitment and resources at subnational, national and global levels, but provides an opportunity for identifying and taking gender-responsive action on health inequities. As a first step the global health community should recognise, value and support the importance of sex-disaggregated data for identifying and tackling an inequitable pandemic.

Prevalence and Risk Factors for Women's Reports of Past-Year Intimate Partner Violence: A Comparative Analysis of Six East African National Surveys.

Violence against women, in all its forms, has been acknowledged as a violation of basic human rights and research evidence shows that it could lead to adverse health consequences. In this study we aimed to determine the prevalence and coexistence of different forms of IPV as well as examine individual-level factors associated with ever experiencing any form of IPV in the 12 months preceding the survey using the most recent Demographic Health Survey data from six East African countries. Results show that the prevalence ranged between 16.5% (Burundi) and 29.3% (Uganda) for emotional, 16.8% (Ethiopia) and 26.6% (Tanzania) for physical, and 8.3% (Rwanda and Ethiopia) and 18.4% (Burundi) for sexual IPV. The prevalence of any IPV ranged from 26.7% to 39.3%. In terms of coexistence, 15.6% to 19.0% of women reported experiencing all the three forms of IPV, with higher proportions reporting experiencing two of the three forms of IPV. The prevalence of both physical and emotional IPV was highest in Tanzania (49.1%), both emotional and sexual IPV in Uganda (28.0%), and both physical and sexual IPV in Burundi (26.2%). A partner's use of alcohol and a woman's justification of wife beating were both statistically significant common risk factors for IPV across the six countries. Women whose partners got drunk often were found to be up to nine times more likely to experience IPV compared to those whose partners did not drink. Younger women and those with larger families were at an increased risk of experiencing IPV, while other significant factors were country specific. In conclusion, our findings highlight the need for integrated and context-specific approaches that deconstruct gendered norms related to power dynamics and patriarchal nuances at household and community level in order to holistically address different forms of IPV.

Factors associated with HIV testing among youth in Tanzania based on the 2016-2017 Tanzania HIV Impact Survey (THIS).

HIV testing continues to be a challenge among the young population in Tanzania. As of 2017, only 30% of 15-19-year-olds reported getting tested and receiving their results. This study will examine the demographic and socio-behavioral characteristics associated with HIV testing among adolescents and young adults in Tanzania. Interview data from the 2016-2017 Tanzania HIV Impact Survey (THIS) were analyzed on 10,128 adolescents and young adults 15-24 years of age, representing 10.5 million youth in Tanzania. Weighted logistic regression was used to model the relationship of HIV testing with demographic and socio-behavioral characteristics. Half (50%) of respondents reported ever having been tested for HIV. HIV testing was significantly lower among males compared with females (AOR = 0.5;95% confidence interval [CI] = 0.5-0.6; p<0.001), 15-19 year olds compared with 20-24 year olds (AOR = 0.4;95% CI = 0.4-0.5; p<0.001), no education compared with secondary or post-secondary education (AOR = 0.4;95% CI = 0.3-0.6; p<0.001), rural residents compared with urban residents (AOR = 0.7;95% CI = 0.6-0.9; p<0.001) and those who don't use condoms during sexual intercourse compared with those who do (AOR = 0.6;95% CI = 0.5-0.8; p<0.001). Among HIV-infected youth, younger age group, rural residents, education less than primary, single, high income, and sex workers were significantly associated with never testing for HIV. This study highlights the majority of characteristics affecting HIV testing among young people in Tanzania have not changed over the years, thus it is necessary to re-examine the current approaches to HIV testing. The COVID-19 pandemic will add to this challenge as it collides with the ongoing HIV epidemic and competes for needed medical supplies and health care provider resources. In light of this current situation, intensified and targeted HIV testing programs for at risk young populations in Tanzania should be prioritized.

Understating the barriers to achievement of the UNAIDS 90-90-90 goal in Tanzania using a population-based HIV impact assessment survey 2016-2017.

The Joint United Nations Programme on HIV/AIDS (UNAIDS) and partners launched the 90-90-90 targets. We used Tanzania HIV Impact Survey (THIS) data in 2017 to study the barriers to achieve 90-90-90 targets. THIS was a population-based survey with a stratified multistage stage sampling design. We used weighted logistic regression to associate three targets with socio-demographics, HIV-related discrimination, fear and shame. We defined HIV awareness by a combination of self-reported of HIV status positive and detected antiretroviral (ARV) in blood among PLWH. On ARV was defined as those who self-reported among awareness. Viral load suppression was defined as 400 copies/ml or less in the blood sample. The three targets were estimated at 61-90-85 in Tanzania from the weighted analysis. The first target was far from being achieved. The weighted regression showed that being female, having attained higher education, married, having insurance, and living in urban areas were associated with a high likelihood of having ever tested for HIV. The results indicated that intervention programmes in Tanzania should focus on the first target. Intervention programmes should be designed for each target separately. Integrated strategies in the context of low and middle-income countries are needed to achieve these targets.

Eave and swarm collections prove effective for biased captures of male Anopheles gambiae mosquitoes in Uganda.

BACKGROUND: Traditional malaria vector sampling techniques bias collections towards female mosquitoes. Comprehensive understanding of vector dynamics requires balanced vector sampling of both males and females. Male mosquito sampling is also necessary for population size estimations by male-based mark-release-recapture (MRR) studies and for developing innovations in mosquito control, such as the male-targeted sterile insect technique and other genetic modification approaches. This study evaluated a range of collection methods which show promise in providing a more equal, or even male-biased, sex representation in the sample. RESULTS: Swarms were found at all study sites and were more abundant and larger at the peak of the wet season. Swarm sampling caught the most males, but when man/hour effort was factored in, sampling of eaves by aspiration was the more efficient method and also provided a representative sample of females. Grass-roofed houses were the most productive for eave collections. Overall few mosquitoes were caught with artificial resting traps (clay pots and buckets), although these sampling methods performed better at the start of the wet season than at its peak, possibly because of changes in mosquito ecology and an increased availability of natural resting sites later in the season. Aspiration of bushes was more productive at the peak of the wet season than at the start. CONCLUSIONS: The results of this study demonstrate that eave aspiration was an efficient and useful male mosquito collection method at the study sites and a potentially powerful aid for swarm location and MRR studies. The methods evaluated may together deliver more sex-balanced mosquito captures and can be used in various combinations depending on the aims and ecological parameters of a given study.

Effect of women's intra-household bargaining power on postnatal and infant healthcare in rural Uganda-Results from a cross sectional survey in Kyenjojo district.

Decision-making power and access to and control over resources are key elements of women's bargaining power within a household, and plays an important role in improving healthcare seeking behaviours for women and their children, which in turn augment maternal and child health outcomes. We examined the relationship between intra-household bargaining power and utilization of postnatal and child healthcare services within 6 months after delivery, based on cross-sectional survey data from Kyenjojo district, Tooro sub-region of Western Uganda. We assessed independent associations between women's intra-household bargaining autonomy and postnatal care attendance using a modified Poisson approach for common outcomes. We found that women who contributed to the decision-making processes on child healthcare, personal healthcare, and how to raise money for healthcare of family members were about 20% more likely to attend postnatal and child healthcare within 6 months of delivery, compared with women who were unable to make such decisions. Therefore, contributing to efforts that empower women to have greater control over child and personal healthcare through gender transformative approaches and policy engagements in important.

HIV-1 superinfection can occur in the presence of broadly neutralizing antibodies.

BACKGROUND: Superinfection of individuals already infected with HIV-1 suggests that pre-existing immune responses may not adequately protect against re-infection. We assessed high-risk female sex workers initially infected with HIV-1 clades A, D or A/D recombinants, to determine if HIV-1 broadly neutralizing antibodies were lacking prior to superinfection. METHODS: Six superinfected female sex workers previously stratified by HIV-1 high-risk behavior, infecting virus clade and volunteer CD4 counts were evaluated at baseline (n = 5) and at 350 days post-superinfection (n = 6); one superinfected volunteer lacked pre-superinfection plasma. Retrospective plasmas were assessed for neutralization of a multi-clade panel of 12 HIV-1 viruses before superinfection, and then at quarterly intervals thereafter. Similarly stratified singly infected female sex workers were correspondingly assessed at baseline (n = 19) and 350 days after superinfection (n = 24). Neutralization of at least 50% of the 12 viruses (broad neutralization), and geometric means of the neutralization titers (IC50) were compared before and after superinfection; and were correlated with the volunteer HIV-1 superinfection status, CD4 counts, and pseudovirus clade. RESULTS: Preexisting broad neutralization occurred in 80% (4/5) of the superinfected subjects with no further broadening by 350 days after superinfection. In one of the five subjects, HIV-1 superinfection occurred when broad neutralization was lacking; with subsequent broadening of neutralizing antibodies occuring within 9 months and plateauing by 30 months after detection of superinfection. Clade B and C pseudoviruses were more sensitive to neutralization (13; [87%]); and (12; [80%]) than the locally circulating clades A (10; [67%]) and D (6; [40%]), respectively (p = 0.025). Low antibody titers correlated with clade D viruses and with >500 CD4 T cell counts, but not with the superinfection status. CONCLUSION: These data demonstrate that HIV-1 superinfection can occur both in the presence, and in the absence of broadly neutralizing antibodies.

HIV-1 transmission networks in high risk fishing communities on the shores of Lake Victoria in Uganda: A phylogenetic and epidemiological approach.

BACKGROUND: Fishing communities around Lake Victoria in sub-Saharan Africa have been characterised as a population at high risk of HIV-infection. METHODS: Using data from a cohort of HIV-positive individuals aged 13-49 years, enrolled from 5 fishing communities on Lake Victoria between 2009-2011, we sought to identify factors contributing to the epidemic and to understand the underlying structure of HIV transmission networks. Clinical and socio-demographic data were combined with HIV-1 phylogenetic analyses. HIV-1 gag-p24 and env-gp-41 sub-genomic fragments were amplified and sequenced from 283 HIV-1-infected participants. Phylogenetic clusters with ≥2 highly related sequences were defined as transmission clusters. Logistic regression models were used to determine factors associated with clustering. RESULTS: Altogether, 24% (n = 67/283) of HIV positive individuals with sequences fell within 34 phylogenetically distinct clusters in at least one gene region (either gag or env). Of these, 83% occurred either within households or within community; 8/34 (24%) occurred within household partnerships, and 20/34 (59%) within community. 7/12 couples (58%) within households clustered together. Individuals in clusters with potential recent transmission (11/34) were more likely to be younger 71% (15/21) versus 46% (21/46) in un-clustered individuals and had recently become resident in the community 67% (14/21) vs 48% (22/46). Four of 11 (36%) potential transmission clusters included incident-incident transmissions. Independently, clustering was less likely in HIV subtype D (adjusted Odds Ratio, aOR = 0.51 [95% CI 0.26-1.00]) than A and more likely in those living with an HIV-infected individual in the household (aOR = 6.30 [95% CI 3.40-11.68]). CONCLUSIONS: A large proportion of HIV sexual transmissions occur within house-holds and within communities even in this key mobile population. The findings suggest localized HIV transmissions and hence a potential benefit for the test and treat approach even at a community level, coupled with intensified HIV counselling to identify early infections.

Impact of decentralisation of antiretroviral therapy services on HIV testing and care at a population level in Agago District in rural Northern Uganda: results from the Lablite population surveys.

Background: We conducted unlinked cross-sectional population-based surveys in Northern Uganda before and after antiretroviral therapy (ART) provision (including Option B+ [lifelong ART for pregnant/breast-feeding women]) at a local primary care facility (Lira Kato Health Centre [HC]). Prior to decentralisation, people travelled 56-76 km round-trip for ART; we aimed to evaluate changes in uptake of HIV-testing, ART coverage and access to ART following decentralisation. Methods: A total of 2124 adults in 1351 households in two parishes closest to Lira Kato HC were interviewed using questionnaires between March and April 2013 and 2123 adults in 1229 households between January and March 2015. Results: Adults reporting HIV-testing in the last year increased from 1077/2124 (50.7%) to 1298/2123 (61.1%) between surveys (p<0.001). ART coverage increased from 74/136 (54.4%) self-reported HIV-positive adults in 2013 to 108/133 (81.2%) in 2015 (p<0.001). Post-decentralisation, 47/108 (43.5%) of those on ART were in care at Lira Kato HC (including 37 new initiations). Most of the remainder (47/61, 77%) started ART prior to any ART provision at Lira Kato HC; the most common reason given for not accessing ART locally was concern about drug-stock-outs (30/59, 51%). Conclusions: HIV-testing and ART coverage increased after decentralisation combined with Option B+ roll-out. However, patients on ART before decentralisation were reluctant to transfer to their local facility.

Prevalence, incidence and predictors of peripheral neuropathy in African adults with HIV infection within the DART trial.

OBJECTIVES: We investigated the prevalence, incidence and predictors of new peripheral neuropathy episodes in previously untreated, symptomatic HIV-infected Ugandan/Zimbabwean adults initiating zidovudine-based antiretroviral therapy (ART). DESIGN: An open-label, multicentre, randomized trial. METHODS: Peripheral neuropathy was self-reported at 12-weekly clinic visits. Cox regression models (excluding participants reporting preexisting peripheral neuropathy at ART initiation), considered sex; pre-ART WHO stage, age and CD4(+) cell count; CD4(+) cell count versus no CD4(+) cell count monitoring; and time-updated CD4(+) cell count, weight and use of stavudine, isoniazid and didanosine. RESULTS: Four hundred and twenty-one out of 3316(13%) patients reported preexisting peripheral neuropathy at ART initiation. Median (interquartile range, IQR) follow-up in 2895 participants without preexisting peripheral neuropathy was 4.9 (4.7-5.4) years. Three hundred and fifty-four (12%) took stavudine as first-line substitution and 518 (18%) took isoniazid during follow-up. Two hundred and ninety (11%) participants developed a new peripheral neuropathy episode, an incidence of 2.12 per 100 person-years. Eighteen (0.1%) had a grade 3/4 episode. Independent predictors of peripheral neuropathy were current stavudine use [adjusted hazard ratio (a)HR 4.16 (95% confidence interval, 95% CI 3.06-5.66], current isoniazid use [aHR 1.59 (95% CI 1.02-2.47)] and current didanosine use [aHR 1.60 (95% CI 1.19-2.14)]. Higher risks were independently associated with higher pre-ART weight [aHR (per+5 kg) 1.07 (95% CI 1.01-1.13)] and older age aHR (per 10 years older) 1.29 (95% CI 1.12-1.49), but there was no significant effect of sex (P = 0.13), pre-ART CD4(+) cell count (P = 0.91) or CD4(+) cell count monitoring (P = 0.73). CONCLUSION: Current stavudine, didanosine or isoniazid use continue to increase peripheral neuropathy risks, as does older age and weight at ART initiation; however, we found no evidence of increased risk in women in contrast to previous studies. The incidence of peripheral neuropathy may now be lower in ART programmes, as stavudine and didanosine are no longer recommended. All patients receiving isoniazid, either as part of antituberculosis (TB) chemotherapy or TB-preventive therapy, should receive pyridoxine as recommended in national guidelines.

Dynamic logistic regression model and population attributable fraction to investigate the association between adherence, missed visits and mortality: a study of HIV-infected adults surviving the first year of ART.

BACKGROUND: Adherence is one of the most important determinants of viral suppression and drug resistance in HIV-infected people receiving antiretroviral therapy (ART). METHODS: We examined the association between long-term mortality and poor adherence to ART in DART trial participants in Uganda and Zimbabwe randomly assigned to receive laboratory and clinical monitoring (LCM), or clinically driven monitoring (CDM). Since over 50% of all deaths in the DART trial occurred during the first year on ART, we focussed on participants continuing ART for 12 months to investigate the implications of longer-term adherence to treatment on mortality. Participants' ART adherence was assessed by pill counts and structured questionnaires at 4-weekly clinic visits. We studied the effect of recent adherence history on the risk of death at the individual level (odds ratios from dynamic logistic regression model), and on mortality at the population level (population attributable fraction based on this model). Analyses were conducted separately for both randomization groups, adjusted for relevant confounding factors. Adherence behaviour was also confounded by a partial factorial randomization comparing structured treatment interruptions (STI) with continuous ART (CT). RESULTS: In the CDM arm a significant association was found between poor adherence to ART in the previous 3-9 months with increased mortality risk. In the LCM arm the association was not significant. The odds ratios for mortality in participants with poor adherence against those with optimal adherence was 1.30 (95% CI 0.78,2.10) in the LCM arm and 2.18 (1.47,3.22) in the CDM arm. The estimated proportions of deaths that could have been avoided with optimal adherence (population attributable fraction) in the LCM and CDM groups during the 5 years follow-up period were 16.0% (95% CI 0.7%,31.6%) and 33.1% (20.5%,44.8%), correspondingly. CONCLUSIONS: Recurrent poor adherence determined even through simple measures is associated with high mortality both at individual level as well as at the ART programme level. The number of lives saved through effective interventions to improve adherence could be considerable particularly for individuals monitored without using CD4 cell counts. The findings have important implications for clinical practice and for developing interventions to enhance adherence.

Clustering based on adherence data.

Adherence to a medical treatment means the extent to which a patient follows the instructions or recommendations by health professionals. There are direct and indirect ways to measure adherence which have been used for clinical management and research. Typically adherence measures are monitored over a long follow-up or treatment period, and some measurements may be missing due to death or other reasons. A natural question then is how to describe adherence behavior over the whole period in a simple way. In the literature, measurements over a period are usually combined just by using averages like percentages of compliant days or percentages of doses taken. In the paper we adapt an approach where patient adherence measures are seen as a stochastic process. Repeated measures are then analyzed as a Markov chain with finite number of states rather than as independent and identically distributed observations, and the transition probabilities between the states are assumed to fully describe the behavior of a patient. The patients can then be clustered or classified using their estimated transition probabilities. These natural clusters can be used to describe the adherence of the patients, to find predictors for adherence, and to predict the future events. The new approach is illustrated and shown to be useful with a simple analysis of a data set from the DART (Development of AntiRetroviral Therapy in Africa) trial in Uganda and Zimbabwe.