Adherence to Antiretroviral Therapy During and After Pregnancy: Cohort Study on Women Receiving Care in Malawi's Option B+ Program.

BACKGROUND: Adherence to antiretroviral therapy (ART) is crucial to preventing mother-to-child transmission of human immunodeficiency virus (HIV) and ensuring the long-term effectiveness of ART, yet data are sparse from African routine care programs on maternal adherence to triple ART. METHODS: We analyzed data from women who started ART at 13 large health facilities in Malawi between September 2011 and October 2013. We defined adherence as the percentage of days "covered" by pharmacy claims. Adherence of ≥90% was deemed adequate. We calculated inverse probability of censoring weights to adjust adherence estimates for informative censoring. We used descriptive statistics, survival analysis, and pooled logistic regression to compare adherence between pregnant and breastfeeding women eligible for ART under Option B+, and nonpregnant and nonbreastfeeding women who started ART with low CD4 cell counts or World Health Organization clinical stage 3/4 disease. RESULTS: Adherence was adequate for 73% of the women during pregnancy, for 66% in the first 3 months post partum, and for about 75% during months 4-21 post partum. About 70% of women who started ART during pregnancy and breastfeeding adhered adequately during the first 2 years of ART, but only about 30% of them had maintained adequate adherence at every visit. Risk factors for inadequate adherence included starting ART with an Option B+ indication, at a younger age, or at a district hospital or health center. CONCLUSIONS: One-third of women retained in the Option B+ program adhered inadequately during pregnancy and breastfeeding, especially soon after delivery. Effective interventions to improve adherence among women in this program should be implemented.

Developing a point-of-care electronic medical record system for TB/HIV co-infected patients: experiences from Lighthouse Trust, Lilongwe, Malawi.

BACKGROUND: Implementation of user-friendly, real-time, electronic medical records for patient management may lead to improved adherence to clinical guidelines and improved quality of patient care. We detail the systematic, iterative process that implementation partners, Lighthouse clinic and Baobab Health Trust, employed to develop and implement a point-of-care electronic medical records system in an integrated, public clinic in Malawi that serves HIV-infected and tuberculosis (TB) patients. METHODS: Baobab Health Trust, the system developers, conducted a series of technical and clinical meetings with Lighthouse and Ministry of Health to determine specifications. Multiple pre-testing sessions assessed patient flow, question clarity, information sequencing, and verified compliance to national guidelines. Final components of the TB/HIV electronic medical records system include: patient demographics; anthropometric measurements; laboratory samples and results; HIV testing; WHO clinical staging; TB diagnosis; family planning; clinical review; and drug dispensing. RESULTS: Our experience suggests that an electronic medical records system can improve patient management, enhance integration of TB/HIV services, and improve provider decision-making. However, despite sufficient funding and motivation, several challenges delayed system launch including: expansion of system components to include of HIV testing and counseling services; changes in the national antiretroviral treatment guidelines that required system revision; and low confidence to use the system among new healthcare workers. To ensure a more robust and agile system that met all stakeholder and user needs, our electronic medical records launch was delayed more than a year. Open communication with stakeholders, careful consideration of ongoing provider input, and a well-functioning, backup, paper-based TB registry helped ensure successful implementation and sustainability of the system. Additional, on-site, technical support provided reassurance and swift problem-solving during the extended launch period. CONCLUSION: Even when system users are closely involved in the design and development of an electronic medical record system, it is critical to allow sufficient time for software development, solicitation of detailed feedback from both users and stakeholders, and iterative system revisions to successfully transition from paper to point-of-care electronic medical records. For those in low-resource settings, electronic medical records for integrated care is a possible and positive innovation.

Prevalence of HIV drug resistance before and 1 year after treatment initiation in 4 sites in the Malawi antiretroviral treatment program.

Since 2004, the Malawi antiretroviral treatment (ART) program has provided a public health-focused system based on World Health Organization clinical staging, standardized first-line ART regimens, limited laboratory monitoring, and no patient-level monitoring of human immunodeficiency virus drug resistance (HIVDR). The Malawi Ministry of Health conducts periodic evaluations of HIVDR development in prospective cohorts at sentinel clinics. We evaluated viral load suppression, HIVDR, and factors associated with HIVDR in 4 ART sites at 12-15 months after ART initiation. More than 70% of patients initiating ART had viral suppression at 12 months. HIVDR prevalence (6.1%) after 12 months of ART was low and largely associated with baseline HIVDR. Better follow-up, removal of barriers to on-time drug pickups, and adherence education for patients 16-24 years of age may further prevent HIVDR.

Evaluating the systems used to monitor HIV populations accessing therapy and care in low-income and lower-middle-income countries.

OBJECTIVES: The scale-up of delivery of antiretroviral therapy (ART) in low-income and middle-income countries has been coupled with the collection of data aimed at monitoring the welfare of the HIV-positive and treated populations in those countries. We aimed to compare the data items collected and reported and the degree of harmonization achieved following the publication of WHO tools for collection and reporting of these data in 2006, and of two United Nations General Assembly Special Session (UNGASS) indicators relating to the health of patients on ART in 2008. DESIGN: Retrospective examination of monitoring tools used in four countries in 2008 and 2010. METHODS: We examined and compared the type of information collected and reported from treatment and care programmes in Malawi, Uganda, Tanzania and Ukraine. We also assessed the effect of the publication of the WHO-recommended data capture and reporting tools and the UNGASS-recommended indicators on harmonizing data in these four countries 2 years following the publication of each of these tools and indicators. RESULTS: : Although the majority of WHO-recommended data items were included in patient record cards, clinic ART registers and in reports submitted to the ministries of health in the countries by 2010, there remains little concordance between the four countries examined on the specific items included in patient records and monitoring reports. Furthermore, numerous additional items, which differ by country, and which are not included in WHO recommendations, are still recorded and reported. CONCLUSION: The differences and diversity of data reported across countries continues to challenge our ability to make international comparisons and to assess programme performance.

Large-scale learning of structure-activity relationships using a linear support vector machine and problem-specific metrics.

The goal of this study was to adapt a recently proposed linear large-scale support vector machine to large-scale binary cheminformatics classification problems and to assess its performance on various benchmarks using virtual screening performance measures. We extended the large-scale linear support vector machine library LIBLINEAR with state-of-the-art virtual high-throughput screening metrics to train classifiers on whole large and unbalanced data sets. The formulation of this linear support machine has an excellent performance if applied to high-dimensional sparse feature vectors. An additional advantage is the average linear complexity in the number of non-zero features of a prediction. Nevertheless, the approach assumes that a problem is linearly separable. Therefore, we conducted an extensive benchmarking to evaluate the performance on large-scale problems up to a size of 175000 samples. To examine the virtual screening performance, we determined the chemotype clusters using Feature Trees and integrated this information to compute weighted AUC-based performance measures and a leave-cluster-out cross-validation. We also considered the BEDROC score, a metric that was suggested to tackle the early enrichment problem. The performance on each problem was evaluated by a nested cross-validation and a nested leave-cluster-out cross-validation. We compared LIBLINEAR against a Naïve Bayes classifier, a random decision forest classifier, and a maximum similarity ranking approach. These reference approaches were outperformed in a direct comparison by LIBLINEAR. A comparison to literature results showed that the LIBLINEAR performance is competitive but without achieving results as good as the top-ranked nonlinear machines on these benchmarks. However, considering the overall convincing performance and computation time of the large-scale support vector machine, the approach provides an excellent alternative to established large-scale classification approaches.

Implication of new WHO growth standards on identification of risk factors and estimated prevalence of malnutrition in rural Malawian infants.

BACKGROUND: The World Health Organization (WHO) released new Child Growth Standards in 2006 to replace the current National Center for Health Statistics (NCHS) growth reference. We assessed how switching from the NCHS to the newly released WHO Growth Standards affects the estimated prevalence of wasting, underweight and stunting, and the pattern of risk factors identified. METHODOLOGY/PRINCIPAL FINDINGS: Data were drawn from a village-informant driven Demographic Surveillance System in Northern Malawi. Children (n = 1328) were visited twice at 0-4 months and 11-15 months. Data were collected on the demographic and socio-economic environment of the child, health history, maternal and child anthropometry and child feeding practices. Weight-for-length, weight-for-age and length-for-age were derived in z-scores using the two growth references. In early infancy, prevalence estimates were 2.9, 6.1, and 8.5 fold higher for stunting, underweight, and wasting respectively using the WHO standards compared to NCHS reference (p<0.001 for all). At one year, prevalence estimates for wasting and stunting did not differ significantly according to reference used, but the prevalence of underweight was half that with the NCHS reference (p<0.001). Patterns of risk factors were similar with the two growth references for all outcomes at one year although the strength of association was higher with WHO standards. CONCLUSIONS/SIGNIFICANCE: Differences in prevalence estimates differed in magnitude but not direction from previous studies. The scale of these differences depends on the population's nutritional status thus it should not be assumed a priori. The increase in estimated prevalence of wasting in early infancy has implications for feeding programs targeting lactating mothers and ante-natal multiple micronutrients supplementation to tackle small birth size. Risk factors identified using WHO standards remain comparable with findings based on the NCHS reference in similar settings. Further research should aim to identify whether the young infants additionally diagnosed as malnourished by this new standard are more appropriate targets for interventions than those identified with the NCHS reference.