In 2004, the Malawi Ministry of Health (MoH) began the rapid expansion of the national HIV treatment program, initiating 271,105 and retaining 198,846 patients on antiretroviral therapy (ART) at 377 sites (279 static and 98 mobile/outreach) by October 2009 . The successful scale-up of the Malawi ART program was made possible through a public health approach . ART eligibility is determined by WHO clinical staging or a low CD4 count. In 2009, these criteria were Stage III or IV disease or CD4 < 250 cells/mm3 (CD4 < 350 cells/mm3 for pregnant women). One standard, fixed-dose first-line regimen (stavudine, lamivudine, nevirapine in 2009) is used for >90% of patients and standard alternative regimens are used for patients who develop side effects to the first-line regimen . Experienced sites (generally district hospitals that have been offering ART for at least three years) also provide standard second-line regimens for eligible patients. Patients visit ART clinics monthly until stable on therapy (usually after six to ten months). Two- or three-monthly appointments are scheduled thereafter.
Two critical components of the national ART program are 1) a standardized system for registering and monitoring patients, including individual patient treatment cards with registration and follow-up data and a clinic register summarizing critical data for each patient  and 2) quarterly supervision at all ART sites in Malawi . Every quarter, ART clinic personnel perform a standardized patient cohort analysis that includes aggregation of case-finding details of patients registered during the previous quarter and since ART was first begun, as well as treatment outcomes for the cumulative cohort. The latter analysis requires a review of all treatment cards in order to update the clinic register and then a tally-score on current patient outcomes. The primary outcome is patient status, namely, alive on ART, died, stopped ART (with clinician’s knowledge), transferred to another ART clinic, or lost to follow-up. Secondary outcomes for patients alive on ART include 1) ART regimen, 2) drug adherence, 3) ART side effects, and 4) current tuberculosis status. A rolling cohort survival analysis is also performed by counting primary outcomes of patients who registered during specified previous quarters .
During quarterly supervision, teams of at least two individuals (including a representative from the MoH national or regional office along with a MoH clinical officer or nurse) visit each site for up to a full day. Additional representatives from the health sector, including other MoH staff or non-governmental organization partners, join teams on occasion. Currently, quarterly supervision requires ten teams to be in the field for two to three weeks each at a cost of approximately $60,000 per quarter (excluding staff salaries). The quarterly supervision activity is coordinated by the monitoring and evaluation (M&E) team at the Department for HIV and AIDS in the MoH and funded through the U.S. President's Emergency Plan for AIDS Relief (PEPFAR) and the Global Fund to Fight AIDS, Tuberculosis and Malaria. These visits provide an opportunity to identify any problems at the sites, to discuss difficult cases with medical staff, and to monitor drug stocks, thus improving the overall quality of care .
The supervision teams also review the patient cohort analyses produced by the ART staff and clean patient data and aggregate quarterly and cumulative reports on 44 data elements. These data are used for drug forecasting, monitoring overall program performance, identifying gaps in care, and as early warning indicators for HIV drug resistance [6–18], and therefore maintaining high quality data is of utmost importance. Like other national treatment programs, Malawi has strived since its inception to provide complete, accurate and up-to-date data. Common errors include incomplete registers, failure to update follow-up outcomes, duplicate entries, and incorrect/missing records . In 2008, Makombe et al reported that unchecked site-produced reports resulted in a 12% undercount of patients on ART, affecting the programs ability to precisely forecast future drug needs .
Currently every patient treatment card is reviewed and compared against the entries in the register during supervision visits to ensure up-to-date and accurate data in the register prior to data extraction. Inconsistencies or incomplete data are updated as appropriate. This process can take many hours at each visit, competing with other important aspects of supervision and mentoring. Further, this full audit of patient records by the supervision team may not yield any changes, especially at sites where experienced clinic staff have correctly updated treatment cards and clinic registers in preparation for the supervision visit.
While the quality of recording for certain key elements (such as outcome status or regimen) seem to have improved with repetition of this intensive data cleaning process, there remain significant challenges in quality of secondary elements, such as pill counts and side effects. The goal of this study was to analyze the types and frequency of data errors currently observed in the national HIV treatment program and to evaluate their impact on management and monitoring. Based on these findings we evaluated the performance of a lot quality assurance sampling (LQAS) algorithm to prioritize ART sites with poor data quality for a full review of their records before aggregation of quarterly reports. We then estimated the potential amount of time saved by omitting an exhaustive review of all treatment cards at sites that pass the LQAS assessment. Further, we estimated secondary outcomes from treatment cards during the sampling process.