At the time of the study, the following authors were employed by Health Alliance International: Barrot Lambdin, Mark Micek, Kenneth Sherr, Joe Lara, James Pfeiffer and Stephen Gloyd. All other authors declare that they have no competing interests.
BL and JL performed the literature search. BL, AS, MM, TK, JH, JP, KS and SG contributed to the study design. BL, JL and MK assisted with data collection. BL, JH, TK, AS and MM informed the data analysis. All authors contributed to the interpretation of study results as well as manuscript preparation. In addition, the final version of the manuscript was read and approved by all authors.
Since the rapid scale-up of antiretroviral therapy (ART) programs in sub-Saharan Africa, electronic patient tracking systems (EPTS) have been deployed to respond to the growing demand for program monitoring, evaluation and reporting to governments and donors. These routinely collected data are often used in epidemiologic and operations research studies intended to improve programs. To ensure accurate reporting and good quality for research, the reliability and completeness of data systems need to be assessed and reported. We assessed the completeness and reliability of EPTS used in 16 HIV care and treatment clinics in Manica and Sofala provinces of Mozambique.
We conducted a cross-sectional study to assess the completeness and reliability of key variables in the electronic data system for patients enrolling in 16 public sector HIV treatment clinics between 1 July 2004 and 30 June 2008. Data from the electronic database was compared with data abstracted from a stratified random sample of 520 patient charts. Percent agreement, kappa scores and concordance correlation coefficients were calculated for specified variables. Percentile bootstrap confidence intervals were calculated to account for the stratified nature of our sampling.
A total of 16,149 patients with a median age of 33 years and a median CD4 count of 151 enrolled in these 16 clinics between 1 July 2004 and 30 June 2008. The level of completeness was high for most variables with height (18.6%) and weight (11.5%) having the highest amount of missing data. The level of agreement for available data was also high with reliability statistics of 0.95 (95% CI: 0.92-0.98) for gender, 0.91 (95% CI: 0.80-1.00) for pre-ART CD4 value and 0.97 (95% CI: 0.95-0.99) for patient retention.
Electronic patient tracking systems have been deployed to respond to the growing monitoring, evaluation and reporting requirements. In our cross-sectional study of clinics in Manica and Sofala provinces of Mozambique, we found high levels of completeness and reliability for key variables indicating that these electronic databases provided adequate data not only for monitoring and evaluation but also for research. Routine evaluations of the completeness and reliability of these databases need to occur to ensure high quality data are being used for reporting and research.