PropertyValue
is nif:broaderContext of
nif:broaderContext
is schema:hasPart of
schema:isPartOf
nif:isString
  • We conducted a retrospective cohort study at a large urban HIV clinic in the south western Ugandan town of Mbarara. The clinic is based within Mbarara Regional Referral Hospital, a teaching hospital for Mbarara University medical school. The clinic was established in 1998 and by December 2012, there were 22,643 cumulative and 9,329 active entries of patients receiving ART. The retrospective cohort was constructed using routinely collected data at the clinic. The data for analysis was extracted from the International Epidemiological Database to evaluate AIDS (IeDEA) of Mbarara Regional Referral Hospital. The hospital joined the IeDEA network in January 2006, which now exists at 3 other sites in Uganda. Our analysis only uses data from Mbarara Hospital. All patients enrolled in care at the regional referral hospital are entered into this database. Patients initiated on ART before January 2006 were not entered in IeDEA. Therefore, we considered patients initiated on ART from January 1, 2006, when the IeDEA database started to December 31, 2012. We extracted several variables including socio-demographic factors, clinical and laboratory data like CD4 counts, viral load, BMI at initiation of ART, WHO clinical staging, ART start date and ART regimen. Patients in the IeDEA database were eligible for analysis if they were aged ≥18 years old, and had initiated ART on or before December 31, 2012. Laboratory biomarkers such as CD4 counts, Complete Blood Count and Hemoglobin for patients at initiation of ART were collated with the patients’ routine follow-up visits information. The data were cleaned to ensure consistency in the coding of variables and for missing data. Patients with missing age were excluded from the analysis. Patients who were missing other variables were included in the analysis but were automatically excluded in any analysis where the missing variable was involved. No data imputation was done. Ages for patients initiated on ART were stratified in three categories namely; 18 to 34 years (young adults), 35 to 49 years (mid-age) and ≥50 years (older adults). The age categories were considered as the main exposure variable. CD4 change as a predictor was created by obtaining the difference between the most recent and CD4 cell count at ART initiation. The variable CD4 change was broken into 4 categories: CD4 decline, CD4 increase of <100, CD4 increase of 100–350 and CD4 increase >350. Baseline characteristics across the age strata were compared using chi square and those that presented differences were included in the multivariable analysis. The primary endpoint for this analysis was mortality after initiation of ART and the secondary endpoint was immunologic response. We used Survival analysis to determine the effect of age on mortality using the Kaplan Meier plot and the Log rank test. We report the median survival and the inter quartile range (IQR). Poisson regression analysis was used to determine the rate ratios for effect of age on mortality and immunological response. Immunological response was a secondary outcome and was classified by dichotomizing the most recent CD4 into either CD4 count below 350 or ≥350. The patients with CD4 cell count less than 350 were classified as having a poor immunological outcome. A multi variable analysis was done using Poisson regression modeling with the predictor variables that were considered confounders entered into the model simultaneously, using a cut off of p<0.1 in the bivariate analysis as consideration for inclusion in the multiple regression model. All analyses were completed using STATA version 11 (College Station, TX). The study protocols were approved by the Department of Community Health, the HIV clinic data sharing committee, the Faculty of Medicine Research ethics committee and the Mbarara University of Science and Technology Research Ethics Committee. All the datasets extracted from the IeDEA database were kept under password lock. Data were de-identified and codes instead of names were used and all the information was kept confidential.
rdf:type