Predictors of loss to follow-up before HIV treatment initiation in Northwest Ethiopia: a case control study

Background: In Ethiopia, there is a growing concern about the increasing rates of loss to follow-up (LTFU) in HIV programs among people waiting to start HIV treatment. Unlike other African countries, there is little information about the factors associated with LTFU among pre-antiretroviral treatment (pre-ART) patients in Ethiopia. We conducted a case-control study to investigate factors associated with pre-ART LTFU in Ethiopia. Methods: Charts of HIV patients newly enrolled in HIV care at Gondar University Hospital (GUH) between September 11, 2008 and May 8, 2011 were reviewed. Patients who were "loss to follow-up" during the pre-ART period were considered to be cases and patients who were "in care" during the pre-ART period were controls. Logistic regression analysis was used to explore factors associated with pre-ART LTFU. Results: In multivariable analyses, the following factors were found to be independently associated with pre-ART LTFU: male gender [Adjusted Odds Ratio (AOR) = 2.00 (95% CI: 1.15, 3.46)], higher baseline CD4 cell count (251-300 cells/mu l [AOR = 2.64 (95% CI: 1.05, 6.65)], 301-350 cells/mu l [AOR = 5.21 (95% CI: 1.94, 13.99)], and >350 cells/mu l [AOR = 12.10 (95% CI: 6.33, 23.12)] compared to CD4 cell count of <= 200 cells/mu l) and less advanced disease stage (WHO stage I [AOR = 2.81 (95% CI: 1.15, 6.91)] compared to WHO stage IV). Married patients [AOR = 0.39 (95% CI: 0.19, 0.79)] had reduced odds of being LTFU. In addition, patients whose next visit date was not documented on their medical chart [AOR = 241.39 (95% CI: 119.90, 485.97)] were more likely to be LTFU. Conclusion: Our study identified various factors associated with pre-ART LTFU. The findings highlight the importance of giving considerable attention to pre-ART patients' care from the time that they learn of their positive HIV serostatus. The completeness of the medical records, the standard of record keeping and obstacles to retrieving charts also indicate a serious problem that needs due attention from clinicians and data personnel.