BackgroundDue to the limited availability of viral load testing for treatment outcome monitoring in resource limited settings, identifying predictive factors of antiretroviral treatment failure will help in selecting clients who will benefit most from the targeted use of viral load monitoring. Little is known about the predictors of treatment failure in the study area. This study was conducted to determine factors that predict first-line antiretroviral therapy failure among HIV-infected adult clients at Woldia Hospital, Northeast Ethiopia. For this study, antiretroviral therapy treatment failure was defined as the fulfillment of clinical and/or immunological criteria set by WHO.MethodsCase-control study was carried out from November to December 2014. Cases were adult clients who were on failing first line regimen and on active follow up while controls were those adult clients on a non-failing first-line regimen for 36 months and above and on active follow up. Data was entered in to Epi Info version 7 and was exported to SPSS version 20 for analysis. Binary logistic regression model was used to identify predictors of ART failure.ResultsA total of 59 cases and 245 controls were included in the analysis. Sixty three percent of the participants were females and the median age at ART enrollment was 33 years (IQR; 28, 40). The median baseline CD4count was not significantly different among cases and controls (105 (IQR = 60–174)vs.131 (IQR = 72.5–189.0); p = 0.301). The median peak CD4 count in the failure group (230 (IQR = 123–387)) was significantly low compared to the non-failure group (463 (IQR = 348.5–577)) [p < 0.001]. High peak CD4count (AOR = 0.993; 95% CI 0.990, 0.996) and longer duration on ART (AOR = 0.923; 95% CI 0.893, 0.954) were protective of treatment failure. In addition stavudine based regimen (AOR = 3.47; 95% CI 1.343, 10.555), low baseline BMI (AOR = 2.75; 95% CI 1.012, 7.457), unemployment (AOR = 4.93; 95% CI 1.493, 16.305) and formal educational level (AOR = 5.15; 95% CI 1.534, 17.276) were independently significant predictors of treatment failure.ConclusionsIn this setting low peak CD4count, shorter duration on first line ART, d4T based regimen, low baseline BMI, unemployment and formal educational level were significantly associated with increased treatment failure. Retaining patients on their initial first line regimen with appropriate follow up and improving their socioeconomic status through various livelihood initiatives should be strengthened.