G/dL; grade 3: Hb 6.5 to,7 g/dL and grade 4: Hb,6.5 g/dL.Data collection and statistical analysisAll available clinical and laboratory data were prospectively recorded using standardized data collection tools and entered prospectively in a local electronic database. Quality control of the entered data was done at regular intervals using standard of procedure for quality monitoring [19]. The primary outcome was time to AZT discontinuation due to anemia (treatment-limiting or `severe’ anemia) within the first year after AZT initiation. Follow-up time was censored at the date of AZT discontinuation, death, last visit within the first year after AZT initiation and 12926553 August 15, 
2011 for the remainder. Cumulative incidence of AZT-related anemia within the first year after AZT initiation was estimated using Kaplan-Meier methods. Body weight at the time of AZT initiation was taken as main exposure, categorized into Anlotinib site clinically meaningful categories(.60 kg, 50?0 kg, 40?0 kg and,40 kg). We constructed a Cox proportional hazard models to study the association between body weight at the time of AZT initiation and risk of AZT-related anemia within the first year of AZT use, adjusted for confounding factors. The following factors were considered a priori for inclusion: MNS web hemoglobin levels and CD4 cell count at time of AZT initiation, time on ART prior to AZT initiation. A number of additional factors were considered as potential confounders for inclusion in the model: use of cotrimoxazole and fluconazole concurrent with AZT initiation, age, sex, baseline WHO clinical stage. Starting from the full model including all co-variates, a backward selection process was performed by observing the effect on the outcome of removing every individual predictor (besides the main exposure and the a priori identified confounders) one by one, starting with the variable with the weakest association with the outcome. Subsequently, all co-variates were added again in the model in a forward selection process to observe whether joint effects of co-variates existed. Co-variates were retained in the model if their removal/inclusion induced a change of.10 in the measure of effect of the main exposure or they were significantly associated with the outcome in adjusted analysis. Interactions were explored guided by the bi-variate analysis and current knowledge. Since the proportional hazard assumption – tested graphically and formally using Schoenfeld residuals – was violated for theMethods Study design and study populationThis was a retrospective study using data routinely collected at each consultation at SHCH between March 2003 and August 2011. The SHCH is a tertiary hospital run by a non-governmental organization, situated in the capital, Phnom Penh, Cambodia. The 15755315 hospital provides free care to poor patients, including treatment of opportunistic infections and ART for HIV-infected adult patients. All consecutive ART-naive patients starting ART with a D4Tbased regimen and substituting AZT for D4T due to D4Tintolerance between March 2003 and July 15, 2011 with follow-up till August 15, 2011 were included. Individuals with ART exposure prior to initiation of D4T-based ART, and patients missing baseline hemoglobin 
and without at least one follow-up hemoglobin result were excluded.HIV treatment and monitoringIn line with WHO and national guidelines, [15,18] the following ART eligibility criteria were used: 1) CD4 cells count#200 cells/ mL; 2) WHO clinical stage 3 with CD4 cells coun.G/dL; grade 3: Hb 6.5 to,7 g/dL and grade 4: Hb,6.5 g/dL.Data collection and statistical analysisAll available clinical and laboratory data were prospectively recorded using standardized data collection tools and entered prospectively in a local electronic database. Quality control of the entered data was done at regular intervals using standard of procedure for quality monitoring [19]. The primary outcome was time to AZT discontinuation due to anemia (treatment-limiting or `severe’ anemia) within the first year after AZT initiation. Follow-up time was censored at the date of AZT discontinuation, death, last visit within the first year after AZT initiation and 12926553 August 15, 2011 for the remainder. Cumulative incidence of AZT-related anemia within the first year after AZT initiation was estimated using Kaplan-Meier methods. Body weight at the time of AZT initiation was taken as main exposure, categorized into clinically meaningful categories(.60 kg, 50?0 kg, 40?0 kg and,40 kg). We constructed a Cox proportional hazard models to study the association between body weight at the time of AZT initiation and risk of AZT-related anemia within the first year of AZT use, adjusted for confounding factors. The following factors were considered a priori for inclusion: hemoglobin levels and CD4 cell count at time of AZT initiation, time on ART prior to AZT initiation. A number of additional factors were considered as potential confounders for inclusion in the model: use of cotrimoxazole and fluconazole concurrent with AZT initiation, age, sex, baseline WHO clinical stage. Starting from the full model including all co-variates, a backward selection process was performed by observing the effect on the outcome of removing every individual predictor (besides the main exposure and the a priori identified confounders) one by one, starting with the variable with the weakest association with the outcome. Subsequently, all co-variates were added again in the model in a forward selection process to observe whether joint effects of co-variates existed. Co-variates were retained in the model if their removal/inclusion induced a change of.10 in the measure of effect of the main exposure or they were significantly associated with the outcome in adjusted analysis. Interactions were explored guided by the bi-variate analysis and current knowledge. Since the proportional hazard assumption – tested graphically and formally using Schoenfeld residuals – was violated for theMethods Study design and study populationThis was a retrospective study using data routinely collected at each consultation at SHCH between March 2003 and August 2011. The SHCH is a tertiary hospital run by a non-governmental organization, situated in the capital, Phnom Penh, Cambodia. The 15755315 hospital provides free care to poor patients, including treatment of opportunistic infections and ART for HIV-infected adult patients. All consecutive ART-naive patients starting ART with a D4Tbased regimen and substituting AZT for D4T due to D4Tintolerance between March 2003 and July 15, 2011 with follow-up till August 15, 2011 were included. Individuals with ART exposure prior to initiation of D4T-based ART, and patients missing baseline hemoglobin and without at least one follow-up hemoglobin result were excluded.HIV treatment and monitoringIn line with WHO and national guidelines, [15,18] the following ART eligibility criteria were used: 1) CD4 cells count#200 cells/ mL; 2) WHO clinical stage 3 with CD4 cells coun.
