The objectives of this study were to examine the association of nutritional status prior to ART initiation, using height-normalized measures of body mass, fat-free mass, fat mass, and skin fold thickness, with the change in CD4 count at an average of 6, 12, and 24 months after ART initiation in a cohort of Rwandan women with a high prevalence of malnutrition at study entry. We hypothesized that poorer nutritional status would be associated with smaller gains in CD4 count in Rwandan women initiating ART.
The Rwanda Women's Interassociation Study and Assessment (RWISA) is an observational prospective cohort of 710 ART-naïve (at enrollment) HIV-infected and 226 HIV-uninfected Rwandan women enrolled in 2005 
. Briefly, participants met the following inclusion criteria: age 25 years or older at study entry, willingness to give informed consent, presence in Rwanda during 1994. Participants were excluded if there was a prior history of receiving antiretroviral treatment other than single-dose nevirapine to prevent mother-to-child transmission of HIV. At study entry and at six-month interval visits, participants provided historical information including socio-demographics, medical history and symptoms, anthropometric measurements and blood specimens.
At each follow up visit, participants were asked whether they had initiated ART. Women reporting ART initiation provided written documentation of the exact date of ART initiation and medication regimen. From 2005 through 2007, following WHO and Rwandan guidelines, women in this cohort were eligible for ART if they had: WHO Stage IV disease, irrespective of CD4 cell count; WHO Stage III disease with CD4 <350 cells/µL; or CD4 <200/µL regardless of clinical stage. In 2008 Rwandan guidelines expanded the CD4 criterion to include ART initiation in all patients with CD4<350 cells/µl.
Included in this analysis are all HIV-positive RWISA participants who initiated ART after study entry, had a CD4 count obtained within one year prior to ART initiation (pre-ART CD4 count), nutritional measures at the pre-ART CD4 visit, and at least 6 months of follow-up with a CD4 measurement.
Body composition and anthropometric measurements
Height and weight were measured while the participant was wearing light clothing and no shoes. Body impedance analysis (BIA) was performed twice using a standard tetrapolar electrode placement on the hand and foot with resistance and reactance recorded. Skin-fold measurements were obtained by study nurses who were trained on techniques for standardized anthropometric measurement, using correct anatomic location of skin folds.
CD4 counts were determined with a FACS counter (Becton and Dickinson, Immunocytometry Systems, San Jose, CA, USA).
The main outcome of interest was change in CD4 count from the pre-ART to the 6, 12, and 24 month follow-up visits. Pre-ART values were defined as those measured at the study visit which fell between 1 day and 12 months prior to the exact date of ART initiation. If there were multiple visits in this time window, we chose the date closest to the exact date of ART initiation. The 6-, 12- and 24-month follow up visits were defined as the study visits which fell between 3 and 9 months, 9 and 15 months, and 21 and 27 months respectively after the exact date of ART initiation. Because some participants did not complete their follow-up at exact 6 month intervals, it was possible to have more than one visit in the follow up range, in which case the visit closest to the 6, 12, or 24 month follow up date was used.
The change in CD4 was calculated as the absolute CD4 count at the follow up visit minus the pre-ART CD4 count. Since the pre-ART CD4 count visit was on average 91 days (3 months) prior to the exact date of ART initiation, the change in CD4 at 6 months reflects on average a change over 9 months (with on average 6 of these months on ART). The change in CD4 at 12 months reflects on average, a change over 15 months (with on average 12 months on ART), and the change in CD4 at 24 months reflects on average a change over 27 months (with on average 24 months on ART).
Nutritional predictor variables
Four nutritional indicators were used, all from the pre-ART visit: body mass index (BMI); fat free mass index (FFMI); fat mass index (FMI); and the sum of skinfold measurement at the mid triceps, front thigh, and sub-scapular regions.
BMI, obtained from standing height and weight measurements, was calculated as weight divided by height-squared (kg/m2
). We used WHO-established BMI cutoffs for nutritional status: malnourished (BMI < 18.5 kg/m2
), normal (BMI 18.5–25 kg/m2
), and overweight (BMI≥25 kg/m2
FFMI was obtained from BIA as described above. Resistance and reactance were entered into standard formulae to calculate fat free mass in kg; these formulae have been previously validated in a multi-ethnic HIV and non-HIV population in the United States, and have been used in several African-studies 
. For <5% of the participants the calculated FFM exceeded weight and was thus set to equal the weight. We then standardized the calculated fat free mass by dividing it by height in meters, squared, to obtain the FFMI (kg/m2
). The FMI was calculated as weight in kg minus fat free mass in kg, and was standardized for height by dividing by height-squared (kg/m2
Skinfolds measurements were taken twice by the same person at each site of mid triceps, front thigh, and sub-scapular region, and the average of the two measurements was used for analysis. If the two measurements differed by ≥2.0 mm, a third measure was taken and the closest two were averaged. The sum of the skinfolds measurement was the sum of the triceps, thigh and sub-scapular measurements in centimeters.
We included the following covariates obtained at the pre-ART visit because of their potential to confound the relationship between malnutrition and change in CD4 count: age, income, education, CD4 count, and self-reported prior occurrence of a Stage 4 WHO AIDS defining illness (ADI). Smoking was not included in the analysis as less than 3% of the women smoked.
Age was included as age per 5 years. Income was categorized as <10,000 Rwandan francs (FRW) per month (in 2005, this was equivalent to < $17), 10,000 to 35,000 FRW per month, and greater than 35,000 FRW per month. Education was categorized as none, some primary school, completed primary school, or some secondary school or higher. Pre-ART CD4 count was included in the analysis per 100 cells/µL increment. We determined the presence of Stage 4 WHO illness from participant self-report at all visits prior to and including the pre-ART study visit.
Each participant provided written informed consent after viewing a video demonstrating study procedures, and discussing the study with research personnel. Study protocols were approved by the Rwanda National Ethics Committee and the Institutional Review Board of Montefiore Medical Center.
The primary outcomes were changes in CD4 count (as a continuous variable), from the pre ART visit at 6, 12, and 24 months of follow up, as defined above. Univariate linear regression analysis was performed for the change in CD4 count at the 6-, 12-, and 24-months post ART. All nutritional measures were analyzed as continuous variables. Other covariates were analyzed as described above.
For the multivariate analysis, we created four multivariate linear regression models each containing one of the nutritional variables alone in the model (BMI, FFMI, FMI or skinfold sum) with the covariates listed above. This was performed 3 times using change from pre-ART visit to 6, 12, and 24 months post-ART as the outcome. We used backwards selection analysis with a p-value of p
0.1 to stay. Variables that did not reach statistical significance were removed from the model until we obtained a final model. SAS software, version 9.1.3 (Cary, North Carolina) was used for the analysis.