Table 2: Pre-treatment serum phosphate and odds ratios of death prior to 12 weeks among study participants, stratified by BMI ( )*.

BMI rangePatients at risk (deaths at 12 weeks)Median pretreatment serum phosphate, mmol/L (IQR)Odds ratio of mortality, per 0.1 mmol/L decrease in pre-treatment serum phosphate (95% CI)** value

All patients305 (40)1.2 (1.0, 1.4)1.13 (1.00, 1.28)0.06
≥18.5 kg/m2161 (12)1.2 (1.0, 1.4)0.96 (0.76, 1.21)0.74
<18.5 kg/m2144 (28)1.3 (1.1, 1.4)1.24 (1.05, 1.47)0.01
<17 kg/m286 (21)1.3 (1.1, 1.4)1.25 (1.02, 1.52)0.03
<16 kg/m249 (13)1.3 (1.1, 1.4)1.06 (0.79, 1.42)0.70

Two patients with known vital status at 90 days (one alive and one deceased) were missing baseline BMI and excluded from the analysis.
**Adjusted for sex, age, and lymphocyte count at ART initiation.
The effect of lower serum phosphate on the odds ratio of mortality remained significant for participants with BMI <18.5 kg/m2 when the model was further adjusted for pretreatment serum hsCRP, ferritin, and albumin levels (OR 1.22, 95% CI: 1.01 to 1.48; ). When adjusted for concomitant tuberculosis treatment the odds ratio of mortality also remained significant for participants with BMI <18.5 kg/m2 (OR 1.27, 95% CI: 1.06 to 1.51; ).