Skip to main content
. 2018 Mar 4;66(Suppl 2):S140–S146. doi: 10.1093/cid/cix1142

Table 2.

Independent Predictors of Mortality in the First 48 Weeks on Antiretroviral Therapy Initiated With CD4 Count <100 Cells/µL

Factor Univariable Effects of Factors Selected for Multivariable Model in REALITY Trial Participants Final Multivariable Model in REALITY Trial Participants Multivariable Model From Bisson et al [10], Fitted to REALITY Trial Participants
HR (95% CI) P Value HR (95% CI) P Value HR (95% CI) P Value
Age (per 5 y older) 1.09 (1.03–1.16) .005 1.12 (1.04–1.20) .002 1.13 (1.06–1.21) <.001
CD4 (per 10 cells/µL higher) 0.89 (.85–.94) <.001 0.90 (.85–.95) <.001 0.87 (.82–.92) <.001
Hemoglobin (per g/dL higher) 0.77 (.72–.82) <.001 0.90 (.83–.99) .01 0.87 (.80–.94) <.001
Albumin (per g/L higher) 0.92 (.90–.93) <.001 0.96 (.94–.98) .001 0.94 (.92–.97) <.001
WBC count (per 109/L higher) 1.17 (1.08–1.26) <.001 1.08 (1.00–1.17) .054 1.16 (1.06–1.25) .001
Grip strength (per kg higher) 0.93 (.91–.95) <.001 0.98 (.96–1.00) .03
Previous healthcare contact 1.16 (.75–1.79) .50 0.63 (.39–1.03) .067
WHO stage 3/4 weight loss 1.90 (1.42–2.53) <.001 1.43 (1.01–2.03) .04
Patient-reported fever 3.37 (2.54–4.47) <.001 1.67 (1.19–2.35) .003
Patient-reported vomiting 3.72 (2.64–5.26) <.001 1.55 (1.00–2.39) .048
EQ-5D Mobility: some problems vs no problems 4.42 (3.36–5.81) <.001 1.88 (1.21–2.93) .005
Confined to bed vs no problems 11.87 (7.00–20.12) <.001 2.47 (1.00–6.08) .050
EQ-5D Self-care: some problems vs no problems 3.73 (2.78–5.02) <.001 1.32 (.82–2.13) .26
Unable to wash/dress vs no problems 8.59 (5.91–12.47) <.001 2.78 (1.42–5.43) .003
Enhanced prophylaxis 0.75 (.57–.97) .03 0.72 (.54–.96) .02
Sex (female vs male) 1.23 (.95–1.60) .12 1.19 (.89–1.60) .23
BMI <18.5 vs >25 kg/m2 2.41 (1.30–4.48) .005 0.90 (.52–1.55) .70
BMI 18.5–25 vs >25 kg/m2 1.38 (.74–2.58) .32 0.70 (.41–1.19) .19
HIV VL (per log10 higher) 1.52 (1.22–1.89) <.001 1.37 (1.09–1.74) .008
Neutrophil % (per 5% higher) 1.07 (1.02–1.12) .003 0.99 (.95–1.04) .64

Final multivariable model in REALITY trial participants also adjusted for center. Backwards elimination (exit P < .1) on factors in Table 1, refitting final model to all observations with complete data for chosen factors. Factors with P > .05 are included to adjust for confounding. All continuous factors were selected in final models with linear effects, ie, each unit increase had the same impact on mortality at all levels of the factor.

Abbreviations: BMI, body mass index; CI, confidence interval; EQ-5D, EuroQol 5 Dimensions (score); HIV, human immunodeficiency virus; HR, hazard ratio; REALITY, Reduction in EArly MortaLITY (trial); VL, viral load; WBC, white blood cell; WHO, World Health Organization.