Table 2.

Multivariable model for viral load <400 copies per mL at week 144 in protease inhibitor and nucleoside reverse-transcriptase inhibitor group

	Unadjusted odds ratio (95% CI)	p value	Adjusted odds ratio (95% CI)	p value
Genomic susceptibility score of second-line regimen (per 0·5 higher)	0·78 (0·61–0·99)	0·04	0·61 (0·46–0·81)	0·001
Proportion of non-adherent visits (per 10% higher)*	0·66 (0·55–0·79)	<0·0001	0·61 (0·49–0·77)	<0·0001
Unemployed or student vs employed	0·39 (0·21–0·72)	0·003	0·22 (0·07–0·63)	0·005
Hours worked by employed and students (per 10 h higher)	1·03 (0·92–1·17)	0·6	0·83 (0·70–0·99)	0·04
Baseline viral load per doubling	0·82 (0·70–0·95)	0·01	0·80 (0·67–0·97)	0·02
Baseline CD4 cell count per doubling	1·24 (1·02–1·50)	0·03	1·33 (1·04–1·71)	0·02

n=317, excluding those with missing week 144 viral load, baseline genotype or baseline employment status. Baseline refers to switch to second-line therapy (enrolment into the trial). Univariable analyses are in the appendix. Adjusted odds ratio adjusted for the factors given in the table. Factors not selected (p>0·05): sex, age, viral subtype, years on first-line antiretroviral treatment, diabetes, family history of cardiovascular disease, previous CNS disease, previous tuberculosis, smoking, alcohol consumption, household income, food availability, years of education, estimated glomerular filtration rate, haemoglobin, glucose, presence of individual mutations in the baseline genotype (where >10% prevalence), presence of combinations of 2 mutations in the baseline genotype (where >10% prevalence).

^*Scheduled visit that was either missed or where the participant self-reported missing pills since the last visit.