Skip to main content
. 2023 Feb 6;57:101836. doi: 10.1016/j.eclinm.2023.101836

Table 1.

Characteristics of study cohort and predictors of switching from efavirenz to dolutegravir (exposed) vs remaining on efavirenz (unexposed) among ART patients at the Themba Lethu Clinic in Johannesburg South Africa (n = 6948).

Variable at time of switch Unexposed
(n = 6154)
Exposed
(n = 794)
Adjusted Odds Ratioa
(95% CI)
Sex of patients
 Female 3819 (90.9) 384 (9.14) ref.
 Male 2335 (85.1) 410 (14.9) 1.94 (1.64, 2.31)
Age (years) at ART initiation
 18–24.9 343 (76.9) 103 (23.1) 2.50 (1.80, 3.48)
 25–29.9 781 (93.4) 55 (6.58) 0.54 (0.38, 0.77)
 30–39.9 2531 (90.5) 265 (9.48) 0.71 (0.56, 0.92)
 40–49.9 1776 (87.4) 256 (12.6) 0.89 (0.69, 1.15)
 ≥50 723 (86.3) 115 (13.7) ref.
CD4 count (cells/mm3) at time of switch
 <200 585 (90.3) 63 (9.72) ref.
 200–299 823 (87.2) 121 (12.8) 1.05 (0.74, 1.49)
 300–399 1067 (88.7) 136 (11.3) 0.86 (0.61, 1.22)
 400–499 1022 (89.3) 122 (10.7) 0.76 (0.53, 1.08)
 ≥500 2657 (88.3) 352 (11.7) 0.82 (0.60, 1.14)
Viral load suppression (<1000 copies/mL)
 No 319 (98.2) 6 (1.85) ref.
 Yes 5819 (88.1) 788 (11.9) 6.86 (0.82, 1.27)
Time (months) on ART at time of switch
 0–11.9 1159 (92.9) 89 (7.13) ref.
 12–23.9 917 (86.2) 147 (13.8) 2.03 (1.52, 2.70)
 24–35.9 771 (88.6) 99 (11.4) 1.55 (1.14, 2.13)
 36–47.9 643 (89.3) 77 (10.7) 1.57 (1.12, 2.20)
 48–59.0 686 (90.6) 71 (9.38) 2.45 (1.72, 3.48)
 ≥60 1978 (86.4) 311 (13.6) 8.85 (6.48, 12.1)
Haemoglobin (ug/dL) at time of switch
 <10.0 261 (94.6) 15 (5.43) ref.
 ≥10.0 5896 (88.3) 779 (11.7) 1.44 (0.83, 2.51)
Body mass Index (kg/m2) at time of switch
 <18.5 413 (94.3) 25 (5.71) ref.
 18.5–24.9 2449 (89.6) 285 (10.4) 1.67 (1.08, 2.59)
 25–29.9 1701 (88.3) 225 (11.7) 2.13 (1.36, 3.34)
 30–39.9 1591 (86.0) 259 (14.0) 3.03 (1.93, 4.76)
Hypertension at time of switch
 No 5271 (88.8) 663 (11.2) ref.
 Yes 883 (87.1) 131 (12.9) 1.02 (0.82, 1.27)
Nucleoside reverse transcriptase inhibitor
 emtricitabine 3322 (94.0) 213 (6.03) ref.
 lamivudine 2832 (83.0) 581 (17.0) 8.61 (6.80, 10.9)
a

Odds ratios calculated from a logistic regression model; STATA (v. 15.0) psmatch procedure was used to create propensity scores to match patients for analyses of our outcomes.