Univariable and multivariable logistic regression model of clinical characteristics associated with incidence of diabetes (n=703)
| OR (95% CI) | Adjusted OR (95% CI) | |
| Sex at birth | ||
| Female | 0.73 (0.38 to 1.39) | 0.53 (0.23 to 1.22) |
| Male (ref) | 1.00 | 1.00 |
| Calendar year of first antiretroviral therapy | ||
| 1997–1999 | 4.11 (2.34 to 7.19) | 48.90 (21.32 to 112.17) |
| 2000–2004 | 3.57 (2.01 to 6.32) | 7.22 (3.7 to 14.1) |
| 2005–2009 (ref) | 1.00 | 1.00 |
| 2010–2015 | 0.17 (0.04 to 0.75) | 0.05 (0.01 to 0.24) |
| CD4 nadir (per 100 cells/mm3 increase) | 0.64 (0.53 to 0.78) | 0.79 (0.61 to 1.02) |
| Time from antiretroviral therapy initiation to either the date of diabetes diagnosis or the end of study (per year increase) | 0.94 (0.90 to 0.98) | 0.69 (0.64 to 0.74) |
| Proportion of follow-up time with plasma viral load ≥500 copies/mL (per 10% increase) | 1.12 (1.04 to 1.22) | 0.93 (0.83 to 1.03) |
| Proportion of follow-up time on antiretroviral therapy agents (per 10% increase) | ||
| Stavudine | 1.39 (1.27 to 1.53) | Not selected |
| Zidovudine | 1.19 (1.08 to 1.31) | |
| Lopinavir | 1.19 (1.11 to 1.28) | |
| Indinavir | 1.44 (1.20 to 1.73) | |
| Saquinavir | 1.17 (0.94 to 1.46) | |
| Nelfinavir | 1.63 (1.19 to 2.24) |