Table 4. Complimentary log-log regression results.
| Haiti, 15–24 years (N = 3744) | Mozambique, 15–24 years (N = 2173) | Eswatini, 15–29 years (N = 2528) | ||||
|---|---|---|---|---|---|---|
| Risk factor | ME | SE | ME | SE | ME | SE |
| History of STI in the past 12 months | 0.00616 | 0.0063 | 0.0469* | 0.0248 | N/A | -- |
| Sexually active in the past 4 weeks | 0.00174 | 0.0044 | -0.0132 | 0.0161 | N/A | -- |
| Early sexual debut | 0.0115** | 0.0049 | 0.0354** | 0.0163 | 0.133*** | 0.0223 |
| Multiple sex partners in the past 12 months | -0.00536 | 0.0125 | 0.0501* | 0.0279 | 0.045 | 0.0354 |
| Age disparate sex | 0.00635 | 0.0057 | 0.0643*** | 0.0155 | 0.1080*** | 0.0202 |
| Inconsistent condom use in last 12 months | Omitted | -- | Omitted | -- | 0.00767 | 0.0171 |
| Transactional sex | Omitted | -- | N/A | -- | 0.0292 | 0.0469 |
| Alcohol misuse | 0.00839* | 0.0051 | N/A | -- | 0.1460 | 0.0956 |
| Uses tobacco | 0.00022 | 0.0132 | N/A | -- | N/A | -- |
| Never attended or not currently enrolled in school | Omitted | -- | 0.0321 | 0.0196 | N/A | -- |
| Empirical optimal cut point estimation (Youden method) | ||||||
| Empirical optimal cut point | 0.0124 | 0.0668 | 0.1840 | |||
| AUROC curve at cut point | 0.65 | 0.60 | 0.62 | |||
| HIV prevalence | 1.1% | 9.4% | 21.2% | |||
ME = marginal effect, SE = standard error, AUROC = area under the receiving operating characteristics.
N/A: Data on risk factor was not collected in the country survey.
Risk factors were omitted from regression models due to missing values, insufficient response rates, and lack of variation among HIV-positive and HIV-negative AGYW.
*p<0.1,
**p<0.05,
***p<0.01.