Table 3.
Cox Proportional Hazard Regression for the Impact of Point-of-Care Versus Central Laboratory–Based Testing on Time to Sexually Transmitted Infection Treatment Initiation Among Women
| Variable | Category | Model A: NG/CT Outcome | Model B: TVa Outcome | ||||
|---|---|---|---|---|---|---|---|
| Treatments/PD | aHR (95% CI) | P Value | Treatments/PD | aHR (95% CI) | P Value | ||
| Age group, y | 18–24 | 413/8913.7 | 1 | 55/1312.2 | 1 | ||
| 25–34 | 138/3080.6 | 1.02 (.83–1.25) | .840 | 37/174.9 | 1.48 (.96–2.29) | .075 | |
| ≥35 | 21/230.4 | 1.51 (1.12–2.06) | .008 | 13/4.4 | 1.05 (.65–1.68) | .844 | |
| School level completed | High school | 354/7786.9 | 1 | 59/1208.1 | 1 | ||
| Primary school | 216/4341.4 | 1.16 (.45–2.98) | .753 | 44/460.3 | 0.99 (.66–1.47) | .955 | |
| No school | 2/96.4 | 1.21 (.47–3.13) | .689 | 2/0.3 | 0.96 (.61–1.51) | .863 | |
| CRSb | Verulam/Isipingoc | 402/11 396.3 | 1 | 78/1124.0 | 1 | ||
| eThekwinid | 170/828.3 | 39.62 (15.13–103.74) | <.001 | 27/534.8 | 0.93 (.59–1.48) | .770 | |
| Study visit type at which STI treatments were initiatedb | Scheduled | 406/9430.6 | 1 | 100/1634.9 | 1 | ||
| Unscheduled | 166/2794.1 | 0.76 (.52–1.11) | .152 | 5/33.9 | 0.71 (.38–1.32) | .281 | |
Denominators that do not equal the sample sizes are due to missing data.
Abbreviations: aHR, adjusted hazard ratio; CI, confidence interval; CRS, clinical research site; CT, Chlamydia trachomatis; NG, Neisseria gonorrhoeae; PD, person-days; STI, sexually transmitted infection; TV, Trichomonas vaginalis.
All CRSs used point-of-care (POC) assays for TV testing.
Variable specified as a time-varying covariate in Model A to satisfy proportional hazard assumption. Model B satisfied the proportional hazard assumption (Schoenfeld test P = .1570).
Central laboratory–based testing for NG/CT was conducted at the Isipingo and Verulam CRSs.
POC testing for NG/CT was conducted at the eThekwini CRS.