Table 4.
Longitudinal Analysis of HIV/STD Risk Behaviors, Adjusting for Clustering Within Randomized Group (Unadjusted and Adjusted for Baseline Response)a
| Treatment Effects | Proportion of Condom-Protected Sex |
Consistent (100%) Condom Use |
(log)Unprotected Sex |
Concurrent Partners |
||||
|---|---|---|---|---|---|---|---|---|
| RR (95% CI)b | P Value | RR (95% CI)b | P Value | Difference (95%CI)c | P Value | RR (95% CI)b | P Value | |
| Unadjusted for Baseline Response | ||||||||
| Baseline | 0.98 (0.77 to 1.24) | .84 | 0.81 (0.52 to1.27) | .36 | 0.35 (–0.21 to 0.90) | .22 | 1.06 (0.74 to 1.52) | .74 |
| Over entire FU | 1.36 (1.16 to 1.59) | <.001 | 1.23 (1.02 to 1.50) | .03 | –0.93 (–1.46 to –0.41) | <.001 | 1.04 (0.80 to 1.34) | .81 |
| IPT | 1.89 (1.49 to 2.40) | <.001 | 1.47 (1.17 to 1.85) | .003 | –1.44 (–2.18 to –0.70) | <.001 | 0.95 (0.64 to 1.42) | .81 |
| 6 mo | 1.37 (1.10 to 1.72) | .008 | 1.44 (1.13 to 1.83) | .006 | –1.65 (–2.41 to –0.90) | <.001 | 1.05 (0.73 to 1.49) | .81 |
| 12 mo | 1.34 (1.04 to 1.72) | .02 | 1.35 (1.07 to 1.71) | .02 | –0.99 (–1.76 to –0.22) | .01 | 1.09 (0.82 to 1.46) | .81 |
| ICC | 0.48 | 0.31 | 0.42 | 0.42 | ||||
|
Adjusted for Baseline Response | ||||||||
| Over entire FU | 1.24 (1.09 to 1.41) | .006 | 1.45 (1.24 to 1.70) | <.001 | –1.52 (–2.07 to –0.98) | <.001 | 1.01 (0.81 to 1.25) | .95 |
| IPT | 1.49 (1.13 to 1.95) | .009 | 1.39 (1.13 to 1.71) | .002 | –1.63 (–2.30 to –0.95) | <.001 | 1.06 (0.76 to 1.49) | .95 |
| 6 mo | 1.22 (1.05 to 1.41) | .01 | 1.57 (1.27 to 1.94) | <.001 | –1.79 (–2.50 to –1.08) | <.001 | 0.96 (0.71 to 1.29) | .95 |
| 12 mo | 1.05 (0.85 to 1.30) | .64 | 1.40 (1.13 to 1.75) | .003 | –1.15 (–1.88 to –0.42) | .002 | 1.01 (0.78 to 1.30) | .95 |
| ICC | 0.34 | 0.37 | 0.41 | 0.31 | ||||
Abbreviations: CI, confidence interval; ICC, estimated intraclass correlation coefficient from exchangeable working correlation matrix; FU, follow-up; HIV/STD, human immunodeficiency virus/sexually transmitted disease; IPT, immediate posttest; RR, risk ratio.
All P values were adjusted.
Empirical RR (risk reduction vs health promotion) estimates examining treatment effects for behavioral outcomes of interest with “independence” working correlation specified.
Difference (risk reduction minus health promotion) estimates examining treatment effects for behavioral outcomes of interest with “independence” working correlation specified.