Table 3. General Estimating Equation Empirical Significance Tests and Effect Size Estimates for the Overall Intervention Effect Unadjusted for Baseline Prevalence and Adjusted for Baseline Prevalence: Eastern Cape Province, South Africa, 2007–2010.

Outcome	ICC	Unadjusted for Baseline			Adjusted for Baseline
		Estimatea (95% CI)	P	d	Estimatea (95% CI)	P	d
Consistent condom use	0.006	1.18 (0.96, 1.46)	.123	0.10	1.32 (1.03, 1.71)	.008	0.17
Proportion condom-protected vaginal intercourse	0.024	0.04 (-0.01, 0.08)	.1	0.12	0.06 (0.01, 0.10)	.014	0.20
Used a condom at last vaginal intercourse	0.028	1.31 (1.04, 1.64)	.02	0.16	1.40 (1.08, 1.82)	.011	0.20
Frequency of condom use	0.006	1.34 (1.11, 1.63)	.003	0.18	1.41 (1.13, 1.76)	.002	0.21
Talked to steady partner about condom use	0.017	1.37 (1.09, 1.73)	.006	0.19	1.50 (1.16, 1.93)	.002	0.24
Unprotected vaginal intercourse	0.001	0.95 (0.81, 1.13)	.592	0.02	0.95 (0.81, 1.13)	.578	0.03
Anal intercourse	–0.001	0.86 (0.60, 1.23)	.409	0.07	0.85 (0.59, 1.23)	.385	0.10
Multiple vaginal intercourse partners	0.016	0.95 (0.78, 1.16)	.618	0.03	0.89 (0.73, 1.10)	.297	0.07

Note. CI = confidence interval; ICC = intraclass correlation coefficient. d is the effect size estimate in standard deviation units based on the mean difference or Cox transformation of the odds ratio.³¹

Intervention effect is average over the 6- and 12-month follow-up assessments.

^aEstimates are odds ratios (intervention vs health control) for all outcomes except proportion condom-protected vaginal intercourse where it is the mean difference (intervention—control).