Table 3.
Comparison of Outcomes, Assessed Three Months Post Intervention, Between 266 Young African American Men Randomly Assigned to Intervention and Control Conditions
| Intervention (n = 141) |
Control (n = 125) |
Univariate Measure of Effect Estimate; [95% CI] p-value |
Multivariable Measure of Effect Estimate; [95% CI] p-value |
|
|---|---|---|---|---|
| Reinfection1 | 45 (31.9%) |
63 (50.4%) |
0.46; [0.28, 0.76] 0.002 |
0.32; [0.12, 0.86] 0.02 |
| Condom use skills2 complete case analysis |
5.35 ± 1.21 [104] |
2.18 ± 1.30 [84] |
3.17; [2.81, 3.53] < 0.0001 |
3.21; [2.80, 3.63] < 0.0001 |
| Condom use skills2 multiple imputation |
3.17; [2.79, 3.54] < 0.0001 |
3.19; [2.81, 3.56] < 0.0001 |
||
| Partners in last 3 months3, 4 complete case analysis |
2.06 ± 1.65 [105] |
4.15 ± 5.59 [91] |
−2.10; [−3.22, −0.98] 0.0003 |
−2.09; [−3.18, −0.99] 0.0002 |
| Partners in last 3 months4 multiple imputation |
−1.85; [−2.97, −0.74] 0.002 |
−1.87; [−2.96, −0.79] 0.001 |
||
| Unprotected acts, last 3 months5,6 complete case analysis |
12.3 ± 25.8 [99] |
29.4 ± 79.3 [84] |
−17.1; [−33.6, −0.5] 0.045 |
−13.4; [−35.6, 8.8] 0.23 |
| Unprotected acts, last 3 months6 multiple imputation |
−14.9; [−31.0, 1.3] 0.07 |
−11.9; [−31.3, 7.5] 0.21 |
||
| Condom used at last act7 complete case analysis |
76 (72.4%) [105] |
49 (53.9%) [91] |
2.25; [1.24, 4.07] 0.008 |
2.20; [1.08, 4.48] 0.03 |
| Condom used at last act7 multiple imputation |
2.27; [1.23,4.19] 0.009 |
2.06; [1.07,3.96] 0.03 |
Entries in the “Intervention” and “Control” columns are Mean ± SD for quantitative variables and Number (Percent) for dichotomous variables.
For quantitative variables, the measure of effect is a mean difference (expected score for intervention participant minus expected score for control participant, adjusted in the multivariable analyses for covariates specified below) and is estimated by linear regression.
For dichotomous variables, the measure of effect is an odds ratio (odds in favor for intervention participant divided by odds in favor for control participant, adjusted in the multivariable analyses for covariates specified below) and is estimated by logistic regression.
Complete case analyses use only those subjects for whom there are no missing values on variables in the regression model. Multiple imputation analyses use all subjects. All results pertain to men who self-identified as heterosexual. For variables on which not all subjects had data, the numbers in square brackets identify how many subjects did have data.
Multivariable analysis controls for monthly income level, having one versus two or more STDs diagnosed at study enrollment (“mixed STDs”), follow-up values for condom skills, and follow-up values for condom use at last sex
Multivariable analysis controls for income, mixed STDs, and the baseline value of condom skills
Median and interquartile range are 2.0, 3.0 for control group and 2.0, 2.0 for intervention group. Excluding four subjects in control group and three in intervention group who claimed more than 100 unprotected acts at baseline or follow-up or who claimed more than 25 partners at baseline or follow-up, mean and standard deviation are 3.52, 4.04 for control group and 2.00, 1.47 for intervention group. See also Table 3.
Multivariable analysis controls for income, mixed STDs, and the baseline value for number of female sex partners in the past 3 months
Median and interquartile range are 4.5, 21.0 for control group and 1.0, 11.0 for intervention group. Excluding four subjects in control group and three in intervention group who claimed more than 100 unprotected acts at baseline or follow-up or who claimed more than 25 partners at baseline or follow-up, mean and standard deviation are 17.24, 28.77 for control group and 11.12, 21.96 for intervention group. See also Table 3.
Multivariable analysis controls for income, mixed STDs, and baseline values for skill, unprotected sex, and condom use at last sex
Multivariable analysis controls for income, mixed STDs, and the baseline values for skills and condom use at last sex