Table 2. Analysis of primary and secondary outcomes comparing difference in scores between intervention arms and survey time period among female and male 13–17 year olds, San Juan de Miraflores, Lima, Peru.
| Group Mean Estimates | Estimated Subject-Specific Mean Difference | ||||||
|---|---|---|---|---|---|---|---|
| Outcome | Baseline Group Mean Score (SE) | Endline Group Mean Score (SE) | Follow-up Group Mean Score (SE) | Unadjusted Mean Difference (MeanΔ1, 95%CI) | p-value | Adjusted2 Mean Difference (MeanΔ1, 95%CI) | p-value |
| Contraception myths and misconceptions index score (endline—baseline assessment) N = 659 | |||||||
| Arm 1: On-demand (n = 209) | 46.6 (0.84) | 43.1 (0.87) | -3.55% (-5.36%, -1.73%) | <0.001 | |||
| Arm 2: Push (n = 222) | 46.9 (0.85) | 41.0 (0.90) | -5.83% (-7.72%, -3.94%) | <0.001 | |||
| Arm 3: Control (n = 228) | 47.0 (0.71) | 45.0 (0.77) | -2.15% (-3.73%, -0.57%) | 0.008 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 3) | -1.40% (-3.79%, 1.00%) | 0.254 | -1.40% (-3.80%, 1.00%) | 0.254 | |||
| Mean (Δ Arm 2)—Mean (Δ Arm 3) | -3.68% (-6.14%, -1.22%) | 0.003 | -3.68% (-6.14%, -1.22%) | 0.003 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 2) | 2.28% (-0.34%, 4.91%) | 0.088 | 2.30% (-0.33%, 4.93%) | 0.087 | |||
| Retention of knowledge outcome (follow-up—endline assessment) N = 599 | |||||||
| Arm 1: On-demand (n = 188) | 42.9 (0.92) | 40.2 (0.94) | -2.63% (-4.51%, -0.74%) | 0.006 | |||
| Arm 2: Push (n = 199) | 41.3 (0.94) | 37.6 (0.89) | -3.72% (-5.64%, -1.81%) | <0.001 | |||
| Arm 3: Control (n = 212) | 45.2 (0.80) | 42.4 (0.79) | -2.77% (-4.43%, -1.12%) | 0.001 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 3) | 0.15% (-2.35%, 2.64%) | 0.908 | 0.13% (-2.37%, 2.64%) | 0.916 | |||
| Mean (Δ Arm 2)—Mean (Δ Arm 3) | -0.95% (-3.47%, 1.57%) | 0.460 | -0.89% (-3.41%, 1.64%) | 0.492 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 2) | 1.10% (-1.59%, 3.79%) | 0.424 | 1.09% (-1.60%, 3.79%) | 0.426 | |||
| Content exposure outcome (endline—baseline assessment) N = 478 | |||||||
| Arm 1: On-demand (n = 28) | 51.8 (4.35) | 40.2 (4.90) | -11.58% (-19.79%, -3.36%) | 0.006 | |||
| Arm 2: Push (n = 222) | 48.0 (1.02) | 39.9 (1.09) | -8.08% (-10.20%, -5.96%) | <0.001 | |||
| Arm 3: Control (n = 228) | 47.0 (0.71) | 45.0 (0.77) | -2.15% (-3.73%, -0.57%) | 0.008 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 3) | -9.43% (-14.76%, -4.09%) | 0.001 | -9.47% (-14.83%, -4.11%) | 0.001 | |||
| Mean (Δ Arm 2)—Mean (Δ Arm 3) | -5.93% (-8.57%, -3.29%) | <0.001 | -5.93% (-8.57%, -3.29%) | <0.001 | |||
| Mean (Δ Arm 1)—Mean (Δ Arm 2) | -3.50% (-10.14%, 3.14%) | 0.302 | -3.54% (-10.23%, -3.14%) | 0.299 | |||
Notes:
1Δ refers to the subject-specific change in the outcome from timepoint 1 to timepoint 2. 95%CI refers to the 95% confidence interval.
2Adjusted estimates control for the time in which the endline and follow-up surveys were conducted. A generalized linear model (GLM) using a normal distribution and identity link was used to compare scores. There was a delay in the timing of the endline survey for many participants, which ranged from 0 to 72 days (mean = 13.30, SD = 11.28). There was also delay in the timing of the follow-up survey for many participants, which ranged from 13 to 121 days (mean = 64.10, SD = 8.97).