Impact of Two Teen Pregnancy Prevention Interventions on Condom and Birth Control Use: Preliminary Results of a Three-Arm, Cluster Randomized Control Trial

Background and Purpose: The purpose of the grant funded study was to test the efficacy of two interventions on teen pregnancy prevention among high risk adolescents using a cluster randomized control trial design.  The hypothesis was that youth exposed to both interventions would gain in knowledge and use condoms and birth control more than youth in the control condition.

Method:  The program was called CHAMPS (Creating Healthy Adolescents through Meaningful Prevention Services). CHAMPS Camps were offered 38 times in 23 community-based organizations between September 2011 and March 2014 to youth between the ages of 14 and 19, who had never been pregnant or impregnated anyone.

Design:  The study is a three-arm (Reducing the Risk, Love Notes interventions and Power of We counterfactual), cluster randomized controlled trial (CRCT) that collected data at six points (pre-training baseline, immediate post-training, 3, 6, 12, and 24 months post-training). Measures included pre-post knowledge tests and use of condoms and other forms of birth control during the last three months.

Sample:  1,446 youth (515 RtR, 512 LN, and 419 POW) were recruited and 1378 youth received the entire intervention or control for an overall dosage rate of 95%.  82% were on free and reduced lunch, 89% were youth of color, 9% were African refugees and 21% were youth living in out of home care. 

Analysis:  To analyze data, the basic approach was Hierarchical Linear Modelling followed by repeated measures ANOVA or χ2. The analytic sample consisted of all trial participants who met program eligibility and data consistency requirements. Tests for baseline equivalence was conducted. Control for clustering is included as a Level-2 variable in HLM. When the impact of the clusters is non-significant in HLM, additional analyses are conducted with repeated measures ANOVA for greater interpretability of the means and interaction effects. Control for multiple comparison is done using appropriate Bonferroni and LSD statistics.

Results: Time 3 and 4 response rates were 76% and 69%, respectively. We found that youth in LN scored higher in the LN knowledge post-test (M= 55.44) than did RTR (50.34) or POW (45.07), F(2, 1352) = 58.28, p < .0001. Similarly, youth in RTR scored higher on the RTR knowledge post-test (M = 69.95) than did youth in LN (64.24) or POW (54.88), F(2,1352) = 75.89, p < .0001. 

There were no differences at the 3 month mark due to a spike in condom and birth control use in the POW control group at that point which was gone by the 6 month mark. By the six month mark, use of condoms had increased among intervention condition youth, but not enough to be higher than POW.   However, both RTR and LN were significantly more likely to use some other form of birth control at 6 months than the POW control group, 66% vs. 47%, F(1,156) 5.58, p < .02 and 67% vs. 47% F(1,142) = 5.59, p < .02.

Conclusions: Preliminary findings support the impact of RTR and LN on immediate knowledge and birth control use. Further explanatory analyses are forthcoming.