Strategies for improving postpartum contraceptive use: evidence from non-randomized studies.

Background: Nearly two-thirds of women in their first postpartum year have an unmet need for family planning. Adolescents often have repeat pregnancies within a year of giving birth. Women may receive counseling on family planning both antepartum and postpartum. Decisions about contraceptive use made right after counseling may differ considerably from actual postpartum use. In earlier work, we found limited evidence of effectiveness from randomized trials on postpartum contraceptive counseling. For educational interventions, non-randomized studies may be conducted more often than randomized trials.
Objectives: We reviewed non-randomized studies of educational strategies to improve postpartum contraceptive use. Our intent was to examine associations between specific interventions and postpartum contraceptive use or subsequent pregnancy.
Search Methods: We searched for eligible non-randomized studies until 3 November 2014. Sources included CENTRAL, PubMed, POPLINE, and Web of Science. We also sought current trials via ClinicalTrials.gov and ICTRP. For additional citations, we examined reference lists of relevant reports and reviews.
Selection Criteria: The studies had to be comparative, i.e., have intervention and comparison groups. The educational component could be counseling or another behavioral strategy to improve contraceptive use among postpartum women. The intervention had to include contact within six weeks postpartum. The comparison condition could be another behavioral strategy to improve contraceptive use, usual care, other health education, or no intervention. Our primary outcomes were postpartum contraceptive use and subsequent pregnancy.
Data Collection and Analysis: Two authors evaluated abstracts for eligibility and extracted data from included studies. We computed the Mantel-Haenszel odds ratio (OR) for dichotomous outcomes and the mean difference (MD) for continuous measures, both with 95% Confidence Intervals (CI). Where studies used adjusted analyses for continuous outcomes, we presented the results as reported by the investigators. Due to differences in interventions and outcome measures, we did not conduct meta-analysis. To assess the evidence quality, we used the Newcastle-Ottawa Quality Assessment Scale.
Main Results: Six studies met our inclusion criteria and included a total of 5143 women. Of three studies with self-reported pregnancy data, two showed pregnancy to be less likely in the experimental group than in the comparison group (OR 0.48, 95% CI 0.27 to 0.87) (OR 0.60, 95% CI 0.41 to 0.87). The interventions included a clinic-based counseling program and a community-based communication project.All studies showed some association of the intervention with contraceptive use. Two showed that treatment-group women were more likely to use a modern method than the control group: ORs were 1.77 (95% CI 1.08 to 2.89) and 3.08 (95% CI 2.36 to 4.02). In another study, treatment-group women were more likely than control-group women to use pills (OR 1.78, 95% CI 1.26 to 2.50) or an intrauterine device (IUD) (OR 3.72, 95% CI 1.27 to 10.86) but less likely to use and injectable method (OR 0.23, 95% CI 0.05 to 1.00). One study used a score for method effectiveness. The methods of the special-intervention group scored higher than those of the comparison group at three months (MD 13.26, 95% CI 3.16 to 23.36). A study emphasizing IUDs showed women in the intervention group were more likely to use an IUD (OR 1.79, 95% CI 1.20 to 2.69) and less likely to use no method (OR 0.48, 95% CI 0.31 to 0.75). In another study, contraceptive use was more likely among women in a health service intervention compared to women in a community awareness program at four months (OR 1.79, 95% CI 1.40 to 2.30) or women receiving standard care at 10 to 12 months (OR 2.08, 95% CI 1.58 to 2.74). That study was the only one with a specific component on the lactational amenorrhea method (LAM) that had sufficient data on LAM use. Women in the health service group were more likely than those in the community awareness group to use LAM (OR 41.36, 95% CI 10.11 to 169.20).
Authors' Conclusions: We considered the quality of evidence to be very low. The studies had limitations in design, analysis, or reporting. Three did not adjust for potential confounding and only two had sufficient information on intervention fidelity. Outcomes were self reported and definitions varied for contraceptive use. All studies had adequate follow-up periods but most had high losses, as often occurs in contraception studies.