Stata software version 15

We used descriptive statistics to produce counts and percentages of characteristics of the study population overall and by intervention arm. We performed univariate comparisons with a mixed effects regression adjusted for cluster of these characteristics to ensure randomization was effective, which it was; p-values were not significant and not shown. We also performed comparisons of implant use at 3 months (primary outcome) and overall method use at 3 months (secondary outcome) by intervention arm, using a mixed effects regression adjusted for cluster that converted the comparisons to risk ratios with 95% confidence intervals. We continued to use descriptive statistics to produce counts and percentages of contraceptive uptake and use by intervention arm and by study timepoints, describing initial method choices in the intervention clusters as well as use of methods by 3 months in all study participants. These same methods were used to describe secondary outcomes of continuation and satisfaction and reasons for contraceptive choices among study participants. STATA software version 15.2 (StataCorp LP, College Station, TX, USA) was used for analysis.

This analysis includes a description of the population of women observed during the study timeframe divided into the population of women who delivered a LBW infant compared to women with a normal weight infant. In bivariate comparisons we analyzed sociodemographic, antepartum, and intrapartum characteristics using Pearson’s chi-squared test for nominal categorical variables unless there was a low cell size, in which case Fisher’s exact test was used. The Kruskal-Wallis test was used for comparison of continuous variables. Variables with a P<0.05 were considered statistically significant, and then analyzed in a multivariable logistic regression. STATA software version 15.2 (StataCorp LP, College Station, TX, USA) was used for analysis.

Meta-analysis was performed on the primary and secondary outcome measures. Summary estimates, including 95% confidence intervals (CIs), were calculated. For discontinuous outcomes, such as proportion of patients who were diagnosed with ERM secondary to vitrectomy and the proportion of patients with recurrence of retinal detachment, a Risk Ratio (RR) with 95% CIs was calculated. Unfavorable events (ERM formation and RD recurrence) were used as the risk numerator. For mean BCVA change, which was analyzed as a continuous variable, Weighted Mean Difference (WMD) with 95% CIs was reported. Heterogeneity was assessed by using the chi-square test and I² statistic. An I² value > 50% was considered as moderate to high heterogeneity. Random-effects models were used as the primary approach to meta-analysis, except when only two studies were included (fixed effects used). Harbord’s modified test was used to test asymmetry within dichotomous outcomes, whereas Egger’s regression within continuous outcomes [17 (link)]. A P value < 0.05 was considered significant for all analyses. Meta-analysis was conducted using Stata software, version 15.2 (StataCorp, College Station, TX).

The data were collected, entered in Microsoft Excel (Microsoft, Redmond, WA, USA), and analyzed using the STATA software, version 15.2 (StataCorp, College Station, TX, USA). Descriptive statistics were used to summarize the results. Categorical variables were presented as frequencies and percentages, while continuous data were expressed as means and standard deviations suitable for normally distributed data and median and interquartile range (IQR) for nonnormal data. Correlation between two continuous variables was assessed using the Spearman correlation test on those variables, which were not normally distributed and the Pearson correlation test for those normally distributed. Normality was assessed using the Shapiro-Wilk test. The level of statistical significance was set at a P value of .05.