Mortality and clinical outcomes in peripartum cardiomyopathy: A systematic review and meta-analysis

Figures

• 

  Fig. 1. PRISMA flow diagram of study selection. Records were identified through PubMed, Embase, and Web of Science, followed by removal of duplicates and screening of titles and abstracts. Full-text articles were assessed for eligibility, with exclusions based on predefined criteria.

• 

  Fig. 2. Forest plot of pooled all-cause mortality in peripartum cardiomyopathy. Each square represents the estimate from an individual study, with horizontal lines indicating 95% confidence intervals. The diamond represents the pooled estimate derived using a random-effects model. Overall pooled all-cause mortality was 7% (95% CI 5–9%), with substantial heterogeneity across studies (I2 = 96.4%).

• 

  Fig. 3. Timing-stratified analysis of all-cause mortality in peripartum cardiomyopathy. Studies were stratified according to reported mortality assessment timing as early mortality (≤6 months) or later mortality (>6 months). Seventeen studies comprising 39,091 patients were included in the early mortality subgroup, with a pooled mortality rate of 3.4% (95% CI 1.7%–6.4%; I2 = 97.3%). Twenty-six studies comprising 3847 patients were included in the later mortality subgroup, with a pooled mortality rate of 8.1% (95% CI 5.9%–10.9%; I2 = 77.1%). Mortality was significantly higher in the >6-month subgroup (p for subgroup difference = 0.009). Studies without a clearly defined mortality assessment time point were excluded from this stratified analysis but retained in the overall mortality synthesis.

• 

  Fig. 4. Forest plot of heart failure hospitalization in peripartum cardiomyopathy. Three studies comprising 316 patients contributed to the analysis. The pooled HF hospitalization rate was 11% (95% CI 1–63%), with substantial heterogeneity (I2 = 90.7%). Given the limited number of studies and wide confidence interval, these findings should be interpreted cautiously.

• 

  Fig. 5. Leave-one-out sensitivity analysis for pooled all-cause mortality in peripartum cardiomyopathy. Each iteration represents the pooled estimate after sequential exclusion of a single study. The pooled mortality estimate remained approximately 7% across all iterations, with 95% confidence intervals consistently overlapping the overall estimate, indicating that no individual study disproportionately influenced the primary result.

• 

  Fig. 6. Funnel plot assessing potential publication bias for all-cause mortality in peripartum cardiomyopathy. Each point represents an individual study plotted against its standard error. Visual asymmetry was observed, and Egger's regression test demonstrated evidence of small-study effects (t = 2.34, df = 58, p = 0.023). Given the substantial between-study heterogeneity, this finding should be interpreted cautiously, as funnel plot asymmetry may reflect underlying clinical and methodological variability rather than publication bias alone.