FMT for chronic constipation: systematic review and meta-analysis using a tri-methodological framework
FMT favors clinical total response rate in chronic constipation (adjusted RR = 1.21; 95% CI: 1.10–1.33), but all included RCTs carry high risk of bias and the effect is attenuated after publication bias correction.
| Population | Adults with chronic constipation (RCTs: n=1040; single-arm studies: n=403; cohort studies: n=422) |
|---|---|
| Intervention | Fecal microbiota transplantation (FMT), regardless of administration route or protocol |
| Comparator | Control (placebo, sham, laxatives, or no intervention) in RCTs; no comparator in single-arm studies |
| Outcome | Clinical total response rate (trim-and-fill adjusted); Clinical total response rate (unadjusted); Wexner score (constipation severity); Remission/improvement rate (single-arm studies); Safety (serious adverse events); Constipation-related quality of life (PAC-QOL); Stool consistency (Bristol Stool Form Scale — BSFS) |
Summary of findings
| Outcome | Effect | 95% CI | Certainty | Clinical relevance | Notes |
|---|---|---|---|---|---|
| Clinical total response rate (trim-and-fill adjusted) | RR 1.21, 95% CI 1.10-1.33 | — | Moderate | — | 12 studies |
| Clinical total response rate (unadjusted) | RR 1.36, 95% CI 1.21-1.52, P<0.001 | — | Moderate | — | 12 studies |
| Wexner score (constipation severity) | MD -2.08, 95% CI -3.35 to -0.81, P=0.003 | — | Moderate | — | |
| Remission/improvement rate (single-arm studies) | proportion 0.517, 95% CI 0.403-0.629 | — | Low | — | 8 studies |
| Safety (serious adverse events) | no serious adverse events reported, follow-up up to 24w | — | Low | — | |
| Constipation-related quality of life (PAC-QOL) | pre-post improvement reported, in the between-group effect size available | — | Low | — | |
| Stool consistency (Bristol Stool Form Scale — BSFS) | pre-post improvement reported, in the between-group effect size available | — | Low | — |
Context
Chronic constipation affects 12–17% of adults globally; conventional treatments fail in refractory cases. FMT emerges as a gut microbiome restoration strategy, but its efficacy lacked robust systematic evaluation integrating multiple study designs.
What the study showed
In 12 RCTs, the trim-and-fill-adjusted clinical total response rate was RR = 1.21 (95% CI: 1.10–1.33); the unadjusted estimate was RR = 1.36 (95% CI: 1.21–1.52, P < 0.001), reduced after publication bias correction (Harbord P = 0.003). Wexner score improved by a mean of −2.08 points (95% CI: −3.35 to −0.81, P = 0.003) in favor of FMT. In 16 single-arm studies, the remission rate was 51.7% (95% CI: 40.3–62.9%), but these data do not establish causality. No serious adverse events were reported during follow-up of up to 24 weeks.
How it was done
Systematic meta-analysis with a 'three-in-one' framework: (1) random-effects meta-analysis of RCTs; (2) single-arm proportional meta-analysis with Freeman-Tukey transformation; (3) paired pre-post sensitivity analysis. Databases searched: CENTRAL, PubMed, Embase, CNKI. Included: 12 RCTs (n=1040), 16 single-arm studies (n=403), and 4 cohort studies (n=422). Maximum follow-up: 24 weeks.
Effect magnitude
Adjusted RR = 1.21 (95% CI: 1.10–1.33) for clinical total response rate; Wexner MD = −2.08 (95% CI: −3.35 to −0.81). Both effects are statistically significant but of modest magnitude, with high heterogeneity not fully quantified in the available summary.
Risk of bias
All included RCTs carry high risk of bias (tool not explicitly named, inferred as RoB 2). Publication bias confirmed by Harbord test (P = 0.003), requiring trim-and-fill correction that reduced RR from 1.36 to 1.21. High heterogeneity across studies, heterogeneous FMT protocols (route, dose, donor), absence of long-term safety data (>24 weeks), and lack of standardized sham comparators limit interpretation.
What this study does NOT prove
This study does not prove causality between FMT and improvement in chronic constipation, as all RCTs have high risk of bias and publication bias is confirmed. Single-arm and cohort data do not allow causal inference and do not apply to pediatric populations or those with structural gastrointestinal comorbidities.
In clinical practice
FMT may be considered in chronic constipation refractory to conventional treatments, with a modest expected response (RR ~1.21 vs. control). Clinicians should inform patients about the absence of safety data beyond 24 weeks and the variability of available protocols. Evidence is insufficient to recommend a specific administration protocol.
Limitations
All included RCTs carry high risk of bias (tool not explicitly named, inferred as RoB 2). Publication bias confirmed by Harbord test (P = 0.003), requiring trim-and-fill correction that reduced RR from 1.36 to 1.21. High heterogeneity across studies, heterogeneous FMT protocols (route, dose, donor), absence of long-term safety data (>24 weeks), and lack of standardized sham comparators limit interpretation.
What is still missing
Sham-controlled RCTs with adequate blinding, standardized FMT protocols, and follow-up ≥12 months are needed to establish causal efficacy and long-term safety profile.
Technical appendix
Version history
- 1.0 · 2026-09-19 — Auto-generated under Evidence Standard v1.0