Chinese herbal formula trials with TCM pattern inclusion reported significant primary outcomes more often
In an analysis of 378 published placebo-controlled trials, selecting patients by TCM pattern had an adjusted risk ratio of 1.14 for significant results; few trials had low bias risk.
A meta-research analysis published in Acta Pharmaceutica Sinica B examined 378 randomized, placebo-controlled trials of Chinese herbal formulas. The trials were published between January 1, 2020, and October 21, 2025, and were identified through searches of eight databases. The primary outcome analyses included 92,691 participants doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗.
What the analysis compared
Traditional Chinese medicine (TCM) pattern differentiation groups patients with the same disease according to their symptoms and signs to guide treatment selection. Some researchers have pointed out that leaving patterns out of inclusion criteria may reduce the alignment between treatment and enrolled patients, which could affect observed efficacy. Whether using patterns for eligibility is associated with reported outcomes remained unclear doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗.
Of the 378 trials, 197 (52.1%) used TCM patterns as an inclusion criterion. Statistically significant primary outcomes were reported by 324 trials (85.7%) overall doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗. By group, 94.9% of trials using patterns reported significant primary outcomes, compared with 75.7% of those that did not.
Result and limitations
After adjustment for four prespecified trial characteristics, multivariable Poisson regression showed that pattern-based inclusion was associated with a higher likelihood of a statistically significant primary outcome: adjusted risk ratio 1.14 (95% CI 1.06–1.23; P = 0.001). The association was generally consistent across prespecified subgroups and sensitivity analyses doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗.
The authors note several limits. Only 52 trials (13.8%) were assessed as low risk of bias with the Cochrane Risk of Bias 2.0 tool. The analysis covered only published trials, and few published pattern-based trials reported statistically nonsignificant primary outcomes. Publication bias, selective outcome reporting, residual confounding and differences in research and reporting practices may have influenced the findings doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗.
The authors state that this trial-level association does not establish that pattern-based selection improves treatment efficacy. They call for further prospectively registered, methodologically rigorous trials with standardized pattern assessment and transparent reporting doi.org
Acta Pharmaceutica Sinica Bdoi.orgOpen source ↗.
