Journal of evidence-based medicine · 2026
Reproducibility of Published Meta-Analyses on Acupuncture for Irritable Bowel Syndrome.
СтатьяНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Wang Yuan, Zou Kezhen, Liu Dan, Guo Qiong, Deng Ya, Long Youlin
Аннотация
OBJECTIVE: Published meta-analytic results should be reproducible from information available to readers. We assessed the numerical reproducibility of pairwise meta-analytic estimates of acupuncture for irritable bowel syndrome (IBS).
METHODS: We searched six English- and Chinese-language databases. Eligible systematic reviews reported at least one direct pairwise meta-analysis with recoverable study-level inputs. Each eligible pooled estimate was recomputed. The primary outcome was result-level numerical reproducibility. A prespecified standardized reanalysis was a key secondary analysis. Proportions used review-clustered robust variance estimation.
RESULTS: Twenty-three systematic reviews contributed 176 pooled estimates. Of these, 144 (81.8%, 95% confidence interval [CI] 71.4%-92.2%) reproduced exactly, 21 (11.9%, 95% CI 4.2%-19.7%) approximately, and 11 (6.3%, 95% CI 1.4%-11.1%) showed substantive numerical discrepancies. All 11 exceeded the 5% threshold for CI width, whereas 10 retained point estimates within 5% of the published value. No original-method recomputation changed whether the 95% CI excluded the null value. I2 reproduced within one percentage point for 162 estimates (92.0%). The text and forest plot disagreed on the analytic model for 12 of 168 evaluable estimates (7.1%). Under the standardized framework, 26 of 174 estimable results (14.9%; 95% CI 7.5%-22.3%) changed from statistically significant to nonsignificant.
CONCLUSIONS: Published pooled estimates were generally reproducible when study-level inputs were recoverable, although uncertainty estimates were more sensitive than point estimates. Complete, internally consistent reporting and availability of data and code would facilitate independent verification.
METHODS: We searched six English- and Chinese-language databases. Eligible systematic reviews reported at least one direct pairwise meta-analysis with recoverable study-level inputs. Each eligible pooled estimate was recomputed. The primary outcome was result-level numerical reproducibility. A prespecified standardized reanalysis was a key secondary analysis. Proportions used review-clustered robust variance estimation.
RESULTS: Twenty-three systematic reviews contributed 176 pooled estimates. Of these, 144 (81.8%, 95% confidence interval [CI] 71.4%-92.2%) reproduced exactly, 21 (11.9%, 95% CI 4.2%-19.7%) approximately, and 11 (6.3%, 95% CI 1.4%-11.1%) showed substantive numerical discrepancies. All 11 exceeded the 5% threshold for CI width, whereas 10 retained point estimates within 5% of the published value. No original-method recomputation changed whether the 95% CI excluded the null value. I2 reproduced within one percentage point for 162 estimates (92.0%). The text and forest plot disagreed on the analytic model for 12 of 168 evaluable estimates (7.1%). Under the standardized framework, 26 of 174 estimable results (14.9%; 95% CI 7.5%-22.3%) changed from statistically significant to nonsignificant.
CONCLUSIONS: Published pooled estimates were generally reproducible when study-level inputs were recoverable, although uncertainty estimates were more sensitive than point estimates. Complete, internally consistent reporting and availability of data and code would facilitate independent verification.