Scholarlink

<p>Dose-response meta-analysis is central to translating epidemiological evidence on continuous exposures into public health guidance. Its validity, however, depends on two closely connected domains: the reporting of the primary studies that provide the data and the methodological choices made by the meta-analysts who synthesise them. This article distinguishes limitations originating in primary studies, such as incomplete exposure-category cut-offs, missing category-specific precision estimates, conflation of former and never-exposed participants, unreported exposure ranges, and incompatible transformations, from limitations in meta-analyses themselves, such as reliance on high-versus-low contrasts, unsupported linearity assumptions, inadequate handling of correlated estimates, and insufficiently reproducible analytic workflows. The consequences also differ; reference-group contamination and extrapolation beyond observed exposure ranges can induce bias, whereas missing standard errors, cases, person-time, or cut-offs often lead to imprecision, study exclusion, or unverifiable covariance assumptions. We propose a minimum reporting set for primary studies and a complementary analysis standard for dose-response meta-analyses. Primary studies should report exact lower and upper category limits, observed category means or medians, the number of participants, cases and person-time, crude and adjusted effect estimates with standard errors or confidence intervals where relevant, explicit reference-group definitions, exposure transformations, and the observed exposure range. Meta-analyses should use flexible dose-response models where data permit, examine non-linearity, account for correlated contrasts, avoid pooling incomparable transformed estimates without a valid conversion, and provide reproducible code. Examples from alcohol, ambient particulate matter, cadmium exposure, and potassium supplementation show how the shape of the curve, the exposure range, and the reference group selection can alter interpretation. A dose-response-specific extension to existing reporting guidelines would improve transparency, reproducibility and policy translation.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Recommendations to improve reporting and methodological rigour in primary studies and dose response meta analyses

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年7月4日

查看全文
  • Osama Abdelhay

摘要

Dose-response meta-analysis is central to translating epidemiological evidence on continuous exposures into public health guidance. Its validity, however, depends on two closely connected domains: the reporting of the primary studies that provide the data and the methodological choices made by the meta-analysts who synthesise them. This article distinguishes limitations originating in primary studies, such as incomplete exposure-category cut-offs, missing category-specific precision estimates, conflation of former and never-exposed participants, unreported exposure ranges, and incompatible transformations, from limitations in meta-analyses themselves, such as reliance on high-versus-low contrasts, unsupported linearity assumptions, inadequate handling of correlated estimates, and insufficiently reproducible analytic workflows. The consequences also differ; reference-group contamination and extrapolation beyond observed exposure ranges can induce bias, whereas missing standard errors, cases, person-time, or cut-offs often lead to imprecision, study exclusion, or unverifiable covariance assumptions. We propose a minimum reporting set for primary studies and a complementary analysis standard for dose-response meta-analyses. Primary studies should report exact lower and upper category limits, observed category means or medians, the number of participants, cases and person-time, crude and adjusted effect estimates with standard errors or confidence intervals where relevant, explicit reference-group definitions, exposure transformations, and the observed exposure range. Meta-analyses should use flexible dose-response models where data permit, examine non-linearity, account for correlated contrasts, avoid pooling incomparable transformed estimates without a valid conversion, and provide reproducible code. Examples from alcohol, ambient particulate matter, cadmium exposure, and potassium supplementation show how the shape of the curve, the exposure range, and the reference group selection can alter interpretation. A dose-response-specific extension to existing reporting guidelines would improve transparency, reproducibility and policy translation.