Evidence Consistency via a Study Quality Lens in Systematic Reviews: a case-study of formaldehyde exposure and respiratory associations
Citation:
Glenn, B. AND E. Radke-Farabaugh. Evidence Consistency via a Study Quality Lens in Systematic Reviews: a case-study of formaldehyde exposure and respiratory associations. ISEE 2019, Utrecht, NETHERLANDS, August 25 - 28, 2019.
Impact/Purpose:
The evidence on exposure-health associations for chemicals with an extensive research history often includes studies with heterogeneous results. These studies may include various study designs examining different outcome and exposure definitions and may be influenced to varying degrees by sources of bias and other factors that affect the magnitude, direction, and precision of effect estimates. We analyzed the impact of bias and other quality criteria that could influence our confidence in the associations from studies of indoor formaldehyde exposure. The literature search (through February 2018) was based on population, exposure, comparator, and outcome (PECO) criteria that included studies of children or adults that analyzed associations with current asthma. Potential bias (e.g., selection, information, confounding) and other aspects of study quality (e.g., sensitivity, precision) were evaluated using a set of a priori criteria within domains of participant selection, exposure, outcome, confounding, analysis and sensitivity. Consistency was examined via forest plots stratifying by population, exposure (low vs high), overall study confidence, and specific domain ratings. A total of 32 population-based and occupational studies were identified that appeared to have considerable heterogeneity across studies for specific outcome. However, when the effect estimates were stratified by exposure level and setting, and overall confidence in the exposure-outcome association, greater consistency was indicated. Study quality considerations are essential to analyses of evidence consistency as part of the integration of evidence in systematic reviews.
Description:
In this poster, consistency across study results is examined via forest plots stratifying by population, exposure (low vs high), overall study confidence, and specific domain ratings. Although the effect estimates vary considerably, when the effect estimates are stratified by exposure level and setting, and overall confidence in the exposure-outcome association, greater consistency is indicated. Generally, limitations varied across multiple domains for the studies rated as having low confidence, potentially with opposing influence on the direction of bias, but stratification by domains clarified some of the greater heterogeneity observed among these exposure-outcome associations.