Zhang, H., & Schuster, T. (2018, August). The usefulness of prediction intervals in interpreting single-study effect estimates. Poster session presented at the Joint International Society for Clinical Biostatistics and Australian Statistical Conference 2018 (ISCB ASC 2018), Melbourne, Australia.
Download poster here
Abstract
In meta-analyses, results of research studies across various subpopulations are aggregated to make inference about population (effect) parameters of interest. The pooled estimate may represent either a common or average effect among all subpopulations.
In the recent literature it has been proposed that meta-analyses should report prediction intervals in addition to conventional confidence intervals. First, prediction intervals properly reflect the heterogeneity of effects across study populations as they depict a pre-specified range (e.g. 95%) of expected effects in future studies. Furthermore, prediction intervals are interpreted on the scale of the effect measure and are more sensitive than established heterogeneity indices such as the inconsistency index (I2).
We demonstrate that the same concept is applicable in large individual research studies that include subpopulations that may differ in their effects. We suggest reporting prediction intervals for single study effect estimates based on pre-specified subgroups such as strata used for randomization or purposive sampling in the context of clinical trials.
In meta-analyses, results of research studies across various subpopulations are aggregated to make inference about population (effect) parameters of interest. The pooled estimate may represent either a common or average effect among all subpopulations.
In the recent literature it has been proposed that meta-analyses should report prediction intervals in addition to conventional confidence intervals. First, prediction intervals properly reflect the heterogeneity of effects across study populations as they depict a pre-specified range (e.g. 95%) of expected effects in future studies. Furthermore, prediction intervals are interpreted on the scale of the effect measure and are more sensitive than established heterogeneity indices such as the inconsistency index (I2).
We demonstrate that the same concept is applicable in large individual research studies that include subpopulations that may differ in their effects. We suggest reporting prediction intervals for single study effect estimates based on pre-specified subgroups such as strata used for randomization or purposive sampling in the context of clinical trials.
In meta-analyses, results of research studies across various subpopulations are aggregated to make inference about population (effect) parameters of interest. The pooled estimate may represent either a common or average effect among all subpopulations.
In the recent literature it has been proposed that meta-analyses should report prediction intervals in addition to conventional confidence intervals. First, prediction intervals properly reflect the heterogeneity of effects across study populations as they depict a pre-specified range (e.g. 95%) of expected effects in future studies. Furthermore, prediction intervals are interpreted on the scale of the effect measure and are more sensitive than established heterogeneity indices such as the inconsistency index (I2).
We demonstrate that the same concept is applicable in large individual research studies that include subpopulations that may differ in their effects. We suggest reporting prediction intervals for single study effect estimates based on pre-specified subgroups such as strata used for randomization or purposive sampling in the context of clinical trials.
Comments (0)
You don't have permission to comment on this page.