Reporting and Analysis of Process-of-Care Time Measures for Hyperacute Stroke Interventions

Area of Science:

• Medical Interventions
• Clinical Trial Analysis
• Systematic Reviews

Background:

• Process-of-care time measures are crucial for evaluating hyperacute stroke interventions and care systems.
• Mean-based statistics can be misleading for skewed time data.
• Percentile-based statistics, like the median, offer unbiased summarization of treatment times.

Purpose of the Study:

• To investigate the reporting and statistical comparison of process-of-care time measures in hyperacute stroke trials.
• To provide a methodological basis for future meta-analyses involving these time measures.

Main Methods:

• A systematic scoping review of 146 studies reporting hyperacute stroke care time measures.
• Analysis of statistical agreement between transformation-based methods and actual process-of-care time data.
• Graphical and statistical comparison of mean- and percentile-based meta-analysis techniques.

Main Results:

• 34% of studies reported time measures using only mean/SD, potentially yielding impossible negative values.
• Transformation-based imputation methods showed moderate agreement with the true standard deviation (CCC=0.88).
• Mean- and median-based meta-analyses yielded comparable results for between-arm effects but differed for individual-arm measures.

Conclusions:

• Stroke researchers should use median/interquartile range for summarizing time measures and employ rank-based analysis.
• Caution is advised when using transformation-based methods for imputing mean/SD of time measures.
• Median-based meta-analytical techniques are recommended for synthesizing process-of-care time measures.