Intracluster correlation coefficient in multicenter childhood trauma studies.

Area of Science:

• Trauma research
• Pediatric critical care
• Biostatistics

Background:

• Multicenter childhood injury studies require accurate statistical power calculations.
• Intracluster correlation coefficients (ICCs) are essential for cluster randomized trials but vary by outcome and population.
• Previous ICC values for pediatric trauma outcomes were not well-established.

Purpose of the Study:

• To calculate the intracluster correlation coefficient (ICC) for emergency department (ED) shock rate, early trauma death, and in-hospital trauma death rate in multicenter childhood injuries.
• To provide data for sample size calculations in pediatric trauma research.

Main Methods:

• Utilized the National Trauma Data Bank (5th revision), a large US multicenter trauma registry.
• Calculated ICCs for in-hospital trauma death rate using data from 80 trauma centers.
• Calculated ICCs for ED shock and early trauma death rates using data from 33 states.

Main Results:

• Analyzed data from 952,242 patients (2000-2004), with 13% under 15 years old.
• Calculated ICCs: 0.005 for ED shock rate, 0.014 for early trauma death, and 0.023 for in-hospital death.
• ICCs were stratified by sex and injury type (blunt vs. penetrating).

Conclusions:

• The calculated ICCs are vital for sample size determination in clustered childhood trauma studies.
• Failure to adjust for ICCs in cluster randomized trials can lead to significant underpowering.
• These findings will improve the design and reliability of pediatric trauma research.