Effect Measure Modification and Confounding of Severe Head Injury Mortality by Age and Multiple Organ Injury Severity

Abstract

The study aim is to assess confounding and effect measure modification of the relationship between head injury severity (measured using the Glasgow Coma Scale [GCS]) and mortality by age and multiple organ injury (measured using the Injury Severity Score [ISS]). Head injury-related mortality is affected by head injury severity, as well as age and multiple organ injury. However, little is known about the effect modification of these relationships. Stratifications and multiple logistic regression analyses examined the association of mortality with GCS score less than 9, controlled for ISS of 16 and higher or age of 55 years and older. Receiver operating curve (ROC) analyses were used to explore the relationship of GCS cutoff values and prediction of death by the ISS and age categories mentioned. Both age and ISS modified head injury mortality in a similar direction: there was a negative interaction between age and ISS and more severe head injury. Lower GCS values (indicating more severe head injury) related more strongly to mortality in younger persons. Lower GCS values related more strongly to mortality when extracranial injuries were less severe. These data indicate effect modification. Data indicate negative confounding of the association between GCS and mortality by age and positive confounding by ISS. Multivariate analyses indicate that only younger age modifies the effect of more severe head injury with statistical significance when both age and multiple organ injury are considered. ROC analyses conducted with stratified and logistic regression analyses indicate that GCS is a better predictor of death in those with younger age. The relationship of head injury severity and mortality is modified and confounded by age and ISS. GCS is a better predictor of death in younger patients.