Demographic, clinical, and social data predictive of 1-month mortality post-aneurysmal subarachnoid hemorrhage

Attempts to develop a multidimensional predictive model for outcomes after aneurysmal subarachnoid hemorrhage have largely failed. Other demographic, clinical, and social information may enhance the predictive value of clinical scales for the prognostication of outcomes. This study investigates other factors that may also be influential for 1-month mortality. An existing quality control database was interrogated for ICD-9/10 codes of 430/I60. Missing data were handled by multiple imputations with predictive mean method. The probability of mortality at 1-month was regressed on each predictor variables with results pooled to estimate regression coefficients; a multivariable model was constructed by a backward model selection procedure. The 1286 subjects averaged 53 years old, were 51.5% Caucasian and 71.2% female. Mortality rate was 13.7% at 1 month. Of the 24 factors interrogated, 10 were significantly ( p < 0.05) associated with death (age; race; marital status; employment status; cardiac history; diabetes; treatment method; Glasgow Coma Scale score [GCS]; modified Fisher Group score [mFGS]; Hunt Hess Grade). The final model had four predictors: age (OR 1.08; p <0.0001), employment status (OR 0.29; p = 0.0051), treatment method (OR 13.11; p =0.0007), and Hunt Hess Grade (OR 2.45; p < 0.0001). Our model, constructed from a fairly large and diverse sample, demonstrates that the inclusion of additional demographic, clinical, and social variables improves predictive power beyond that offered by traditional clinical grading scales. Future work should examine the utility of this or a similar model for prediction of morbidity and long-term functional outcomes.