Impact of statins on validation of ICH mortality prediction models.

Background: Intracerebral hemorrhage (ICH) has the highest mortality rate of all strokes. Hemphill's ICH score is commonly used to predict mortality after ICH. More recently, the ICH grading scale (ICH-GS) has been shown to improve sensitivity of 30 day mortality prediction in this patient group.

Objective: To assess the impact of admission variables not included in prediction models, such as coagulopathy, hyperglycemia, seizures and previous aspirin or statin use on 30 day mortality prediction using two contemporary prediction models.

Methods: Records of consecutive ICH patients from 1999 to 2006 were reviewed. Patients with ICH secondary to trauma or underlying lesions (e.g. brain tumors, aneurysms, arteriovenous malformations) and of infratentorial location were excluded. We dichotomized patients into a 'predicted survival group' and 'predicted death group' based on a <50% or >50% probability of death, respectively. The predicted mortality using ICH score and ICH-GS prediction models was calculated and was compared with the observed mortality in all patients and then separately in patient subgroups differentiated based on the presence or absence of coagulopathy, hyperglycemia (blood glucose> 180), seizures on presentation and previous exposure to aspirin or statins. Chi-square test was used for comparison of predicted and observed outcomes.

Results: One hundred and twenty-five patients were included in the analysis. The overall observed mortality was 23.2% (29/125), which was significantly lower than the 34.4% mortality predicted by ICH-GS (p=0.03). Hemphill's ICH score overestimated overall mortality by 7.2% (30.4-23.2%); however, this difference was not statistically significant (p=0.14). In patients using statins before ICH, observed mortality was 38% (5/13) and 42% (5/12) of the predicted mortality using ICH-GS (p=0.03) and ICH score (p=0.04), respectively; this difference was not seen in patients not previously exposed to statins. ICH-GS (but not ICH score) significantly overestimated mortality in patients with a serum glucose <180 (p=0.02); none of the other factors analysed significantly impacted the two mortality prediction models.

Conclusion: The significant difference between predicted and observed mortality using ICH-GS and the ICH score in the statin cohort suggests a protective effect of statins in the setting of ICH. Such observation warrants prospective validation.