Editors' Note: Novel Score for Stratifying Risk of Critical Care Needs in Patients With Intracerebral Hemorrhage

Abstract

In their analysis of 451 consecutive adults with nontraumatic, spontaneous intracranial hemorrhage (ICH), Dr. Faigle et al. derived and internally validated a simple score for predicting critical care needs. This INTRINSIC score was also externally validated in a separate population and was found to have high sensitivity and specificity for identifying whether a patient with acute ICH may require transfer to an intensive care unit. Dr. Patel et al. comment on the limitations of scoring systems such as this, with an outcome of “intensive care interventions” being a subjective endpoint. The investigators acknowledge that interinstitutional practice variability may limit the transportability of their findings; however, the model was externally validated in a center remote from their own practice. Dr. Foroush et al. note that the estimated positive likelihood ratios (PLRs) for each cut point in the INTRINSIC score render it less useful at higher cut points. Dr. Faigle recognizes that high PLRs may not be possible for individual cut points and reference several examples of scoring systems (e.g., RACE score and LAMS), which are presently in use for stroke triage which do not meet the PLR standards cited by Dr. Foroush et al. Examining the cut point of individual scores may not provide as meaningful a measure as compared with a continuous composite score. On a separate note, readers may find that the score has considerable overlap with the traditional ICH score, which is predictive of early mortality. Both scores assign points for ICH volume, presence of intraventricular hemorrhage, and Glasgow Coma Scale. However, the INTRINSIC score is unique in that it may prove useful for early triage of patients with unclear dispositions after hospital arrival. In their analysis of 451 consecutive adults with nontraumatic, spontaneous intracranial hemorrhage (ICH), Dr. Faigle et al. derived and internally validated a simple score for predicting critical care needs. This INTRINSIC score was also externally validated in a separate population and was found to have high sensitivity and specificity for identifying whether a patient with acute ICH may require transfer to an intensive care unit. Dr. Patel et al. comment on the limitations of scoring systems such as this, with an outcome of “intensive care interventions” being a subjective endpoint. The investigators acknowledge that interinstitutional practice variability may limit the transportability of their findings; however, the model was externally validated in a center remote from their own practice. Dr. Foroush et al. note that the estimated positive likelihood ratios (PLRs) for each cut point in the INTRINSIC score render it less useful at higher cut points. Dr. Faigle recognizes that high PLRs may not be possible for individual cut points and reference several examples of scoring systems (e.g., RACE score and LAMS), which are presently in use for stroke triage which do not meet the PLR standards cited by Dr. Foroush et al. Examining the cut point of individual scores may not provide as meaningful a measure as compared with a continuous composite score. On a separate note, readers may find that the score has considerable overlap with the traditional ICH score, which is predictive of early mortality. Both scores assign points for ICH volume, presence of intraventricular hemorrhage, and Glasgow Coma Scale. However, the INTRINSIC score is unique in that it may prove useful for early triage of patients with unclear dispositions after hospital arrival.