Identifying optimal cut points of National Institutes of Health Stroke Scale to Predict Mortality: A Population-based Assessment
Robert Stanton1, David Robinson2, Mathew Reeves3, Lili Ding2, Jane Khoury2, Mary Haverbusch2, Kathleen Alwell1, Opeolu Adeoye4, Elisheva Coleman5, Felipe De Los Rios La Rosa6, Stacie Demel7, Simona Ferioli8, Matthew Flaherty2, Adam Jasne9, Pooja Khatri10, Jason Mackey11, Sharyl Martini12, Eva Mistry2, Sabreena Slavin13, Michael Star14, Daniel Woo2, Kyle Walsh2, Brett Kissela15, Dawn Kleindorfer16
1Neurology, University of Cincinnati, 2University of Cincinnati, 3Michigan State University, 4Washington University in St. Louis, 5University of Chicago Medical Center, 6The Neurology Group, 7University of Cincinnati Medical Center, 8UCMC, 9Yale, 10Univ of Cincinnati/Dept of Neuro, 11Indiana University, 12VHA Neurology, 13University of Kansas Hospital, 14Star Neurology, 15University of Cincinnati Hospital, 16University of Michigan Department of Neurology
Objective:
We sought to identify the optimal cut-points of NIHSS at initial presentation that correlate with higher 30-day mortality.
Background:
Ischemic stroke is the 5th leading cause of death in the US. As a measure of stroke severity, initial NIHSS has been used to predict clinical outcome.
Design/Methods:
In 2005, 2010, and 2015 all hospitalized, first acute ischemic stroke events occurring within the Greater Cincinnati 5-county area were ascertained. Potential cases underwent chart abstraction and physician adjudication, including assignment of a retrospective NIHSS score based on clinical findings at initial presentation. Descriptive statistics for NIHSS were estimated by study year, demographics, and medical history. Data regarding mortality was obtained from the National Death Index. The Contal and O’Quigley method based on a modified log-rank test statistic was used to determine cut-points of the NIHSS score associated with 30-day mortality, and hazard ratios were obtained from Cox models with adjustment for sex, race, and age.
Results:
In 2005, 2010, and 2015 there were 1704, 1818, 1852 ischemic stroke events, respectively. Thirty-day mortality rates were 10.5%, 9.6%, 9.0%, respectively. Optimal cut-points of NIHSS <9, 9-16 and >16 were identified. Across all 3 periods, 3431 (84.5%) of cases had NIHSS 0-8, 352 (8.7%) had NIHSS 9-16 and 274 (6.8%) >16. Kaplan Meier Survival Curves for the 3 NIHSS groups are shown in the Figure. NIHSS >16 at initial presentation was associated with a 15 fold (HR with 95% CI: 13, 19) increase in the risk of death at 30-days compared to those with NIHSS <9.
Conclusions:
NIH Stroke Scale scores are a predictor of mortality, with higher NIHSS scores having higher risk of death. The cut points reported identify subgroups of stroke patients with dramatically different prognoses. Future studies should assess if this excess mortality risk among severe strokes persists after more widespread implementation of thrombectomy beyond 2015.