Admitting the Patient With Acute Stroke to the Right House-Lessons From the Sorting Hat of Hogwarts.

Abstract

The goal of any prehospital sorting function is to allocate patients to themost appropriate destination that will maximize their outcomes as defined by their preferences, goals, needs, and resources. Inmuch the sameway, the SortingHat at Hogwarts School of Witchcraft and Wizardry in J. K. Rowling’s world of Harry Potter seeks toplaceeach student in the proper “house.” Just as there are competing schools of wizardry, there are also multiple organizations1 that seek to certify stroke centers by differing criteria, although The Joint Commission (JC) Primary Stroke Center (PSC) program is by far the oldest and largest. In their article in this issue of JAMA Internal Medicine, Bekelis and colleagues2 examine the effect of patients with strokebeing “sorted” to admission at a JCPSCvs anon–JCPSC on the outcome of death by 7 and 30 days. Their main interest was in determining the additional travel distance necessary beyond the hospital nearest to a patient’s home for admission to a PSC afterwhichnodifference in outcomeswould be evident. Among patients who entered this complex maze of stroke care from2010 to 2013, they found highermortality aftermultivariable adjustment forpatients assigned toahouse that was JC PSC certified vs one that was not, but this effect reversed when using instrumental variable analysis to account forunmeasuredconfounding.After90minutesofadded travel, nobenefitwas gainedby admission to a JCPSC.Within the limits of theirMedicare (Centers forMedicare &Medicaid Services [CMS]) fee-for-service claimsdata source, theydidan elegant job of trying to control for measured and unmeasured confounding introduced by the nonrandom allocation of patients. Ischemic stroke and hemorrhagic stroke have different clinical trajectories: hemorrhagic stroke has greater mortality, less diagnostic uncertainty, greater likelihood of transfer, and relatively fewpatients forwhomtreatment in the “golden hour” dramatically alters the outcome. Unlike many serious acutediseases, strokemortality is largelypredictedbyone (ie, stroke severity) rather than many covariates. The CMS riskstandardized model of hospital rankings on 30-day ischemic stroke mortality does not adjust for stroke severity. Addition of this unmeasured confounder has a dramatic effect on reclassificationofhospitalperformance,3andallmeasuresofperformance that relate to functional outcomesormortalitymust include severity. Inmodels of in-hospitalmortality usingdata in theGetWithTheGuidelines–StrokeProgram,4 almost all of the fully adjustedmodel’s C statistic is due to theNational Institutes ofHealth Stroke (NIHSS) score alone. Fortunately, the International Statistical Classification of Diseases, 10th Revision will soon include a mechanism for collecting NIHSS severity into claims data, and the CMS has proposed modified versions of its 30-day risk-standardized measures to include these data. Although frequentlydisabling, stroke is ahigh-impact lowfrequency event in the prehospital or emergency department (ED) setting.Estimates suggest that stroke represents less than 5% of emergencymedical services (EMS) transports.5 Recent data suggest that 15% of ED visits resulted in hospital Related article page 1361 Research Original Investigation Primary Stroke Center Hospitalization for Elderly Patients With Stroke