Abstract DP279: Hospital Variation in Documentation of NIHSS in Medicare Claims

Background: The NIHSS is critical for adequate risk adjustment in claims-based analyses of patients with acute ischemic stroke (AIS). In 2016, an ICD-10-CM code was introduced and in 2023, the Centers for Medicare and Medicaid Services (CMS) began usings NIHSS in risk adjustment for hospitals’ 30-day mortality performance measure for stroke. This application may incentivize documentation of higher scores more than lower scores. We sought to examine hospital-level variation in documentation rates, changes in hospitals’ documentation patterns over time, and the extent to which hospitals used the full range of possible scores. Methods: We used traditional Medicare claims data from 2016-2019 to identify AIS encounters. We included all hospitals with at least 5 AIS encounters (emergency department or inpatient) in each year of the study period. For each hospital we determined the proportion of encounters with NIHSS documented, by year, and used a linear regression model to examine the hospital-level trend in the proportion of cases with NIHSS documented over time. We examined the proportion of NIHSS <6 for each hospital by year, and how this varied by hospital documentation rates. Results: In the 2,331 hospitals that met inclusion criteria there were 885,967 AIS encounters from 2016-2019, with a documented NIHSS in 295,796 (33.4%) and scores ranging from 0-42 (Median: 5; 25th-75th: 2-11). From 2016 to 2019 the percentage of hospitals documenting at least one NIHSS increased from 54.8% to 92.6%. The median percentage of cases with a documented NIHSS by hospital in 2016 was 1.1% (25th-75th: 0%-4.9%) increasing to 46.2% in 2019 (25th-75th: 22.7%-68.4%; p-value for trend<0.001; Figure 1). Among encounters with any NIHSS documented, 48.4% were <6 in 2016 and 55.9% in 2019 (Figure 2). Among hospitals with an overall documentation percentage greater than 50%, the median percentage <6 was 33.3% (25th-75th: 27.7%-38.9%). Among hospitals with an overall documentation percentage less than 25% and at least one NIHSS documented, the median percentage <6 was 7.2% (25th-75th: 3.70%-10.9%). Conclusions: Documentation of NIHSS has improved over time, however, there remains substantial hospital-level variation. While hospitals’ case-mixes vary, our data suggest that hospitals with lower documentation rates may record NIHSS for severe stroke cases more frequently than for milder ones, thereby biasing the claims-based characterization of stroke severity in the US.