To use the National Institutes of Health Stroke Scale (NIHSS) as reported in Paul Coverdell Stroke Registry (PCSR) to assess the accuracy of the ICD-10-based NIHSS as a measure of Acute Ischemic Stroke (AIS) severity.

NIHSS score is used to evaluate stroke severity in clinical settings, typically close to admission, whereas the ICD-10 NIHSS is used for billing. The validity of stroke severity from billing registries remains uncertain. 

In this observational study, we analyzed data from a dataset linking the PCSR and Medicare databases, sampling 148,759 patients aged 65 and above admitted for AIS from 2016 to 2019.  We examined the concordance between clinical NIHSS scores documented in PCSR and ICD-10-based NIHSS from Medicare claims data, treating both as continuous variables. Discordance was measured by calculating the absolute difference between the PCSR and ICD-10-based NIHSS scores.  We explored demographic and clinical factors associated with score discordance across nine different states. 

64,684 patients had documentation of both clinical and ICD-10 NIHSS scores. The mean NIHSS score from PCSR was 7.25 (SD=7.80), and the mean from ICD-10 was 7.55 (SD = 7.82). There was a high concordance between the two scores with Concordance Correlation Coefficient of 0.92 (95% CI: 0.914 - 0.920, P < 0.001), and a linear regression beta coefficient of 0.91 (95% CI: 0.9159 - 0.9220, P < 0.001). The average discordance was 1.2 points (SD = 2.9). State-specific discordance ranged from a minimum of 0.69 (SD = 2.27) in Massachusetts to a maximum of 1.73 (SD = 3.65) in California.

The observed high concordance between clinical and ICD-10 NIHSS scores signals that ICD-10 NIHSS scores can be used as a valid proxy in assessing stroke severity, offering valuable clinical, academic, and administrative implications.