Abstracts

Rationale: Epilepsy is a serious, chronic neurological condition, for which there is limited information on patient trajectories after diagnosis. Some patients could have well-controlled disease after treatment, whereas others have more severe or intractable disease despite active treatment. The new ICD-10 coding scheme with its larger set of codes (compared with the prior version) offers a potential data source for assessing patient trajectories. We compared physician diagnoses (using ICD-10 codes) with patient-reported physical and mental health scores (PROMIS-10) among patients seen in an outpatient clinic within a major academic medical center. Methods: We administered iPad-based surveys to consecutive adult patients at the MGH Epilepsy Outpatient Clinic between Jan-Dec of 2016. We conducted our analyses on the 473 subjects with complete diagnostic and PROMIS data (Figure 1). We classified patients’ epilepsy status as intractable or not using ICD-10 claims-based diagnoses and defining intractable epilepsy as having the qualifying term “intractable” (e.g., G40.A11 for absence epileptic syndrome, intractable, with status epilepticus) or being a known intractable syndrome (e.g., Lennox-Gastaut syndrome), as in table 1. We then compared these diagnosis-based classifications with PROMIS scores. Finally, we examined the association between the diagnoses and PROMIS scores using linear regression, and adjusting for patient baseline characteristics and number of prescribed drugs. Results: Among eligible unique subjects, 263 subjects (56%) were classified with non-intractable epilepsy, 90 (19%) with intractable epilepsy, and in 120 (25%) severity could not be reliably determined. Subjects with non-intractable epilepsy were similar to those with intractable epilepsy with respect to age, gender, ethnicity, preferred language, and insurance type (Table 2, all p>0.05). Compared to subjects with non-intractable epilepsy, subjects who had intractable epilepsy were more likely to report worse mental and physical health T-scores (Figure 2, p Conclusions: Physician diagnoses under the ICD-10 scheme appear to be well correlated with patient perceptions of their health. Claims-based studies using the enhanced ICD-10 codes could help provide better estimates of epilepsy severity using claims data, and help monitor both patient trajectories and care quality. Future work on a diagnostic algorithm weighting multiple codes by their accuracy may be necessary to enhance the performance of new or existing claims-based algorithms. Funding: Harvard Catalyst Fellowship