Abstracts

Rationale: In 2018, the American Academy of Neurology and Child Neurology Society published a quality measures for infantile spasms (IS): proportion of infants with new onset IS who receive standard first line therapy (ACTH, oral steroids, or vigabatrin) within 1 week of diagnosis. We report our experience implementing this measure. Methods: The Pediatric Epilepsy Learning Healthcare System (PELHS) is a consortium of U.S. academic pediatric epilepsy centers that aggregate electronic health data to support quality improvement and clinical research. Centers submit administrative data (visit history, demographics, medications, diagnoses) to a data coordinating center (Weill Cornell) for analysis. For the IS quality measure, we used data from the first year of life to identify candidate cases that had (1) an ICD9/10 code for IS (345.x or G40.6) and (2) at least one anti-seizure medication. If ACTH, prednisolone, or vigabatrin were prescribed or administered within 1 week of the first diagnosis of IS, we considered that the infant had received recommended care. If not, we asked sites to review charts for additional information. We used logistic regression and ANOVA to examine center-to-center and year-to-year variation.  Results: We identified 434 candidate cases at eleven centers (median 32 per center, range 2 to 109) from 2015 - 2017. Of these 434, administrative data indicated that 263 (61%) had received recommended care. Chart review of the remaining 171 found 40 (23%) had IS but had not received recommended care. The most commonly cited reasons were (a) an outside provider had started a non-standard first medication (n = 9), (b) the diagnosis of IS was uncertain (n = 8), and (c) the infant was very sick or medically complex (n=8). Thirty-four of these 40 (85%) had a known etiology, a significantly higher percentage than historical controls (e.g., compared to 65% in Wirrell et al 2015; p < 0.01). Chart review found 94 (55% of 171 chart reviews) had indeed received recommended care. The most common reasons for the mismatch between the administrative data and the chart review were (a) 1stline therapy had been given at another center (n = 19) and (b) the correct drug had been called into a pharmacy, and so the prescription information was not in the administrative data (n = 15). The remaining 35 (8% of the 434 candidate cases) did not have IS -- 23 had a different diagnosis (e.g., early infantile epileptic encephalopathy) and 14 were admitted to rule out IS but did not have IS (Figure 1). Median center-level yearly performance was 88% [IQR 82% - 100%; minimum 67%]. Overall, 89% of patients with IS (357 of 399) received recommended therapy: 90% (128 of 142) in 2015, 88% (122 of 139) in 2016, and 91% (107 of 118) in 2017. Of note, in 2017, 6 of the 11 centers treated all cases according to recommendations, up from 3 of 11 in 2015 (Figure 2). Center-to-center differences explained a statistically significant amount of variation in performance (change in deviance 32.9 for 10 degrees of freedom, p < 0.0005; McFadden pseudo R2 = 12%). There were not statistically significant year-to-year differences (p = 0.7).  Conclusions: Roughly 1 in 10 infants with IS did not receive recommended care. There were significant center-to-center variations. Performance was stable over time. Of importance, these centers were early adopters of PELHS, potentially limiting generalizability to the rest of the U.S. Strategies to improve rates of recommended care should be tailored to each center, and may include education of providers outside academic centers and/or focus on infants with a known cause of IS.  Funding: No funding