Abstract 14816: The Impact of Practice Facilitation and Seasonal Variation on Blood Pressure Control in Small Independent Practices.

Introduction: HealthyHearts NYC (HHNYC), a cooperative in the Agency for Healthcare Research and Quality EvidenceNOW initiative, tested the effect of practice facilitation (PF) on practice-level blood pressure (BP) control in 255 small independent practices (SIPs). Few studies of interventions to improve BP control have considered the implications of seasonal variation when analyzing BP outcomes Methods: HHNYC used a cluster randomized stepped-wedge design to test the effect of PF (12 month intervention) on the proportion of patients meeting the Million Hearts BP target (≥70% BP controlled) among SIPs in NYC. BP data were extracted from SIPs' electronic health records. We used Poisson generalized linear mixed-effects models, with repeated measures nested within one of 255 SIPs and randomly varying intercepts across SIPs to assess the intervention effect. We did a second analysis that adapted the Poisson model to include (1) an observation-level random effect to address overdispersion (more variance in the counts than expected for a Poisson distribution) and (2) two terms in the fixed-effects portion of the model to separately contrast quarter one and then quarter three with quarters two and four, to address BP seasonality. Results: The first model found a significant negative association between PF and BP control (i.e., BP control declined) (IRR=0.98, 95% CI=0.96-0.99, p=<.001). In the full model (i.e., addressing seasonal changes) PF had no effect on BP control (IRR = 0.99, 95% CI=0.97-1.0, p=.19). The rate of BP control was significantly lower in quarter one (IRR = 0.97; 95% CI: 0.96-0.98, p<.01) and higher in quarter three (IRR=1.02; 95% CI: 1.01-1.03, p<.01) when compared with quarters two and four. Discussion: Many SIPs already met or exceeded national estimates of BP control and the Million Hearts BP performance target before the intervention, perhaps challenging further improvement without additional patient level intervention components that were not part of HHNYC. BP seasonality has strong study design implications. For stepped-wedge trials, sufficient waves are required to ensure the expected timing of intervention impact is spread evenly over quarters so the seasonality and any intervention impact can be more easily disentangled. [ABSTRACT FROM AUTHOR]