Comparing reports of smoking prevalence from electronic health records and a community health survey

Background: The New York City health department launched a clinician incentive program, eQuits, to increase the documentation and delivery of smoking cessation interventions. Obtaining reliable baseline estimates of current smokers is needed for determining incentive amounts and to improve on the delivery of cessation aides or counseling. Furthermore, it is important to validate EHR data to track and improve performance.

Objective: Assess the reliability of smoking prevalence across Community Health Centers (CHCs) in neighborhoods with medically under-served patients.

Methods: Data abstracted from 16 CHCs across 68 facilities using electronic health records (EHR) were compared with population based estimates by neighborhood from the 2009 New York City Community Health Survey (CHS).

Results: The average difference in smoking prevalence between EHR derived data and CHS estimates was 29.2% (range 0.14% to 65.6%). Smoking rates derived from CHC EHRs varied widely, ranging from 2.3% to 80.3%; median smoking prevalence was 42.9%. In contrast, smoking rates found in the CHS ranged from 8.2% to 20.2% with a median of 15.3%. Using EHR data, 34 facilities had significantly higher rates; 8 sites were significantly lower than the CHS estimates.

Conclusions: When implementing programs using EHR data, accurate and consistent documentation is critical for tracking progress and determining outcomes. Comparing survey data to those collected by CHC EHRs is an important quality check that the data accurately represents the population served. Additional training is needed and the use of incentives may help CHCs improve their documentation of patient's smoking status and subsequent receipt of cessation interventions.