Methods
Practices were included if they adopted the PCIP subsidized eClinicalWorks EHR system at least 3 months before the baseline data analysis period (October 2009). Practices were excluded if data were not transmitted at any of the 3 data analysis periods: October 2009 (T1), October 2010 (T2), and October 2011 (T3). Community health centers (CHCs) were defined as any PCIP practice that was either a federally qualified health center, a New York State-defined diagnosis and treatment center, or had unusual special needs such as serving a foster care, mentally disabled, or other special population. All other practices were classified as small practices (SPs).
Data shown in the analysis were derived from practice EHRs that automatically transmit quality measurement data monthly to PCIP. From these transmissions, averages of practice performance data were assembled from October 2009 through October 2011. Four key quality measures of relevance to the TCNY agenda were included in the analytical data set: antithrombotic therapy among patients aged 18 years or older with ischemic vascular disease or aged 40 years or older with diabetes; blood pressure (BP) control among patients aged 18 to 75 years with essential hypertension and no diagnosis of ischemic vascular disease or diabetes; hemoglobin A1c (HbA1c) testing in the past 6 months among patients aged 18 to 75 years with diabetes; and smoking cessation intervention in the past 12 months among patients aged 18 years or older who were "current smokers" (Table 1). These measures are similar to National Quality Forum endorsed measures (no. 0631 [antithrombotic therapy], no. 0018 [BP control], no. 0057 [HbA1c testing], and no. 0028 [smoking cessation intervention]), with the exception of age ranges and the fact that patients must have had at least 1 office visit with the primary care practice to be eligible. No other rules were applied for patient inclusion to the denominator. Additional practice and provider variables included number of encounters and number of unique patients seen per month, practice location, number of providers, number of practice sites, and practice-reported percentage of patients on Medicaid or self-insured. Practice date of EHR implementation was obtained from a customer relationship management database maintained by PCIP staff.
All statistical analyses were conducted using SAS version 9.2. Comparisons between groups of practices, time points, or trends were considered significant if the observed P value for a statistical test was less than .05. Practice characteristics were measured at the baseline period (October 2009). Chi-square tests, t tests and nonparametric tests were used to compare characteristics between SPs and CHCs.
Trend graphs were generated for each quality measure over time and show average monthly practice rates. Separately, a comparison of practice level performance rates on each of the 4 quality measures was conducted for 3 points: October 2009 (T1), October 2010 (T2), and October 2011 (T3); and by practice characteristics. Generalized estimating equation models with logistic link function were used to analyze the association between time and practice characteristics and performance on the selected quality measures.