A total of 23 of the 27 primary care practices approached were enrolled in this project. Table 1 summarizes practice characteristics. The majority of practices were located in urban areas, and 2 of these practices hosted primary care resident physicians. Four separate EHR systems were used across the 23 practices. A total of 210 clinical providers attended the ADS at their respective practices; per-practice attendance can be found in Table 1. The largest group of ADS attendees (37%) was physicians, followed by nurse practitioners, physician assistants, nursing staff, and administrative staff.
The cumulative time devoted to practice facilitation activities by the PFs was approximately 889 hours. Each facilitator was assigned either 5 or 6 practices. Practice facilitation activities included efforts to address documentation procedures for screening referral and completion within practice EHR systems, provider feedback and assessment activities, streamlining of provider reminder systems and office workflows, and patient education and outreach interventions.
Cancer Screening Rates and TRANSLATE Evaluations
The average before- and after-intervention screening rates for the 23 practices can be found in Table 2. One-way repeated measures ANOVA revealed that the difference between mean screening rates before and after the intervention was statistically significant for breast cancer (36.96% vs 49.96%; P = .001) and CRC (32.74% vs 38.30%; P = .001). There was a wide degree of variation in the cancer screening rates across practices, with standard deviations ranging from roughly 16 to 21 percentage points. The cancer screening rates reported by the practices participating in this project fall, on average, well below the estimated 2014 New York state rates of 68.1% for CRC, 78.6% for breast cancer, and 82.6% for cervical cancer (estimates are based on self-reported data). It is important to note, though, that a small number of practices did meet or exceed the statewide screening rates for each cancer. Individual practice screening rates are not listed here but are available upon request.
Table 3 presents the average TRANSLATE model scores before and after the intervention for the 23 practices. The practices, on average, improved in each of the 9 elements measured under the TRANSLATE model. However, it is important to note that, individually, some practices did not make measurable improvements across the 2 measurement periods. One-way mixed ANOVA showed statistically significant improvement in scores for the following elements: setting an improvement target, gaining administrative buy-in, using network information systems, conducting audit and feedback activities, and providing staff education, as well as the overall TRANSLATE score for the practices. Two elements were affected by PF assignments: 1 PF rated practices with a lower score for target, and another PF rated practices with a higher score for education, compared with other PF practice groups. These differences may have been the result of subjective evaluation differences between the PFs or because of the characteristics of the practices evaluated. Spearman correlation results indicated that higher practice scores in site coordinator engagement and staff education were associated with higher cancer screening rates for all 3 cancer targets. In addition, a team approach was positively correlated with cervical cancer screening rates, and administrative buy-in and audit and feedback were positively correlated with CRC screening rates. Results of the correlation analysis are shown in Table 4.
Staff Surveys and Focus Groups
A total of 144 individuals responded to the staff surveys; of these, only 56 respondents had linked before/after data from the 23 practices. The majority of the nonresponse occurred during the data collection period after the project because of staff turnover and absence at several of the participating practices. The majority of survey respondents were physicians; other respondents included nurse practitioners, physician assistants, registered nurses, licensed practical nurses, and practice/clinic managers. Comparisons of mean responses to Likert-scale questions were conducted only among the 56 linked before/after surveys; most of these differences were not statistically significant. The most important patient-related and system-related barriers to increasing cancer screening as perceived by practice staff centered on a lack of follow-through by patients on screening recommendations and the inability to track patients receiving services at outside offices. Respondents identified the lack of personnel support to both maintain and use registries as among the top barriers to using EHR-based patient registries, and they highlighted provider reminder systems, patient education, and patient reminders as the top QI strategies that would most benefit their practices' ability to increase cancer screening.
Focus groups were only conducted in 3 practices, with 3 to 5 individuals participating in each group (a total of 13 participants). Because of scheduling conflicts, key informants were interviewed at the remaining 20 practices. Participants in the 3 focus groups included practice medical directors (n = 3), practice managers (n = 3), care coordinators (n = 2), and clinical providers (n = 5). Key informants were primarily practice managers, practice QI specialists, and practice medical directors.
Content analysis of the focus group and key informant interview transcripts identified several themes related to cancer screening, which fall broadly within the overarching concepts of barriers to increasing cancer screening and supports for increasing cancer screening at the patient, provider, and practice levels. Table 5 displays a breakdown of these concepts. Other common themes that emerged from analysis focused on the working relationship practices had with the PFs, as well as dynamics within practices that supported systems-level change.
Consistent engagement by the PFs with their site coordinators helped practices keep a focus on QI initiatives, and those practices that experienced the greatest change in both cancer screening rates and policies had fully engaged staff at several levels within the practice. The ADS oriented practice staff to the goals and targets of the project, increasing buy-in for interventions developed later with the PF. Introducing regular provider performance assessment and feedback mechanisms and increasing point-of-care reminders and prompts for cancer screening helped clinic staff become more invested in QI projects.
Personnel and funding limitations necessitated the implementation of QI project tasks within existing practice structures and priorities, and workflow assessments guided by PFs were viewed as a valuable tool for achieving this end. Several practices chose to overlap project improvement activities with existing requirements and priorities, including patient-centered medical home and meaningful use targets. PFs were also viewed as a valuable resource for connecting practices with patient outreach resources and community services, including mobile mammography providers and transportation services.
All the practices felt that EHR-based reports measuring patient screening rates did not represent accurate data, but rather reflected only those screening tests that were recorded as structured data within their EHRs. PFs were able to provide guidance to practices on altering data entry procedures and support for data mining, data correction, and chart auditing to improve the validity and reliability of EHR registry data used by the practices.
J Am Board Fam Med. 2016;29(5):533-542. © 2016 American Board of Family Medicine