For most indicators, overall center performance was higher at the one-year assessment than it was at baseline. In addition, two of the three CHCs having year-two assessments showed generally improved results from year one. The third center's year-two results showed worsening performance in most areas compared with year one. Given the cross-sectional nature of the samples, these results should be interpreted with caution.
Only two of the five centers prioritized the same five diabetes indicators, and only two indicators were selected by all five CHCs — namely, annual assessment of urine for microalbuminuria and HbA1c testing. For HbA1c testing, there were differences in the desired frequency, with some CHCs wanting at least two HbA1c tests per year and others wanting three tests per year. Factors that seemed to influence indicator selection and adherence included whether the selection was based on consensus or made by the medical director; provider training, experience, and beliefs; and CHC staff and organizational issues. In one CHC where the medical director chose the indicators, we later revised them midstream to reflect the priorities of a physician who had not been involved in the initial process and was the sole CHC physician participant in our initiative. From this experience and others, we learned that it was important to recognize and address local issues that could adversely affect indicator selection and/or staff buy-in and participation.
All five of our CHCs used paper records. While there is evidence that provider reminder systems such as diabetes flow sheets helped improve diabetes care, not all of our centers used them (10
). Some were understandably resistant to adding yet another flow sheet to their already complicated charts. One CHC already had incorporated its diabetes measures into its adult health maintenance flow sheet. The majority of its patients did not have diabetes, and providers were only infrequently using that portion of the flow sheet. After some discussion, we arrived at the solution of placing colored stickers inside the charts on the adult health maintenance/diabetes flow sheets of their patients with diabetes. This change resulted in improved recognition of patients with diabetes and improved performance on the indicators. The use of flow sheets, in general, was associated with improved recognition and performance.
Any new initiative dependent on the participation of providers must compete with many other demands on their time during usual patient encounters (e.g., patient expectations and requests, professional concerns, diverse and sometimes conflicting practice guidelines and prevention recommendations, local and national initiatives, interruptions, emergencies). For example, even though four of our five partner CHCs were participating in the HRSA/CDC Diabetes Collaborative — whose members agreed to adopt local shared quality-improvement measures consistent with national guidelines — the level of participation still varied considerably from site to site. This taught us that participation in other diabetes programs was no guarantee of success.
In most centers, providers reacted to our initial presentation of results with disbelief, as both their individual and CHC levels of compliance were typically lower than they expected. During our meeting, they appeared to be comparing their results with center results and sometimes with another provider's results. We addressed the skepticism in several ways. First, we described our methods during our presentation (i.e., the comprehensiveness of the chart reviews, the use of two reviewers for quality control, the levels of interrater reliability). Second, we also asked the reviewers, who could be project and/or local office staff, to be present to respond to any questions. Third, we put the results in context by comparing each center's results with available national statistics that were typically about the same or worse. Generally, these strategies overcame barriers to acceptance, and we were able to move on to a more substantive discussion on what steps could be taken for improvement. We then facilitated discussions on what behavioral and structural changes providers could make as a staff to improve their results, and we offered technical assistance, such as help with the development of flow sheets or telephone consultation. By the end of the meeting, centers had usually developed a tentative plan for improvement. From this, we concluded that while obtaining provider-level data was more work, it generated a healthy interest and sense of competition among participants.
Given how busy providers often are, we looked for other ways to improve care. When feasible, we recommended implementing measures via "systems change" as an alternative to assigning a new responsibility to already overburdened providers. In one center, the medical director agreed with our recommendation that medical assistants take more responsibility for charting and ordering certain diabetes screening tests under standing orders, such as annual urine testing for microalbumin, annual lipid panel, and periodic HbA1c testing. We conducted a special training session for those staff. However, it took several visits before we noticed a change, and we were not confident that it would persist. We learned from this and other experiences that systems change at the practice level can be quite difficult to achieve and sustain.