Practice facilitation (PF) is an especially promising approach to supporting primary care redesign [1–3]. PF occurs when a trained facilitator provides support services to a primary care practice for an improvement initiative. The PF approach enables teams to overcome challenges encountered when implementing changes in the office setting by building their internal capacity to engage in redesign or improvement efforts [4]. Facilitators assist teams within the practice as they identify and prioritize areas of change as well as help them develop tailored action plans for improvement. PF has also been referred to as quality improvement coaching or practice enhancement assistance [2]. A recent systematic review suggests that PF is a robust intervention for improving the adoption of evidence-based preventive care guidelines in primary care [1].

The Chronic Care Model (CCM) describes elements of an approach to providing chronic care in ambulatory care settings with proven effectiveness [5–7]. Studies of CCM implementation and other primary care redesign efforts suggest that practice change/redesign is difficult [7]. Change can be especially challenging in small, autonomous primary care practices where resources are scarce and where the environment external to the practice is not supportive of change [8]. PF is emerging as a promising approach for assisting practices in overcome these challenges. Although effective at improving guideline adoption for preventive care, such as colon cancer screening [1], the ability of PF to change how the primary care team organizes itself around chronic illness care remains unknown.

Here we report the results of group-randomized trial of a PF intervention to improve the organization and delivery of diabetes care in small, autonomous primary care practices. The specific objective is to examine changes in the degree to which care is organized around the CCM at the conclusion of the one-year intervention, its sustainability one year after withdrawal of the PF intervention, and subsequent improvement in the delayed intervention (control) practices.

This was a group-randomized trial launched in the fall of 2007 in the South Texas region of the United States. The study design of this trial and details about the intervention have been previously reported [9]. Briefly, the study was conducted in small, autonomous primary care clinics or ‘practices’ in South Texas. These urban, suburban, and rural practices, each with one to three clinicians, serve a population of primary care patients diverse in demographic characteristics, insurance coverage, and healthcare needs. Subjects for this study were the clinicians and staff at participating practices.

ACIC scores for intervention and control practices for all three waves of data collection are shown in Table 3 and the Figure 1. The repeated measures ANOVA revealed a significant within-subjects effect accounting for a significant increase in ACIC over time. The between subjects (e.g., group) effect was not significant, however the time by intervention interaction term was significant, (F = 4.26, p = 0.046), indicating a significant difference in ACIC scores between the initial intervention and delayed intervention groups. Table 3 also provides the number and proportion of clinics in each ACIC category (limited, basic, good, full) across the three waves of data collection. Within the initial intervention group, 10 of the 19 clinics were in the ‘good’ or ‘fully’ supported ACIC category at baseline and 16 of the 19 clinics were in these categories after the PF intervention (wave two data). A similar pattern can be seen for the delayed (control) intervention clinics from wave two to wave three, after the PF intervention.

A visual inspection of the ACIC means shown in the Figure 1 suggests that the significant time by intervention term may have occurred from wave one (baseline) to wave two. This was confirmed by a paired t-test that showed a significant difference in ACIC scores for the initial intervention group from baseline to wave two. The mean difference in CCM score from wave one to wave two for the intervention practices was 0.75 (95% CI of the difference 0.09, 1.40; p = 0.02), mean difference. Although the proportional improvement in CCM score (13.9%) in the delayed group from wave two to wave three (before and after the PF intervention) was almost equal to the change in the initial intervention group between wave one and wave two (14.4%), the improvement did not reach statistical significance. Of note is the sustained and slight increase in ACIC score in the initial intervention group one year after withdrawal of the PF intervention.

There was significant improvement in the degree to which care was consistent with the CCM in small primary care practices randomized to the PF intervention compared to those in the delayed intervention (control group). This improvement was sustained one year after completion of the PF intervention. These findings are consistent with prior studies which demonstrated the effectiveness of PF [1]. An analysis of sub-scores from the ACIC revealed the biggest change occurred in the area of self-management support score, which increased from 4.39 to 5.83 among practices in the initial intervention group, compared to a slight decline from 5.22 to 4.95 in the delayed intervention (control) group. Such a sub-score change is consistent with the finding that 16 of the 19 intervention practices made improvement efforts within this area as shown in Table 2.

Why should PF result in such sustained improvement? When an organization and its members encounter a change intervention, such as redesigning the way they deliver chronic illness care, they must often shift goals and learn to solve non-routine problems, often under time pressure and turbulent conditions. Primary care practices and the diverse individuals in them must continually make sense of new environmental cues as a team and develop a collective meaning of new or stressful situations if they are to successfully and consistently implement change [16]. Research across different healthcare settings has demonstrated the importance of improving relationships and sense-making activities within teams when designing interventions [17–20]. It is possible that the PF intervention was effective because the facilitation improved relationships among the practice members and enabled conversations within teams during their team meetings that allowed them to collectively make sense of the changes they were asked to make [19].

It is also important to note that the ACIC scores were sustained one year after withdrawal of the PF intervention, and in fact, improved slightly. This suggests that the PF intervention may have changed how information is shared and used in the practice and how practice members accomplished tasks that proved enduring over time [21]. That is, PF changed practice member relationships in ways that improved the ability of the practice team to make sense and learn from the intervention, leading to sustainability [19].

The change in self-assessed chronic illness care, although was significant, was relatively small, from 5.59 at baseline to 6.60 one year after the PF intervention. This raises questions about its relevance for improved clinical outcomes. Prior work has demonstrated a relationship between ACIC score and control of HbA1c [22–24]. In one study, a one-point increase in the ACIC score was associated with a 0.144% decrease in HbA1c; thus, it is possible that the observed one-point improvement in ACIC scores seen one year after the PF intervention may result in improved clinical outcomes [22]. One explanation for the small observed change in ACIC scores is that development of a common understanding of the desired goals and required changes needed to deliver better chronic illness care requires a longer period of time than one year. In fact, ACIC scores continued to improve in the year after the completion of the PF intervention, although the rate of change was less than during the year of facilitation.

It is possible that larger improvements in chronic illness care within a primary care practice are possible with a more intense intervention than the six to seven facilitated team meetings over a one-year period delivered in this study. Although the original intent of the study design was to have monthly facilitation visits over the 12-month intervention period, competing demands within the practices such as EMR implementation and staff turnover, prevented this higher number of visits. In addition, one may need to combine facilitation with other components, such as a learning collaborative, repeated performance reporting, and educational outreach visits [25]. Finally, one would be remiss not to mention the importance of the lack of financial incentives for these practices to make substantive change in the manner in which they deliver chronic illness care. These practices are small businesses and as such must consider the possible ‘return on investment’ in any change effort they undertake.

Change efforts within some practices were stymied by other changes that occurred during the intervention. Staff turnover often created problems by delaying facilitation visits or suspended visits all together. For example, in one practice, the staff ‘champion’ for improving diabetes care left, and the practice ‘stalled’ in its efforts to make changes until several months later when another staff member took on this role near the end of the facilitation intervention. In addition to staff turnover, several practices experienced other significant changes during the one-year PF intervention such as moving the practice location (three practices in the initial intervention group and three practices in the delayed intervention group), and implementation of an electronic health record (three practices in the initial intervention group and four practices in the delayed intervention group).

This study is also limited by the geographic restriction to a narrow region of the United States, and potential for selection bias inherent the selection of practices utilized for the study. In addition, the participating primary care practices were small, and different results might have been obtained using larger primary care practices, especially those embedded within integrated healthcare systems. The potential for selection bias favoring practices that are eager to improve also exists. However, these practices also tend to be the ones that are higher performing at the start of the study, making it more difficult to demonstrate significant and sustained improvements.

The study also has a number of strengths, including a diverse sample of ‘real-world’ practices, the use of theory to inform the intervention design, the randomization of practices to intervention and delayed intervention (control) groups, and the low rate of attrition of practices from the study. The impact of the PF intervention is demonstrated by the finding that 16 out of 19 practices showed improvement on the self-managemetn sub-scale.

Practice facilitation resulted in significant and sustained improvement in the degree to which chronic illness care was consistent with the CCM in small primary care practices, although the clinical impact of the degree of improvement on patient outcomes is uncertain. Given the current challenges and ongoing efforts to redesign primary care into Patient-Centered Medical Homes across the United States, many are calling for the development of an infrastructure that supports the use of practice facilitators or coaches should [26, 27]. This study adds additional support for these initiatives as it adds to the growing body of evidence that practice facilitation is an effective intervention for sustained improvement in the how care is delivered in small primary care practices. Next steps include linking organizational and process changes that result from practice facilitation to improvements in clinical outcomes, patient experience, and cost-containment.