Quality Care Improvement Program in a Community-Based Participatory Research Project: DISCUSSION


This evaluation of Project DIRECT showed that at the end of the four-year study period, 40% of enrolled providers still participated in the program.

Among these providers, major patterns of DC (e.g., HbAlc testing, dilated eye exam, foot care, and microalbuminuria) significantly improved, but indicators of glycemic control were unchanged. In addition, office system change as evaluated by use of reminder systems also improved. There are no comparison studies using community-based participatory CQIP in the current literature.

Several program implementation factors may explain some of the results. Turnover in practices and provider drop-out was notable. Turnover in Project DIRECT personnel also played a role. The months-long lack of an evaluator may explain the differential tracking of follow-up activities and variability in the periodicity of follow-up contacts with providers. Snapshots, which were greatly appreciated by the providers, were not provided after 1998 because of turnover in the DC co-PI, DC coordinator, and evaluator positions. Finally, in 1997, new requirements for IRB submission caused interruptions of project implementation. Characteristics of providers who dropped out may become part of the exclusion criteria for community-based projects, i.e., retirement (upper age limit), migration out of target area, and practice disbandment.

A few methodological limitations are related to provider recruitment and chart selection. A provider’s location appears a posteriori to be strongly associated with drop out, as none of the providers who dropped out and about 86% of the providers who stayed in the program were from southeast Raleigh (Table 2). We note that the original inclusion criteria—location in southeast Raleigh—was modified to include providers practicing outside southeast Raleigh. This modification occurred because several practices and/or providers within targeted practices had migrated from the southeast Raleigh to other parts of Wake County over the past few years while still maintaining their stable client base from within the target region. In addition, other participating practices were affiliated with those located within the target area, and therefore, in order to recruit the southeast Raleigh sites, we needed to open participation to the other satellite sites as well. generic cialis 20mg

Chart selection criteria were modified for one practice (a teaching hospital) because it was difficult to assign patients to a single provider, and random chart selection did not apply in all cases. In addition, it is difficult to know how representative of a provider’s usual care practice are data derived from 10 charts. Mobility of providers, e.g., recruitment of new practitioners who did not participate in the intervention, may have diluted the effect of the intervention in later years. Providers’ selection affects the generalizability of results, and the chart selection process and providers’ drop-out/drop-in limit internal validity of results. The latter may dilute the program effect through lack of program continuity or inclusion of other ethnic groups that typically receive better DC, e.g., whites.

Retention is an issue in community-based participatory studies; it varied between 70% and 30% in one diabetes intervention. The high attrition rate of providers/practices raises the sampling unit issue in community interventions: the practice (up-down hierarchy) or individual providers (down-up hierarchy). In the former, the practice management team adopts the intervention, particularly system-changing components, and then relays them to providers; the outcomes may be favorable if the adoption process is participatory. In the latter, the responsibility for running the intervention lies with the providers. A similar issue involves single- vs. multi-practitioner institutions because drop-out of the latter has more consequences for sample size. These issues are important because high and differential attrition rates challenge the project’s sustained implementation, and hence the validity of observed results. Despite the methodological threat, the characteristics of providers more likely to stay in the program can be used as inclusion criteria in community-based replication studies, i.e., single, community-based practitioners with high minority and diabetes patient populations and if multipracti-tioner institution, high minority practitioners. canada viagra online

Competing interventions were not documented; hence, their effect cannot be evaluated. We can nevertheless compare our results with those of similar community-based projects that used DQIP measures. Baseline levels of care in DIRECT were generally greater than those described in Project IDEAL (Improving Diabetes Education, Access to Care, and Living), a statewide initiative to improve the quality of DC for underserved North Carolinians.

In DIRECT, we observed a positive change in providers’ documented patterns of care but not in indicators of patient glycemic control over time. Renders et al. came to a similar conclusion in a nonrandomized trial that contrasted QIP with usual care. The provision of DC improved by all indicators, but none of the indicators for patient outcomes (HbAlc, BP, lipid profile, and weight control) had significantly changed 42 months later. Conversely, the Indian Health Services (IHS) adopted the DQIP measures to monitor their Diabetes Care and Outcomes Audit, and an upward, improving trend was observed for patient care and patient outcomes measures over time, perhaps owing to the fact that the IHS is a more limited system of care with a long-standing surveillance system in a smaller community.

Paradoxically, we did not observe improvement in glycemic control, despite improved measures of care. Variability in lab testing may not account for this finding because the population and laboratories were geographically stable over time. Lack of power is not a reason, as the number of charts audited each year provided enough power to detect significant difference in HbAlc as low as 0.3% ± 0.7%. As an outcome measure, glycemic control is subject to many patient-related factors (e.g., patient knowledge and understanding of the disease and risk-factor reduction strategies and application) other than quality of care that we did not measure. Apcalis Oral Jelly

Another important reason may be the lack of a patient-centered intervention component because a recent review of DQIP interventions indicated that multifaceted interventions that added patient education to professional (healthcare professionals education) and/or organizational (e.g., systems change and nurse role enhancement) appeared to improve both processes of care and patient outcomes. In addition to improving clinical care, DIRECT providers were encouraged to refer diabetes patients to other DIRECT preventive activities. Referral rates were poor all four years, perhaps because providers did not document this in the chart.

These observations call for the addition of a formal patient education component to the DIRECT CQIP, or any diabetes QIP, emphasizing not only diabetes education but prevention of cardiovascular complications. Supporting this suggestion are preliminary results from a small subset of participants in the DIRECT diabetes self-management workshops (n=41). Their pre- and postintervention data showed a downward, though not significant, trend of HbAlc values (8.7% [8.1, 9.4] to 8.2% [7.4, 9.1]; D. Porter-field, unpublished, poster presentation at the DIRECT

National Forum, Raleigh, NC, Mars 2002). Patient education is now understood to be so important that the DQIP second version included patient survey of self-management skills, interpersonal care from providers, and satisfaction with and access to care.
online pharmacy uk


In conclusion, this first multilevel community-based participatory project to reduce the burden of diabetes in African-American adults has demonstrated that a community CQI program, though challenging, is feasible. An effort to maintain or add providers is warranted for increased intervention effectiveness. Among providers who sustained the effort, short- and long-term improvements were observed in diabetes quality care but not in indicators of patient glycemic control. Lack of glycemic control is the pathophysiological basis for costly and disabling complications, thus, a patient-centered component, also strongly suggested by current research, may enhance a clinical QIP program.


So Many Advances in Medicine, So Many Yet to Come