Quality Care Improvement Program in a Community-Based Participatory Research Project: OUTCOMES DEFINITION AND MEASUREMENT

Posted by James

Process Outcome

Retention plan. The plan was evaluated by the annual number of contacts with primary care providers recorded on DC practice records. These contacts were defined as a continuous variable and used to assess the objective’s implementation by the DIRECT staff. Overall attendance at education events was monitored, but attendance of specific CQIP providers was not transcribed.

Objective Outcome

Providers’ retention. Practice and provider characteristics reported by providers on the baseline survey instrument were used to describe the study population. The quantitative objective was to retain 70% of the providers yearly; the choice of this cutpoint was based on the investigators’ previous experience in practice-based research. Outcomes were defined as year-to-year and overall retention. Year-to-year retention was measured as the proportion of providers still actively participating in the program at the end of each fiscal year over the number of providers present at the beginning of the year. Overall retention was measured as the proportion of providers present at the end of each fiscal year over the original number of providers.
antibiotics online pharmacy

Patterns of care. This outcome was the main measure of effectiveness of the intervention and was assessed by medical chart data abstraction during four audit sessions.

Chart reviews were conducted by trained chart reviewers. The criteria for selecting records for review were patient age between 18 and 75, diagnosed diabetes, African-American ethnicity, at least one visit within the last year, at least 50% of visits over the past two years with the named provider, and absence of terminal disease. The initial plan was for the practice managers each to pull 20 records meeting the above criteria, for each provider, and from which the reviewers would randomly select 10 records per provider for review. If the practice staff was unable to find enough charts that met study requirements, they pulled all available charts. In three instances in 1996-1997 and 1999-2000, the reviewers pulled <10 charts/provider. In the second case, it was because the providers had just joined the practice and did not have enough charts. In 1997-1998, when it appeared that it was difficult to relate charts to providers because the providers’ turnover in this teaching hospital was very rapid, the DIRECT investigators decided to abstract data from a fixed number of 50 charts per year for the whole institution for the remaining study period.

The Project DIRECT Record Review Abstraction Form encompassed nine sections: demographics, smoking status, vitals (weight, blood pressure), glucose control (medications, self-management, hypoglycemia assessment, home glucose monitoring), dietary assessment or counseling, physical activity assessment, depression screen, physical exam (vascular, peripheral nerve, foot, ophthalmologic referral), and lab testing (HbAlc and its most recent value, proteinuria, microalbuminuria, blood urea nitrogen and creatinine, generic sodium and potassium, high-density- and low-density-lipopro-tein, sickle cell disease, keto-acidosis), and the number of times these tests were performed during the current and previous years. Documentation of selected care was defined as “whether there was documented evidence that the specific exam/procedure had been ordered, discussed, recommended, or performed within the last two years.” Regarding the ethnicity criterion, instructions were given to practices to pull charts of African Americans, and patients were assumed to be African Americans. Midway through the project it was noted by the reviewers that occasionally a non-African-American chart was provided. In such cases, the reviewers then handwrote in the race on the abstraction form. Because of the inconsistent abstraction of this variable, it was not coded, and we do not have information on patient race. It is believed, however, the proportion of non-African-American residents included in the sample size is low.
buy antibiotics canada

The CQIP effectiveness was defined as a 10% incremental annual change in the prevalence of selected DC measures. Although the project started prior to Diabetes Quality Improvement Program (DQIP), for the sake of comparison, we organized these into measures of accountability and of quality improvement as defined by the DQIP. Accountability measures included the percentage of the population with at least one HbAlc performed in the year, poor HbAlc control (HbAlc >9.5%), documented annual dilated eye exam, lipid profile (HDL and LDL cholesterol) performed in the year, monitoring for nephropathy (proteinuria, microalbuminuria), and controlled blood pressure (BP <140/90 mmHg). The percentage of persons with controlled lipid level was not available in DIRECT CQIP dataset because HDL/LDL cholesterol values were not abstracted. However, monitoring of lipid profile values began in July 2000.

The quality improvement measures included the percentage of patients with well-documented foot exams and distribution of HbAlc and BP. We included other measures suggested in the latest version of DQIP, although they were measured via chart abstraction rather than by patient survey as recommended in the DQIP. These measures are smoking cessation counseling for the accountability set, and diabetes self-management, diet, and physical activity education for the quality improvement set. Accountability measures allow comparisons across healthcare settings; quality improvement measures allow internal QI monitoring.

Glycemic control. The most recent HbAlc (%) values were recorded on the abstraction form. They were analyzed as a continuous variable and categorized according to DQIP criteria of <7, 7-7.9, 8-8.9, 9-9.9, and >10%
silagra tablets

ANALYSES

Provider Retention

An analysis of retention was performed for a one-group pre- and post-test design with repeated measures over time on the same practices and providers. Baseline characteristics of providers who remained in the project were contrasted with those of providers who did not. Differences in the distribution were tested by Chi-squared statistics for categorical variables and by t-test for continuous variables. Unadjusted estimates of year-to-year prevalence and overall retention rate were derived from a repeated-measures regression model of retention status on year.

Patterns of Care and Glycemic Control

Practices were the first stage of a two-stage sampling and charts were the unit of observation. The types of correlation examined are within-practice/ physician correlation for each year, and across time. The former were accounted for using generalized linear mixed models to model random effects for practitioners; the latter was treated as a fixed effect (GLIMMIX macro, SAS Inc., Cary, NC). Preva-lence-of-care measures and mean HbAlc were computed for each year, and tests for linear trends of these values were obtained from these models. For technical and conceptual reasons, all analyses were carried out on providers who stayed in the program the entire study period. Analyses for the QI measures with more than two levels are not supported by GLIMMIX and were analyzed by Chi-squared statistics for the association test and the Mantel-Haentsel test for the trend test. Conceptually, analyzing only those who stayed four years eliminates a bias introduced by the fact that practices that dropped out had lower HbAlcs at baseline than did those who stayed—a finding noted in our exploratory analyses. Analyses were performed with Statistical Analyses software, version 8 (SAS Inc., Cary, NC). The nominal p-value for statistical significance was set to <0.05 for all analyses.
tadacip 20 mg