Search

PURPOSE: There is growing interest in incorporating area indicators into epidemiologic analyses. Using data from the 1990 U.S. Census linked to individual-level data from three epidemiologic studies, we investigated how different area indicators are interrelated, how measures for different sized areas compare, and the relation between area and individual-level social position indicators. METHODS: The interrelations between 13 area indicators of wealth/income, education, occupation, and other socioenvironmental characteristics were investigated using correlation coefficients and factor analyses. The extent to which block-group measures provide information distinct from census tract measures was investigated using intraclass correlation coefficients. Loglinear models were used to investigate associations between area and individual-level indicators. RESULTS: Correlations between area measures were generally in the 0.5–0.8 range. In factor analyses, six indicators of income/wealth, education, and occupation loaded on one factor in most geographic sites. Correlations between block-group and census tract measures were high (correlation coefficients 0.85–0.96). Most of the variability in block-group indicators was between census tracts (intraclass correlation coefficients 0.72–0.92). Although individual-level and area indicators were associated, there was evidence of important heterogeneity in area of residence within individual-level income or education categories. The strength of the association between individual and area measures was similar in the three studies and in whites and blacks, but blacks were much more likely to live in more disadvantaged areas than whites. CONCLUSIONS: Area measures of wealth/income, education, and occupation are moderately to highly correlated. Differences between using census tract or block-group measures in contextual investigations are likely to be relatively small. Area and individual-level indicators are far from perfectly correlated and provide complementary information on living circumstances. Differences in the residential environments of blacks and whites may need to be taken into account in interpreting race differences in epidemiologic studies.

We sought to evaluate the predictive accuracy of four bypass surgery mortality clinical risk models and to examine the extent to which hospitals' risk-adjusted surgical outcomes vary depending on which risk-adjustment method is applied.Cardiovascular "report cards" often compare risk-adjusted surgical outcomes; however, it is unclear to what extent the risk-adjustment process itself may affect these metrics.As part of the Cooperative Cardiovascular Project's Pilot Revascularization Study, we compared the predictive accuracy of four bypass clinical risk models among 3,654 Medicare patients undergoing surgery at 28 hospitals in Alabama and Iowa. We also compared the agreement in hospital-level risk-adjusted bypass outcome performance ratings depending on which of the four risk models was applied.Although the four risk models had similar discriminatory abilities (C-index, 0.71 to 0.74), certain models tended to overpredict mortality in higher-risk patients. There was high correlation between a hospital's risk-adjusted mortality rates regardless of which of the four models was used (correlation between risk-adjusted rating, 0.93 to 0.97). In contrast, there was limited agreement in which hospitals were identified as "performance outliers" depending on which risk-adjustment model was used and how outlier status was defined.A hospital's risk-adjusted bypass surgery mortality rating, relative to its peers, was consistent regardless of the risk-adjustment model applied, supporting their use as a means of provider performance feedback. Designation of performance outliers, however, can vary significantly depending on the benchmark and methods used for this determination.

Health care financing is changing rapidly in the United States. We investigated whether and how health care access is changing concurrently with changes in financing, with special attention to a minority population.We examined a longitudinal biracial (half African-American, half White) urban cohort of 3,565 individuals, aged 25-37 years old, in 1992-93 and again in 1995-96. We measured access by self-reported (1) health insurance status, (2) regular source of medical care, and (3) lack of care due to financial problems.In 1992-93, 30.3% of the cohort experienced at least one access barrier, with a decline to 26.8% in 1995-96 (P.005). However, access improved more for Whites than for African Americans; and access improved for higher, but not for lower, income groups (7% improvement for high income, vs 2% deterioration for lower income, P.01). In addition, there was an 11% to 19% absolute increase in individuals making co-payments for health care utilization across all race/sex groups, with African Americans having markedly higher proportions of cost-sharing. African-American, low income, and unemployed individuals reported more acute care, but fewer outpatient visits. Income and employment explained racial differences.While access has improved or stabilized for higher income groups, there is a widening gap according to income, accompanied by an acute care pattern for low income groups that may be both inadequate and cost inefficient.

There is an inverse relation between mortality from cardiovascular causes and the number of elective cardiac procedures (coronary angioplasty, stenting, or coronary bypass surgery) performed by individual practitioners or hospitals. However, it is not known whether patients with acute myocardial infarction fare better at centers where more patients undergo primary angioplasty or thrombolytic therapy than at centers with lower volumes.We analyzed data from the National Registry of Myocardial Infarction to determine the relation between the number of patients receiving reperfusion therapy (primary angioplasty or thrombolytic therapy) and subsequent in-hospital mortality. A total of 450 hospitals were divided into quartiles according to the volume of primary angioplasty. Multiple logistic-regression models were used to determine whether the volume of primary angioplasty procedures was an independent predictor of in-hospital mortality among patients undergoing this procedure. Similar analyses were performed for patients receiving thrombolytic therapy at 516 hospitals.In-hospital mortality was 28 percent lower among patients who underwent primary angioplasty at hospitals with the highest volume than among those who underwent angioplasty at hospitals with the lowest volume (adjusted relative risk, 0.72; 95 percent confidence interval, 0.60 to 0.87; P0.001). This lower rate, which represented 2.0 fewer deaths per 100 patients treated, was independent of the total volume of patients with myocardial infarction at each hospital, year of admission, and use or nonuse of adjunctive pharmacologic therapies. There was no significant relation between the volume of thrombolytic interventions and in-hospital mortality among patients who received thrombolytic therapy (7.0 percent for patients in the highest-volume hospitals vs. 6.9 percent for those in the lowest-volume hospitals, P=0.36).Among hospitals in the United States that have full interventional capabilities, a higher volume of angioplasty procedures is associated with a lower mortality rate among patients undergoing primary angioplasty, but there is no association between volume and mortality for thrombolytic therapy.

Benchmarking is generally considered to be an important tool for quality improvement. Traditional approaches to benchmarking have relied on subjective identification of 'leaders in the field'. We derive an objective, reproducible and attainable Achievable Benchmark of Care (ABC) by measuring and analysing performance on process-of-care indicators. Three characteristics of the ABC that we deem essential are: (1) benchmarks represent a measurable level of excellence; (2) benchmarks are demonstrably attainable; (3) benchmarks are derived from data in an objective, reproducible and predetermined fashion. From these characteristics it follows that (4) providers with high performance are selected to define a level of excellence in a predetermined fashion, but (5) providers with high performance on small numbers of cases do not influence unduly benchmark levels. We use the 'pared mean' to operationalize the ABC. Roughly, the pared mean summarizes the performance of top-ranked providers whereby at least 10% of the patient pool across all providers is included. Bayesian estimators for adjustment of performance of providers with small sample sizes are used to rank providers. Randomized controlled trials to assess the independent effect of the ABC in quality improvement projects are under way. We have developed a methodology objectively and reproducibly to derive a level of excellent, attainable performance, based on measured performance by a group of providers. The ABC can be applied to groups of providers in communities, to institutions and departments within them, or to individual practitioners.

Webster's Dictionary defines a benchmark as 'something that serves as a standard by which others can be measured'. Benchmarking pervades the health care quality improvement literature, and benchmarks are usually based on subjective assessment rather than on measurements derived from data. As such, benchmarks may fail to yield an achievable level of excellence that can be replicated under specific conditions. In this paper, we provide an overview of benchmarking in health care. We then describe the evolution of our data-driven method for identifying an Achievable Benchmark of Care (ABC) on the basis of process-of-care indicators. Here, our experience leads us to postulate the following premises for sound benchmarks: (i) benchmarks should represent a level of excellence; (ii) benchmarks should be demonstrably attainable; (iii) providers with high performance should be selected from among all providers in a predefined way using reliable data; (iv) all providers with high performance levels should contribute to the benchmark level; and (v) providers with high performance levels but small numbers of cases should not unduly influence the level of the benchmark. An example of an ABC applied to the cooperative cardiovascular project leads the reader through the computation of an ABC. Finally, we consider several refinements of the original ABC concept that are in progress, e.g. how to approach the special problems posed by very small denominators. The ABC methodology has been well accepted in multiple quality improvement projects. This approach lends objectivity and reliability to benchmarks that have been a widely used, but until now, arbitrarily defined tool.

To determine associations among health care access, cigarette smoking, and change in cigarette smoking status over 7 years.A cohort of 4,086 healthy young adults was followed from 1985-1986 through 1992-1993. Participants were recruited from four urban sites balanced on gender, race (African Americans and whites), education (high school or less, and more than high school), and age (18-23 and 24-30). Outcome measures were smoking status at Year 7, as well as 7-year rates of smoking cessation and initiation.For each of three access barriers reported at Year 7 (lack of health insurance, lack of regular source of medical care, and expense), participants experiencing the barrier had a higher prevalence of smoking, quit smoking less frequently, and started smoking more frequently; e.g., only 15% of participants with health insurance lapses quit smoking over the 7-year period, compared with 26% of those with insurance (P0.001). Results were similar for each race/gender stratum, and persisted after adjustment for usual markers of socioeconomic status: education, income, employment, and marital status.Health care access was associated with lower prevalence of smoking and beneficial 7-year changes in smoking, independent of socioeconomic status. The possibility that this is a causal relationship has implications in the prevention of cardiovascular disease, cancer and multiple other smoking-related diseases, and deserves further exploration.

In 1991, Medicare began covering screening mammograms subject to copayment and deductible. This study evaluated the effectiveness of Medicare in removing financial barriers to screening mammography among low-income older women.In an inner-city public hospital's General Medicine Clinic, 119 consecutive, eligible, and consenting Medicare-enrolled women without known risk factors for breast cancer other than age, and no mammogram in the previous 2 years, were entered into a randomized controlled trial with follow-up after 2 months. The mean age was 71 years; 77% were black, 92% had an annual income below $10,000, and 52% had had a previous mammogram. All patients were counseled concerning indications for screening mammograms and Medicare coverage, and all were referred to a low-cost mammography facility. Sixty-one subjects were randomly assigned a voucher for a free screening mammogram at the referral facility. Obtaining a mammogram within 60 days of study entry was the main outcome measure.Of the women given vouchers, 27 (44%) obtained screening mammograms, compared with six (10%) of those without vouchers (P.001). Adjustment by multiple logistic regression confirmed this association, yielding an adjusted odds ratio of 7.4 (95% confidence interval, 2.5 to 21.4). Knowledge concerning mammography and breast cancer increased significantly overall (and within randomization groups) between initial interview and follow-up; fear did not change. For women without the voucher, the main reason for not obtaining a mammogram was financial; the main reason for women with the voucher was transportation.In a low-income, inner-city population of older women, financial barriers to screening mammography persist despite Medicare coverage.