Risk-Adjusted Mortality Rates as a Potential Outcome Indicator for Outpatient Quality Assessments
ABSTRACT The quality of outpatient medical care is increasingly recognized as having an important impact on mortality. We examined whether a clinically credible risk adjustment methodology can be developed for outpatient quality assessments.
This study used data from the 1998 National Survey of Ambulatory Care Patients, a prospective monitoring system of outcomes of patients receiving ambulatory care in the Veterans Affairs (VA) integrated service networks.
Thirty-one thousand eight hundred twenty-three patients were followed for 18 months.
The main study outcome measures were observed and risk-adjusted mortality rates.
Of the 31,823 patients, 1559 (5%) died during the 18-months of follow-up. Observed mortality rates across the 22 VA integrated service networks varied significantly from 3.3% to 6.7% (P <0.001). Age, gender, comorbidities (Charlson Index), physical health, and mental health were significant predictors of dying. The resulting risk-adjusted mortality model performed well in cross-validated tests of discrimination (c-statistic = 0.768; 95% CI, 0.749-0.788) and calibration. Analysis of variance confirmed that the 22 integrated service networks differed in their average level of expected risk (P <0.001). Risk-adjusted rates and ranks of the networks differed considerably from unadjusted ratings.
Risk-adjusted mortality rates may be a useful outcome measure for assessing quality of outpatient care. We have developed a clinically credible risk adjustment model with good performance properties using sociodemographics, diagnoses, and functional status data. The resulting risk adjustment model altered assessments of the performance of the integrated service networks when compared with the unadjusted mortality rates.
- SourceAvailable from: Peter Kokkinos
[Show abstract] [Hide abstract]
- "Second, the probability of reverse causality was likely reduced for the following reasons: First, the long follow-up period reduces the potential for confounding due to underlying illness, low BMI, and low fitness. Second, the equal access to care independent of a patient’s financial status provided by the Veterans Health Administration (32) permits epidemiologic evaluations while minimizing the influence of disparities in medical care (33). Finally, the existence of electronic health records within the VA Healthcare System enables detailed observation of history and alterations in health status, coupled with the exclusion of individuals with low exercise capacity (≤5 METs) who died during the first 2 years of follow-up, minimizes the potential impact of pre-existing disease on our findings. "
ABSTRACT: To assess the association between BMI, fitness, and mortality in African American and Caucasian men with type 2 diabetes and to explore racial differences in this association. We used prospective observational data from Veterans Affairs Medical Centers in Washington, DC, and Palo Alto, California. Our cohort (N = 4,156; mean age 60 ± 10.3 years) consisted of 2,013 African Americans (mean age, 59.5 ± 9.9 years), 2,000 Caucasians (mean age, 60.8 ± 10.5 years), and 143 of unknown race/ethnicity. BMI, cardiac risk factors, medications, and peak exercise capacity in metabolic equivalents (METs) were assessed during 1986 and 2010. All-cause mortality was assessed across BMI and fitness categories. There were 1,074 deaths during a median follow-up period of 7.5 years. A paradoxic BMI-mortality association was observed, with significantly higher risk among those with a BMI between 18.5 and 24.9 kg/m(2) (hazard ratio [HR] 1.70 [95% CI 1.36-2.1]) compared with the obese category (BMI ≥ 35 kg/m(2)). This association was accentuated in African Americans (HR 1.95 [95% CI 1.44-2.63]) versus Caucasians (HR 1.53 [1.0-2.1]). The fitness-mortality risk association for the entire cohort and within BMI categories was inverse, independent, and graded. Mortality risks were 12% lower for each 1-MET increase in exercise capacity, and ~35-55% lower for those with an exercise capacity >5 METs compared with the least fit (≤ 5 METs). CONCLUSIONS A paradoxic BMI-mortality risk association was observed in African American and Caucasian patients with diabetes. The exercise capacity-mortality risk association was inverse, independent, and graded in all BMI categories but was more potent in those with a BMI ≥ 25 kg/m(2).Diabetes care 03/2012; 35(5):1021-7. DOI:10.2337/dc11-2407 · 8.57 Impact Factor
[Show abstract] [Hide abstract]
- "The first is to restrict the types of patients in the data, e.g., patients with acute myocardial infarction (AMI) are likely to be admitted to the nearest hospital rather than a hospital of choice due to the urgency of their conditions (Shen, 2003). A variant to this strategy is to account for the severity of patients' conditions in the sample using various medical status variables (Selim et al., 2002). "
ABSTRACT: This paper proposes a method of deriving a quality indicator for hospitals using mortality outcome measures. The method aggregates any number of mortality outcomes into a single indicator via a two-stage procedure. In the first stage, mortality outcomes are risk-adjusted using a system of seemingly unrelated regression equations. These risk-adjusted mortality rates are then aggregated into a single quality indicator in the second stage via weighted least squares. This method addresses the dimensionality problem in measuring hospital quality, which is multifaceted in nature. In addition, our method also facilitates further analyses of determinants of hospital quality by allowing the resulting quality estimates be associated with hospital characteristics. The method is applied to a sample of heart-disease episodes extracted from hospital administrative data from the state of Victoria, Australia. Using the quality estimates, we show that teaching hospitals and large regional hospitals provide higher quality of care than other hospitals and this superior performance is related to hospital case-load volume.Health Economics 01/2008; 19(12):1404-24. DOI:10.1002/hec.1560 · 2.14 Impact Factor