Risk-Adjusted Mortality Rates as an Indicator for Outpatient Quality
Alfredo J. Selim MD MPH
VA New England Health Care System
Bedford, MA
Funding Period: January 2004 - December 2005

BACKGROUND/RATIONALE:
In an environment of health care restructuring and shifting from inpatient and to outpatient care, the Veteran Health Administration (VHA), acknowledges that the quality of outpatient care needs to be monitored. We have recently suggested that risk-adjusted mortality rates have potential for assessing quality of outpatient care. Our work showed that it is feasible to develop a clinically credible risk adjustment model with good statistical performance for monitoring the outcome of mortality in ambulatory populations. The resulting risk-adjusted mortality rates altered assessments of performance of the Veterans Integrated Service Networks (VISNs) when compared with unadjusted mortality rates. However, this work is preliminary and requires additional evaluation before we can advocate its use in profiling VISNs and facilities and their quality of care.

OBJECTIVE(S):
To explore the performance of the risk-adjusted model and the extent to which this model captures quality of care as opposed to random variation. To evaluate the association of the risk-adjusted mortality rates with process measures and other performance indicators that the VHA uses for profiling VISNs and facilities.

METHODS:
To evaluate the model performance model discrimination and calibration using a cohort of approximately 800,000 veterans. To examine the stability of risk-adjusted mortality rates over time, we will compare model summaries of patients for two time periods of 18 months each. To examine the association between mortality rates and processes of care, we will use two analytical approaches. First, we will perform a correlation analysis between VISN's and facilities’ Clinical Practice Guidelines and risk adjusted mortality data. Second, twenty percent of the patients in the Large Health Survey (160,000 veterans) were surveyed about their perception of the quality of the care, access to health care, coordination /continuity of care, and interaction with caregivers. We are interested in comparing how many and which facilities are considered outliers using
patients’ ratings of health care quality and how risk-adjusted mortality rates relate to these findings.

FINDINGS/RESULTS:
The multistage multivariate models showed a gain in the explanatory power of the models for mortality by adding comorbidities and baseline health status. Older patients, males, whites, and those with a higher number of comorbidities were more likely to die. In contrast, married patients, those with higher incomes, those employed, and those with higher baseline PCS and MCS scores were less likely to die. When we compared the model performance using the 1999 cohort of 705,719 VA users with the 1998 cohort of 31,825, the c-statistics of the models were similar, 0.77 and 0.79 respectively. Our results showed that the risk adjusted mortality were stable over time using a two-way analysis of variance, with VISN as one factor, and time period as the other. Looking at the correlation analysis, there is some association between the rates during the two subsequent 18 months-periods (r=0.65, p=0.001). The patient perceptions of not being able to get a referral to a specialist and poor quality of care were significantly associated with prospective mortality at the VISN level (r-0.50, p=0.04 and r=-0.45, p=0.03; respectively). There were no significant associations between the adjusted mortality rates and the EPRP or the Preventive Services scores at the VISN level.

IMPACT:
Risk-adjusted models that are validated and predict mortality are important evidence of outcomes. Thus, this proposal will expand and greatly strengthen risk adjustment models for comparisons of outcomes at the VISN-level. The VHA performance measurement system can incorporate this methodology into their process of evaluation.

PUBLICATIONS:

Journal Articles

1. Selim AJ, Kazis LE, Rogers W, Qian S, Rothendler JA, Lee A, Ren XS, Haffer SC, Mardon R, Miller D, Spiro A, Selim BJ, Fincke BG. Risk-adjusted mortality as an indicator of outcomes: comparison of the Medicare Advantage Program with the Veterans' Health Administration. Medical Care. 2006; 44(4): 359-65.
2. Selim AJ, Berlowitz D, Fincke G, Rogers W, Qian S, Lee A, Cong Z, Selim BJ, Ren XS, Rosen AK, Kazis LE. Use of risk-adjusted change in health status to assess the performance of integrated service networks in the Veterans Health Administration. International journal for quality in health care : journal of the International Society for Quality in Health Care / ISQua. 2006; 18(1): 43-50.
3. Selim AJ, Fincke G, Berlowitz DR, Cong Z, Miller DR, Ren XS, Qian S, Rogers W, Lee A, Rosen AK, Selim BJ, Kazis LE. No racial differences in mortality found among Veterans Health Administration out-patients. Journal of Clinical Epidemiology. 2004; 57(5): 539-42.

DRA: Health Services and Systems
DRE: Technology Development and Assessment, Quality of Care
Keywords: Outpatient, Quality assessment, Risk adjustment
MeSH Terms: none