In this study, we assessed care for a vulnerable group of patients in the VA health care system. Examining a variety of process-of-care measures, including guideline-based angiography procedures and medication use, we found that patients with and without mental illness were marginally less likely to receive angiography or revascularization, and equally likely to receive medications of known benefit after acute myocardial infarction. Mortality at one year may have been higher, although in contrast with other studies (e.g., Druss, Bradford et al. 2001
), this finding did not reach statistical significance. Advantages of this study include the national sample, the diverse age group, and the availability of both inpatient and outpatient clinical data sources for assessment of mental illness.
In contrast to other studies of those with mental illness, which have reported significantly higher mortality (Druss, Bradford et al. 2001
; Frasure-Smith 1991
; Frasure-Smith, Lesperance, and Talajic 1993
; Felker, Yazel, and Short 1996
; Black 1998
; Tsuang, Perkins, and Simpson 1983
; Penninx et al. 1999
; Wulsin, Vaillant, and Wells 1999
; Barefoot and Schroll 1996
; Barefoot et al. 1996
) and lower rates of use of cardiac procedures (Druss, Bradford et al. 2000
), we found minimal differences in both process and outcomes of care in this VA sample. Since some of these other studies were carried out among older patients, and older patients are less likely than younger patients to receive therapies for acute myocardial infarction (Giugliano et al. 1998
; Gatsonis et al. 1995
; Rosenthal and Fortinsky 1994
), we wondered whether restricting our analyses to patients aged 65 and older would yield findings similar to those of other authors.
In the subset of patients aged 65 and older, adjusted analyses showed that those with mental illness were significantly less likely to undergo coronary angiography in the 90 days after the index acute myocardial infarction event (RR 0.80 [0.70,0.91]). However, conditional upon undergoing coronary angiography within 90 days, there were no differences in those with and without mental illness in use of revascularization in the 90 days following admission (age-adjusted RR 1.07 [0.90, 1.26]). There were also no differences in risk-adjusted 30-day or one-year mortality in the subset of these older patients, nor in use of medical therapies. Findings from these subset analyses suggest that our results are not merely due to the younger age of the VA patient population.
Jha et al. (2001)
showed lower mortality for African Americans admitted to the VA health care system for six common medical diagnoses (angina, congestive heart failure, pneumonia, chronic obstructive pulmonary disease, diabetes, and chronic renal failure). We also found smaller mortality differentials but also smaller variation in quality-of-care indicators for patients with mental disorders. The relatively uniform quality of the care delivered in the VA health care system may be an important factor underlying the lack of mortality differences seen in the vulnerable populations treated.
Why might disparities in health care use and mortality for vulnerable patients be attenuated in VA health care settings? In addition to being a system where the delivery of medical care and mental health care is inte-grated by proximity and leadership structure (Druss, Rohrbaugh et al. 2001
), the VA is a national system with a common electronic information system for patient data, and national care and quality monitoring standards (Kizer, Demakis, and Feussner 2000
). In the VA, there is dissemination of information on best practices to practitioners (Feussner, Kizer, and Demakis 2000
), collection and monitoring of data (Daley et al. 1997
), and provision of feedback on performance measures to clinicians (Kizer, Demakis, and Feussner 2000
). Furthermore, access to VA health care requires fewer financial resources than non-VA health care. In fact, one criterion for free care in the VA is an income qualification. This feature may be particularly important for the patient group assessed here. All of these may limit variation in the measures we assessed.
Another possible explanation is that the VA health care system treats large numbers of patients with mental illness (Rosenheck and Dilella 2000
; Norquist et al. 1990
), and mental health care accounts for 14 percent of VA expenditures (Rosenheck and Dilella 2000
). Because of the volume of patients with mental illness treated in the VA, there is greater experience and overall quality of medical care for patient groups characterized by social instability, financial barriers, and stigma (Gelberg, Andersen, and Leake 2000
). Indeed, for other conditions, hospitals that provide high volume care for particular conditions tend to have better outcomes (Thiemann et al. 1999
; Dudley et al. 2000
). In this situation, the VA health care system is delivering high-volume medical care to patients with mental disorders.
In general, why might patients with mental illness be less likely to under-go diagnostic angiography or to receive other types of beneficial treatment? Possible explanations for the differential use of health care in patients with mental illness include difficulties with obtaining informed consent (Shander 2000
), patient preferences and lack of trust, poor compliance (DiMatteo, Lepper, and Croghan 2000
), physicians’ inaccurate estimates of disease prevalence (Graber et al. 2000
), and physician bias (Schulman et al. 1999
Our study is limited by the lack of detailed clinical information on patient symptoms of mental illness. Thus, our findings generalize only to patients with well-documented mental illness, those who are likely to have the most serious mental illnesses, and those with the greatest frequency of behavioral comorbidities (Kessler et al. 1999
). Also, the acute condition we chose to study may be relatively insensitive to disparities in care for patients with mental illness. Had we chosen a chronic disease, such as diabetes, or a preventive care service requiring multiple steps for completion (Druss et al. 2002
), we might have found greater disparities in care.
We had more than 90 percent power to detect an absolute difference of 10 percent or greater for the comparisons presented in , except for the comparison of angioplasty in , where we had 77 percent power to show such a difference. In terms of medication use, we were sufficiently powered for all comparisons except for beta-blockers, thrombolytic therapy, and ACE inhibitors in ideal candidates. Lastly, because this is a retrospective, observational study, there is always the possibility that unmeasured confounders may have biased our findings.
In summary, we found similar process of care and outcome for patients with and without mental illness treated in an integrated health care system. Our findings are not explained solely by the younger age of patients in our sample. These findings are consistent with studies demonstrating reduced health care disparities for other vulnerable groups, such as racial minorities (Jha et al. 2001
; Petersen et al. 2002
; Taylor et al. 1997
), and suggest that an integrated health care system such as the VA may attenuate such disparities for vulnerable groups of patients. The features that attenuate these disparities could be due to the lack of financial barriers for health care, the integrated nature of health care delivery and the strong systems in place for quality monitoring and improvement, the volume of patients with mental health treated, or other possibilities. Further work should address the mechanism by which various health care systems might foster processes of care that narrow disparities in health care use and outcome for vulnerable patients such as these.