Health-Related Quality of Life and Utilities Among Patients With Depression in the Department of Veterans Affairs

Depression is common, chronic, costly, and deadly ( 1 ). Despite the development of effective depression treatment strategies, there have been few rigorous evaluations of the cost-effectiveness of treatment for depression, particularly among veterans ( 2 ). Although clinical guidelines outline the minimum recommended basic care for a given health condition, such as depression, additional data are needed to determine optimal resource allocation from a health economic perspective. In the absence of data, health care providers, policy makers, and administrators may lack critical information that could guide the depression treatment policies and practices at a health system or a population level.

Having measures of health-related quality of life for specific health conditions is necessary for formal cost-utility evaluations, which are one form of cost-effectiveness analysis. Utilities are values that are given to a health state by assigning a value that ranges from 1, for being in perfect health, to 0, for being dead ( 3 ). Utilities provide a common metric to compare the health effects of different health conditions. Utility values allow morbidity and mortality improvements to be combined into a single weighted measure, incremental quality-adjusted life years (QALYs) gained ( 3 ). QALYs are generated by multiplying the utility measure by a person's life expectancy, summed across the duration of life. In calculating cost-utilities, the costs of changes in resource use are compared with those of a relevant alternative treatment and summed in the numerator, and changes in health effects, measured in QALYs, are compared to those from the alternative treatment and summed in the denominator ( 4 ). Resource allocation decisions can then be made by comparing QALYs for individual interventions and diseases.

Cost-effectiveness studies require evaluation of disease-specific utilities. Currently, there are no studies of utilities for depression among veterans receiving care in the Department of Veterans Affairs (VA) health system. The VA is the largest national health system in the United States ( 5 ), and it provides services to a complex patient population; patients tend to be older, poorer, and have more medical and psychiatric comorbidities than individuals in the general population ( 6 ), and individuals receiving VA services are more likely to have mental disorders ( 7 ). In particular, compared with the general population, VA patients have lower scores on the 12-Item Short-Form Health Survey for all eight mental and physical health dimensions (role limitations due to physical problems, physical function, bodily pain, general health perceptions, vitality, social functioning, role limitations due to emotional health, and mental health) ( 7 ), and they have poorer health status, more hospital admissions and longer lengths of stay, and more outpatient physician visits than non-VA patients ( 6 ).

This study breaks new ground by establishing veteran-specific utility measures for patients with and those without depression, which will be available for future cost-utility analyses of VA depression care. More accurate and relevant utility measures are crucial for veterans because many treatments developed outside the VA may have different effects on VA patients. A more accurate utility measure is needed in order to make good decisions regarding resource allocation in the VA.

Methods

We conducted a cross-sectional study of patients from the VA's National Registry for Depression, which was developed and maintained by the national Serious Mental Illness Treatment Research and Evaluation Center (SMITREC) in Ann Arbor, Michigan. The National Registry for Depression includes comprehensive administrative data on diagnoses, utilization, and cost for all VA patients nationwide who have had a depression diagnosis recorded as part of a VA services encounter from fiscal year (FY) 1997 to the present. We identified patients on the basis of ICD-9 diagnoses of depression ( ICD-9 codes 293.83, 296.2, 296.3, 298.0, 300.4, 301.12, 309.0, 309.1, 309.28, and 311). Additionally, SMITREC maintains data for a random sample (N=100,000) of all VA patients who had at least one inpatient or outpatient encounter during FY 1999. We included patients from the random sample who did not have a diagnosis of depression.

Patients were included in the analyses if they received VA care in FY 1999 (either with a diagnosis of depression, as indicated in the National Registry for Depression, or from SMITREC's random sample of VA patients) and if they also completed the SF-12 portion of the VA's Large Health Survey of Veteran Enrollees (LHSV). The LHSV is the largest survey ever conducted of veterans using VA health services. It was administered in 1999 using a national random sample of veteran enrollees who were mailed a survey including a comprehensive assessment of health and demographic variables ( 8 ). Overall, 87,797 patients met these criteria. This study received approval from local institutional review boards.

Although there is little systematic information available regarding health-related quality of life among VA patients with depression, utility weights can be estimated for veterans with depression by using data from the LHSV. The LHSV includes data from the SF-12, a survey including eight scales measuring physical and mental health. These scales can be combined into two summary measures: the physical component summary (PCS) and the mental component summary (MCS) ( 9 ).

By using the method developed by Brazier and colleagues ( 10 ), SF-12 measures can be converted into utilities. In this algorithm, Brazier and colleagues created the SF-6D from the SF-12, which includes seven items across six dimensions: physical functioning, role limitations, social functioning, pain, mental health, and vitality. Each dimension had between two and five levels with a total of 7,500 possible health states. Then they conducted a preference-based valuation survey of the general population of the United Kingdom (N=836) with a sample of 249 health states in which each participant ranked and valued six health states by using the standard gamble utility elicitation technique, which forces respondents to make explicit tradeoffs between a given health state and the possibility of either full health or death ( 11 ). Finally, they estimated a model to predict values for all health states from the SF-6D by using econometric techniques, including ordinary least squares regression. Parameter estimates derived from each health state from Brazier's model were then used to calculate utilities for each respondent's reported health state.

Other measures of interest drawn from the LHSV include patient age, gender, marital status, race, education level, and the PCS and MCS scores from the SF-12. We also included responses to the following question: "Has a doctor ever told you that you have depression?" Furthermore, we created a count of self-reported comorbidities for all conditions specified in the LHSV for each respondent, including arthritis, benign prostatic hypertrophy, cancer, chronic low back pain, chronic lung disease, congestive heart failure, diabetes, heart attack, hypertension, posttraumatic stress disorder (PTSD), schizophrenia, spinal cord injury, and stroke.

Other measures of interest drawn from VA claims data for all patients in our study included whether during the study period the patient had a diagnosis of alcohol dependence or abuse (including ICD-9 codes 291, 303.0, 303.9, and 305.0) and a diagnosis of drug dependence or abuse (including opioids, barbiturates, cocaine, cannabis, amphetamines, hallucinogens, and other drug abuse with ICD-9 codes 292, 304, and 305.2–305.9).

Results

The findings from our analyses are presented in Table 1 . All comparisons of patients with depression and those without depression were statistically significant. Overall, the average age of participants in the study was 60 years, with patients with depression being slightly younger than patients without depression. More than 90% of the sample was male, more than 80% was white, slightly more than half were married, and approximately 75% had graduated from high school. On average, study respondents had 3.9 chronic medical comorbidities (3.5 associated with physical conditions and .4 associated with mental conditions other than depression). Veterans with a diagnosis of depression had 4.1 chronic medical comorbidities (3.6 physical and .5 mental conditions other than depression), and veterans without a diagnosis of depression had 3.5 chronic medical comorbidities (3.3 physical and .2 mental conditions other than depression). Although almost 69% of the overall sample reported having depression, 86% of the sample with a diagnosis of depression reported having depression (meaning that 14% of patients identified using the VA depression registry did not report having depression) and only 36% of the sample without a diagnosis of depression reported having depression. According to VA claims data, 7% of respondents reported any alcohol abuse or dependence (8% in the sample with depression and 4% in the sample without depression), and 3% reported drug abuse or dependence (4% in the sample with depression and 2% in the sample without depression).

Enlarge table

All MCS and PCS scores were well below national averages of 50 (MCS and PCS scores are normalized to 50 in the general population) ( 9 ), with greater variation between groups for the MCS scores than for the PCS scores. Veterans with depression had an average MCS score of 32.32 and an average PCS score of 32.63. The comparison sample had an average MCS score of 43.86 and an average PCS score of 34.51. Finally, utilities were lower for veterans with depression (.54), compared with veterans in the comparison sample (.63).

Discussion

Study findings add to the literature on utilities among VA patients with versus those without a diagnosis of depression. The sample of veterans with depression had higher rates of psychiatric, physical, and substance abuse comorbidities than a comparison sample of the veteran population without depression treated at the VA. This is the first study to examine on a nationally representative level utilities associated with depression among VA patients, and we observed that the VA population with depression had relatively low utilities. To put these findings in context, Pyne and colleagues ( 12 ) found that a .10 change in utility could represent the difference between responding and not responding to depression treatment, thereby indicating that the observed difference in utilities between veterans with depression (.54) and veterans without depression (.63) is also likely a clinically significant difference.

Our goals were to both quantify utilities among veterans with depression and those without depression and to place our findings in the context of existing literature on health-related quality of life among individuals with and without depression in the general population. A review of research on this topic reveals that our findings of health-related quality of life among veterans are consistent with the results of other studies, such as the Veterans Health Study ( 6 , 7 , 13 ). However, none of these prior studies calculated VA-specific utilities—they calculated only MCS and PCS scores. Furthermore, utilities for depression range from .30 to .70 in non-VA populations ( 14 ). Our work demonstrates that not only do veterans with depression fall in the middle of that range (.54), but veterans without depression have a utility value that is similar to that of nonveterans with mild depression (.64) ( 15 ). Therefore, even relatively healthy veterans (for example, those without depression) still rate their health-related quality of life as being worse than that of the general population, and in particular, worse than that of people from the general population who actually have depression. This reinforces existing literature that veterans (with and without depression) are sicker and have poorer health-related quality of life than nonveterans, thereby also indicating that this population may be in greater need of psychiatric and medical services to improve their health and well-being than nonveteran populations.

Although our findings are consistent with the literature, we note that this study included veterans who received some VA services for depression and who completed the LHSV; therefore, our findings may not be representative of the veteran population overall. Also, because administrative data were used, some of the cases of depression may not necessarily be identified. If this is the case, some members of the group without depression may actually be a group with undetected depression. If so, the differences between the groups may be a conservative estimate. Furthermore, study data did not include detailed measures of illness severity, treatment, and course of depression or other health conditions. Finally, Brazier's algorithm is based on the preference for health states observed in the U.K. general population. These may be different than the preferred health states found in the U.S. veteran or general population, and future research should obtain similar preference-based valuation survey estimates from these U.S. populations. That being said, this study is unique because it includes nationally representative data and because it combines survey responses, including the SF-12 that we can use for utility calculation, and linked medical claims data (many studies usually include only one of these sources of data). It would, however, be helpful to replicate this study with newer data (after 1999) and to have repeated measures of utilities, as well as patient health information over time.

Conclusions

This work lays the foundation for more in-depth research on health-related quality of life for veterans with and those without depression. Future research should investigate how treatment or interventions effect changes in health-related quality of life and build these findings into broader cost-effectiveness models of VA depression care. The VA is currently focused on implementing integrated mental health care into primary care clinics, as part of the Comprehensive Veterans Health Administration Mental Health Strategic Plan ( 16 ). Researchers and policy makers need to be able to determine both the costs and the effectiveness associated with these interventions, which may include services such as providing telephone disease management calls conducted by nurses or social workers to improve adherence to depression treatment. The application of more accurate utility measures is therefore paramount for advancing research and practice in this group. Making greater use of data on physical and mental health of a nationally representative sample of veterans with depression can only help providers and policy makers maximize the impact of these changes and improve patients' quality of life.