Veterans with psychiatric illness were identified from administrative data by having at least one visit with an ICD-9-CM code25
for any of seven psychiatric diagnoses (as a primary or secondary diagnosis): schizophrenia or schizoaffective disorder (295.xx), bipolar disorder (296.0×, 296.1×, 296.40–296.89), major depressive disorder (296.2–296.39), other depression (300.4×, 296.9×, 311.xx, 301.10–301.19), posttraumatic stress disorder or PTSD (309.81), alcohol use disorder (303.xx or 305.00), or drug use disorder (292.01–292.99 or 304.xx or 305.20–305.99). The “other depression” category was comprised almost entirely of dysthymia (300.4×) and depressive disorder not otherwise specified (311.xx).
Medical co-morbidity Medical co-morbidity was evaluated based on patient self-report of medical diagnoses in the LHS survey. Veterans were asked about 12 common medical conditions in the survey, and the count of the diagnoses was used in the analyses.
Physical health status
Physical health status was evaluated by the physical component scale (PCS) of the Short Form-36 for Veterans, an adapted form of the Medical Outcomes Study Short Form 3622
designed specifically for use with veterans23
. It consists of the same 36 items and eight sections as the MOS SF-36: physical functioning, role limitations due to physical problems, bodily pain, general health perceptions, energy/vitality, social functioning, role limitations due to emotional problems, and mental health. Responses in the two role functioning scales are on a five-point ordinal scale, which differs from the dichotomous responses in the MOS SF-3626
. The PCS is standardized to the national U.S. population with a mean of 50 and a standard deviation of 10, where higher scores denote better health status.
VA Site characteristics
Characteristics of the VA facilities where subjects obtained their medical care were evaluated to identify differences in patterns of primary care utilization in terms of geographic location, size of the VA facility, and emphasis on mental health care. We used Rural-Urban Commuting Area (RUCA) codes developed in 1998 at the University of Washington (http://depts.washington.edu/uwruca/about.html
) to identify veterans living in settings ranging from large urban locations to isolated rural ones, using the zip code of residence in the first outpatient encounter record. Facility characteristics included the size of the VA hospital (number of employees), and emphasis on mental health care (the percentage of the budget dedicated to mental health treatment, and the percentage of the mental health budget dedicated to research and education).
Primary care utilization The primary outcome of the study was receipt of any VA outpatient primary care services during FY2000. This was determined using clinic stop codes in the VA Outpatient Care File, a computerized database of all ambulatory care encounters at VA facilities represented by stop codes 301, 322, 323, 348, and 350. Sensitivity analyses expanded the definition of primary care to include all medical specialty clinics and mental health primary care clinics.
Four separate analyses were conducted. First, the representativeness of the LHS sample in terms of sociodemographic and diagnostic characteristics, and outpatient mental health and medical service utilization was evaluated by comparison to data on all VA health service users in FY 2000 (
Second, multivariate logistic regression analyses were conducted to evaluate whether veterans with specific psychiatric diagnoses use primary care services differently than patients without psychiatric illness. Three regression models were evaluated, all with the outcome of any primary care visit in FY 2000. The first model evaluated the odds of any primary care visit across the seven psychiatric diagnoses controlling for sociodemographic characteristics (age, race, gender, education, and service connection). Service connection reflects compensation for a disability connected to military service, and may affect priority for services and motivation to seek services. In the second model, covariates for medical co-morbidity (the count of self-reported diagnoses in the LHS survey) and physical health status (SF-36V PCS) were added. In the third model, VA site characteristics were added: rurality of residence, size of VA facility, the facility’s emphasis on mental health care, and the percentage of the mental health budget spent on research and education.
Third, interaction terms between facility characteristics and mental health diagnoses were evaluated. These analyses required the use of general estimation equations (GEE) to adjust the standard errors of the coefficients because observations within facilities are not independent.
Fourth, the number of primary care visits was evaluated across the seven psychiatric diagnoses (among those veterans who had any primary care visit). Negative binomial regression technique was used because the count of the number of primary care encounters did not meet the distributional assumptions of ordinary least squares analysis. Analyses were conducted in Proc Genmod of SAS®.