Diabetes Treatment Among VA Patients With Comorbid Serious Mental Illness

Abstract

Back to table of contents Previous article Next article ArticleFull AccessDiabetes Treatment Among VA Patients With Comorbid Serious Mental IllnessSarah L. Krein Ph.D., R.N.C. Raymond Bingham Ph.D.John F. McCarthy Ph.D., M.P.H.Allison Mitchinson M.P.H.Jonathan Payes M.A.Marcia Valenstein M.D., M.S.Sarah L. Krein Ph.D., R.N.Search for more papers by this authorC. Raymond Bingham Ph.D.Search for more papers by this authorJohn F. McCarthy Ph.D., M.P.H.Search for more papers by this authorAllison Mitchinson M.P.H.Search for more papers by this authorJonathan Payes M.A.Search for more papers by this authorMarcia Valenstein M.D., M.S.Search for more papers by this authorPublished Online:1 Jul 2006AboutSectionsPDF/EPUB ToolsAdd to favoritesDownload CitationsTrack Citations ShareShare onFacebookTwitterLinked InEmail Individuals with serious mental illnesses, such as schizophrenia, bipolar disorder, and other psychoses, represent a high-risk patient subpopulation, with complex, ongoing treatment service needs. Many of these individuals also have comorbid medical illnesses and are at increased risk for poor health outcomes. High rates of morbidity and mortality have been reported among patients with serious mental illness and have sometimes been attributed to health system characteristics, including inadequate integration of the delivery of psychiatric and nonpsychiatric services. Patients with serious mental illness also may be at increased risk for poor outcomes because of decreased access to providers or poorer health habits and treatment adherence.The U.S. Department of Veterans Affairs (VA) health care system provides the ideal context for examining characteristics of health services treatment among large samples of patients with serious mental illness. The VA information system is regarded as one of the best in the nation, and VA administrative data include detailed diagnosis and treatment measures. Moreover, the VA has transformed itself over the past decade, substantially enhancing its integration of services. We evaluated the hypothesis that patients with serious mental illness are less likely to receive adequate treatment for nonpsychiatric illnesses. Specifically, we compared processes of care for diabetes and intermediate disease outcomes (hemoglobin A1c levels), which influence later health outcomes (for example, risk of blindness) among VA patients with diabetes with and without serious mental illness. As a chronic condition requiring ongoing care and often complex treatment, diabetes can pose substantial management challenges for patients and health care providers. Conditions associated with reduced functioning, especially mental, cognitive, and emotional functioning, may particularly complicate diabetes treatment and worsen health and other outcomes. For example, several previous studies have reported that depression is associated with poorer adherence to dietary and self-management recommendations ( 1 , 2 ), poorer glycemic control ( 3 ), greater social and vocational impairments ( 4 ), and increased health care cost and utilization of services ( 2 , 5 , 6 , 7 , 8 ). Diabetes is a prevalent condition among patients with schizophrenia ( 9 , 10 , 11 ), and second-generation antipsychotic medications commonly used to treat psychosis may further increase patients' risk for diabetes ( 12 , 13 , 14 , 15 ). Among VA patients who received a diagnosis of serious mental illness in a given year, one in five also received a diagnosis of diabetes. Serious mental illnesses may cause additional difficulties in diabetes management and make positive outcomes harder to achieve. However, two recent studies ( 16 , 17 ) found little evidence of poorer quality of diabetes care for patients with a co-occurring serious mental illness. These studies focused only on the receipt of recommended care processes, such as having a foot sensory exam, or on a single intermediate outcome (A1c). In addition, they did not explore mechanisms that might account for or promote good-quality care for these patients. In this study, we compared large national samples of patients who had diabetes and a serious mental illness with patients who had diabetes and no diagnosis of mental illness using several diabetes-specific process measures and intermediate outcomes. We also examined inpatient and outpatient health care utilization and patterns of care to better understand how the organization of care delivery might influence these processes and outcomes.MethodsStudy design and setting Approval for this study was obtained from the institutional review board at the Ann Arbor VA Healthcare System. Data were obtained from two VA health services registries: the National Psychosis Registry ( 18 ) and the Healthcare Analysis and Information Group/QUERI-DM (Quality Enhancement Research Initiative for Diabetes Mellitus) diabetes registry ( 19 ). We examined data for patients receiving care in fiscal year (FY) 1998 (October 1, 1997-September 30, 1998). Serious mental illness was operationally defined as having a diagnosis in one of five ICD-9-CM ( 20 ) mental health diagnostic categories: all diagnoses of schizophrenia except latent schizophrenia, schizoaffective disorder, bipolar disorder, other nonorganic psychoses, paranoid states, and affective psychoses. The National Psychosis Registry contains utilization data, including diagnoses and procedures for inpatient stays and outpatient visits. The FY 1998 diabetes registry contains pharmacy, laboratory, and utilization information for individuals with diabetes who used VA health services during the fiscal year. Patients with diabetes were identified as those who received a prescription for insulin, an oral hypoglycemic agent, or blood glucose monitoring supplies or who had at least one inpatient encounter or two outpatient encounters with a diabetes-related diagnosis code ( ICD-9-CM codes 250.0-250.9, 357.2, 362.0, and 366.41). The two registries were used to identify two mutually exclusive groups of patients. The first group included 18,273 patients with both diabetes and a comorbid serious mental illness in FY 1998. The second group consisted of a randomly selected age-matched set of 18,273 patients with diabetes who did not have a diagnosis of serious mental illness. To screen out individuals who might have used the VA to receive medication only, all patients needed to have had at least one primary care or mental health care visit during the year to be included in the study. Also, because the utilization part of the study primarily focused on ambulatory care, patients with an inpatient stay longer than 150 contiguous days were excluded.MeasuresIndicators of diabetes management. Our selection of diabetes process and outcome measures for the analysis was based on recommended clinical standards for diabetes management. Specifically, we used the set of measures (both technical process and intermediate outcome) proposed for diabetes care monitoring and quality improvement through the Diabetes Quality Improvement Project ( 21 ). The indicators used for this study were the percentage of patients who had at least one hemoglobin A1c level obtained in the past year, the hemoglobin A1c level and the percentage of patients with a high-risk hemoglobin A1c level (A1c ≥9.5 percent) based on the last value obtained, the percentage of patients who had a low-density lipoprotein (LDL) cholesterol measure in the past year, the LDL level and the percentage of patients with high-risk LDL values based on the most recent test using a cutoff of 130 mg/dl, and the percentage of patients who had a total cholesterol test (because an LDL level could not be calculated for some patients because of elevated triglyceride levels). For quality measures related to having had a test in the past year, higher percentages indicated higher quality, whereas for measures based on test values, such as mean A1c value, lower values or percentages indicated better quality. Inpatient and outpatient utilization. Inpatient utilization measures included the number and percentage of patients with at least one inpatient stay and the average length of stay per patient. Outpatient care was measured by the total number of outpatient visits and the average number of visits per patient. An outpatient visit was defined as all outpatient encounters occurring on the same day, and each visit was classified according to the types of clinics visited by the patient: primary care only; mental health care only; specialty care only; or a multiclinic visit, which was a visit to more than one type of clinic on the same day—for example, visits to both primary care and mental health care clinics. Primary care included general internal medicine, diabetes clinic, geriatrics clinic, and primary care. Mental health care included encounters for mental illness treatment (individual and group), substance abuse treatment and rehabilitation, intensive case management, and intensive individual programs. Specialty care included encounters at clinics offering specialized care, such as dermatology, ophthalmology, and neurology. Outpatient encounters that did not fall into one of the three classes, such as visits for laboratory tests, were disregarded. Statistical analysis Group comparisons were initially conducted using one-way analysis of variance (ANOVA) and chi square tests. Because of the large sizes of the groups being compared in this study, however, conventional tests of statistical significance were not necessarily clinically meaningful, as small differences were often statistically significant. Thus, instead of using statistical significance as a primary tool in identifying potentially important results, estimates of effect size were used to aid in data interpretation. Cohen's d ( 22 ) was used to estimate effect sizes for the ANOVA models, and odds ratios (ORs) ( 23 ) represented the effect size for chi square tests. Small, medium, and large effects were .20, .50, and ≥.80 for Cohen's d; 1.2, 1.5, and ≥2.0 for positive ORs; and .83, .67, and ≤.5 for negative ORs. All comparisons of diabetes care processes and intermediate outcomes were stratified by treatment regimen as a proxy for disease severity. Specifically, patients were categorized as to whether they were treated without any hypoglycemic agents (indicating milder disease), treated with oral hypoglycemic agents only, or treated with insulin (indicating more severe disease). ResultsThe study sample was predominantly male (97 percent in the diabetes-only group and 96 percent in the diabetes and serious mental illness group) with a mean age of 58±12 years. A higher proportion of those in the diabetes-only group were married (29 percent compared with 21 percent). Among patients with an identified race or ethnicity classification, those in the diabetes-only group were slightly more likely than patients with comorbid serious mental illness to be classified as white (69 percent compared with 64 percent) and less likely to be classified as Hispanic (8 percent compared with 11 percent) or black (22 percent compared with 24 percent). Within the diabetes and serious mental illness group, 73 percent were diagnosed as having schizophrenia, 18 percent as having bipolar disorder, and 9 percent as having another type of psychosis.Quality of diabetes careTable 1 presents the comparison of diabetes care processes and intermediate outcome measures between the two patient groups stratified by diabetes treatment regimen. Only one difference in the quality-of-care measures and intermediate outcomes between patients with diabetes and those with diabetes and serious mental illness reached a small effect size. Among patients who did not receive hypoglycemic medications (who presumably had less severe diabetes), patients with only diabetes had slightly higher A1c levels than their counterparts who also had a serious mental illness. Otherwise, within each treatment stratum, the two groups of patients had very similar amounts of diabetes care and levels of control. Table 1 Diabetes care processes and intermediate outcomes stratified by treatment regimen for 36,546 Veterans Affairs patientsTable 1 Diabetes care processes and intermediate outcomes stratified by treatment regimen for 36,546 Veterans Affairs patientsEnlarge tableUtilization of inpatient and outpatient care Differences in inpatient and outpatient health care utilization between the two groups are shown in Table 2 . Patients in the diabetes and serious mental illness group had more total contact with the health care system than patients with solely diabetes and received both more inpatient and outpatient services. Patients with diabetes and serious mental illness were 2.8 times more likely than the diabetes-only group to have had an inpatient stay and also had a slightly longer length of stay when hospitalized. Patients with comorbid serious mental illness also made far more visits to outpatient care. On average, patients with diabetes and serious mental illness had more than double the number of outpatient visits per person compared with diabetes-only patients (36.7 compared with 16.4; Cohen's d=.62, a moderate effect size). Table 2 Utilization of inpatient and outpatient health care services among patients with diabetes only and patients with diabetes and a comorbid serious mental illnessTable 2 Utilization of inpatient and outpatient health care services among patients with diabetes only and patients with diabetes and a comorbid serious mental illnessEnlarge tableType of outpatient careTable 2 also shows the types of outpatient care received by each patient group. Notably, patients with diabetes and serious mental illness not only had more total contact with the health system and more mental health visits than patients with diabetes alone, they also had more primary care, medical specialty, and multiclinic visits than did patients with diabetes only. On average, patients with diabetes and serious mental illness had 1.0 more visit to primary care than patients with diabetes only (mean of 5.8 visits compared with 4.8), 1.1 more specialty medical visits (mean of 5.3 compared with 4.2), and 1.3 more multiclinic visits (mean of 3.5 compared with 2.2). The frequency of multiclinic visits made by the patients with diabetes and serious mental illness is especially striking, with a total of 43,171 multiclinic visits, compared with 14,186 for the diabetes-only group. On a per-patient basis, 69 percent of those in the group with diabetes and serious mental illness had a multiclinic outpatient visit, compared with 36 percent in the diabetes-only group, and nearly half of the multiclinic visits for patients with diabetes and serious mental illness involved both primary care and mental health services ( Table 3 ). Table 3 Multiclinic visits by type for patients with diabetes with and without comorbid serious mental illnessTable 3 Multiclinic visits by type for patients with diabetes with and without comorbid serious mental illnessEnlarge tableDiscussion In general, patients with multiple chronic conditions, especially diabetes and comorbid serious mental illness, have challenging health care needs and subsequently may be more likely to receive poorer-quality diabetes care and have poorer outcomes. However, in this large-sample study, diabetes care processes and intermediate outcomes of patients with diabetes and serious mental illness did not differ appreciably from those of patients with diabetes alone. Performance on several of the measures might be considered poor for both groups, as compared with current performance and quality standards in the VA ( 24 ), although it is important to recognize that there have been substantial improvements in VA-delivered diabetes care over the past several years ( 24 , 25 , 26 ). Our overall findings of comparable diabetes care processes and A1c values were somewhat unexpected, given prior research that has demonstrated a clear negative effect of concurrent depression on diabetes outcomes ( 2 , 3 ), even though our findings are consistent with two recent studies of individuals with diabetes and comorbid serious mental illness ( 16 , 17 ). Although the causal processes underlying these results remain unclear, this study offers some possible explanations. Patients with diabetes and serious mental illness had much more contact with the health care system than patients with diabetes alone. Patients with diabetes and serious mental illness used more inpatient care than patients with diabetes only and had much higher levels of outpatient use, both of which, even if not focused specifically or solely on diabetes treatment, provide an increased opportunity to complete important diabetes care processes. Patients who also had serious mental illness not only had a greater number of total outpatient visits, they also had more primary care and specialty medical visits than patients with diabetes only. Patients with diabetes and serious mental illness made, on average, 36.7 outpatient visits compared with 16.4 visits for those with diabetes only. They had one additional primary care visit, 1.1 additional specialty medical care visits, and 1.3 more multiclinic visits during the year compared with visits for those with diabetes only.The substantial number of visits involving multiple types of health care providers is especially noteworthy. In particular, 69 percent of patients with comorbid serious mental illness had visits involving multiple types of clinic encounters, compared with 27 percent of those with diabetes only, and nearly 50 percent of multiclinic visits for this group involved encounters with both primary care and mental health services. These results indicate that patients with diabetes and serious mental illness not only had more contact with health care providers but also may have received more intensive and integrated outpatient care. Unlike other health care systems that might provide only medical care and contract out for mental health care, the VA is an integrated health care system that provides comprehensive access to health care across the full spectrum of health needs. In many areas this integration includes the actual physical integration of health care services, which allows patients with diabetes and comorbid serious mental illness to visit multiple clinics, receive treatment for both conditions, and have their prescriptions filled during a single visit day.Integration also may be achieved through better communication among providers and referrals across specialties. The VA maintains an electronic patient record system that contains both medical and mental health care records for all patients treated at that medical center. Seamless access to patients' records gives providers greater insight into their patients' needs and potential conflicts between different aspects of their care, which creates opportunities for improving treatment. In addition, primary care providers can often refer patients for subspecialty care and mental health providers can refer their patients for medical care without requiring the patient to travel to another facility or burdening the patient with the stresses of reconciling conflicts that arise from referrals outside their health care system. We note several study limitations. First, this analysis was based on a population of patients who were using the VA health care system. This population was primarily male and likely had a poorer health status than populations in other health care settings ( 27 ). Thus these results may not be generalizable to all populations. Second, with the exception of age and type of mental health condition, this analysis did not take into account specific sociodemographic factors, such as race, that may affect diabetes outcomes or utilization. However, in a separate analysis we compared treatment utilization by the groups with diabetes with an age-matched group of patients with serious mental illness alone for each mental health diagnosis category and found few effects that appeared to be specific to a particular type of mental illness. Also, among patients with staff-identified race and ethnicity information, modest differences were observed by race and ethnicity, with those in the diabetes-only group more likely than patients with comorbid serious mental illness to be classified as white (69 percent compared with 64 percent) and less likely to be classified as Hispanic (8 percent compared with 11 percent) or black (22 percent compared with 24 percent).Also, we were unable to assess the use of health services outside the VA system. We used visit-based criteria to restrict our analysis to those who had used the VA system, but this restriction could not ensure that patients did not also use services outside the VA. Consequently, if outside use was more likely among those with diabetes only, then our results may be somewhat overstated. Nonetheless, even with more outside use among those with diabetes only, we feel these results are robust in describing the potential benefits of medical and mental health services integration. Finally, we examined care processes and intermediate outcomes but not costs in this study. Achieving comparable outcomes for vulnerable diabetes patients with serious mental illness may require more health care resources and greater funding.ConclusionsThis study suggests that VA patients with diabetes and serious mental illness are receiving diabetes care of comparable clinical quality and have similar intermediate outcomes as VA patients with diabetes only. This finding could be due, in part, to individuals' with both conditions using considerably more health care services, meaning that there is more opportunity for delivering recommended services. Likewise, the integration of medical and mental health services, which is an inherent part of the VA health care delivery system, may also play an important role in ensuring that patients with diabetes and comorbid serious mental illness receive recommended diabetes care. Further research in this area is warranted, however, including longitudinal analyses to examine patterns of care over time. Adherence to diabetes and antipsychotic medications should also be compared to determine how much adherence might contribute to better diabetes outcomes as well as the interplay between diabetes treatment and mental health treatment. A more thorough understanding of these issues is paramount to help inform the design of health care systems, both inside and outside the VA, that best meet the needs of patients with multiple complex chronic conditions.AcknowledgmentsThis study was supported by grant DIB-98-001 from the Department of Veterans Affairs (VA), Health Services Research and Development Service; by the VA National Serious Mental Illness Treatment Research and Evaluation Center; and by grant NIH-5060 DK-20572 to Michigan Diabetes Research and Training. Dr. McCarthy was funded by Career Development Award Merit Review Entry Program MRP-03-320 from the VA Health Services Research and Development program. The authors thank the VA Healthcare Analysis and Information Group for their assistance with data collection. The views expressed in this article are the opinions of the authors and not necessarily the supporting agencies.Dr. Krein is affiliated with the Center for Practice Management and Outcomes Research, Department of Veterans Affairs Ann Arbor Healthcare System, and with the Department of General Medicine, University of Michigan. Dr. Bingham is with the Transportation Research Institute, University of Michigan. Dr. McCarthy and Dr. Valenstein are with National Serious Mental Illness Treatment Research and Department of Veterans Affairs, Ann Arbor, and with the Department of Psychiatry, University of Michigan. Ms. Mitchinson is with the Center for Practice Management and Outcomes Research, Department of Veterans Affairs Ann Arbor Healthcare System. Mr. Payes is with the Department of Psychiatry, Case Western Reserve University, Cleveland, Ohio. Send correspondence to Dr. Krein at HSR&D Center of Excellence, HSR&D/SMITREC, Department of Veterans Affairs, P.O. Box 130170, Ann Arbor, Michigan 48113-0170 (e-mail, [email protected]).