- Email: [email protected]
Medical-Surgical Hospitalization Among Veterans With Psychiatric and Substance Use Disorders
Psychosomatics 2019:-:-–-
Published by Elsevier Inc. on behalf of Academy of Consultation-Liaison Psychiatry.
Original Research Article Medical-Surgical Hospitalization Among Veterans With Psychiatric and Substance Use Disorders David T. Moore, M.D., Ph.D., Robert A. Rosenheck, M.D.
Background: Mental illness is associated with an increased risk for medical hospitalizations. Objective: This study investigates the degree to which nonpsychiatric factors account for these hospitalizations. Methods: Using National Veterans Health Administration (VHA) fiscal year 2012 data for 2 million veterans under the age of 60 years, hospitalization risks were compared for veterans with and without mental illnesses. Bivariate analyses identified factors associated with mental illnesses. Multiple logistic regression was used to calculate adjusted psychiatric risk for medical hospitalization, controlling for these factors. Results: Veterans carrying mental health diagnoses were at increased risk for hospitalizations (odds ratio [OR] = 2.52, 2.48–2.55). Among individual diagnoses, alcohol use disorder (AUD) (OR = 3.84, 3.78–3.91) and drug use disorders
(OR = 4.58, 4.50–4.66) were associated with the highest risk. After adjusting for nonpsychiatric medical, addiction-related, and care utilization factors and the use of outpatient medical services, veterans with mental illnesses were at increased risk for medical hospitalization (OR = 1.43, 1.41–1.45). After further adjustment for AUD and drug use, hospitalization risk decreased further (OR = 1.23, 1.21–1.26) while the association of AUD and hospitalizations remained high (OR = 1.77, 1.73–1.81). Conclusions: Medical comorbidities and service use accounted for most, but not all, of the increased risk of medical hospitalizations associated with mental illness. Even after accounting for poor health, AUD remained strongly associated with medical hospitalization. (Psychosomatics 2019; -:-–-)
Key words: alcohol addiction, alcohol/drug abuse, internal medicine, health care services, comorbidity, substance use
INTRODUCTION Medical-surgical hospitalizations account for roughly one-third of all health care expenditures.1 There is also a concern that the growing scarcity of inpatient psychiatric beds may have resulted in increased use of general emergency room services among psychiatric patients,2 potentially resulting in unnecessary hospitalizations or “boarding” on general medical units for what are ultimately psychiatric problems. Adults diagnosed with psychiatric or substance use disorders (SUDs) have been reported to be far more likely to Psychosomatics
-:-, -
2019
have medical hospitalizations,3–7 have longer lengths of stay once admitted,8–10 and have greater readmission rates after discharge.7,11–13 Adults with mental illness Received June 3, 2018; revised April 21, 2019; accepted April 22, 2019. From the Department of Psychiatry (D.T.M., R.A.R.), Yale University, New Haven, CT; VA Connecticut Healthcare System (D.T.M.), West Haven, CT; Veterans Affairs New England Mental Illness Research, Education, and Clinical Center (R.A.R.), West Haven, CT; Yale University School of Public Health (R.A.R.), New Haven, CT. Send correspondence and reprint requests to David T. Moore, MD, PhD, 300 George Street, Suite 901, New Haven, CT 06511; e-mail: [email protected] Published by Elsevier Inc. on behalf of Academy of Consultation-Liaison Psychiatry.
www.psychosomaticsjournal.org
1
Medical Hospitalizations and Mental Illness are also more likely to develop a broad range of acute and chronic medical conditions,14,15 most dramatically reflected in their greater risk of dying prematurely from these ailments.16 In the face of increased medical morbidity, it is unclear whether high hospitalization rates in medical-surgical inpatient programs are simply due to worse physical health or whether they may reflect independent effects of mental illnesses and, perhaps, inadequate access to appropriate mental health services. To better understand the increased risk of medicalsurgical hospitalizations among psychiatric patients, it is necessary to evaluate inpatient service use in the context of sociodemographic factors, medical comorbidities, and the utilization of ambulatory and emergency medical services. The correlates of medicalsurgical hospital care would be best examined using administrative data from integrated health systems that provide inpatient, outpatient, and emergency services under one roof for medical, psychiatric, and substancerelated illnesses. The Veterans Health Administration (VHA) is the largest integrated health system in the United States, providing comprehensive services to and maintaining administrative data for nearly 6 million veterans annually. Two studies of VHA service delivery have demonstrated significantly increased inpatient medicalsurgical service utilization among veterans with mental illnesses. The first reported that veterans returning from Iraq and Afghanistan were much more likely to be hospitalized on medical services if they had a psychiatric or addiction diagnosis.17 A second study limited to veterans who had visited primary care clinics in Southern California at least twice in one year found that diagnosis of mental illness was associated with greater risk of avoidable hospitalizations for ambulatory care–sensitive conditions even after accounting for medical comorbidity.6 Because ambulatory care– sensitive condition hospitalizations represent only a fraction of hospitalizations in the VHA18,19 and veterans recently returning from combat represent less than 10% of VHA patients, a broader analysis of inpatient medical services use among patients with mental health problems using VHA administrative data is needed. The present study used national VHA data from approximately two million veterans younger than 60 years, a population selected because of its limited access to Medicare-funded services and non-VHA nursing home services, neither of which would be documented 2
www.psychosomaticsjournal.org
in VHA records. To evaluate an association between mental illness and the risk for medical-surgical hospitalizations, we first evaluated unadjusted relative odds of medical-surgical hospitalizations among veterans with any mental health diagnosis and with several specific mental health diagnoses. We then used multiple logistic regression analysis to determine the extent to which the likelihood of such hospitalizations is accounted for by sociodemographic characteristics, medical comorbidities, the use of ambulatory and emergency medical services, and addiction-related diagnoses. MATERIALS AND METHODS Sample VHA administrative data from fiscal year (FY) 2012 (October 1, 2011, to September 30, 2012) were used to identify all 2,016,392 service users younger than 60 years. Veterans aged 60 years and older were excluded to avoid confounding our analyses of psychiatric diagnoses with diagnoses of dementias and of inpatient service utilization with the use of nursing home services and non-VA hospitalizations funded by Medicare and Medicaid, important components of service utilization among older veterans.20 A waver of written informed consent was obtained as the study used administrative data without individual patient identifiers. This study was approved by the Institutional Review Board of the VA Connecticut Healthcare System. Measures Sociodemographic characteristics included age, gender, race (white vs non-white), service and income in Operation Enduring Freedom or Operation Iraqi Freedom, receipt of a VA pension, service-connected disability, or homelessness during FY 2012. Homelessness status was determined by the International Classification of Diseases-9 diagnosis code V60 for use of specialized VA homeless services. Medical, psychiatric, and addiction-related diagnoses were identified by International Classification of Diseases-9 codes. Mental illness was defined as having either a psychiatric or a SUD diagnosis during FY 2012 (International Classification of Diseases-9 codes 290.00–312.99, 310.xx, 331.xx, excluding 305.1). Individual psychiatric diagnoses included schizophrenia or schizoaffective disorder (295.xx), bipolar disorder Psychosomatics
-:-, -
2019
Moore and Rosenheck (296.0x, 296.1x, 296.40–296.89), major depressive disorder (MDD) (296.2–296.39), other depressive disorders or dysthymic disorders (300.4x, 296.9x, 311.xx, 301.10–301.19), posttraumatic stress disorder (PTSD) (309.81), and anxiety disorders (300.xx, excluding 300.4). SUDs included alcohol use disorder (AUD) (300.xx or 305.00) and drug use disorder (292.01–292.99, 304.xx, 305.20–305.99), which included specific diagnoses of opioid use disorder, cocaine use disorder, sedative-hypnotic use disorder (barbiturates or benzodiazepines), cannabis use disorder, and hallucinogen use disorder. Individual medical diagnoses included diagnoses that compose the Charlson Comorbidity Index.21 These diagnoses included myocardial infarction, congestive heart failure, chronic obstructive pulmonary disease, peripheral vascular disease, diabetes, renal disease, seizures, stroke, peptic ulcer, liver disease, and nausea and vomiting, HIV, cancer, musculoskeletal pain, headache, and pneumonia. The Charlson Comorbidity Index itself was used to assess overall medical comorbidity.21 We also included eight pain diagnoses, based on International Classification of Diseases-9 diagnostic codes: any pain diagnosis, herpetic pain (053.12 or 729.2), fibromyalgia pain (729.1), musculoskeletal pain (338.xx, 719.4, 780.96), skeletal-spasm pain (728.85, 781.0), pain from diabetes (250.6, 357.2, 337.1), migraine and headache (346.x, 784.0), and pain and neuropathy (250.6, 357.2, 337.1, 338.x, 719.4, 780.96, 729.1, 728.85, 781.0, 053.12, 729.2, 352.1, 350.1). Inpatient health service utilization was evaluated by using bed section discharge codes for inpatient medical-surgical discharges. The numbers of outpatient medical, primary care, and emergency department services were assessed using clinic stop codes.
Next, to identify potential nonpsychiatric correlates of medical-surgical hospitalization among veterans with psychiatric diagnoses, a set of bivariate analyses were used to compare veterans with psychiatric and addiction diagnoses to those without such diagnoses on multiple nonpsychiatric indicators that could be associated with inpatient medical-surgical hospitalization including sociodemographic characteristics, medical diagnoses (as described previously), outpatient medical and emergency service utilization, and opioid prescription fills. Because extremely weak relationships can emerge as statistically significant in the analysis of large samples with high statistical power, effect sizes were used for comparisons rather than P values. Substantial effect sizes were defined by Cohen’s d (difference between means divided by the pooled standard deviation) .0.2 or ,20.2 for continuous variables or a relative risk .1.5 or ,0.67 for dichotomous variables.22 Multiple logistic regression analysis of inpatient medical-surgical hospitalization was used to adjust for sociodemographic characteristics, health factors, and outpatient and emergency utilization patterns that differentiated veterans with and without mental health diagnoses in the bivariate analyses described previously. In each case, the dependent variable was a dichotomous measure of whether the veteran had been hospitalized during FY 2012. Multiple logistic regression models were constructed to compare veterans with mental illnesses to those without mental health diagnoses, as well as those with eight specific mental health diagnoses to those without those diagnoses.
Analysis
Between October 1, 2011, and September 30, 2012, there were 2,016,392 veterans younger than 60 years who received care in the VHA, of which 47% (N = 952,282) were diagnosed with a mental illness (psychiatric or SUD). Altogether, 4.97% (N = 100,191) of veterans under the age of 60 years had a medicalsurgical hospitalization. Veterans with mental illnesses were more than twice as likely to be hospitalized than veterans without mental illnesses, odds ratio = 2.52 (2.48–2.55). Although more likely to be hospitalized, those with mental illnesses who were admitted did not spend any more days hospitalized than other veterans during the same FY (7.84 d, standard
Analysis proceeded in several steps. First the proportion of veterans with medical-surgical hospitalizations among those with mental health diagnoses were compared with the proportion among those without such diagnoses with unadjusted odds ratios (OR). Then, veterans with eight specific mental health diagnoses (schizophrenia, bipolar disorder, MDD, other depressions, generalized anxiety disorder, PTSD, AUD, drug use disorder) were compared with those without any of these diagnoses on the relative odds of medical-surgical hospitalization. Psychosomatics
-:-, -
2019
RESULTS
www.psychosomaticsjournal.org
3
Medical Hospitalizations and Mental Illness
TABLE 1.
Characteristics of Veterans Health Administration Patients Younger Than 60 Years Diagnosed With Mental Illnesses During FY 2012
Demographics Male Homeless during the year VA pension Service connected 50% or more Service connected less than 50% Total income White race Age Health factors Charlson Index Seizures Myocardial infarction CHF Peripheral vascular disease Cerebrovascular accident Chronic obstructive airway disease Peptic ulcer disease Hepatic disease Renal disease Cancer Moderate/severe liver Pneumonia Nausea/vomiting Insomnia Congestive heart failure Connective tissue disease Diabetes mellitus Complications of diabetes Paraplegia HIV Metastatic cancer Any pain diagnosis Tobacco use Opiate prescriptions Service use Primary care visits Outpatient specialty medicine visits ER visits
No mental illness, N/mean (%/SD)
Mental illness, N/mean (%/SD)
(N = 1,064,110) 922,068 (86.7) 31,285 (2.9) 14,425 (1.4) 174,153 (16.4) 292,485 (27.5) 23,502 (42,115) (N = 871,509) 592,942 (68.0) (N = 1,064,075) 45.74 (10.90) (N = 1,064,110) 1.22 (1.49) 4474 (0.4) 5624 (0.5) 316,377 (29.7) 13,655 (1.3) 15,135 (1.4) 76,866 (7.2) 3808 (0.4) 22,032 (2.1) 19,149 (1.8) 27,760 (2.6) 1610 (0.2) 6552 (0.6) 8704 (0.8) 23,012 (2.2) 316,377 (29.7) 7488 (0.7) 136,737 (12.8) 22,807 (2.1) 5765 (0.5) 7728 (0.7) 2796 (0.3) 267,937 (25.2) 141,240 (13.3) 1.03 (3.59)
(N = 962,282) 813,098 (85.4) 116,897 (12.3) 30,800 (3.2) 283,653 (29.8) 200,013 (21) 28,023 (19,574) (N = 871,509) 598,186 (70.0) (N = 952,280) 45.42 (10.98) (N = 962,282) 1.51 (1.7) 10,563 (1.1) 6429 (0.7) 40,5121 (42.5) 16,862 (1.8) 23,156 (2.4) 110,436 (11.6) 6495 (0.7) 44,036 (4.6) 18,759 (2) 30,266 (3.2) 4450 (0.5) 11,509 (1.2) 20,434 (2.1) 80,077 (8.4) 405,121 (42.5) 7424 (0.8) 137,068 (14.4) 25,636 (2.7) 7389 (0.8) 9194 (1.0) 3072 (0.3) 359,870 (37.8) 270,786 (28.4) 2.44 (5.73)
1.91 (2.13) 2.81 (5.52) 0.32 (0.89)
3.19 (3.33) 4.99 (7.99) 0.76 (1.96)
Cohen’s d
RR 0.99 4.18 2.39 1.82 0.76
20.16 1.03 20.04 0.25 2.64 1.28 1.43 1.38 1.71 1.61 1.91 2.23 1.09 1.22 3.09 1.96 2.62 3.89 1.43 1.11 1.12 1.26 1.43 1.33 1.23 1.50 2.14 0.42 0.65 0.45 0.41
Bold values indicate a significant effect size (Cohen’s d . 0.20, risk ratio (RR) . 1.5, or RR , 0.67). FY = fiscal year; SD = standard deviation; CHF = congestive heart failure; ER = emergency room; VA = Veterans Administration.
deviation = 16.10, vs 7.96 d, standard deviation = 19.99; Cohen’s d = 0.01). To identify factors that might contribute to the apparent increased risk for hospitalizations, the characteristics of veterans with mental illnesses were compared with those of veterans without mental illnesses (Table 1). Veterans diagnosed with mental health issues were more likely to have been homeless, to have a service-connected disability with a disability
4
www.psychosomaticsjournal.org
rating greater than 50%, and to receive a VA pension. Several medical diagnoses were more common among veterans with mental illnesses than others including seizures, stroke, peptic ulcer disease, pneumonia, liver disease, and nausea and vomiting. Likewise, the Charlson Comorbidity Index, an indicator of overall health and expected mortality, was substantially higher among veterans with mental health diagnoses with a Cohen’s d of 0.25. High-risk behaviors, such as
Psychosomatics
-:-, -
2019
Moore and Rosenheck
TABLE 2.
Risk of Admission to Medical-Surgical Units for Psychiatric and Addiction Diagnoses During FY 2012
Diagnosis
Number hospitalized (%)
OR*
No MH diagnosis (N = 1,064,110) Any MH diagnosis (N = 952,282) Schizophrenia (N = 51,752) Bipolar disorder (N = 77,839) Major depression (N = 179,474) Other depression (N = 429,799) Anxiety disorder (N = 261,212) PTSD (N = 303,793) Alcohol use (N = 222,854) Drug use (N = 173,529)
31,750 68,441 4348 7151 15,353 35,735 19,702 16,835 23,550 21,425
2.52 2.98 3.29 3.04 2.95 2.65 1.91 3.84 4.58
(2.98) (7.19) (8.40) (9.19) (8.55) (8.31) (7.54) (5.54) (10.57) (12.35)
OR† (2.48–2.55) (2.89–3.08) (3.20–3.38) (2.98–3.10) (2.90–2.99) (2.60–2.70) (1.87–1.94) (3.78–3.91) (4.50–4.66)
1.43 0.87 1.02 1.00 1.18 1.04 0.98 1.95 1.38
(1.41–1.45) (0.83–0.90) (0.99–1.06) (0.98–1.02) (1.16–1.20) (1.01–1.06) (0.96–1.00) (1.91–2.00) (1.36–1.41)
FY = fiscal year; MH = mental health; OR = odds ratio; PTSD = posttraumatic stress disorder; SD = standard deviation. * Unadjusted OR comparing the risk of hospitalization for veterans with a mental illness (psychiatric or substance use) to that of those without mental illnesses. † Adjusted OR controlling for sociodemographic, health, and service characteristics identified as significant in bivariate analyses.
tobacco use and receipt of opioid prescriptions, were also more frequently observed among those with mental health diagnoses. Veterans with mental health diagnoses used considerably more outpatient primary care and specialty medical services and had more emergency department visits than veterans without such diagnoses. Veterans diagnosed with individual diagnoses of schizophrenia, bipolar disorder, MDD, other depressive disorders, generalized anxiety disorder, PTSD, AUD, and illicit drug use were more likely to be hospitalized than veterans without mental illness (Table 2). After adjustment for sociodemographic factors, medical comorbidities, and service use differences identified in bivariate analyses (Table 2), the OR for medicalsurgical hospitalization in association to mental illness declined from 2.52 to 1.43. Among individual diagnoses, adjusted ORs for alcohol use and drug use remained most strongly associated with medicalsurgical hospitalizations after covarying sociodemographic, medical, and service use (Table 2). In view of the robustness of the relationship of hospitalization with substance use, one further logistic regression analysis was conducted, adding adjustment for alcohol and drug use, which resulted in further reduction of the adjusted OR for any mental health diagnoses from 1.43 (95% confidence interval: 1.41–1.45) to 1.24 (95% confidence interval: 1.21–1.26). In this final model, the OR for drug use dropped further (OR = 1.04, 95% confidence interval: 1.02–1.07), while the relationship between AUD and hospitalizations Psychosomatics
-:-, -
2019
remained robust (OR = 1.77, 95% confidence interval: 1.73–1.81).
DISCUSSION In this study of 2 million veterans younger than 60 years who were treated in the VHA during FY 2012, we found evidence that worse physical health likely accounts for most—but not all—of the over 2-fold increased odds of medical-surgical hospitalization among veterans with mental health diagnoses compared with veterans without such diagnoses. SUDs involving both illicit drugs and alcohol were the mental health diagnoses most strongly associated with medical hospitalization and accounted for one-third of all hospitalized patients in this sample. Among all mental health diagnoses, AUD was unique in that chronic medical comorbidities did not account for the high hospitalization rates. Together, these results suggest two major paths to reduce costly medical hospitalizations. First, we must better understand the details of how mentally ill adults become prematurely physically ill to mitigate this risk. Second, we must consider focused and intense efforts to assist populations at particularly high risk (e.g., those using alcohol). The poor physical health of VHA patients diagnosed with mental illness cannot be understated. We confirmed that a number of medical conditions including headaches and chronic pain, chronic obstructive pulmonary disease, liver disease, and stroke www.psychosomaticsjournal.org
5
Medical Hospitalizations and Mental Illness were more prevalent among veterans with psychiatric and SUDs than others. While the current study did not have access to mortality data, others have found that veterans with mental illnesses die prematurely,23 and the Charlson Comorbidity Index, a predictor of future mortality, was substantially higher in our sample of veterans diagnosed with mental illnesses than that of other VHA patients. Factors thought to contribute to the increased risk for medical problems and higher mortality associated with mental illnesses include decreased adherence to prescribed treatments,24 poorer access to outpatient medical services,25 and poorer overall quality of medical care.5,26 Data used in this study did not include evaluation of quality of medical care, although there was no evidence of decreased access to outpatient medical services. Rather, veterans with mental health diagnoses were more likely to use services of both primary care and specialty medical clinics. Mental illness is also strongly associated with a range of psychosocial and health factors that are linked to poor health outcomes including tobacco use homelessness,27 tobacco use, and receipt of potentially dangerous medications—especially opioids.28 In our sample, one in ten veterans with mental illnesses had been homeless during the prior year, more than a quarter used tobacco products, and nearly a third had received opioid prescriptions. In the face of the evolving US crisis of opioid use and dependence, there has been significant focus on overprescription of opioids to veterans for diverse pain syndromes.29 Beyond the risk of overdose, opioid medications are increasingly recognized to adversely affect physical health28 and to predispose toward medical hospitalizations.30 The VHA has made significant efforts to reduce overprescribing of opioids through its national Opioid Safety Initiative, and there have been some promising results. In a recent analysis, VHA opioid prescriptions were shown to have decreased significantly since 2012,31 the same year the data examined here were recorded. Further research is needed to determine whether the harm of excess opioid prescriptions affects veterans with mental illnesses and whether hospitalization rates might be reduced as opioid prescription use is reduced. Among individual diagnoses, alcohol and drug use were most strongly associated with inpatient admissions. This likely reflects the particularly high rates of medical illnesses among adults with addictions.14 However, alcohol use appears to impart additional 6
www.psychosomaticsjournal.org
risk even after accounting for the shared variance with drug use, other mental health disorders, and chronic medical problems. Nearly 10% of veterans diagnosed with AUD had medical-surgical hospitalizations, accounting for 23% of all veterans hospitalized on medical-surgical units. This finding is consistent with prior estimates that roughly 20–40% of medically hospitalized patients have AUD.32,33 It is possible that our data set, which focuses on chronic medical conditions, may not capture acute medical problems associated with alcohol use such as withdrawal, intoxication and poisoning, dehydration, arrhythmias, and falls.34 Further research is needed to better understand the balance between hospitalizations associated with chronic medical conditions and hospitalizations when alcohol consumption or cessation was the direct precipitant. Because AUD was the single strongest predictor of hospitalizations, it is necessary to consider treatment options as they relate to medical illnesses. The VHA offers extensive substance use treatment through mental health service lines that can reduce alcohol use but may not support coordination with medical teams. There have also been preventive efforts based on early identification of AUD in the primary care setting using the Alcohol Use Disorder Identification Test-C as a screening tool.35 VA has developed Primary Care Mental Health Integration programs that provide mental health services directly within the context of primary care service delivery.36 Our present study did not evaluate the use of Primary Care Mental Health Integration, but other studies have found that such services can improve overall health outcomes.37 Integration of primary care and addiction treatments has been found to improve abstinence in severe AUDs and to reduce medical complications.38 VHA has piloted an alcohol care management initiative, which has been shown to improve treatment engagement, increase naltrexone adherence, and decrease days of heavy drinking.39 Discussion is also warranted related to why comorbid sociodemographic characteristics and medical illnesses did not account for all the increased risks of medical-surgical hospitalization among VHA patients with psychiatric diagnoses. On the one hand, illnesses not documented in this analysis may have explained some of the unexplained increased risks. On the other hand, it is possible that some degree of unmet need for psychiatric hospital care was reflected in the increased Psychosomatics
-:-, -
2019
Moore and Rosenheck risk of medical-surgical hospitalization. There has been substantial reduction in psychiatric bed availability in VHA in recent years, even as there has been an increase in the number of patients with psychiatric diagnoses.40 A recent study reported that between 1999 and 2007, the number of VA psychiatric bed declined 28% from 6570 to 4745 while the total number of patients increased by 50% from 3.4 million to 5.1 million.40 This decrease in psychiatric beds per patient from 1.9 of 1000 veterans to 0.9 of 1000 veterans may have reflected justifiable changes in practice in an overbedded system, but it may also have left some need for inpatient care unmet, resulting in spillovers to medicalsurgical beds. Unfortunately, we cannot precisely determine the reasons for the unaccounted increase in the risk for veterans with psychiatric diagnoses. A number of methodological limitations of this study require comment. First, administrative data do not always identify diagnoses accurately and can identify functional impairments only through proxy measures of utilization of homeless or rehabilitative services and through disability payments. In addition, data were not available on whether specific mental health or medical diagnoses were identified in outpatient, inpatient, or emergency department service settings—data on which would be valuable in understanding where these conditions had been treated by specialized providers and whether they were temporarily associated with medical-surgical hospitalizations. In addition, subjective measures of physical health or mental health were lacking in administrative files. References 1. Health, United States, 2016: with chartbook on long-term trends in health. Hyattsville, MD: Centers for Disease Control; 2017 2. Pitts SR, Vaughns FL, Gautreau MA, Cogdell MW, Meisel Z: A cross-sectional study of emergency department boarding practices in the United States. Acad Emerg Med 2014; 21:497–503 3. Himelhoch S, Weller WE, Wu AW, Anderson GF, Cooper LA: Chronic medical illness, depression, and use of acute medical services among medicare beneficiaries. Med Care 2004; 42:512–521 4. Li Y, Glance LG, Cai X, Mukamel DB: Mental illness and hospitalization for ambulatory care sensitive medical conditions. Med Care 2008; 46:1249–1256 5. Mai Q, Holman CD, Sanfilippo FM, Emery JD: The impact of mental illness on potentially preventable hospitalisations: a population-based cohort study. BMC Psychiatry 2011; 11:163
Psychosomatics
-:-, -
2019
Finally, this study was limited to veterans treated by the VHA. While national in scope, the cohort studied may not be representative of veterans not using VHA services or nonveteran populations. In particular, patients in this study were older than the US average, and relatively few women were included. CONCLUSIONS This study has shown that VHA patients diagnosed with mental illnesses are at higher risk for developing serious medical conditions than other VHA patients and that these indicators of poor physical health account for much, although not all, of the high risk of medical-surgical hospitalization associated with mental illness, especially in the case of AUD which poses a substantial independent risk for medical hospitalization. Our results found no evidence that poor health and high utilization of hospital services were associated with limited access to outpatient medical services. Rather, all services were used at higher rates among those diagnosed with mental illnesses. Further research is needed to identify more effective treatment for these high-risk patients. Conflicts of Interest: The authors declare that they have no conflict of interest. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
6. Yoon J, Yano EM, Altman L, et al: Reducing costs of acute care for ambulatory care-sensitive medical conditions: the central roles of comorbid mental illness. Med Care 2012; 50:705–713 7. Davydow DS, Ribe AR, Pedersen HS, et al: Serious mental illness and risk for hospitalizations and rehospitalizations for ambulatory care-sensitive conditions in Denmark: a nationwide population-based cohort study. Med Care 2016; 54:90–97 8. Levenson JL, Hamer RM, Rossiter LF: Relation of psychopathology in general medical inpatients to use and cost of services. Am J Psychiatry 1990; 147:1498–1503 9. Hochlehnert A, Niehoff D, Wild B, Junger J, Herzog W, Lowe B: Psychiatric comorbidity in cardiovascular inpatients: costs, net gain, and length of hospitalization. J Psychosom Res 2011; 70:135–139 10. Thorpe K, Jain S, Joski P: Prevalence and spending associated with patients who have a behavioral health disorder and other conditions. Health Aff 2017; 36:124–132
www.psychosomaticsjournal.org
7
Medical Hospitalizations and Mental Illness 11. Albrecht JS, Hirshon JM, Goldberg R, et al: Serious mental illness and acute hospital readmission in diabetic patients. Am J Med Qual 2012; 27:503–508 12. Daratha KB, Barbosa-Leiker C, M HB, et al: Co-occurring mood disorders among hospitalized patients and risk for subsequent medical hospitalization. Gen Hosp Psychiatry 2012; 34:500–505 13. Chwastiak LA, Davydow DS, McKibbin CL, et al: The effect of serious mental illness on the risk of rehospitalization among patients with diabetes. Psychosomatics 2014; 55: 134–143 14. Dickey B, Normand SL, Weiss RD, Drake RE, Azeni H: Medical morbidity, mental illness, and substance use disorders. Psychiatr Serv 2002; 53:861–867 15. Scott KM, Lim C, Al-Hamzawi A, et al: Association of mental disorders with subsequent chronic physical conditions: world mental health surveys from 17 Countries. JAMA Psychiatry 2016; 73:150–158 16. Miller BJ, Paschall CB 3rd, Svendsen DP: Mortality and medical comorbidity among patients with serious mental illness. Psychiatr Serv 2006; 57:1482–1487 17. Cohen BE, Gima K, Bertenthal D, Kim S, Marmar CR, Seal KH: Mental health diagnoses and utilization of VA nonmental health medical services among returning Iraq and Afghanistan veterans. J Gen Intern Med 2010; 25:18–24 18. Nelson KM, Helfrich C, Sun H, et al: Implementation of the patient-centered medical home in the Veterans Health Administration: associations with patient satisfaction, quality of care, staff burnout, and hospital and emergency department use. JAMA Intern Med 2014; 174:1350–1358 19. Department of Veterans Affairs: VHA facility quality and safety report fiscal year 2012 data. Washington, D.C: Department of Veterans Affairs; 2013 20. Miller EA, Rosenheck RA: Risk of nursing home admission in association with mental illness nationally in the Department of Veterans Affairs. Med Care 2006; 44:343–351 21. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM: How to measure comorbidity. A critical review of available methods. J Clin Epidemiol 2003; 56:221–229 22. Ferguson CJ: An effect size primer: a guide for clinicians and researchers. Prof Psychol Res Pract 2009; 40:532–538 23. Chwastiak LA, Rosenheck RA, Desai R, Kazis LE: Association of psychiatric illness and all-cause mortality in the national department of veterans affairs health care system. Psychosom Med 2010; 72:817–822 24. DiMatteo MR, Lepper HS, Croghan TW: Depression is a risk factor for noncompliance with medical treatment: metaanalysis of the effects of anxiety and depression on patient adherence. Arch Intern Med 2000; 160:2101–2107 25. Druss BG, Rosenheck RA: Mental disorders and access to medical care in the United States. Am J Psychiatry 1998; 155:1775–1777
8
www.psychosomaticsjournal.org
26. Druss BG, Rosenheck RA, Desai MM, Perlin JB: Quality of preventive medical care for patients with mental disorders. Med Care 2002; 40:129–136 27. O’Connell JJ: Premature mortality in homeless populations: A review of the literature. Nashville, TN: National Health Care for the Homeless Council; 2005 28. Paulozzi LJ, Zhang K, Jones CM, Mack KA: Risk of adverse health outcomes with increasing duration and regularity of opioid therapy. J Am Board Fam Med 2014; 27:329–338 29. Petrakis IL, Sofuoglu M, Rosenheck R: VA patients with high numbers of opioid prescriptions: a national study of sociodemographic and diagnostic characteristics, health service, and psychotropic medication use. Addict Disord Their Treat 2015; 14:167–175 30. Liang Y, Turner BJ: National cohort study of opioid analgesic dose and risk of future hospitalization. J Hosp Med 2015; 10:425–431 31. Hadlandsmyth K, Mosher H, Vander Weg MW, Lund BC: Decline in prescription opioids attributable to decreases in long-term use: a retrospective study in the veterans health administration 2010-2016. J Gen Intern Med 2018; 33:818–824 32. Coder B, Freyer-Adam J, Bischof G, et al: Alcohol problem drinking among general hospital inpatients in northeastern Germany. Gen Hosp Psychiatry 2008; 30:147–154 33. de Wit M, Jones DG, Sessler CN, Zilberberg MD, Weaver MF: Alcohol-use disorders in the critically ill patient. Chest 2010; 138:994–1003 34. Mannelli P, Pae CU: Medical comorbidity and alcohol dependence. Curr Psychiatry Rep 2007; 9:217–224 35. Bradley KA, Williams EC, Achtmeyer CE, Volpp B, Collins BJ, Kivlahan DR: Implementation of evidence-based alcohol screening in the veterans health administration. Am J Manag Care 2006; 12:597–606 36. Johnson-Lawrence V, Zivin K, Szymanski BR, Pfeiffer PN, McCarthy JF: VA primary care-mental health integration: patient characteristics and receipt of mental health services, 2008-2010. Psychiatr Serv 2012; 63:1137–1141 37. Trivedi RB, Post EP, Sun H, et al: Prevalence, comorbidity, and prognosis of mental health among US veterans. Am J Public Health 2015; 105:2564–2569 38. Weisner C, Mertens J, Parthasarathy S, Moore C, Lu Y: Integrating primary medical care with addiction treatment: a randomized controlled trial. JAMA 2001; 286:1715–1723 39. Oslin DW, Lynch KG, Maisto SA, et al: A randomized clinical trial of alcohol care management delivered in Department of Veterans Affairs primary care clinics versus specialty addiction treatment. J Gen Intern Med 2014; 29:162–168 40. Bowersox NW, Szymanski BJ, McCarthy JF: Associations between psychiatric inpatient bed supply and the prevalence of serious mental illness in Veterans Affairs nursing homes. Am J Public Health 2013; 103:1325–1331
Psychosomatics
-:-, -
2019
Published by Elsevier Inc. on behalf of Academy of Consultation-Liaison Psychiatry.
Original Research Article Medical-Surgical Hospitalization Among Veterans With Psychiatric and Substance Use Disorders David T. Moore, M.D., Ph.D., Robert A. Rosenheck, M.D.
Background: Mental illness is associated with an increased risk for medical hospitalizations. Objective: This study investigates the degree to which nonpsychiatric factors account for these hospitalizations. Methods: Using National Veterans Health Administration (VHA) fiscal year 2012 data for 2 million veterans under the age of 60 years, hospitalization risks were compared for veterans with and without mental illnesses. Bivariate analyses identified factors associated with mental illnesses. Multiple logistic regression was used to calculate adjusted psychiatric risk for medical hospitalization, controlling for these factors. Results: Veterans carrying mental health diagnoses were at increased risk for hospitalizations (odds ratio [OR] = 2.52, 2.48–2.55). Among individual diagnoses, alcohol use disorder (AUD) (OR = 3.84, 3.78–3.91) and drug use disorders
(OR = 4.58, 4.50–4.66) were associated with the highest risk. After adjusting for nonpsychiatric medical, addiction-related, and care utilization factors and the use of outpatient medical services, veterans with mental illnesses were at increased risk for medical hospitalization (OR = 1.43, 1.41–1.45). After further adjustment for AUD and drug use, hospitalization risk decreased further (OR = 1.23, 1.21–1.26) while the association of AUD and hospitalizations remained high (OR = 1.77, 1.73–1.81). Conclusions: Medical comorbidities and service use accounted for most, but not all, of the increased risk of medical hospitalizations associated with mental illness. Even after accounting for poor health, AUD remained strongly associated with medical hospitalization. (Psychosomatics 2019; -:-–-)
Key words: alcohol addiction, alcohol/drug abuse, internal medicine, health care services, comorbidity, substance use
INTRODUCTION Medical-surgical hospitalizations account for roughly one-third of all health care expenditures.1 There is also a concern that the growing scarcity of inpatient psychiatric beds may have resulted in increased use of general emergency room services among psychiatric patients,2 potentially resulting in unnecessary hospitalizations or “boarding” on general medical units for what are ultimately psychiatric problems. Adults diagnosed with psychiatric or substance use disorders (SUDs) have been reported to be far more likely to Psychosomatics
-:-, -
2019
have medical hospitalizations,3–7 have longer lengths of stay once admitted,8–10 and have greater readmission rates after discharge.7,11–13 Adults with mental illness Received June 3, 2018; revised April 21, 2019; accepted April 22, 2019. From the Department of Psychiatry (D.T.M., R.A.R.), Yale University, New Haven, CT; VA Connecticut Healthcare System (D.T.M.), West Haven, CT; Veterans Affairs New England Mental Illness Research, Education, and Clinical Center (R.A.R.), West Haven, CT; Yale University School of Public Health (R.A.R.), New Haven, CT. Send correspondence and reprint requests to David T. Moore, MD, PhD, 300 George Street, Suite 901, New Haven, CT 06511; e-mail: [email protected] Published by Elsevier Inc. on behalf of Academy of Consultation-Liaison Psychiatry.
www.psychosomaticsjournal.org
1
Medical Hospitalizations and Mental Illness are also more likely to develop a broad range of acute and chronic medical conditions,14,15 most dramatically reflected in their greater risk of dying prematurely from these ailments.16 In the face of increased medical morbidity, it is unclear whether high hospitalization rates in medical-surgical inpatient programs are simply due to worse physical health or whether they may reflect independent effects of mental illnesses and, perhaps, inadequate access to appropriate mental health services. To better understand the increased risk of medicalsurgical hospitalizations among psychiatric patients, it is necessary to evaluate inpatient service use in the context of sociodemographic factors, medical comorbidities, and the utilization of ambulatory and emergency medical services. The correlates of medicalsurgical hospital care would be best examined using administrative data from integrated health systems that provide inpatient, outpatient, and emergency services under one roof for medical, psychiatric, and substancerelated illnesses. The Veterans Health Administration (VHA) is the largest integrated health system in the United States, providing comprehensive services to and maintaining administrative data for nearly 6 million veterans annually. Two studies of VHA service delivery have demonstrated significantly increased inpatient medicalsurgical service utilization among veterans with mental illnesses. The first reported that veterans returning from Iraq and Afghanistan were much more likely to be hospitalized on medical services if they had a psychiatric or addiction diagnosis.17 A second study limited to veterans who had visited primary care clinics in Southern California at least twice in one year found that diagnosis of mental illness was associated with greater risk of avoidable hospitalizations for ambulatory care–sensitive conditions even after accounting for medical comorbidity.6 Because ambulatory care– sensitive condition hospitalizations represent only a fraction of hospitalizations in the VHA18,19 and veterans recently returning from combat represent less than 10% of VHA patients, a broader analysis of inpatient medical services use among patients with mental health problems using VHA administrative data is needed. The present study used national VHA data from approximately two million veterans younger than 60 years, a population selected because of its limited access to Medicare-funded services and non-VHA nursing home services, neither of which would be documented 2
www.psychosomaticsjournal.org
in VHA records. To evaluate an association between mental illness and the risk for medical-surgical hospitalizations, we first evaluated unadjusted relative odds of medical-surgical hospitalizations among veterans with any mental health diagnosis and with several specific mental health diagnoses. We then used multiple logistic regression analysis to determine the extent to which the likelihood of such hospitalizations is accounted for by sociodemographic characteristics, medical comorbidities, the use of ambulatory and emergency medical services, and addiction-related diagnoses. MATERIALS AND METHODS Sample VHA administrative data from fiscal year (FY) 2012 (October 1, 2011, to September 30, 2012) were used to identify all 2,016,392 service users younger than 60 years. Veterans aged 60 years and older were excluded to avoid confounding our analyses of psychiatric diagnoses with diagnoses of dementias and of inpatient service utilization with the use of nursing home services and non-VA hospitalizations funded by Medicare and Medicaid, important components of service utilization among older veterans.20 A waver of written informed consent was obtained as the study used administrative data without individual patient identifiers. This study was approved by the Institutional Review Board of the VA Connecticut Healthcare System. Measures Sociodemographic characteristics included age, gender, race (white vs non-white), service and income in Operation Enduring Freedom or Operation Iraqi Freedom, receipt of a VA pension, service-connected disability, or homelessness during FY 2012. Homelessness status was determined by the International Classification of Diseases-9 diagnosis code V60 for use of specialized VA homeless services. Medical, psychiatric, and addiction-related diagnoses were identified by International Classification of Diseases-9 codes. Mental illness was defined as having either a psychiatric or a SUD diagnosis during FY 2012 (International Classification of Diseases-9 codes 290.00–312.99, 310.xx, 331.xx, excluding 305.1). Individual psychiatric diagnoses included schizophrenia or schizoaffective disorder (295.xx), bipolar disorder Psychosomatics
-:-, -
2019
Moore and Rosenheck (296.0x, 296.1x, 296.40–296.89), major depressive disorder (MDD) (296.2–296.39), other depressive disorders or dysthymic disorders (300.4x, 296.9x, 311.xx, 301.10–301.19), posttraumatic stress disorder (PTSD) (309.81), and anxiety disorders (300.xx, excluding 300.4). SUDs included alcohol use disorder (AUD) (300.xx or 305.00) and drug use disorder (292.01–292.99, 304.xx, 305.20–305.99), which included specific diagnoses of opioid use disorder, cocaine use disorder, sedative-hypnotic use disorder (barbiturates or benzodiazepines), cannabis use disorder, and hallucinogen use disorder. Individual medical diagnoses included diagnoses that compose the Charlson Comorbidity Index.21 These diagnoses included myocardial infarction, congestive heart failure, chronic obstructive pulmonary disease, peripheral vascular disease, diabetes, renal disease, seizures, stroke, peptic ulcer, liver disease, and nausea and vomiting, HIV, cancer, musculoskeletal pain, headache, and pneumonia. The Charlson Comorbidity Index itself was used to assess overall medical comorbidity.21 We also included eight pain diagnoses, based on International Classification of Diseases-9 diagnostic codes: any pain diagnosis, herpetic pain (053.12 or 729.2), fibromyalgia pain (729.1), musculoskeletal pain (338.xx, 719.4, 780.96), skeletal-spasm pain (728.85, 781.0), pain from diabetes (250.6, 357.2, 337.1), migraine and headache (346.x, 784.0), and pain and neuropathy (250.6, 357.2, 337.1, 338.x, 719.4, 780.96, 729.1, 728.85, 781.0, 053.12, 729.2, 352.1, 350.1). Inpatient health service utilization was evaluated by using bed section discharge codes for inpatient medical-surgical discharges. The numbers of outpatient medical, primary care, and emergency department services were assessed using clinic stop codes.
Next, to identify potential nonpsychiatric correlates of medical-surgical hospitalization among veterans with psychiatric diagnoses, a set of bivariate analyses were used to compare veterans with psychiatric and addiction diagnoses to those without such diagnoses on multiple nonpsychiatric indicators that could be associated with inpatient medical-surgical hospitalization including sociodemographic characteristics, medical diagnoses (as described previously), outpatient medical and emergency service utilization, and opioid prescription fills. Because extremely weak relationships can emerge as statistically significant in the analysis of large samples with high statistical power, effect sizes were used for comparisons rather than P values. Substantial effect sizes were defined by Cohen’s d (difference between means divided by the pooled standard deviation) .0.2 or ,20.2 for continuous variables or a relative risk .1.5 or ,0.67 for dichotomous variables.22 Multiple logistic regression analysis of inpatient medical-surgical hospitalization was used to adjust for sociodemographic characteristics, health factors, and outpatient and emergency utilization patterns that differentiated veterans with and without mental health diagnoses in the bivariate analyses described previously. In each case, the dependent variable was a dichotomous measure of whether the veteran had been hospitalized during FY 2012. Multiple logistic regression models were constructed to compare veterans with mental illnesses to those without mental health diagnoses, as well as those with eight specific mental health diagnoses to those without those diagnoses.
Analysis
Between October 1, 2011, and September 30, 2012, there were 2,016,392 veterans younger than 60 years who received care in the VHA, of which 47% (N = 952,282) were diagnosed with a mental illness (psychiatric or SUD). Altogether, 4.97% (N = 100,191) of veterans under the age of 60 years had a medicalsurgical hospitalization. Veterans with mental illnesses were more than twice as likely to be hospitalized than veterans without mental illnesses, odds ratio = 2.52 (2.48–2.55). Although more likely to be hospitalized, those with mental illnesses who were admitted did not spend any more days hospitalized than other veterans during the same FY (7.84 d, standard
Analysis proceeded in several steps. First the proportion of veterans with medical-surgical hospitalizations among those with mental health diagnoses were compared with the proportion among those without such diagnoses with unadjusted odds ratios (OR). Then, veterans with eight specific mental health diagnoses (schizophrenia, bipolar disorder, MDD, other depressions, generalized anxiety disorder, PTSD, AUD, drug use disorder) were compared with those without any of these diagnoses on the relative odds of medical-surgical hospitalization. Psychosomatics
-:-, -
2019
RESULTS
www.psychosomaticsjournal.org
3
Medical Hospitalizations and Mental Illness
TABLE 1.
Characteristics of Veterans Health Administration Patients Younger Than 60 Years Diagnosed With Mental Illnesses During FY 2012
Demographics Male Homeless during the year VA pension Service connected 50% or more Service connected less than 50% Total income White race Age Health factors Charlson Index Seizures Myocardial infarction CHF Peripheral vascular disease Cerebrovascular accident Chronic obstructive airway disease Peptic ulcer disease Hepatic disease Renal disease Cancer Moderate/severe liver Pneumonia Nausea/vomiting Insomnia Congestive heart failure Connective tissue disease Diabetes mellitus Complications of diabetes Paraplegia HIV Metastatic cancer Any pain diagnosis Tobacco use Opiate prescriptions Service use Primary care visits Outpatient specialty medicine visits ER visits
No mental illness, N/mean (%/SD)
Mental illness, N/mean (%/SD)
(N = 1,064,110) 922,068 (86.7) 31,285 (2.9) 14,425 (1.4) 174,153 (16.4) 292,485 (27.5) 23,502 (42,115) (N = 871,509) 592,942 (68.0) (N = 1,064,075) 45.74 (10.90) (N = 1,064,110) 1.22 (1.49) 4474 (0.4) 5624 (0.5) 316,377 (29.7) 13,655 (1.3) 15,135 (1.4) 76,866 (7.2) 3808 (0.4) 22,032 (2.1) 19,149 (1.8) 27,760 (2.6) 1610 (0.2) 6552 (0.6) 8704 (0.8) 23,012 (2.2) 316,377 (29.7) 7488 (0.7) 136,737 (12.8) 22,807 (2.1) 5765 (0.5) 7728 (0.7) 2796 (0.3) 267,937 (25.2) 141,240 (13.3) 1.03 (3.59)
(N = 962,282) 813,098 (85.4) 116,897 (12.3) 30,800 (3.2) 283,653 (29.8) 200,013 (21) 28,023 (19,574) (N = 871,509) 598,186 (70.0) (N = 952,280) 45.42 (10.98) (N = 962,282) 1.51 (1.7) 10,563 (1.1) 6429 (0.7) 40,5121 (42.5) 16,862 (1.8) 23,156 (2.4) 110,436 (11.6) 6495 (0.7) 44,036 (4.6) 18,759 (2) 30,266 (3.2) 4450 (0.5) 11,509 (1.2) 20,434 (2.1) 80,077 (8.4) 405,121 (42.5) 7424 (0.8) 137,068 (14.4) 25,636 (2.7) 7389 (0.8) 9194 (1.0) 3072 (0.3) 359,870 (37.8) 270,786 (28.4) 2.44 (5.73)
1.91 (2.13) 2.81 (5.52) 0.32 (0.89)
3.19 (3.33) 4.99 (7.99) 0.76 (1.96)
Cohen’s d
RR 0.99 4.18 2.39 1.82 0.76
20.16 1.03 20.04 0.25 2.64 1.28 1.43 1.38 1.71 1.61 1.91 2.23 1.09 1.22 3.09 1.96 2.62 3.89 1.43 1.11 1.12 1.26 1.43 1.33 1.23 1.50 2.14 0.42 0.65 0.45 0.41
Bold values indicate a significant effect size (Cohen’s d . 0.20, risk ratio (RR) . 1.5, or RR , 0.67). FY = fiscal year; SD = standard deviation; CHF = congestive heart failure; ER = emergency room; VA = Veterans Administration.
deviation = 16.10, vs 7.96 d, standard deviation = 19.99; Cohen’s d = 0.01). To identify factors that might contribute to the apparent increased risk for hospitalizations, the characteristics of veterans with mental illnesses were compared with those of veterans without mental illnesses (Table 1). Veterans diagnosed with mental health issues were more likely to have been homeless, to have a service-connected disability with a disability
4
www.psychosomaticsjournal.org
rating greater than 50%, and to receive a VA pension. Several medical diagnoses were more common among veterans with mental illnesses than others including seizures, stroke, peptic ulcer disease, pneumonia, liver disease, and nausea and vomiting. Likewise, the Charlson Comorbidity Index, an indicator of overall health and expected mortality, was substantially higher among veterans with mental health diagnoses with a Cohen’s d of 0.25. High-risk behaviors, such as
Psychosomatics
-:-, -
2019
Moore and Rosenheck
TABLE 2.
Risk of Admission to Medical-Surgical Units for Psychiatric and Addiction Diagnoses During FY 2012
Diagnosis
Number hospitalized (%)
OR*
No MH diagnosis (N = 1,064,110) Any MH diagnosis (N = 952,282) Schizophrenia (N = 51,752) Bipolar disorder (N = 77,839) Major depression (N = 179,474) Other depression (N = 429,799) Anxiety disorder (N = 261,212) PTSD (N = 303,793) Alcohol use (N = 222,854) Drug use (N = 173,529)
31,750 68,441 4348 7151 15,353 35,735 19,702 16,835 23,550 21,425
2.52 2.98 3.29 3.04 2.95 2.65 1.91 3.84 4.58
(2.98) (7.19) (8.40) (9.19) (8.55) (8.31) (7.54) (5.54) (10.57) (12.35)
OR† (2.48–2.55) (2.89–3.08) (3.20–3.38) (2.98–3.10) (2.90–2.99) (2.60–2.70) (1.87–1.94) (3.78–3.91) (4.50–4.66)
1.43 0.87 1.02 1.00 1.18 1.04 0.98 1.95 1.38
(1.41–1.45) (0.83–0.90) (0.99–1.06) (0.98–1.02) (1.16–1.20) (1.01–1.06) (0.96–1.00) (1.91–2.00) (1.36–1.41)
FY = fiscal year; MH = mental health; OR = odds ratio; PTSD = posttraumatic stress disorder; SD = standard deviation. * Unadjusted OR comparing the risk of hospitalization for veterans with a mental illness (psychiatric or substance use) to that of those without mental illnesses. † Adjusted OR controlling for sociodemographic, health, and service characteristics identified as significant in bivariate analyses.
tobacco use and receipt of opioid prescriptions, were also more frequently observed among those with mental health diagnoses. Veterans with mental health diagnoses used considerably more outpatient primary care and specialty medical services and had more emergency department visits than veterans without such diagnoses. Veterans diagnosed with individual diagnoses of schizophrenia, bipolar disorder, MDD, other depressive disorders, generalized anxiety disorder, PTSD, AUD, and illicit drug use were more likely to be hospitalized than veterans without mental illness (Table 2). After adjustment for sociodemographic factors, medical comorbidities, and service use differences identified in bivariate analyses (Table 2), the OR for medicalsurgical hospitalization in association to mental illness declined from 2.52 to 1.43. Among individual diagnoses, adjusted ORs for alcohol use and drug use remained most strongly associated with medicalsurgical hospitalizations after covarying sociodemographic, medical, and service use (Table 2). In view of the robustness of the relationship of hospitalization with substance use, one further logistic regression analysis was conducted, adding adjustment for alcohol and drug use, which resulted in further reduction of the adjusted OR for any mental health diagnoses from 1.43 (95% confidence interval: 1.41–1.45) to 1.24 (95% confidence interval: 1.21–1.26). In this final model, the OR for drug use dropped further (OR = 1.04, 95% confidence interval: 1.02–1.07), while the relationship between AUD and hospitalizations Psychosomatics
-:-, -
2019
remained robust (OR = 1.77, 95% confidence interval: 1.73–1.81).
DISCUSSION In this study of 2 million veterans younger than 60 years who were treated in the VHA during FY 2012, we found evidence that worse physical health likely accounts for most—but not all—of the over 2-fold increased odds of medical-surgical hospitalization among veterans with mental health diagnoses compared with veterans without such diagnoses. SUDs involving both illicit drugs and alcohol were the mental health diagnoses most strongly associated with medical hospitalization and accounted for one-third of all hospitalized patients in this sample. Among all mental health diagnoses, AUD was unique in that chronic medical comorbidities did not account for the high hospitalization rates. Together, these results suggest two major paths to reduce costly medical hospitalizations. First, we must better understand the details of how mentally ill adults become prematurely physically ill to mitigate this risk. Second, we must consider focused and intense efforts to assist populations at particularly high risk (e.g., those using alcohol). The poor physical health of VHA patients diagnosed with mental illness cannot be understated. We confirmed that a number of medical conditions including headaches and chronic pain, chronic obstructive pulmonary disease, liver disease, and stroke www.psychosomaticsjournal.org
5
Medical Hospitalizations and Mental Illness were more prevalent among veterans with psychiatric and SUDs than others. While the current study did not have access to mortality data, others have found that veterans with mental illnesses die prematurely,23 and the Charlson Comorbidity Index, a predictor of future mortality, was substantially higher in our sample of veterans diagnosed with mental illnesses than that of other VHA patients. Factors thought to contribute to the increased risk for medical problems and higher mortality associated with mental illnesses include decreased adherence to prescribed treatments,24 poorer access to outpatient medical services,25 and poorer overall quality of medical care.5,26 Data used in this study did not include evaluation of quality of medical care, although there was no evidence of decreased access to outpatient medical services. Rather, veterans with mental health diagnoses were more likely to use services of both primary care and specialty medical clinics. Mental illness is also strongly associated with a range of psychosocial and health factors that are linked to poor health outcomes including tobacco use homelessness,27 tobacco use, and receipt of potentially dangerous medications—especially opioids.28 In our sample, one in ten veterans with mental illnesses had been homeless during the prior year, more than a quarter used tobacco products, and nearly a third had received opioid prescriptions. In the face of the evolving US crisis of opioid use and dependence, there has been significant focus on overprescription of opioids to veterans for diverse pain syndromes.29 Beyond the risk of overdose, opioid medications are increasingly recognized to adversely affect physical health28 and to predispose toward medical hospitalizations.30 The VHA has made significant efforts to reduce overprescribing of opioids through its national Opioid Safety Initiative, and there have been some promising results. In a recent analysis, VHA opioid prescriptions were shown to have decreased significantly since 2012,31 the same year the data examined here were recorded. Further research is needed to determine whether the harm of excess opioid prescriptions affects veterans with mental illnesses and whether hospitalization rates might be reduced as opioid prescription use is reduced. Among individual diagnoses, alcohol and drug use were most strongly associated with inpatient admissions. This likely reflects the particularly high rates of medical illnesses among adults with addictions.14 However, alcohol use appears to impart additional 6
www.psychosomaticsjournal.org
risk even after accounting for the shared variance with drug use, other mental health disorders, and chronic medical problems. Nearly 10% of veterans diagnosed with AUD had medical-surgical hospitalizations, accounting for 23% of all veterans hospitalized on medical-surgical units. This finding is consistent with prior estimates that roughly 20–40% of medically hospitalized patients have AUD.32,33 It is possible that our data set, which focuses on chronic medical conditions, may not capture acute medical problems associated with alcohol use such as withdrawal, intoxication and poisoning, dehydration, arrhythmias, and falls.34 Further research is needed to better understand the balance between hospitalizations associated with chronic medical conditions and hospitalizations when alcohol consumption or cessation was the direct precipitant. Because AUD was the single strongest predictor of hospitalizations, it is necessary to consider treatment options as they relate to medical illnesses. The VHA offers extensive substance use treatment through mental health service lines that can reduce alcohol use but may not support coordination with medical teams. There have also been preventive efforts based on early identification of AUD in the primary care setting using the Alcohol Use Disorder Identification Test-C as a screening tool.35 VA has developed Primary Care Mental Health Integration programs that provide mental health services directly within the context of primary care service delivery.36 Our present study did not evaluate the use of Primary Care Mental Health Integration, but other studies have found that such services can improve overall health outcomes.37 Integration of primary care and addiction treatments has been found to improve abstinence in severe AUDs and to reduce medical complications.38 VHA has piloted an alcohol care management initiative, which has been shown to improve treatment engagement, increase naltrexone adherence, and decrease days of heavy drinking.39 Discussion is also warranted related to why comorbid sociodemographic characteristics and medical illnesses did not account for all the increased risks of medical-surgical hospitalization among VHA patients with psychiatric diagnoses. On the one hand, illnesses not documented in this analysis may have explained some of the unexplained increased risks. On the other hand, it is possible that some degree of unmet need for psychiatric hospital care was reflected in the increased Psychosomatics
-:-, -
2019
Moore and Rosenheck risk of medical-surgical hospitalization. There has been substantial reduction in psychiatric bed availability in VHA in recent years, even as there has been an increase in the number of patients with psychiatric diagnoses.40 A recent study reported that between 1999 and 2007, the number of VA psychiatric bed declined 28% from 6570 to 4745 while the total number of patients increased by 50% from 3.4 million to 5.1 million.40 This decrease in psychiatric beds per patient from 1.9 of 1000 veterans to 0.9 of 1000 veterans may have reflected justifiable changes in practice in an overbedded system, but it may also have left some need for inpatient care unmet, resulting in spillovers to medicalsurgical beds. Unfortunately, we cannot precisely determine the reasons for the unaccounted increase in the risk for veterans with psychiatric diagnoses. A number of methodological limitations of this study require comment. First, administrative data do not always identify diagnoses accurately and can identify functional impairments only through proxy measures of utilization of homeless or rehabilitative services and through disability payments. In addition, data were not available on whether specific mental health or medical diagnoses were identified in outpatient, inpatient, or emergency department service settings—data on which would be valuable in understanding where these conditions had been treated by specialized providers and whether they were temporarily associated with medical-surgical hospitalizations. In addition, subjective measures of physical health or mental health were lacking in administrative files. References 1. Health, United States, 2016: with chartbook on long-term trends in health. Hyattsville, MD: Centers for Disease Control; 2017 2. Pitts SR, Vaughns FL, Gautreau MA, Cogdell MW, Meisel Z: A cross-sectional study of emergency department boarding practices in the United States. Acad Emerg Med 2014; 21:497–503 3. Himelhoch S, Weller WE, Wu AW, Anderson GF, Cooper LA: Chronic medical illness, depression, and use of acute medical services among medicare beneficiaries. Med Care 2004; 42:512–521 4. Li Y, Glance LG, Cai X, Mukamel DB: Mental illness and hospitalization for ambulatory care sensitive medical conditions. Med Care 2008; 46:1249–1256 5. Mai Q, Holman CD, Sanfilippo FM, Emery JD: The impact of mental illness on potentially preventable hospitalisations: a population-based cohort study. BMC Psychiatry 2011; 11:163
Psychosomatics
-:-, -
2019
Finally, this study was limited to veterans treated by the VHA. While national in scope, the cohort studied may not be representative of veterans not using VHA services or nonveteran populations. In particular, patients in this study were older than the US average, and relatively few women were included. CONCLUSIONS This study has shown that VHA patients diagnosed with mental illnesses are at higher risk for developing serious medical conditions than other VHA patients and that these indicators of poor physical health account for much, although not all, of the high risk of medical-surgical hospitalization associated with mental illness, especially in the case of AUD which poses a substantial independent risk for medical hospitalization. Our results found no evidence that poor health and high utilization of hospital services were associated with limited access to outpatient medical services. Rather, all services were used at higher rates among those diagnosed with mental illnesses. Further research is needed to identify more effective treatment for these high-risk patients. Conflicts of Interest: The authors declare that they have no conflict of interest. Funding: This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.
6. Yoon J, Yano EM, Altman L, et al: Reducing costs of acute care for ambulatory care-sensitive medical conditions: the central roles of comorbid mental illness. Med Care 2012; 50:705–713 7. Davydow DS, Ribe AR, Pedersen HS, et al: Serious mental illness and risk for hospitalizations and rehospitalizations for ambulatory care-sensitive conditions in Denmark: a nationwide population-based cohort study. Med Care 2016; 54:90–97 8. Levenson JL, Hamer RM, Rossiter LF: Relation of psychopathology in general medical inpatients to use and cost of services. Am J Psychiatry 1990; 147:1498–1503 9. Hochlehnert A, Niehoff D, Wild B, Junger J, Herzog W, Lowe B: Psychiatric comorbidity in cardiovascular inpatients: costs, net gain, and length of hospitalization. J Psychosom Res 2011; 70:135–139 10. Thorpe K, Jain S, Joski P: Prevalence and spending associated with patients who have a behavioral health disorder and other conditions. Health Aff 2017; 36:124–132
www.psychosomaticsjournal.org
7
Medical Hospitalizations and Mental Illness 11. Albrecht JS, Hirshon JM, Goldberg R, et al: Serious mental illness and acute hospital readmission in diabetic patients. Am J Med Qual 2012; 27:503–508 12. Daratha KB, Barbosa-Leiker C, M HB, et al: Co-occurring mood disorders among hospitalized patients and risk for subsequent medical hospitalization. Gen Hosp Psychiatry 2012; 34:500–505 13. Chwastiak LA, Davydow DS, McKibbin CL, et al: The effect of serious mental illness on the risk of rehospitalization among patients with diabetes. Psychosomatics 2014; 55: 134–143 14. Dickey B, Normand SL, Weiss RD, Drake RE, Azeni H: Medical morbidity, mental illness, and substance use disorders. Psychiatr Serv 2002; 53:861–867 15. Scott KM, Lim C, Al-Hamzawi A, et al: Association of mental disorders with subsequent chronic physical conditions: world mental health surveys from 17 Countries. JAMA Psychiatry 2016; 73:150–158 16. Miller BJ, Paschall CB 3rd, Svendsen DP: Mortality and medical comorbidity among patients with serious mental illness. Psychiatr Serv 2006; 57:1482–1487 17. Cohen BE, Gima K, Bertenthal D, Kim S, Marmar CR, Seal KH: Mental health diagnoses and utilization of VA nonmental health medical services among returning Iraq and Afghanistan veterans. J Gen Intern Med 2010; 25:18–24 18. Nelson KM, Helfrich C, Sun H, et al: Implementation of the patient-centered medical home in the Veterans Health Administration: associations with patient satisfaction, quality of care, staff burnout, and hospital and emergency department use. JAMA Intern Med 2014; 174:1350–1358 19. Department of Veterans Affairs: VHA facility quality and safety report fiscal year 2012 data. Washington, D.C: Department of Veterans Affairs; 2013 20. Miller EA, Rosenheck RA: Risk of nursing home admission in association with mental illness nationally in the Department of Veterans Affairs. Med Care 2006; 44:343–351 21. de Groot V, Beckerman H, Lankhorst GJ, Bouter LM: How to measure comorbidity. A critical review of available methods. J Clin Epidemiol 2003; 56:221–229 22. Ferguson CJ: An effect size primer: a guide for clinicians and researchers. Prof Psychol Res Pract 2009; 40:532–538 23. Chwastiak LA, Rosenheck RA, Desai R, Kazis LE: Association of psychiatric illness and all-cause mortality in the national department of veterans affairs health care system. Psychosom Med 2010; 72:817–822 24. DiMatteo MR, Lepper HS, Croghan TW: Depression is a risk factor for noncompliance with medical treatment: metaanalysis of the effects of anxiety and depression on patient adherence. Arch Intern Med 2000; 160:2101–2107 25. Druss BG, Rosenheck RA: Mental disorders and access to medical care in the United States. Am J Psychiatry 1998; 155:1775–1777
8
www.psychosomaticsjournal.org
26. Druss BG, Rosenheck RA, Desai MM, Perlin JB: Quality of preventive medical care for patients with mental disorders. Med Care 2002; 40:129–136 27. O’Connell JJ: Premature mortality in homeless populations: A review of the literature. Nashville, TN: National Health Care for the Homeless Council; 2005 28. Paulozzi LJ, Zhang K, Jones CM, Mack KA: Risk of adverse health outcomes with increasing duration and regularity of opioid therapy. J Am Board Fam Med 2014; 27:329–338 29. Petrakis IL, Sofuoglu M, Rosenheck R: VA patients with high numbers of opioid prescriptions: a national study of sociodemographic and diagnostic characteristics, health service, and psychotropic medication use. Addict Disord Their Treat 2015; 14:167–175 30. Liang Y, Turner BJ: National cohort study of opioid analgesic dose and risk of future hospitalization. J Hosp Med 2015; 10:425–431 31. Hadlandsmyth K, Mosher H, Vander Weg MW, Lund BC: Decline in prescription opioids attributable to decreases in long-term use: a retrospective study in the veterans health administration 2010-2016. J Gen Intern Med 2018; 33:818–824 32. Coder B, Freyer-Adam J, Bischof G, et al: Alcohol problem drinking among general hospital inpatients in northeastern Germany. Gen Hosp Psychiatry 2008; 30:147–154 33. de Wit M, Jones DG, Sessler CN, Zilberberg MD, Weaver MF: Alcohol-use disorders in the critically ill patient. Chest 2010; 138:994–1003 34. Mannelli P, Pae CU: Medical comorbidity and alcohol dependence. Curr Psychiatry Rep 2007; 9:217–224 35. Bradley KA, Williams EC, Achtmeyer CE, Volpp B, Collins BJ, Kivlahan DR: Implementation of evidence-based alcohol screening in the veterans health administration. Am J Manag Care 2006; 12:597–606 36. Johnson-Lawrence V, Zivin K, Szymanski BR, Pfeiffer PN, McCarthy JF: VA primary care-mental health integration: patient characteristics and receipt of mental health services, 2008-2010. Psychiatr Serv 2012; 63:1137–1141 37. Trivedi RB, Post EP, Sun H, et al: Prevalence, comorbidity, and prognosis of mental health among US veterans. Am J Public Health 2015; 105:2564–2569 38. Weisner C, Mertens J, Parthasarathy S, Moore C, Lu Y: Integrating primary medical care with addiction treatment: a randomized controlled trial. JAMA 2001; 286:1715–1723 39. Oslin DW, Lynch KG, Maisto SA, et al: A randomized clinical trial of alcohol care management delivered in Department of Veterans Affairs primary care clinics versus specialty addiction treatment. J Gen Intern Med 2014; 29:162–168 40. Bowersox NW, Szymanski BJ, McCarthy JF: Associations between psychiatric inpatient bed supply and the prevalence of serious mental illness in Veterans Affairs nursing homes. Am J Public Health 2013; 103:1325–1331
Psychosomatics
-:-, -
2019