Functional Impairment and Co-occurring Psychiatric Disorders in Medically Hospitalized Men FREE

ALTHOUGH previous research has described a high prevalence of psychiatric symptoms among general medical and surgical inpatients, there is limited information on prevalence based on diagnostic criteria as described by the American Psychiatric Association's Diagnostic and Statistical Manual of Mental Disorders.^1- 2 Furthermore, although we have learned much about the functional limitations imposed by major depression, less is known about impairment associated with a range of psychiatric disorders in patients who concurrently have a medical illness. We conducted this study to examine the impairment and functional limitations associated with 13 co-occurring psychiatric disorders in medically hospitalized men.

The majority of prospective studies on psychiatric comorbidity in general medical and surgical inpatients has relied on mental health symptoms, rather than on psychiatric diagnoses, in determining prevalence. Nonetheless, various reports^3- 6 have suggested that as many as 50% of general medical and surgical inpatients have symptoms indicative of psychological distress or disturbance, including symptoms of depression, anxiety, and substance abuse. While documentation of symptoms has made an important contribution to our understanding of the relationship between mental health and medical illness, establishing diagnoses according to standard criteria is a critical requirement for mental health treatment. The psychiatric diagnosis defines a clinically significant behavioral or psychological syndrome or pattern associated with present distress or disability.¹

Community-dwelling individuals with chronic diseases are more likely to have a co-occurring psychiatric disorder compared with those without chronic medical problems. For example, results from 1 site (Los Angeles, Calif) of the National Institute of Mental Health Epidemiologic Catchment Area (ECA) study demonstrated that the risk of having a recent psychiatric disorder was increased 41% by the presence of a chronic medical condition⁷ and that results varied among diseases by recent psychiatric disorder.⁸ The Medical Outcomes Study (MOS) identified a substantial number of patients with major depression in primary care settings, most of whom (about 70%) had a comorbid medical condition.⁹ Therefore, a study of individuals using the health care system should reveal a higher prevalence of psychiatric comorbidity because patients with more complicated medical problems are more likely to seek medical care¹⁰ and individuals for whom mental diagnoses are made have been shown to be high users of medical services.^11- 12

Results from the MOS and other studies^7,13- 16 have documented significant impairment associated with major depression. While lower health and functional status is expected to accompany medical illness and hospitalization, we have limited information on the additional burden presented by the broad range of psychiatric disorders that may be comorbid with a medical illness. Furthermore, most research on the relationship between psychiatric disorders and medical illness has concentrated on the effect of depression^13,17- 18 and has frequently focused on reductions in symptoms rather than on more general measures of health and functioning.^19- 20 Little is known about the effect of anxiety and substance abuse disorders on functioning for the hospitalized medical patient, nor do we have information comparing the loss in functioning among co-occurring psychiatric disorders. Not only is it important to recognize which disorders are associated with the greatest decrements in functioning but we also need to understand which specific dimensions of functioning are the most severely affected by psychiatric disorders in the medically ill inpatient.

Our study had 2 major objectives. The first was to estimate the current and lifetime prevalence of 13 psychiatric disorders in a random sample of medical and surgical inpatients at 3 large university-affiliated midwestern Department of Veterans Affairs (VA) hospitals. Prevalence of co-occurring psychiatric disorders was determined using computerized structured diagnostic interviews. The second objective was to measure the decrements in health-related functioning during hospitalization associated with a co-occurring psychiatric disorder and to compare the relative contributions of mood, anxiety, and substance use disorders to functional impairment. We conducted these analyses controlling for patient sociodemographics, medical diagnosis, and severity of medical illness.

The project was approved by the institutional review boards at all 3 study sites. Patients admitted to acute medical and surgical wards at the 3 participating VA Medical Centers (VAMCs) were requested at random to participate in the study between November 1992 and August 1994. The daily admission sheets were scanned and each new admission was assigned a 2-digit random number from a random number table. Patients were approached in order of random number by a study interviewer until the daily screening workload was completed. Patients were disqualified from participating in the study if they were (1) women, who represent only 1% of VA inpatients and they were excluded because of significant variations in the prevalence of psychiatric disorders from men^21- 22; (2) admitted primarily for psychiatric reasons, including detoxification from alcohol or drug use; (3) admitted primarily for administrative purposes, such as transfer to a nursing home or another VAMC, or admission to determine eligibility for service-connected disabilities; these patients were not eligible because they are not representative of the general medical and surgical patients; or (4) in a coma or otherwise unable to communicate because of cognitive impairment. Individuals with scores lower than 20 on the Mini-Mental State Examination²³ were not eligible for the study. We used a higher cutoff point on the Mini-Mental State Examination than used in the ECA Study²¹ because we were not using the Mini-Mental State Examination to diagnose substantial cognitive impairment but rather to exclude patients with sufficient cognitive impairment to make them unreliable informants.

All patients enrolled in the study received written and oral information about the study and signed informed consent to participate.

Although we considered limiting the study to certain medical conditions, the decision to include all possible diagnoses was made because (1) there is conflicting evidence regarding the association of psychiatric comorbidity with specific diagnoses, (2) prevalence estimates would be more generalizable with unrestricted medical diagnoses, and (3) there is no current information regarding interactions of medical diagnosis with psychopathologic characteristics in relation to functional outcomes.

Most patients were interviewed within 2 to 4 days following hospital admission. Exceptions to this schedule included patients admitted to critical care units who were interviewed after their transfer to a medical or surgical unit, and patients who were initially unable to be interviewed because of illness or impairment but whose long-term rehabilitation during the index hospitalization eventually enabled them to be interviewed.

We used the computerized Quick Diagnostic Interview Schedule Interview (Q-DIS) to identify psychiatric disorders. The Q-DIS was developed based on data from the ECA study²⁴ and interviews for the presence of 16 Diagnostic and Statistical Manual of Mental Disorders, Revised Third Edition (DSM-III-R) disorders.¹ For these diagnoses, lifetime diagnostic decisions are identical with positive and negative results for the full lay-interviewer administered Diagnostic Interview Schedule (DIS).^24- 25 The Q-DIS is not a screening interview because it does not sacrifice accuracy by asking only the "best" questions. Instead, it provides all the questions to make a lifetime diagnosis, but achieves brevity by attempting to classify subjects as cases or noncases as soon as it can be determined whether positive diagnostic criteria are met. Symptoms that the patient attributes to a medical condition are not counted toward a diagnosis. Once the Q-DIS has determined that a patient meets DSM-III-R diagnostic criteria, the program queries the patient for symptoms within the previous 12 months. Recent data have shown the Q-DIS to have good to excellent sensitivity and specificity compared with the DIS, although the Q-DIS may underestimate lifetime generalized anxiety disorder and recent posttraumatic stress disorder (PTSD).²⁵

For this study, we used the Q-DIS for 13 psychiatric disorders: PTSD, panic disorder, generalized anxiety disorder, agoraphobia, social phobia, simple phobia, obsessive-compulsive disorder, depression or dysthymia, mania, schizophrenia or schizophreniform disorder, alcohol abuse or dependence, drug abuse or dependence, and antisocial personality disorder. (The Q-DIS does not make distinctions between alcohol or drug dependence vs abuse, nor between major depression and dysthymia.) The remaining Q-DIS disorders (eating disorders, pathologic gambling, and transsexualism) were judged to be either irrelevant to the veteran population or less essential to the study's objectives. The Q-DIS was administered to patients by the study's research assistants using portable computers.

For measurement of health and functional status, we used the MOS 36-Item Short-Form Health Survey (SF-36). The SF-36 assesses health concepts relevant to functional status and well-being.^26- 27 This interview was developed as a 20-item scale for the MOS,²⁸ was derived from more detailed measures used in the RAND Health Insurance Experiment,²⁹ and has since been expanded to 36 items.²⁶ As currently defined, the SF-36 measures physical functioning, social functioning, physical and emotional role functioning, mental health, general health perceptions, vitality, and pain. The SF-36 was administered during the index hospitalization at the same time as the Q-DIS.

We measured illness severity to avoid confounding the decreased functioning for seriously medically ill patients with any impairment associated with the co-occurring psychiatric disorder. We used an independent assessment of medical illness severity, the Computerized Severity Index (CSI).³⁰ Using clinical data abstracted from medical records, the CSI provides an overall maximum severity score for each patient that is based on the severity of all existing disease states. Information in the medical record, including multiple International Classification of Diseases, Ninth Revision, Clinical Modification codes,³¹ disease complications, and objective clinical findings such as laboratory values and symptoms are used to generate a severity score from 1 to 4.³⁰ The CSI presents several important advantages over other measures of disease severity, including score comparability among all individual and combinations of medical diagnoses and improved prediction of resource use, including costs and length of stay.^30,32 Photocopies of medical records from the index hospitalization were sent to the developers of the CSI for abstraction.

The lifetime and recent (last 12 months) prevalence of DSM-III-R diagnoses was calculated using frequency distributions. The Q-DIS data were also used to classify patients into groups of mood, anxiety, and substance abuse disorders. We also examined the frequencies of multiple psychiatric disorders.

The relationships between health-related functional status and the presence of co-occurring psychiatric disorders were examined using analysis of covariance. We calculated analysis of covariance models in which the effect of comorbidity was examined after adjusting for age, race (white or nonwhite), marital status, medical diagnosis, and severity of illness. First, we tested the associations of the separate SF-36 subscales with a summary indicator of psychiatric comorbidity in which comorbidity was defined in nonoverlapping categories as: (1) no lifetime DSM-III-R psychiatric diagnosis; or (2) a past disorder, any lifetime diagnosis but no recent (last 12 months) disorder; or (3) any recent (last 12 months) disorder. Subsequent models were constructed to contrast 3 broad groups of psychiatric diagnoses (mood, anxiety, and substance abuse) simultaneously in which each diagnostic group was coded as: (1) absence of a diagnosis in the particular diagnostic group; (2) past disorder, a lifetime disorder but no recent disorder in that diagnostic group; and (3) recent disorder in that diagnostic group. Adjustment variables were as above plus the other psychiatric diagnostic groups. Finally, we calculated the adjusted mean SF-36 subscale scores³³ and compared no comorbidity, lifetime but not recent disorder, and recent disorder for each separate diagnostic group after removing the effects of the other independent variables, including the remaining psychiatric disorder groups. In this way, we could estimate the relationship between a disorder group, such as anxiety disorders, while controlling for the effect of mood and substance abuse disorders and demographics, medical diagnosis, and medical illness severity.

One thousand seven male inpatients from the 3 VAMCs completed the Q-DIS and the SF-36. This sample represents 2.2% of the medical and surgical admissions to these 3 VAMCs during the study enrollment period. On average, these patients (all men) were white (86.4%), late middle-aged (mean age, 61 years; SD, 12 years), married (59.7%), and had primary diagnoses of cardiovascular disease (23.7%) and cancer (10.1%) as well as a wide variety of other diagnoses (Table 1). Eighty-six percent of the patients were interviewed within 4 days of admission in accordance with the primary study protocol, an additional 8.6% on days 5 to 7 after admission, and the remaining 5.6% were interviewed later than 1 week after admission (range, 8-33 days; with only 8 patients interviewed >2 weeks after admission). On average, the hospital length of stay was 9.8 days (SD, 11.4 days; range, 1-180 days).

Almost half (46.6%) of the patients had a DSM-III-R diagnosis during their lifetime and almost 1 (17.8%) in 5 had a recent (last 12 months) diagnosis (Table 2). The most prevalent single diagnosis was alcohol abuse or dependence (32.5%), followed by PTSD (10.0%) and major depression or dysthymia (8.9%). The most frequent recent disorder was reported for major depression or dysthymia (7.0%). However, among groups of disorders, recent disorder occurred most frequently for anxiety disorders (10.9%).

A substantial number of patients were diagnosed as having more than 1 psychiatric disorder with the Q-DIS (Table 3). While 54% of the sample had no Q-DIS diagnoses, almost a quarter had more than 1 diagnosis. (In this analysis, a patient meeting DSM-III-R diagnostic criteria for more than 1 drug disorder was counted only once, but alcohol abuse or dependence and drug abuse or dependence were counted separately.) While only 60 patients met criteria for more than 3 disorders, among themselves these patients accounted for 330 lifetime diagnoses and 99 separate reports of recent disorders.

We found significant decrements associated with the presence of any co-occurring psychiatric disorder for all measures of functional status and well-being in these medically hospitalized patients (P<.001 or P= .001) after controlling for age, race, marital status, and medical diagnostic group (Figure 1). Early exploratory analyses also included the measure of severity of illness, the maximum CSI score that had been completed on only a subset of subjects (n = 603) because of resource limitations. However, illness severity was not independently associated with any measures of functional status, other than Physical Functioning. Therefore, we dropped that variable from the final model described herein to retain as large a sample size as possible. Comparisons of the adjusted mean scores on the SF-36 subscales (Figure 1) found that all subscales were significantly lower for individuals with any past disorder and even lower for those with recent disorder. Only for Physical Functioning was there no difference between recent and past disorders. The greatest reduction in functioning and well-being came in the 2 role-functioning scores (Physical and Emotional Role Functioning). The average adjusted scores for these medically hospitalized inpatients with a recent disorder were only 63% and 66%, respectively, of the adjusted scores for the group with no psychiatric disorder. In contrast, the decrements associated with a recent disorder on Physical Functioning, Bodily Pain, General Health, Vitality, Social Health, and Mental Health were less (89%, 74%, 84%, 80%, 76%, and 70%, respectively).

The second set of analyses of covariance examined the unique associations of the 3 diagnostic groups—mood, anxiety, and substance abuse disorders—while controlling again for age, race, marital status, and medical diagnosis, and also for the remaining diagnostic groups (Table 4). As before, severity of medical illness was independently associated with only Physical Functioning, so the analyses were conducted with the full sample without this measure. These analyses allowed us to investigate the varying and unique effects of these groups of disorders recognizing the presence of multiple co-occurring disorders as shown in Table 3. In all cases, the level of functioning and well-being for the absent (ie, no diagnosis in that disorder group) category was lower than for the group with no psychiatric disorder at all described in Figure 1, demonstrating the effect of multiple psychiatric diagnoses (eg, an individual with no anxiety disorder may have had a mood disorder with associated lower functioning). The overall statistical main effects of individual diagnostic groups were less strong with generally no significant difference in mean adjusted SF-36 scores between recent and past disorders. These specific results are probably due to a combination of the lower adjusted SF-36 scores in the no disorder categories, as well as the smaller sample sizes in the remaining categories.

The lower functioning associated with a co-occurring psychiatric disorder was greatest for mood and anxiety disorders. In fact, all measures of functioning for patients with recent mood and anxiety disorders were similar, apart from Emotional Role Functioning in which individuals with recent anxiety disorders had better functioning compared with those with recent mood disorders. The associations of functioning with substance abuse were significant but generally less strong than those with mood and anxiety disorders. While in general individuals with lifetime and current substance abuse reported functioning that was comparable with those with mood and anxiety disorders, those with current substance use disorders reported higher scores on the General Health Ratings, Physical Functioning, and Vitality compared with those with no substance use disorders.

From a large, multisite, random sample of hospitalized male medical and surgical patients, we identified substantial prevalence of co-occurring psychiatric disorders as well as large decrements in health-related functioning associated with all 3 groups of psychiatric disorders: mood, anxiety, and substance abuse. Our study is unique in documenting the associations of functioning with psychiatric disorders that are additional to medical illnesses severe enough to require hospitalization. Medical inpatients with recent co-occurring psychiatric disorders reported substantially reduced functioning, by as much as one third, compared with those with no psychiatric disorder. Lower functioning was particularly found in self-reported impairment in work or regular daily activities due to physical health and emotional problems (Physical and Emotional Role Functioning). Observed decrements in functioning were also substantial for Vitality and Social Health for patients with mood disorders only. On the other hand, co-occurring psychiatric disorders appeared to pose few limitations on the general measure of Physical Functioning, which is a range of activities from vigorous (sports) to stair climbing to mild walking.

Functional impairment associated with depression is well known. Results using the 20-item form of the MOS found that primary care patients with depression had impaired functioning comparable to that for patients with chronic medical illness, both cross-sectionally and longitudinally.^13,15 Again in primary care using the SF-20, the PRIME-MD (Primary Care Evaluation of Mental Disorders) study³⁴ demonstrated substantial and unique impairment in health-related quality of life for a range of psychiatric disorders compared with patients with no psychiatric disorder. Similarly in medically hospitalized inpatients, we were also able to compare the independent or unique impairment in functioning associated with 3 major groups of psychiatric disorders.

According to the Q-DIS, the prevalence of psychiatric comorbidity was high: almost half of our large sample reported at least 1 lifetime psychiatric disorder, with alcohol abuse or dependence occurring most frequently. The high prevalence of alcohol disorders is not surprising, given previous data on hospitalized veterans³⁵ and nonveterans³⁶ and the generally higher rates of alcohol disorders among men.³⁷ Previous research^5,38- 39 consistently suggests that medically hospitalized individuals will demonstrate significant psychological distress, indicated by symptoms, and we demonstrate that diagnostic rates in 3 medical centers are similarly high.

Not surprisingly for a hospitalized population, we found substantially higher prevalence of psychiatric disorders in this hospitalized male population compared with similarly aged community cohorts in the ECA studies.⁴⁰ The ECA data are probably more relevant for our patient group than the more recent community survey, the National Comorbidity Study, because the National Comorbidity Study included only individuals younger than 55 years.²² Although by virtue of veteran status we would expect the prevalence of PTSD to be high in our sample, it is important to note herein that other nonveteran service settings may also treat many patients with PTSD. For example, a study of an urban Michigan Health Maintenance Organization⁴¹ found that 9% of urban young adults met criteria for a DSM-III diagnosis of PTSD and 39% had been exposed to a potentially traumatic event, including exposure to serious accidents, physical assault, and life-threatening events.

Just as MOS participants, who were all individuals using primary or specialty health care outpatient services, were in general sicker and older than those in the earlier Health Insurance Experiment,²⁷ so were our patients significantly more impaired in health-related functional status compared with MOS participants, although similar in age. Compared with the MOS subsample who were administered the SF-36,²⁷ our inpatient subjects, even those with no co-occurring psychiatric disorders, reported greater impairment than those with recent (within the past year) acute myocardial infarction. These differences are undoubtedly not surprising given that our study recruited inpatients severely ill enough to be hospitalized for medical conditions.

Many individuals with mental disorders use general medical services in place of mental health care.^11,42 Reductions in hospital costs and subsequent health care utilization among medical and surgical patients with co-occurring mental disorders are realized when mental health treatment such as consultation-liaison psychiatry is used appropriately.^43- 45 Therefore, lack of attention paid to the mental health needs of inpatients with concurrent psychiatric disorders or psychological symptoms may in the long run result in the inefficient and more costly use of medical services. Previous research has indicated that psychiatric comorbidity is associated with increased cost and use of services and with costs during hospitalization^5,46 and in primary care patients.⁴⁷

Results of this study and others confirm the importance of detecting and treating mental health problems, both in the inpatient and outpatient settings. Many psychiatric disorders, such as major depression, are treatable illnesses with a high likelihood of remission with appropriate treatment.^48- 49 Alcohol disorders are highly amenable to brief interventions provided by physicians or other health care providers in medical settings.^50- 51 Medical and surgical inpatient and outpatient settings are also important places of initial contact between patients and health care providers from which referrals to specialty mental health can be made. Internists and family physicians can be important access points to appropriate services for individuals with psychiatric disorders.

Brief screening for mental health disorders in inpatient and outpatient settings should improve the quality of health care as well as lead to appropriate treatment with accompanying improvement in outcomes. A targeted beginning could be admission screening for major depression and alcohol problems, for which there are excellent brief screening instruments,^52- 55 and this screening can be implemented by nurses, physicians assistants, or office personnel. Patients with positive findings on screening can be followed up more carefully by their primary care physicians with specialty referrals to psychiatrists or other mental health professionals where necessary. Continuity of care is critical to ensure that patients adhere to treatment and move toward remission of symptoms.

The effectiveness of brief psychiatric services such as hospital consultation-liaison psychiatry or psychiatric consultation in primary care settings has been varied,^{20,38,56- 58} perhaps because of the brevity of the intervention. Nonetheless, psychiatric consultation for patients with somatization disorder in primary care has been shown to be effective⁵⁹ and a number of psychosocial interventions have improved outcomes in chronically medically ill patients.⁶⁰ Appropriate medication management^48,61 in conjunction with behavioral interventions should reduce psychiatric symptoms substantially and improve functioning. However, these types of interventions do not necessarily occur in patients seen in routine medical settings, either inpatient or outpatient.^62- 64 A program designed to include a combination of initial screening, primary care physician interventions, psychiatric consultation, and coordinated care over time may be required.

Psychiatric consultation on inpatient medical units is not the only model of treatment in that other health care professionals can contribute to improving patient functioning. For example, difficulties in social health and emotional role functioning can also be assessed and treated by social workers, nurses, and/or psychologists. The contribution of these other professionals, perhaps as part of an interdisciplinary team, can also supplement the work of a consultation psychiatrist, especially in situations limited to a single consultation-liaison visit.

There are some remaining concerns regarding the validity of a lay interviewer–administered structured psychiatric diagnostic interview in individuals with severe or chronic medical illnesses, particularly in the case of depression.^17,65- 66 For example, many of the somatic criteria for major depression such as sleeplessness, fatigue, and loss of weight may be characteristic signs of a medical illness or consequences of treatment such as cancer chemotherapy. The American Psychiatric Association's diagnostic criteria for major depressive episode clearly state that such symptoms can count toward a diagnosis unless they are fully accounted for by a general medical condition,¹ and interviews such as the Q-DIS query this issue specifically for each reported symptom. On the other hand, some individuals, including older adults who represent the majority of our patient sample, may be more likely to attribute psychiatric and psychological symptoms to physical states or at least have greater difficulty in attribution of symptoms.

It has been argued that disentangling medical symptoms and medical treatment from psychiatric criteria is difficult at best for experienced clinicians and cannot be accomplished by structured interviews such as the Q-DIS. However, most studies of depression in primary or inpatient medical care have used instruments that are similar to the Q-DIS, including the DIS,⁶⁷ symptom checklists, or briefer screening interviews.^{15,38,47,55,61} Furthermore, we recently conducted a study to examine the validity of a structured psychiatric diagnostic interview for major depression in medically hospitalized male veterans (unpublished data, 1997). The chance-corrected agreement of the lay-administered interview compared with a physician expert panel was good to very good (κ = 0.43-0.67), with no significant direction in the disagreement with the expert panel.

It is not clear how the acuteness of the medical condition may have affected the nature of the responses to the interview, especially our self-reported measures of functional status, the SF-36 subscales. Even though the SF-36 has been extensively validated in a range of settings and populations²⁷ and has become a "criterion" standard measure of health-related quality of life, it remains a subjective self-report of functioning. However, we found only minimal correlations between the SF-36 subscales and our objective measure of illness severity, the maximum CSI (all Pearson r<0.1), suggesting that the acuteness of the medical condition did not greatly affect self-reported functional status.

This was an all-veteran sample, therefore potentially limiting the generalizability of the findings. There are certainly a number of unique characteristics of the VA health care system, particularly the predominantly male patient population and the substantial amount of resources devoted to psychiatric care. For example, we conducted an analysis of the VA national hospital discharge data for fiscal year 1994, one of the years during which we enrolled patients for this study, and found that substance use, mood, and anxiety disorders accounted for 10.1%, 2.5%, and 2.4% of all hospital discharges, respectively. However, a much earlier study⁶⁸ found no differences between veterans' and nonveterans' use of health services. Furthermore, veterans in the general population do not appear to have more psychiatric disorders than nonveterans, especially in the age groups and service cohorts most frequently represented in our study.⁶⁹ Psychiatric disorders and high levels of psychological distress and problem drinking are also frequent in primary care and general inpatient settings.^{19,51,55,70- 71} For example, it is estimated that major depression occurs in 5% to 10% of primary care patients and in 10% to 14% of medical inpatients,¹⁹ and almost 10% of young adults in an urban health maintenance organization met criteria for PTSD.⁴¹ Therefore, even though this study was based on a VA patient population, the issue of psychiatric comorbidity and associated functional impairment is clearly relevant for nonveteran patient settings.

Given the nature of the VA patient population, we could not include female patients in this study. Certainly the prevalence of psychiatric disorders would have differed from that observed herein had substantial numbers of women been included, given the population sex differences in prevalence of certain psychiatric disorders such as depression (higher in women) and substance abuse (higher in men).^21- 22 The prevalence of psychiatric disorders also varies by age, in that older cohorts demonstrate lower lifetime rates, particularly for substance use disorders and antisocial personality disorders.^22,40

We know little about how functional impairment varies by sex. Older women appear to report somewhat lower functioning on the SF-36 in population surveys but the differences are slight.²⁷ Studies such as the PRIME-MD³⁴ have adjusted for sex rather than reported sex-specific impairments and we know little about sex-specific differences in functioning associated with depression in the MOS. It follows that we need to know more about differential functional impairment associated with co-occurring psychiatric disorders in medically ill women. Similarly, it would be important to replicate our study in a nonveteran sample, although we suspect that the findings would be similar.

Despite difficult issues related to diagnostic certainty and effective interventions for individuals with medical illnesses, it is clear from our study that apparent co-occurring psychiatric disorders, particularly mood and anxiety disorders, are associated with excess impairment and lower health-related functional status over and above what is contributed by the medical illness itself. In a new era of accountability and required outcomes assessment, poorer functional outcomes of medically ill patients associated with psychiatric disorders clearly have the potential to lower overall performance indicators. Medical providers may find that their report cards on functional outcomes are lower than expected or required because many of their patients have co-occurring psychiatric disorders. Furthermore, co-occurring psychiatric disorders in primary or inpatient care are associated with substantially higher health care costs,^47,62,72 which may be a major concern in capitated reimbursement systems. Therefore, the recognition, detection, and appropriate management of psychiatric disorders in medically ill patients has become increasingly important for public accountability as well as for patient care.

Accepted for publication November 6, 1997.

This study was supported by grant IIR 90-077 from the HSR&D Department of Veterans Affairs (all authors), grant P50-MH48197 from the National Institute of Mental Health (Dr Booth), and grant P50-AA07378 from the National Institute on Alcohol Abuse and Alcoholism (Dr Blow), Rockville, Md.

Presented at the Annual Meeting of the Veterans Affairs Health Services Research & Development Service, February 28-March 2, 1995.

We thank G. Richard Smith, MD, for several helpful reviews of earlier drafts.

Reprints: Brenda M. Booth, PhD, HSR&D Field Program for Mental Health (152/NLR), Veterans Affairs Medical Center, 2200 Fort Roots Dr, North Little Rock, AR 72114.