Druss BG, Wang PS, Sampson NA, Olfson M, Pincus HA, Wells KB, Kessler RC. Understanding Mental Health Treatment in Persons Without Mental DiagnosesResults From the National Comorbidity Survey Replication. Arch Gen Psychiatry. 2007;64(10):1196-1203. doi:10.1001/archpsyc.64.10.1196
Copyright 2007 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2007
Epidemiologic surveys have consistently found that approximately half of respondents who obtained treatment for mental or substance use disorders in the year before interview did not meet the criteria for any of the disorders assessed in the survey. Concerns have been raised that this pattern might represent evidence of misallocation of treatment resources.
To examine patterns and correlates of 12-month treatment of mental health or substance use problems among people who do not have a 12-month DSM-IV disorder.
Design and Setting
Data are from the National Comorbidity Survey Replication, a nationally representative face-to-face US household survey performed between February 5, 2001, and April 7, 2003, that assessed DSM-IV disorders using a fully structured diagnostic interview, the World Health Organization Composite International Diagnostic Interview (CIDI).
A total of 5692 English-speaking respondents 18 years and older.
Main Outcome Measures
Patterns of 12-month service use among respondents without any 12-month DSM-IV CIDI disorders.
Of respondents who used 12-month services, 61.2% had a 12-month DSM-IV CIDI diagnosis, 21.1% had a lifetime but not a 12-month diagnosis, and 9.7% had some other indicator of possible need for treatment (subthreshold 12-month disorder, serious 12-month stressor, or lifetime hospitalization). The remaining 8.0% of service users accounted for only 5.6% of all services and even lower proportions of specialty (1.9%-2.4%) and general medical (3.7%) visits compared with higher proportions of human services (18.9%) and complementary and alternative medicine (7.6%) visits. Only 26.5% of the services provided to the 8.0% of presumably low-need patients were delivered in the mental health specialty or general medical sectors.
Most services provided for emotional or substance use problems in the United States go to people with a 12-month diagnosis or other indicators of need. Patients who lack these indicators of need receive care largely outside the formal health care system.
A 2006 report by the Institute of Medicine1 described a gap in the quality of mental health care as great as or greater than that seen in the rest of the health care system. The report described a primary goal for quality improvement of reducing underuse and overuse of services. In documenting problems of underuse of services, the report cited studies indicating that only approximately half of persons with serious mental illnesses receive any treatment and that fewer than half of those treated receive care consistent with evidence-based guidelines.2- 6 Government-sponsored quality improvement initiatives in the United States and abroad are seeking to address the problem of undertreatment of mental disorders through increased public awareness, systematic screening, and implementation of treatment guidelines.7- 9
Overuse, defined as “when a health care service is provided under circumstances in which its potential for harm exceeds the possible benefit,”10(p1002) is an equally important but much less well understood concern in quality assessment. The classic example of overuse in medicine is the prescription of antibiotic agents for the common cold, which is a problem from a clinical perspective (exposure of patients to unnecessary risks) and a public health perspective (potential development of resistance).11 Similarly, overtreatment in mental health would present a clinical problem if it exposed patients to adverse effects or a public health problem if it diverted resources away from those with greater evidence of need.12
Although there has been little formal study of the problem of overuse in mental health services, community mental health surveys have consistently found that a substantial proportion of people in treatment for emotional problems do not meet the criteria for any of the 12-month diagnoses. Data from the 1980 Epidemiological Catchment Area Survey,13 the 1990-1992 National Comorbidity Survey,14,15 and the 2001-2003 National Comorbidity Survey Replication (NCS-R)4 all reported that approximately half of persons using mental health services during the year before interview did not meet the criteria for any of the DSM disorders assessed in the surveys. This pattern is not confined to the United States. Persons without diagnosed disorders make up most treated cases in mental health needs assessment surveys throughout the developed and developing worlds.16- 18 This consistent finding has been interpreted as evidence of potential overuse, and recommendations have been made based on this conclusion to redistribute services from persons without diagnoses to those with more serious conditions.18
If overuse of services implies use of services in the absence of need, then evaluating this problem empirically requires carefully considering what we mean by the term need and what we include in the enumeration of mental health services. Clinical need is a complex concept, best conceptualized not as a single entity but as a series of overlapping constructs, including symptoms, disease burden, treatment effectiveness, and consumer perceptions.19- 22 In assessing potential need it is critical to complement diagnosis with other measures of symptoms and psychosocial burden and with patients' explanations of why they sought care.
In exploring the implications of use of services in the absence of diagnosis, it is also critical to understand where people are receiving their treatment. The types of mental health care that people obtain are heterogeneous, ranging from services delivered in specialty mental health settings, to those provided by general health professionals, to those obtained from nonprofessionals, such as clergy and self-help group facilitators.23 Whereas treatment in the formal health care system (either medical or mental health) contributes to rising health care costs, treatment delivered outside of those sectors has fewer implications for allocation of resources.
This study examines the prevalence and characteristics of use of 12-month mental health services in people without 12-month mental disorders based on the most comprehensive survey ever conducted of the prevalence and correlates of mental disorders in the United States. Extending research from previous epidemiologic surveys, we seek to understand the extent to which patients who do not meet the 12-month criteria for any of the disorders assessed in the survey have other indicators of possible need. We examine the sectors where treatment is received, the number of visits made to each treatment sector, and patient self-reported reasons for seeking treatment.
The NCS-R is a nationally representative face-to-face household survey of respondents 18 years and older in the coterminous United States conducted between February 5, 2001, and April 7, 2003.24,25 Part 1 included a core diagnostic assessment administered to all respondents. Part 2 assessed risk factors, correlates, service use, and additional disorders and was administered to all part 1 respondents with lifetime disorders plus a probability subsample of other respondents. The overall response rate was 70.9%. The part 2 sample, which is the focus of this study, included all part 1 respondents with a disorder and a probability subsample of other part 1 respondents, for a total of 5692 part 2 respondents. Part 2 data were weighted to adjust for differential probability of selection in households, differential nonresponse, differential selection of part 1 respondents into part 2, and residual discrepancies with US Census population data. Recruitment and consent procedures were approved by the human subjects committees of Harvard Medical School and the University of Michigan (Ann Arbor).
The DSM-IV diagnoses were made using the World Health Organization Composite International Diagnostic Interview (CIDI) version 3.0,26 a fully structured lay-administered diagnostic interview. The DSM-IV27 criteria were used to generate diagnoses. The 12-month DSM-IV/CIDI disorders considered herein include mood disorders (bipolar I and II disorders, subthreshold bipolar disorder, major depressive disorder, and dysthymia), anxiety disorders (panic disorder, agoraphobia without panic, specific phobia, social phobia, generalized anxiety disorder, obsessive-compulsive disorder, posttraumatic stress disorder, and adult separation anxiety disorder), impulse control disorders (anorexia, bulimia, binge eating disorder, oppositional defiant disorder, intermittent explosive disorder, and pathological gambling disorder), substance use disorders (alcohol and drug abuse with or without dependence), and nonaffective psychosis. All diagnoses were made using organic exclusions and diagnostic hierarchy rules, except for the substance use disorders, in which abuse was defined with or without dependence. Masked clinical reappraisal interviews using the Structured Clinical Interview for DSM-IV (SCID)25,28 showed generally good concordance between diagnoses based on the CIDI and the SCID for anxiety, mood, and substance use disorders but CIDI underestimation of the prevalence of nonaffective psychosis.29 The CIDI lifetime diagnoses of impulse control disorders were not validated.
Respondents who did not meet the 12-month criteria for any of the previously mentioned DSM-IV disorders were divided into subsamples based on a 3-category gradient of possible need for treatment: (1) those with at least 1 lifetime DSM-IV/CIDI disorder; (2) those with 1 or more indicators of severity; that is, 12-month subthreshold disorders (ie, lacking only 1 criterion for a diagnosis), exposure to a major stressful event (eg, rape or divorce) in the past 12 months, and lifetime hospitalization for a mental disorder; and (3) those without any lifetime diagnosis or other indicator of possible need.
All part 2 respondents were asked whether they had ever received treatment for “problems with your emotions or nerves or your use of alcohol or drugs.” A list of types of treatment professionals and a separate list of settings were presented in a respondent booklet to provide a visual recall aid. Separate assessments were made for different types of professionals, support groups, self-help groups, mental health crisis hotlines, and complementary and alternative medicine (CAM) therapies. Follow-up questions were asked about ages at the first and most recent contacts and about number of visits in the past 12 months.
Reports of 12-month service use were classified into the following predefined categories: psychiatrist, nonpsychiatrist mental health specialist (psychologist or other nonpsychiatrist mental health professional in any setting, social worker or counselor in a mental health specialty setting, and use of a mental health hotline), general medical provider (primary care doctor, other general medical doctor, nurse, and any other health professional not previously mentioned), human services professional (religious or spiritual advisor and social worker or counselor in any setting other than a specialty mental health setting), and CAM (any other type of healer, such as a chiropractor, participation in an Internet support group, or participation in a self-help group).
Basic patterns of service use were examined by computing proportions in treatment and mean numbers of visits among those in treatment. Logistic regression30 analysis was used to study predictors of 12-month treatment and treatment in particular sectors among those receiving any treatment. Standard errors were estimated using the Taylor series method31 and implemented using a software system (SUDAAN version 8.0.1; Research Triangle Institute, Research Triangle Park, North Carolina). Logits and their 95% confidence intervals were transformed into odds ratios (ORs) for ease of interpretation. Multivariate significance tests in the logistic regression analyses were made using Wald χ2 tests based on coefficient variance-covariance matrices that were adjusted for design effects using the Taylor series method. Statistical significance was evaluated using 2-sided design-based tests and the P < .05 level of significance.
As reported previously,4 17.9% of NCS-R respondents obtained some type of treatment for problems with their mental health or substance use in the 12 months before interview. There was a strong monotonic relationship between the gradient of possible need and the probability of treatment, with 39.9% of respondents with 12-month DSM-IV/CIDI disorders obtaining treatment compared with 18.3% of those with lifetime but not 12-month disorders, 12.7% of those without any lifetime disorder but with another indicator of possible need, and 3.8% of those without any indicator of possible need (Table 1). Mean number of visits among patients was significantly related to the gradient as well, with the highest mean (16.5 visits) among patients with 12-month disorders and the lowest means among patients without lifetime disorders either in the presence (8.4 visits) or absence (10.3 visits) of other indicators of possible need. Of all patients who obtained treatment, 61.2% met the criteria for 1 or more of the 12-month DSM-IV/CIDI disorders assessed in the survey. These 12-month cases accounted for 69.0% of all visits. An additional 21.1% of patients had a lifetime disorder (19.9% of all visits), and the remaining 17.7% of patients (11.1% of visits) did not meet the criteria for any lifetime or 12-month disorder.
Patients with 12-month disorders accounted for higher proportions of visits to psychiatrists (73.5%), other mental health professionals (74.3%), and general medical professionals (74.3%) than to human services (60.8%) and CAM (60.7%) professionals (detailed results available on request). Patients with lifetime but not 12-month disorders accounted for a higher proportion of visits to CAM professionals (27.7%) than to other treatment sectors (14.1%-17.8%). Persons without any of the disorders assessed in the survey and without any other indicator of possible need, in comparison, accounted for a higher proportion of visits to the human services sector (18.9%) than to other sectors (1.9%-7.6%).
These differences resulted in variation in treatment profiles across sectors among patients in each subsample of the gradient of possible need (Table 2). Among patients with a 12-month diagnosis, more than half of all visits (55.1%) were delivered by either a psychiatrist (13.7%) or another mental health professional (41.4%), and another 9.7% were delivered in the general medical sector. This means that nearly two-thirds of all visits (64.8%) by patients with a 12-month disorder were made in the formal health care sector. These are all higher proportions than among patients with a lifetime but not a 12-month disorder, for whom 45.5% of visits were delivered by a mental health professional and 6.5% by a general medical professional (52.0% in a health care sector), and among patients without either a lifetime disorder or any of the indicators of potential need (26.5% in a health care sector). Regarding this last result, 73.5% of the visits of patients classified as having the least evidence of need were provided outside of the health care system, in either the human services sector (30.7%) or the CAM sector (42.8%). The highest proportion of visits in a health care sector (70.5%), in comparison, was made by patients without a lifetime disorder but with some other indicator of possible need.
Logistic regression analysis documented a variety of significant predictors of 12-month treatment among respondents who met the criteria for 1 or more lifetime disorders but none of the 12-month disorders (Table 3). These predictors include a high number (≥4) of lifetime disorders (OR, 4.0), lifetime hospitalization for a mental illness or substance use problem (OR, 2.5), most recent episode within 1 year of the 12-month recall period (OR, 3.2), a subthreshold episode within the 12-month recall period (OR, 2.4), and a major stressful event within the 12-month recall period (OR, 2.3). Respondents with lifetime bipolar disorder also had a meaningfully elevated OR of 12-month treatment (OR, 1.9), although this was only of marginal significance (P = .07). Significant predictors of 12-month treatment among respondents who did not meet the criteria for any of the disorders considered herein include lifetime hospitalization (OR, 4.1), a subthreshold episode within the 12-month recall period (OR, 3.1), and a major stressful event within the 12-month recall period (OR, 2.9).
A weighted count was created of individual-level predictors of treatment among respondents who did not meet the criteria for a 12-month DSM-IV/CIDI disorder by summing the indicators of possible need with weights based on the ORs in Table 3. In the subsample of these respondents who met the criteria for 1 or more lifetime disorders, this weighted estimate of need was significantly related in largely monotonic form to the probability of 12-month treatment (χ23 = 144.2; P < .001) and mean number of visits among cases (F3,311 = 4.6; P = .007) (Table 4). The small (2.7%) proportion of respondents in this subsample with the highest estimated level of need had a 63.9% probability of 12-month treatment, had a mean of 54.3 visits, and accounted for more than one-third (36.6%) of all 12-month visits in this subsample. The proportion increased to 55.9% of visits to a psychiatrist when examining the distribution by treatment sector (data available on request). At the other extreme, the 19.8% of respondents in this subsample with the lowest estimated level of need accounted for only 3.8% of all 12-month visits in this subsample.
In the subsample of respondents who did not meet even lifetime criteria for any of the disorders considered herein, the weighted estimate of need was significantly related to probability of 12-month treatment (χ23 = 94.7; P < .001) but not to mean number of visits (F3,157 = 1.2; P = .34) (Table 5). The small (4.0%) proportion of respondents in this subsample with the highest estimated level of need had a 28.7% probability of 12-month treatment and accounted for nearly one-fourth (23.4%) of all 12-month visits in this subsample. The proportion increased to 27.8% of visits to a psychiatrist when examining the distribution by treatment sector (data available on request). The 73.5% of respondents in this subsample with the lowest estimated level of need accounted for 50.4% of all 12-month visits in this subsample. This proportion decreased to 19.8% of visits to a psychiatrist, however, compared with 65.6% of CAM visits when examining the distribution by treatment sector (data available on request).
Respondents who received treatment were asked their reasons for doing so. This question series began by asking patients whether they sought treatment based on their own perceived need for professional help or only because someone else urged them to do so. Most patients in all subsamples (67.3%-76.0%) reported that they perceived themselves as needing treatment (Table 6). However, this percentage varied significantly across subsamples (χ23 = 9.5;P = .02) and was higher among respondents with a 12-month disorder (74.9%) or other indicator of possible need (76.0%) than among respondents with only a lifetime disorder (67.3%) or no indicator of possible need (69.4%). Among patients with perceived need for treatment, significant subsample differences were found in most reasons for seeking treatment, with patients who had a 12-month disorder generally reporting more reasons than other patients. However, the 3 most commonly reported reasons were the same across subsamples: to obtain help for emotions or behavior problems, to cope with stressful events, and to cope with ongoing stress.
The results reported herein show that most people who use services for mental health or substance use problems in the United States have either a DSM diagnosis or some other indicator of possible need for treatment. These findings extend those of previous studies in showing that a substantial percentage of service users did not have any of the 12-month DSM-IV/CIDI diagnoses assessed in the survey4,13-15 despite the fact that a much wider range of conditions was considered herein than in earlier studies. However, when we focused on number of visits, we found that the nearly two-thirds of mental health service users who had a 12-month diagnosis accounted for approximately three-quarters of all professional visits for mental health or substance use problems. These results may help reduce concerns based on the findings of previous epidemiologic surveys that a high proportion of services are provided to patients who do not have a DSM disorder.
We also found that patients with lifetime but not 12-month diagnoses made up most other 12-month service users in the NCS-R. In this subsample of respondents, those with recent episodes were significantly more likely than others to be undergoing 12-month treatment, suggesting that they may be receiving time-limited treatment for recent episodes. Use of services among other patients with lifetime disorders may reflect the growing awareness that maintenance treatment is important for relapse prevention in people with a history of serious conditions.32- 34 The fact that indicators of lifetime severity (number of diagnoses and lifetime history of hospitalization) were significant predictors of 12-month treatment is consistent with this hypothesis. This pattern is consistent with earlier NCS-R findings that allocation of services is significantly associated with burden of illness in patients with 12-month diagnoses.15,17,18
To the extent that maintenance treatment is taking place, asymptomatic patients with lifetime disorders in 12-month treatment may be treatment success stories. The comparatively high proportion of people with lifetime bipolar disorder in 12-month treatment despite not having a 12-month manic-hypomanic or depressive episode is an especially important case in point given the literature suggesting the value of maintenance therapy in this population.33,35 The fact that asymptomatic people with lifetime diagnoses had fewer visits than those with more symptoms or risk factors further suggests an underlying rationality in resource allocation among stable patients receiving maintenance treatment.
The finding that a meaningful proportion of services is provided to patients who do not meet either 12-month or lifetime criteria for any of the DSM-IV disorders assessed in the NCS-R raises a more complex set of clinical and policy concerns. More than three-fourths of these patients had a subthreshold 12-month condition, reported a serious 12-month stressor, or had a history of hospitalization. Subthreshold syndromes are currently less well defined than threshold diagnoses, and relatively little is known about the risks and benefits of treating these conditions.36,37 Nonetheless, particularly in the presence of serious psychosocial stressors, arguments have been made that treatment of subthreshold syndromes can have value not only in reducing present distress and suffering but in preventing the future onset of syndromal disorders.38- 40 The fact that in this subsample a dose-response relationship exists between number of indicators of potential need and service use suggests that treatment decisions are being made based on these types of considerations. More generally, our findings support the notion that need for care may be more appropriately thought of as a continuum than as a categorical construct.
Only a small proportion (8.0%) of service users did not have any of the indicators of need considered herein. For a wide range of medical interventions, it is commonly necessary to have some false-positive rate of treatment. For example, a review of the appendectomy literature found an inverse relationship between the perforation rate and the surgeon's false-positive rate, leading the authors to recommend that a 23% error rate (removal of a normal appendix) would be appropriate.41 For mental disorders, public health efforts such as social marketing and antistigma campaigns may simultaneously increase the rate of care in persons with and without disorders.42 Thus, some level of overtreatment may be an acceptable, and even desirable, consequence of efforts to reduce the problem of undertreatment of mental disorders.
Although service users without potential indicators of need reported similar reasons for using services as other mental health service users, they were much less likely than other patients to receive their care in the formal health care system and much more likely to be treated in the human services sector. This means that these presumably low-need patients are not contributing importantly to formal mental health expenditures, nor do they divert a substantial proportion of professional resources away from patients with diagnosable disorders. They account for only 1.9% to 2.4% of all visits to psychiatrists and other mental health professionals and 3.7% of all visits to general medical professionals for mental health or substance use problems. Note that 44.8% of these patients who used CAM services reported that prayer was the main service provided. This finding is consistent with previous research demonstrating the common use of prayer43 and clergy visits44 in the United States for problems in daily life.
These results should be interpreted with the following 2 limitations in mind. First, the CIDI does not provide a fully comprehensive assessment of all DSM-IV disorders, nor is it completely accurate in the diagnoses it assesses, as it is somewhat conservative relative to the SCID. As a result, some of the respondents classified as not having had a 12-month disorder actually had one. Second, the study relied on self-reported measures of service use. Because reporting bias for mental health services seems to be greatest in persons with high levels of distress,45 such bias may be less of a concern for the population of primary interest in the present study (ie, patients with low evidence of need for treatment).
Although the study's findings should provide some reassurance regarding the magnitude of overtreatment of mental health and substance use problems in the United States, there is still much work to be done to ensure that mental health resources are used effectively and efficiently.45 Overuse can be a problem not only for persons without need who receive services but also for individuals with evidence of need who receive poor-quality services. From an economic perspective, poor quality of care represents wasted resources.46,47 The present study suggests that efforts to reduce waste need to shift from whether the wrong persons are receiving mental health care to ensuring that those who do receive care receive the right services, in the right manner, and at the right time.
Correspondence: Benjamin G. Druss, MD, MPH, Department of Health Policy and Management, Rollins School of Public Health, Emory University, 1518 Clifton Rd NE, Room 606, Atlanta, GA 30322 (firstname.lastname@example.org).
Submitted for Publication: January 9, 2007; final revision received February 20, 2007; accepted March 16, 2007.
Author Contributions: All authors had full access to all the data in the study. Dr Kessler takes responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosures: Dr Druss has been a consultant for Pfizer, Inc. Dr Olfson has received grants from Bristol-Myers Squibb and Eli Lilly & Co; has been a consultant for Bristol-Myers Squibb, Eli Lilly & Co, Pfizer, and McNeil; and is on the speaker's bureau for Janssen. Dr Pincus has been a consultant for Cisco System, Community Care Behavioral Health Organization/University of Pittsburgh Medical Center Health Plan, and Bristol-Myers Squibb and has received speaking fees from Bimark Medical Education, Comprehensive NeuroScience Inc, Medical Information Technologies, Academy of Managed Care Pharmacy Horizons LLC, Cardinal Health Inc, and Health Partners.
Funding/Support: The NCS-R is supported by grant U01-MH60220 from the National Institute of Mental Health (NIMH), with supplemental support from the National Institute on Drug Abuse, the Substance Abuse and Mental Health Services Administration, the Robert Wood Johnson Foundation (grant 0447), and the John W. Alden Trust. Instrumentation, fieldwork, and consultation on data analysis activities were supported by grant R01 MH070884 from the NIMH; the John D. and Catherine T. MacArthur Foundation; the Pfizer Foundation; grants R13-MH066849, R01-MH069864, and R01 DA016558 from the US Public Health Service; grant FIRCA R03-TW006481 from the Fogarty International Center; the Pan American Health Organization; Eli Lilly & Co; Ortho-McNeil Pharmaceutical Inc; GlaxoSmithKline; Bristol-Myers Squibb; and grant 1K24MH075867-01A1 from the NIMH.
Role of the Sponsor: The funding organizations had no role in the design and conduct of the study; the collection, management, analysis, and interpretation of the data; or the preparation, review, or approval of the manuscript.
Disclaimer: The views and opinions expressed in this article are those of the authors and should not be construed to represent the views of any of the sponsoring organizations, agencies, or the US government.
Additional Information: A complete list of NCS publications and the full text of all NCS-R instruments can be found at http://www.hcp.med.harvard.edu/ncs. Send correspondence to email@example.com. The NCS-R is performed in conjunction with the World Mental Health Survey Initiative. A complete list of World Mental Health publications and instruments can be found at http://www.hcp.med.harvard.edu/wmh.
Additional Contributions: Eric Bourke, Jerry Garcia, and Emily Phares assisted with manuscript preparation, and the staff of the World Mental Health Data Collection and Data Analysis Coordination Centres assisted with instrumentation, fieldwork, and consultation on data analysis.