Mental health service use by Americans with severe ... - Springer Link

2 downloads 68 Views 109KB Size Report
D. S. Rae á C. Kennedy á B. Arons. Mental health ... G. Norquist á C. Kennedy á D. S. Rae ...... Haven, Conn., UO1 MH34224 ± Jerome K. Myers, Ph.D., Myrna.
Soc Psychiatry Psychiatr Epidemiol (2000) 35: 147±155

Ó Steinkop€-Verlag 2000

ORIGINAL PAPER

W. E. Narrow á D. A. Regier á G. Norquist D. S. Rae á C. Kennedy á B. Arons

Mental health service use by Americans with severe mental illnesses

Accepted: 4 January 2000

Abstract Background: The aim of this study was to determine the patterns and determinants of service use in severely mentally ill persons drawn from the National Institute of Mental Health Epidemiological Catchment Area (ECA) program, a community-based epidemiologic survey. This information provides a baseline against which to track ongoing changes in the US mental health service system. Methods: Severe mental illness (SMI) was de®ned according to US Senate Appropriations Committee guidelines. Comparisons were made with persons who had a mental disorder that did not meet these criteria (non-SMI). Sociodemographic factors, and 1-year volume and intensity of mental or addictive services use were determined. Di€erences between those who used services and those who did not were examined using logistic regression. Results: Persons with SMI di€ered from persons with non-SMI in most sociodemographic characteristics. A higher proportion of persons with SMI

used ambulatory services, but the mean number of visits per person did not di€er from the non-SMI population. Persons with SMI comprised the bulk of hospital inpatients admitted during a 1-year period. Several signi®cant sociodemographic determinants of service use were found, with di€erent patterns for general medical and specialty service use, pointing out potential barriers to care. Conclusions: As health care reform measures continue to be debated, attention to the service needs of the severely mentally ill is of crucial importance. Pre-managed care (pre-1993) baseline service use benchmarks will be essential to assess the impact of managed care on access to care, particularly for the severely mentally ill. Periodic collection of epidemiologic data on prevalence and service use would thus greatly facilitate service planning and addressing barriers to receiving mental health services in this population.

W. E. Narrow (&) Rm 403, 1400 K St, N.W. Washington, DC 20005 USA Tel.: +1-202-6826129

Introduction

W. E. Narrow National Institute of Mental Health, National Institutes of Health, Bethesda, Maryland, USA D. A. Regier American Psychiatric Institute for Research and Education, Washington, District of Columbia, USA G. Norquist á C. Kennedy á D. S. Rae Division of Services and Intervention Research, National Institute of Mental Health, Rockville, Maryland, USA B. Arons Center for Mental Health Services, Substance Abuse and Mental Health Services Administration, Rockville, Maryland, USA

The mental health treatment needs of US Americans with severe mental illnesses (SMI) have received increasing attention in recent years [1±4]. As health care costs and reform of the health care system continue to play a major role in the social and political agendas of the United States, it is important that mental health care for this vulnerable population receives proper attention in terms of health insurance coverage that allows accessibility to needed services. Toward this end, in the ®scal year 1993 appropriations bill for the Department of Health and Human Services, the US Senate Committee on Appropriations requested a report from the National Advisory Mental Health Council (NAMHC) on the ef®cacy of treatment and the costs of commensurate insurance coverage for mental and physical disorders for persons with SMI. The bill stated that: ``Severe mental illness is de®ned through diagnosis, disability and duration, and includes disorders with psychotic symptoms such as schizophrenia, schizoa€ective disorder, manic

148

depressive disorder, autism, as well as severe forms of other disorders such as major depression, panic disorder, and obsessive compulsive disorder.'' The resulting report, entitled ``Health Care Reform for Americans with Severe Mental Illnesses: Report of the National Advisory Mental Health Council'' [5], was released in March, 1993. In the Council's report, data from the National Institute of Mental Health (NIMH) Epidemiologic Catchment Area Program (ECA) provided the primary epidemiologic framework for the report's estimates of prevalence, service use, and costs. Two subsequent NAMHC reports have extended analyses into ®nancing of mental health services and the effects of parity in a managed care environment on access to and quality of services [6, 7]. The Senate Appropriation Committee's de®nition was also used as part of the US Substance Abuse and Mental Health Services Administration's response to PL 102±321, the ADAMHA Reorganization Act. For that response, data from the cross-sectional National Comorbidity Survey (NCS), conducted from 1990 to 1992, was used to develop estimation methodologies and prevalence estimates for ``serious mental illness'' [4]. No national survey of mental disorder prevalence and service use has been conducted since managed behavioral health care organizations began reducing costs and access to services, around 1993. Building on the work presented in the ®rst NAMHC report, the analyses presented in this article used ECA data to answer the following questions: 1. What are the sociodemographic characteristics of adults with SMI? 2. To which specialty and general medical service settings did they go for problems with mental health or substance use? 3. How many visits did they make to these settings? 4. What was the average individual rate of service use per year? and 5. What factors determined whether or not they sought services in a 1-year period? Answers to such questions are vitally important to health policy and service planning. The process of determining and projecting costs of mental health services under a particular health bene®t plan requires population estimates of disorder prevalence and service use in order to determine the extent of eligibility for proposed coverage and the estimated volume and intensity of services that can be expected. Patterns of service use among various clinical and demographic subpopulations are also helpful in addressing di€erential unmet need and the underlying barriers to care.

Subjects and methods Design of the Epidemiologic Catchment Area Program The NIMH ECA Program has been described at length in previous publications [8±11]. A total of 18,571 household and 2290 institutional residents age 18 and over were sampled and interviewed in

®ve areas: New Haven, Baltimore, Durham, St. Louis, and Los Angeles. The sample size for these analyses is 20,291 after excluding proxy interviews. Because of the complex multistage sampling procedures employed in the ECA, a statistical weighting strategy was devised to account for unequal probabilities of selection for each person sampled [9, 12]. Two face-to-face interviews were conducted 12 months apart (wave I and wave II). A telephone interview (face-to-face in New Haven) of the household respondents was conducted 6 months after wave I. Institutional residents were interviewed in waves I and II only. Respondents were asked about their use of health services at each interview. Diagnostic data were obtained at waves I and II only. Diagnostic and service use instruments The NIMH Diagnostic Interview Schedule (DIS) [13, 14] is an epidemiologic research instrument developed for use in the ECA, in which responses are scored by computer to determine whether the subject meets criteria for a psychiatric diagnosis. In the ECA, it was administered in a single session by a trained lay interviewer. Diagnoses are made according to criteria set forth in DSM-III. The choice of diagnoses and severity criteria for the operationalization of the Senate Appropriations Committee's de®nition was based on previously published work [5]. Respondents with a diagnosis of schizophrenia or bipolar disorder (type I) in the month prior to the wave 1 interview or in the year between the wave 1 and wave 2 interviews were considered severely mentally ill. Persons with a diagnosis of schizophrenia or bipolar disorder (type I) at some other point in their lives were considered severely mentally ill if there was further evidence of severity. For this group, the severity criteria included one or more of the following within the past year: any inpatient psychiatric hospitalization or nursing home placement, any outpatient mental health treatment in a specialty mental health or general medical setting, or psychotic symptoms (corresponding to criterion A of DSM-III schizophrenia). Respondents with major depressive disorder, bipolar disorder (type II), panic disorder, or obsessive compulsive disorder in the month prior to the wave 1 interview or in the year between the wave 1 and wave 2 interviews (or at any point in their lives for bipolar type II) were considered severely mentally ill if one or more of the following severity criteria were met: inpatient psychiatric hospitalization or psychotic symptoms. Because schizoa€ective disorder was not operationalized in DSM-III, it was not included in the DIS, nor was autism. Reports of each respondent's mental health service use from the three waves of data collection were aggregated to correspond most closely with the time of their active mental disorder diagnosis, as previously described [15]. Service use over a 1-year time period was the basis for analysis. Ten ambulatory settings were categorized into either specialty mental and addictive (SMA) or general medical (GM) sectors. (Note that for these analyses, only visits to the General Medical sector for mental health or substance use reasons were used.) Combined, these two sectors are referred to as the `health systems' sector. Seven inpatient settings were analyzed, six of which were in the SMA sector. Sociodemographic variables were chosen on the basis of previous research on service use, as well as to control for potentially confounding e€ects of factors such as age and socioeconomic status in logistic regression analyses. The variables chosen were age, race/ ethnicity, sex, marital status, socioeconomic status, residence (rural vs non-rural), living alone, receipt of disability payments, receipt of welfare payments such as general assistance or AFDC (Aid to Families with Dependent Children), and comorbidity with any substance use disorder [4, 16±20]. Socioeconomic status was calculated by taking an arithmetic mean of household income percentile, education (grade) percentile, and Nam occupational status score [21]; the resulting scores were divided into quartiles. Statistical analysis Persons with any past-year DIS/DSM-III disorder who did not meet criteria for severe mental illness (``non-SMI'') were used as a

149 comparison group in the analyses. Sociodemographic variables were tabulated and di€erences between the SMI and non-SMI groups were tested using chi-square. Ambulatory service use outcomes for each diagnosis were (1) number of persons treated; (2) total number of visits; and (3) average (mean) number of visits per treated person per year. The number of persons admitted at least once to each inpatient setting over the 1-year time frame was determined. (Other volume indicators could not be calculated for inpatient settings due to intersite variations in the nature of the data that were obtained by the ECA sites.) US population estimates were made by weighting the sample population to corresponding sex, race, and age group categories of the 1980 US civilian population. (A rough approximation to 1990 census results can be done by increasing the 1980 census-based ®gures by 14.8%.) Di€erences in service use outcomes were tested using the t statistic. Multivariate logistic regression analyses determined the important variables associated with use of services by the severely mentally ill and non-severely mentally ill. These regressions were performed using SUDAAN [22]. Because of small cell sizes resulting from strati®cation, the race/ethnicity, marital status, age and socioeconomic status variables were collapsed to enhance the stability of estimates. Results of the regressions are expressed as odds ratios (OR), with service use as the dependent variable. Statistically signi®cant odds ratios indicate that a person who used a service is more likely (OR > 1) or less likely (OR < 1) to have the characteristic of interest. Ninety-®ve and 99 percent con®dence intervals were calculated for each odds ratio. Con®dence intervals that did not include 1 were considered to be statistically signi®cant. (For reasons of space, 95 and 99% con®dence intervals are not presented; they can be obtained from the ®rst author.)

Results Prevalence of SMI Using the Senate de®nition, the 1-year prevalence of SMI among US adults aged 18 and over was 2.8% [5]. This represents 4,452,000 persons, using 1980 census ®gures. Schizophrenia was the most prevalent disorder in this group, with a prevalence of 1.5%. The severe a€ective disorders, bipolar disorder and major depression, accounted for 1% and 1.1% of the US population respectively. Panic disorder and obsessive compulsive disorder had prevalences of 0.4% and 0.6%. There was a large degree of comorbidity among these severe disorders, with the individual disorder prevalence rates adding up to 4.6%, compared to the overall prevalence of 2.8%. Characteristics of the SMI population (Table 1) With the exception of sex and residence, the SMI population di€ered signi®cantly from the non-SMI population in all sociodemographic and clinical categories examined. Over 65% of the SMI population was female. The age distribution of the SMI group was younger, with 73% under age 45, and only 6.5% over age 65. Nonwhite persons made up 20% of the SMI group, with Hispanics underrepresented and Blacks overrepresented compared to the non-SMI group. Signi®cantly more persons with SMI received disability compensation or public assistance/AFDC payments. Thirty percent of

Table 1 Sociodemographic and clinical characteristics of persons with mental disorders (Weighted data) (SMI severely mentally ill, SES socioeconomic status, AFDC Aid to Families with Dependent Children) Characteristic

SMI

Non-SMIa

Sex Female

65.4

62.4

Age 18±24 25±44 45±64 65+

21.6 51.3 20.6 6.5

15.7 42.0 25.9 16.4

71.007 (3)***

Race/ethnicity White Black Hispanic

79.8 16.5 3.8

81.0 12.7 6.3

13.035 (2)**

Marital status Married Widowed Separated Divorced Never married

40.2 5.6 6.3 15.0 32.9

51.8 11.6 5.0 10.6 21.0

86.893 (4)***

SES quartile 1 (High) 2 3 4 (Low)

9.6 27.2 39.1 24.1

14.2 31.3 33.1 21.4

21.090 (3)***

17.7 15.4 13.5 11.6 29.5

9.0 6.3 15.5 14.9 13.3

Disability compensation Welfare/AFDC Rural residence Lives alone Comorbid substance disorder

Chi-square (df) 2.385 (1)

50.699 71.851 1.810 5.315 121.378

(1)*** (1)*** (1) (1)* (1)***

* P < 0.05; **P < 0.01; ***P < 0.001 Non-SMI indicates any person with a mental disorder who is not in the SMI group

a

persons with SMI had a comorbid substance use disorder ± over twice the percentage for persons with nonSMI. Use of ambulatory mental health and addictions services Number of persons treated in ambulatory settings (Table 2) Over 2.6 million persons with SMI received services in the health systems sector. This number represents 15% of the total 16.8 million persons who received mental health treatment in this sector [23]. Persons with nonSMI in the past year accounted for 35% of persons receiving treatment, with the remaining 50% largely accounted for by persons with subthreshold conditions and non-severe lifetime disorders [15, 23]. General medical physicians, mental health specialists in private practice, and mental health center outpatient clinics saw the largest numbers of persons with SMI. For persons with non-SMI, general medical physicians and mental

150 Table 2 Number (percentage) of severely mentally ill persons seen in ambulatory settings (000s omitted) (MH mental health, VA Veterans' A€airs, SMA specialty mental and addictive services, GM general medical services) Service setting

SMI

Non-SMI

Psychiatric hospital outpatient clinic 120 (2.7)* MH center outpatient clinic 415 (9.4)*** General hospital outpatient clinic 336 (7.6)*** VA hospital outpatient clinic 104 (2.4) Alcohol/drug unit outpatient clinic 58 (1.3) MH specialist in health plan/clinic 291 (6.6)** MH specialist in private practice 870 (19.7)*** Crisis center 53 (1.2) SMA subtotala 1,828 (41.5)*** General hospital emergency dept 214 (4.8)*** GM physician 1,190 (27.0)*** GM subtotala 1,369 (31.0)*** Health systems total 2,612 (59.2)***

149 439 172 195 213 519 1,591 60 2,983 253 3,464 3,624 5,813

(0.5) (1.5) (0.6) (0.7) (0.7) (1.7) (5.3) (0.2) (10.0) (0.8) (11.6) (12.2) (19.5)

* P < 0.05; **P < 0.01; ***P < 0.001 The subtotals in each column are less than the sum of their component settings, because a person may visit more than one setting in the course of a year

a

health specialists in private practice also saw the largest number of persons, with mental health specialists in health plans or family clinics in third place. Overall, over half (59%) of all persons with SMI were seen in the health systems sector in a 1-year period, with 31% going to a general medical setting and 42% going to a specialty setting. In contrast, a signi®cantly lower percentage of persons with non-SMI (20%) was seen in the health systems sector, with 12% and 10% going to the general medical and specialty settings, respectively. These di€erences in yearly attendance were also re¯ected in the individual treatment settings (except for some of the less frequently used settings in the specialty sector). Number of visits to ambulatory health systems settings (Table 3) Previous ECA analyses showed that the total number of visits made in 1 year to all health systems settings was

158 million [23]. Table 3 shows that 35 million (22%) of these visits were made by persons with SMI and 62 million (39%) were made by persons with non-SMI. The remaining 61 million visits (39%) were made by persons with other conditions as noted above. While a large number of people visit the general medical sector, Table 3 shows that the overwhelming majority of ambulatory mental health visits, close to 80% for both SMI and non-SMI populations, are made to the specialty mental and addictive sector. For both SMI and non-SMI populations, private practice specialists were most frequently used, although a signi®cantly larger percentage of total visits for non-SMI (42%) were made to private practitioners compared to the percentage of total SMI visits (29%). The other signi®cant disparity in visit distribution was in the general hospital outpatient clinics, which accounted for over 10% of SMI visits, but just 1.6% of non-SMI visits. For several settings, the SMI population accounted for more visits than the non-SMI population, despite the overall higher prevalence of persons with non-SMI. Visit rate (Table 4) The total visit rate for the SMI group was 13 visits per treated person per year, compared to 11 visits per year for the non-SMI group, a statistically nonsigni®cant di€erence. As might be expected, the rate was higher in the specialty sector (SMI: 15 visits per person; non-SMI: 16 visits per person) than in the general medical sector (SMI: ®ve visits per person; non-SMI: four visits per person). The only signi®cantly di€erent visit rate was for general hospital outpatient clinics, with the SMI group having a higher rate. In several other settings (e.g., general hospital emergency department, alcohol/drug outpatient unit), large rate di€erences did not meet statistical signi®cance, probably due to small sample size.

Table 3 Number (percentage) of health systems visits to ambulatory settings (000s omitted)

Table 4 Average number of health systems visits per treated person per year in ambulatory settings

Service setting

Service setting

SMI

Non-SMI

Psychiatric hospital outpatient clinic MH center outpatient clinic General hospital outpatient clinic VA hospital outpatient clinic Alcohol/drug unit outpatient clinic MH specialist health plan/clinic MH specialist private practice Crisis center SMA subtotal General hospital emergency dept GM Physician GM subtotal Health systems total

16.7 14.9 10.6* 7.1 40.8 8.1 11.6 3.8 15.0 4.7 5.2 5.2 13.3

11.9 16.0 5.7 7.0 24.3 12.5 16.1 1.7 16.3 1.6 3.7 3.7 10.7

SMI

Psychiatric hospital outpatient clinic 2,006 MH center outpatient clinic 6,175 General hospital outpatient clinic 3,571 VA hospital outpatient clinic 736 Alcohol/drug unit outpatient clinic 2,369 MH specialist health plan/clinic 2,365 MH specialist private practice 10,085 Crisis center 204 SMA subtotal 27,510 General hospital emergency dept 1,009 GM physician 6,145 GM subtotal 7,154 Health systems total 34,664 *P < 0.05; ***P < 0.001

Non-SMI (5.8) 1,773 (2.9) (17.8) 7,011 (11.3) (10.3)*** 987 (1.6) (2.1) 1,356 (2.2) (6.8) 5,170 (8.3) (6.8) 6,481 (10.5) (29.1)* 25,676 (41.5) (0.6) 100 (0.2) (79.4) 48,554 (78.4) (2.9) 415 (0.7) (17.7) 12,964 (20.9) (20.6) 13,379 (21.6) 61,933

* P < 0.05

151

Use of inpatient mental health and addictions services (Table 5) Almost 17% of persons with SMI received inpatient mental health or addictions services in 1 year, compared to 0.9% of persons with non-SMI, a statistically significant di€erence. Of the total 1.4 million persons treated in inpatient settings in 1 year [23], over half (741,000) had SMI, 18% had non-SMI, and the remaining 29% had other conditions. The majority of SMI and nonSMI persons were treated in the specialty sector, with a

Table 5 Number (percentage) of severely mentally ill persons seen in inpatient settings (000s omitted) Service setting

SMI

Non-SMI

General hospital (psychiatric unit and scatterbed) State & county mental hospital (Residential supportive care) Community MH center Private mental hospital VA hospital psychiatric unit Alcohol/drug treatment unit SMA subtotal Nursing home Total

302 (6.8)***

105 (0.4)

270 66 58 90 62 50 727 18 741

88 62 1 40 7 39 240 12 252

(6.1)*** (1.5)* (1.3) (2.0) (1.4) (1.1)* (16.5)*** (0.4)*** (16.8)***

(0.2) (0.2) (