Mental Health Services in General Health Care: A Conference Report, Volume I (1979)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

41 THE NATURE AND SCOPE OF MENTAL HEALTH PROBLEMS IN PRIMARY CARE: VARIABILITY AND METHODOLOGY Darrel A. Regier, M.D., M.P.H.* Director, Division of Biometry and Epidemiology, National Institute of Mental Health The current and future role of primary care providers and settings is an issue of considerable health policy interest. Realistic policy and program development concerning this role must be based, in part, on accurate understanding of the scope and nature of mental disorder with- in the primary care sector e This fact has not been lost on the research community. As demonstrated by the recent literature review by Hankin and Oktay' a striking array of prevalence data is available. 1/ But what is a planner or policy maker or educator to make of mental dis- order prevalence rates ranging from below 1 percent to over 50 percent of the population in primary care settings? Adding to the potential confusion are health services research data in primary care settings which are often focused on visits or volume of: services rather than on the number of persons using cervices e Useful though they may be, visit-based utilization data and person-haled epidemiologic data are difficult to interrelate meaningfully. At present, we can give only a general rather than a definitive estimate of the rate of mental disorder in U.S. primary care settings. But we know that much of the variance in reported rates is directly related to differences in research methodology We will examine some of the effects of methodology on prevalence rates, citing study results from the literature in which single prevalence measures are usually used, as well as results from some recent NIMH-sponsored studies in which multiple measures were used within and across sites e Results from the latter studies, which permit person-based and visit-based rates to be compared, will also be reviewed. We hope that this exercise will serve both as a guide to understanding the current state of knowledge and a goal to more systematic study. *Prepared in collaboration with Anne H. Rosenfeld, Social Science Analyst, Division of Biometry and Epidemiology, National Institute of Mental Health; Barbara Je Burns, Ph.D., Research Psychologist, Primary Care Research Section, Applied Biometrics Research Branch, Division of Biometry and Epidemiology, National Institute of Mental Health; Irving D. Goldberg, M.P.H., Chief, Applied Biometrics Research Branch, National Institute of Mental Health; and Edwin W. Hopper, MeDo ~ Psychiatrist, Marshfield Clinic, and Chief of Staff, St. Joseph's Hospital, Marshfield, Wisconsin.

42 As background to this discussion, let us first sketch in the overall prevalence of mental disorders in the population at large and the distribution of persons with mental disorders among major sectors of the health care system. In the recent reviews prepared by the NIMH Division of Biometry and Epidemiology for the President's Commission on Mental Health, we estimated that about 15 percent of the general population could be diagnosed as having an ICD Section V- defined mental disorder in a given year. 2/ While this estimate is relatively crude, it does provide a general framework for understanding the scope of the problem. Regarding the division of responsibility in the current service system for the care of those with mental disorders, as shown in Figure 1, we have identified some 21 percent in the specialty mental health sector, another 54 percent as being seen only in the outpatient primary care sector (pith 6 percent specialty overlap), 3 percent in the general hospital-nursing home sector, and 22 percent for whom we cannot account. Thus, in discussing the mental health care role of primary practice, we are addressing the health care sector used ex- clusively by better than half of those with mental disorders in a given year. Figure 1 Not in Treatment/Other Human Services Sector* , / 21.5% General Hospital Inpatient/ Nursing Home Sector* 7 Specialty Mental ~ Health Sector 15.0% 2\ Both Specialty Mental Health / ~ Sector & Primary Care/Outpatient '/6.0% ~ Medical Sector (Overlan) _ /I Primary Care/Outpatient / Medical Sector Excludes overlap of an unknown percent of persons also seen in other sectors. NOTE: Data relating to sectors other than the specialty mental health sector reflect the number of patients with mental disorder seen in those sectors without regard to the amount or adequacy of treatment protruded. 1

43 Shifting perspective now to the primary care sector, we might expect the prevalence rate for mental disorder to match the overall population rate of approximately 15 percent. Indeed, reported rates are predominantly in the 10-20 percent range. 1/ But, as we have noted, the variation in reported rates is considerable. Let us look more closely at its methodologic sources. Major differences in reported rates can stem from the choice of Population base (total registered, total users, or consecutive users) from the time period (one point in time, several months, one year, or lifetime), and from the principal unit of analysis (number of persons, number of visits, or allocation of resources and costs for-the treat- ment of persons with mental disorders). Further, differences can stem from the choice of case identification method. Let us take, as an example, the effect of five methods of case identification on reported rates. year had a diagnosis clinical records. 3/ studies of four ___ between 1.3 and ~ Figure 2 shows some of the prevalence rate ranges in primary care settings, grouped by case identification method. Starting with the person-based, more epidemiologically oriented studies, the first method of case identification is routine reporting of mental disorder diagnoses on clinical records of general practitioners. Using this method of case identification, Fink, et al, found that about 5 percent of the Health Insurance Plan of New York (HIP) population in a given of mental disorder routinely recorded on their _ (We have recently repeated this approach on different health programs, demonstrating a range 6.3 percent across programs. 4/ These rates increase to a range of 1.5 to 8.2 percent when the population base is utilizing patients rather than the total population potentially using services; when only patients are used as the population base, the denominator is smaller and prevalence rates correspondingly increase.) a The second case identification method is the routine recording of diagnoses, symptoms, treatments, and referrals O Hoeper, et al, have recently completed a study, using a chart review of adult patients in a prepaid group practice at Marshfield, Wisconsin, which revealed a three- month prevalence rate of some 5 percent. 5/ This rate was somewhat higher than the 2.6 percent rate for the same setting found by routinely re- corded Section V diagnosis of mental disorder. 4/ The third method of case identification, which has received the most attention within epidemiological circles, is the use of survey report form for recording mental disorder diagnoses by general medical physicians. In the classic study of this type, Dr. Michael Shepherd reported that some 14 percent of the patients in 46 general medical practices in London were identified by GP's as having mental dis- orders. 6/ In this country, Locke, Goldberg, Rosen and others of the

44 Figure 2 CASE IDENTIFICATION METHODS IN PRIMARY CARE SETTINGS Person-Based Studies Routine recording of mental disorder diagnoses on clinical records --Fink R. et al: 4.8: --Regier DA, et al: 1.3~6.3: (1.5-8.21) Routine recording of diagnoses, symptoms, treatments, and referrals --Hoeper EW, et al: 5.0: Survey report form used for recording mental disorder diagnoses --Shepherd M, et al: 14.0: - Locke BZ, et al: 16. 9: Routine recording of mental disorder diagnoses on clinical records --NAMCS( 1975 ): 2 .1: --Regier DA, et at: 0. 4-4 .0: Visit-Based Studies Standardized psychiatric interview with general practice patients Rawnsley K: 22X --Hoeper EW, et al: 26. 7: Patient sel f-report on psychiatric symptom questionnaire Shepherd, M, et al: 20-36X —Pedder JR and Goldberg DP: 30% —Hoeper hW, et al: 30% Routine recording of mental health treatments --Bar ter MB, et al: 12: of vi sit s receive psychotropic drugs 18/ Division of Biometry and Epidemiology of NIMH have identified rather comparable rates. 7,8,9/ (However, since a patient population base and not a general population base was used, the rates are somewhat higher.) The fourth case identification method is a standardized psychia- tric interview with general practice patients. There have been several interviews of this type. The interview developed in the University of London, Institute of Psychiatry general practice unit by Drs. David Goldberg, Michael Shepherd, and others, 10/ has been used by Rawnsley to identify some 22 percent as having a mental disorder diagnosis in a primary care practice. 11/ Similarly, in their recently completed study in Marshfield, Wisconsin, Hoeper, et al, using the SADS-L standardized psychiatric interview, reported some 26.7 percent of patients identified with an RDC diagnosis of mental disorder. 5/ The final method of case identification is patient self-report on psychiatric symptom questionnaires. The Cornell Medical Index (CMI) was used in the original Shepherd study in which a range of 20-36 percent

45 scored positive on the MR section of the CMI. 6/ Other studies, using the General Health Questionnaire (GHQ) developed by David Goldberg, 12/ have found rates in the range of 30 percent. 13/ Hoeper and his colleagues found the same range when they used this instrument in the Marshfield area. 5/ In reviewing Figure 2, then, it is important to recognize that the rates tend to be higher with each successive case identification method. Elevated rates are also found when consecutive patients are used, and when only the patient population is counted in the denomina- tor. Shifting now to health services research data on the volume (and cost) of services, these tend to use visits as the unit of analysis. The most common utilization data available are from routine reporting systems on medical records. However, visit-based physician practice surveys are also used. The National Ambulatory Medical Care Survey (NAMCS) of U.S. office- based physicians identified a primary diagnosis of mental disorder in only 2.1 percent of visits to all nonpsychiatrist physicians. General practitioners had a slightly higher rate of 3 percent, internists 3.6 percent, and pediatricians a lower rate of 1 percent. 13/ A similar type of diagnostic reporting was also examined in four organized health care programs, in three settings under contract with NIMH. These settings included the Bunker Hill Health Center in Boston, the Columbia Health Plan in Columbia, Maryland, and the Marshfield Clinic, which has both a fee-for-service and prepaid practice program. The rates across settings ranged from .4 to 4 percent of visits. 14,15,16/ Because multiple measures were used in these settings, it is possible to compare the relationship between visit-based and person- based rates. In all cases, the percent of patients identified with mental disorder in one year is higher than the percent of visits for such diagnoses -- indeed, on the average, about two times greater (see Table 1~. This finding seems somewhat counterintuitive, considering that patients with diagnoses of mental disorder have much higher total visits to their physicians than those without such disorders. (In fact, in the four plans that we studied, patients with mental disorder averaged from 1.4 to 2 times as many visits per patient per year as patients without mental disorder diagnoses.) (See Table 2) However, although patients with mental disorder diagnoses visit more frequently, most of their visits are for diagnoses of other medical illnesses, a fact con- sistent with the finding that patients with mental disorders have higher morbidity rates for all other categories of medical illness. 17/ The interaction among different case identification methods and their relationship with utilization indices may be illustrated by a case example involving the Marshfield Clinic, in Marshfield, Wisconsin. In this large clinic, where multiple measures of primary care practice activities were used, it was found that 2.7 percent of the visits had

46 Table 1. Percent of Visits and Patients, in Non-Mental Health Departments with Diagnoses of Mental Disorder, By Setting, 1975 Percent Visits Associated With Ra tio of Setting Mental Disorder Diagnosis Patients/Visits Percent Patients Columbia Medical Plan 0.4 1.5 3.8 Marshfield Clinic Prepaid 2. 7 3. 7 1.4 Fee for Service 2.3 4.0 1.7 Bunker Bill Health Center 4.0 8.2 2.1 Table 2. Mean Visits to Non-Mental Health Departments Per Patient With and Without Diagnosed Mental Disorder, By Setting, 1975 Setting Mental Disorder Diagnosis Ratio of Means: Present/Absent Present Absent . Columbia Medical Plan 7.1 5.0 1.4 Marshf ield Clinic Prepaid 7.7 4. 2 1.8 Fee For Service 6.5 3.6 1.8 Bunker Hill Health Center 6.4 3.2 2.0

47 a principal diagnosis of mental disorder. This represented 3.7 percent of patients aged O to 65. Patients with mental disorder diagnoses averaged almost twice as many visits per year as patients without such disorders, accounting for 7 percent of the total visits. However, only slightly more than one-third of their visits were for a diagnosis of mental disorder. In a recent three-month survey of adult members of this same popu- lation, it was found that 5 percent of consecutive patients could be diagnosed as having a mental health problem if prescriptions for psychotropic drugs and recorded emotional symptoms are included. Although no survey per se of the Marshfield GP's was performed, such as has been done by Shepherd, Locke, and others,-the GHQ was administered as a patient self-report form, and a standardized psychiatric interview (SADS-L) was conducted with a sample of patients. Preliminary findings indicate that 30 percent of the patients scored positive on the GHQ and approximately 27 percent received diagnoses of specific mental disorder on a standardized psychiatric interview with mental health specialists. Thus, results from the Marshfield study, in which multiple case identification techniques were used in one site, confirm the method- dependent effects noted earlier across many studies and sites: relatively low rates when GP reporting is used, and relatively high rates when self-reports of standardized psychiatric interviews are used. Elucidating precisely the prevalence of mental disorders in primary care practice is but one of the important research tasks required to guide future policy and program development. It is also important to determine the diagnostic and treatment needs of persons identified by different methodological approaches. Likewise, it is necessary to determine both the effectiveness and cost-effectiveness of services provided to such patients in either the primary care or the specialty mental health referred settings. If we are to integrate health and mental health services further, or simply to pay more attention to the role of primary care practi- tioners in treating persons with mental disorder, we will need, first, to have more data on the specific types of disorders they tend to identify, misidentify, treat, decide not to treat, or decide to refer. Second, we need to link the data on specific disorders with service utilization, cost, specific treatments, and outcome. Third, the effect of training methods on improving accurate identification and effective treatment also needs further study. If the primary care sector is to be a full partner in the treat- ment of patients with mental disorder, primary care physicians will need all of the tools of the specialty sector, in somewhat modified form, to function effectively. These include: a classification system for psychosocial problems and mental disorders oriented to the level of specificity expected of a primary care provider; a method of case

48 identification which includes elements of symptom questionnaires in routine history taking; and more detailed structured interviews which can be used when there is doubt about diagnosis or treatment. In short, a combination of descriptive, analytic, and methodolo gically oriented studies is needed to improve understanding of the mental health service role of primary care providers, to aid them in carrying out that role, and to guide the informed development of services policy related to that role. We have a good initial effort; it must be sustained. -

49 References Hankin, J., Oktay, J.S.: Mental Disorder and Primary Medical Care: An Analytical Review of the Literature. Rockville. DHEW Publication No. (ADM) 78-661, 1979. 2. Regier, D.A., Goldberg, I.D., Taube, C.A.: The de facto U.S. mental health services system. Arch. Gen. Psychiatry 35:685-693' 1978. 3. Fink, R., Goldensohn, S., Shapiro, S., and Daily, E.: Changes in family doctors' services for emotional disorders after addition of psychiatric treatment to a prepaid group practice program. Medical Care 7:209-224, 1969. 4. Regier, D.A., Goldberg, I.D., Burns, B.J., Hankin, J., Hoeper, E.W., and Nycz, G.R.: Mental health services in four integrated health/ mental health settings. Presented at Annual Meeting of the American Psychiatric Association, Atlanta, May, 1978. 5. Hoeper, E.W., Nycz, G.Ro ~ Cleary, P.D.: The Quality of Mental Health Services in an Organized Primary Care Setting: Final Report, Marshfield Medical Foundation, Marshfield, Wisconsin (unpublished), 1979. 6. Shepherd, M., Cooper, B., Brown, A.C., Kalton, G.W.: Psychiatric Illness in General Practice. London, Oxford University Press, 1966. Locke, B.Z., Gardner, E.P.: Psychiatric disorders among the patients of general practitioners and internists. Public Health Reports, 84:2:167-173, 1969. 8. Rosen, B.M., Locke, B.Z., Goldberg, I.D., Babigian, H.M.: Identifi- cation of emotional disturbance in patients seen in general medical clinics. Hosp. and Community Psychiatry, 23:364-370, 1972. 9. Locke, B.Z., Krantz, G., Kramer, M.: Psychiatric need and demand in a prepaid group practice program. Am. J. Pub. Hlth. 56:895- 904, 1966. 1O. Goldberg, D.P., Cooper' B., Eastwood, M.R.., Kedward, H.B., Shepherd, M.A.: Standardized psychiatric interview for use in community surveys. Brit. J. Prevent. and Soc. Med. 24:18-23, 1970. 11. Rawnsley, K.: Congruence of independent measures of psychiatric morbidity. J. Psychosomatic Rsch. 10:84-93, 1966.

50 12. Goldberg, D.P.: The Detection of Psychiatric Illness Institute of Psychiatry, Maudsley Monographs No. 21. London, Oxford University Press, 1972. Questionnaire. - 13. Pedder, J.R., Goldberg, D.P.: A survey by questionnaire of psychiatric disturbance in patients attending a venereal disease clinic. Brit. J. of Venereal Disease, 46:58-61, 1970. Jacobson, A.M., Leet, R., Goldner, N.: 14. Burns, B.J., Orso, C., Utilization of Health and Mental Health Outpatient Services in Organized Medical Settings: Final Report, Bunker Hill Health Center of Massachusetts General Hospital, 1978. (Unpublished). 15. - Shapiro, S., Hankin, J., Steinwachs, D.M.: Utilization of Health and Mental Health Outpatient Services in Organized Medical Care Settings: Final Report, Columbia Medical Plan, Health Services Research and Development Center, The Johns Hopkins University, Baltimore, Md., 1977. (Unpublished). 16. Hoeper, E.W., Nycz, G.: Utilization of Health and Mental Health Outpatient Services in Organized Medical Care Settings: Final Report, Marshfield Clinic, Marshfield, Wisconsin, 1977 (Unpublished). 17. Eastwood, M.R.: Toronto: University of Toronto Press, 1975. The Relation Between Physical and Mental Illness. 18. Baiter, M.B.: Coping with Illness: Choices, alternatives and consequences, in Drug Development and Marketing. Helms,R.B. (ed.), American Enterprise Institute for Public Policy Research, Washington, D.C., 1974, pp. 27-46.

51 A NEIGHBORHOOD HEALTH CENTER MODEL OF INTEGRATED AND LINKED HEALTH AND MENTAL HEALTH SERVICES Barbara J. Burns, Ph.D.* Research Psychologist, Primary Care Research Section, Applied Biometrics Research Branch, Division of Biometry and Epidemiology, National Institute of Mental Health Rockville, Maryland Efforts to provide more closely integrated health and mental health services particularly within primary care settings have taken a number of forms in community and neighborhood health centers and health maintenance organizations. A current Federal emphasis encourages closer ties between existing neighborhood or community health centers (CHCs) and community mental health centers (CMHCs). A description is provided of one organized primary care setting, a neighborhood health center (NHC), with integrated mental health services from its inception which are linked to a community mental health center. Despite its unique features, this particular example can serve as one model for the delivery of mental health services in a general health care context while maintaining a close working relationship with the specialty mental health sector. A brief description of the organization will be followed by reported patient benefits and provider issues experienced in this setting. THE MODEL The focal point of this report is the Bunker Hill Health Center (BHHC) of the Massachusetts General Hospital, a NHC serving a primarily low-income Irish-American community of about 17,000 persons who live in Charlestown, Massachusetts, a relatively isolated section of the city of Boston. Following assessment of the community's health and mental health needs, BHHC was opened in 1968 under the strong leader- ship of an Irish pediatrician. The Center is a satellite of the Massachusetts General Hospital (MGH), a teaching hospital in Boston. Its stated goals were to provide comprehensive, coordinated, continuous, personalized, non-fragmented, family-centered health care at a reason- able cost. As a physical extension of the MGH, the Center offered . *In collaboration with Darrel A. Regier, M.D'., M.P.H., Director, Division of Biometry and Epidemiology, National Institute of Mental Health, Rockville, Maryland.

52 medical, pediatric, mental health, social service, dental, nutritional and other specialist services on a scheduled appointment and walk-in basis, with back-up from the MGH emergency room and inpatient services. The mental health services included evaluation, individual and group therapy, day treatment, emergency services, pre-school screening, consultation to community agencies, and community-based prevention programs. In 1975, the mental health staff made up 44 percent of total staff. A more extensive description of the services offered can be found elsewhere. 1/, 2/ The community mental health services were originally designed, and have been continuously delivered, as an integral and essential component of this comprehensive health program. Since 1971, when the catchment area CMHC (the Erich Lindemann Mental Health Center) was founded, BHHC was designated as the provider of outpatient mental health services for Charlestown. Such a model has been characterized by Borus as a "joint endeavor model" in which a health center's mental health program is supported by both the NHC and the CMHC. 3/ Specifically, the CMHC provided the NHC with staff for specialized day care programs for chronic patients and a child psychiatrist consultant to the Center's child mental health program. Affiliation with the CMHC also increased the center's access to other sources of program funds. Organizationally, although there was no interlocking directorate between the NHC and the CMHC (a potential disadvantage to consumer groups), there was considerable administrative and staff interaction among the directors of the CMHC, BHHC, and the BHHC Mental Health Unit The organizational linkages between the NHC and the CMHC fostered many types of interaction around the provision of secondary and tertiary mental health services by the CMHC for patients seen by the NHC. Fre- quent communication between the BHHC and the CMHC developed around coordination of specialized services including those for CMHC inpatients, and the mental retardation and children's programs. Within the neighborhood health center, there was active interaction between general health and mental health providers. Consistent with the stated BHHC goals to provide comprehensive and coordinated care, staff were recruited with a strong commitment to and capacity for inter- disciplinary teamwork. The single medical record and multi-disciplinary health care teams were instituted from the beginning. A number of vehicles for communication developed somewhat sequentially with the growth of the program and understanding of the community, e.g., con- sultation, inservice education, and collaborative programs. These appraoches, undertaken to facilitate health/mental health provider communication, were intended to strengthen the mental health skills of health providers and to facilitate appropriate referral of patients to mental health specialists.

53 FIGURE 1. APPROACHES FOR HEALTH/MENTAL HEALTH INTERACTION AT THE PROVIDER LEVEL Vehicles for Health/Mental Health Coordination Examples of Applications 1. Single Medical Record 2. Multi-discipline health care team 3. Ease of referral to mental health services 4. Case-centered consultation 5. Center Committees 6. Inservice education 7. Joint patient interview 8. Collaborative program 9. Client - centered consultation 10. Liaison to units 11. Collaborative research Formal communication among providers Multi-problem families with health and mental health problems are frequently presented Open intake system by mental health service Health provider requests for advice on managment of patients Medical records, various task forces Series on alcoholism, child development, infant-parent relationships Modeling assessment approaches for health providers Pre-school screening, obesity group, medication clinic Pediatric nurses counseling adolescent mothers meet with psychiatrist regularly A pediatrician attends the child mental health intake meeting A study of internists' patterns for prescribing tranquilizers

54 BENEFITS OF THE MODEL Through an NIMH contract, it was possible to study the utili- zation of health and mental health services at Bunker Hill Health Center for persons with and without diagnosed mental disorder for the year 1975. 4/ That study permitted us to document some of the advantages to patients, 9,233 of whom used the Center in 1975 for a total of over 53,000 visits. First, accessibility and accepta- bility of mental health services delivered by both health providers and specialty mental health providers within BHHC were demonstrated. During 1975, 15.7 percent of Health Center patients received a provider-reported diagnosis of mental disorder, a high rate as compared with other studies of routine recording of mental disorder by providers. 5/, 6/ Health providers identified about half of these patients while mental health providers diagnosed two-thirds (the overlap is accounted for by patients seen and diagnosed by both provider types). Second, comprehensive health care was provided for persons with mental disorder in the sense that 86 percent of such persons received some health services at the Center during the year. The availability and use of health services are particularly important for this affected population since relatively higher medical morbidity was found among persons with mental disorder than those without mental disorder (also reported in the literature 7/, 8/~. Third, specified target groups, namely children and parents, received a high level of specialty mental health services. Among the 1975 Health Center patients from Charlestown, high rates of utilization were seen, e.g., 17 percent of the boys in the 5-14 year age group and 23 percent of women in the 25-44 year age group (many of whom were single parents) were seen in one year in the Mental Health Services. Fourth, patients with chronic severe mental illnesses were served extensively. Persons diagnosed with schizophrenia, affective psychoses and other psychoses comprised l9 percent of those seen by the mental health providers for mental disorder diagnoses, and accounted for 29 percent of the visits to the Mental Health Unit. The latter finding, which is unusually high for a health center, is largely a function of the affiliation with the CMHC and participation by these patients in chronic care programs at BHHC. The linkage with the CMHC also offered these patients numerous other advantages including easy access to inpatient care, sheltered workshops and other specialized mental health programs. Fifth, that specialty mental health outpatient services were delivered at a reasonable cost (an average of $345 per patient per year, which is appreciably less than the $500 required Massachusetts

55 outpatient mental health insurance) tends to give support to the economy of this form of organization. Indeed, 80 percent of the patients seen had their mental health services fully covered at a cost below the $500 limit. ISSUES ASSOCIATED WITH THE MODEL Despite the long heritage of ideological and other barriers separating the health and mental health service systems and their personnel, 9/ it was possible in the BHHC experience to develop a staff that viewed mental health as integral to general health. None- theless, certain tensions and differences along discipline lines became manifest as the organization grew and had to undergo financial and management changes. The initial teamwork served to develop trust and differentiate expertise among the various types of health and mental health providers. Initial limited emphasis on specialist departments or units resulted in close working relationships with minimal boundaries among the various provider types. Yet, working side by side, health and mental health provider differences in pace, role and approach were revealed. For example, health providers tended to see two or three times as many patients a day as mental health providers while the latter spent more time in the community in outreach and agency consultation activities. Although the objectives of both provider groups were to improve the overall health of the community, the approaches varied considerably. Frustrated with the length of mental health treatment for chronic men- tal illness and long-term psychosocial problems, some health providers felt that mental health providers should spend less time with patients and be more authoritative and directive. In later years, as a result of both organizational growth and de- creased grant support, other issues developed. The multi-disciplinary team became too large to function as an efficient vehicle for case coordination. Although other communication vehicles emerged, increased pressures to generate fees reduced the time available for health and mental health provider interaction. These factors contributed to isolation of health and mental health providers until a combination of raising fees and generating further grants and contracts occurred. Other issues arose associated with the special relationship that the Mental Health Unit had to the CMHC. In some respects the mental health services were able to grow more rapidly due to the potential to expand programs through CMHC resources. However, the short-term nature of such staffing or funding created a source of anxiety within the Health Center. Further nonreimbursed activities associated with the mental health unit's involvement with the CMHC (e.g., attending policy and planning meetings concerning the catchment area program),

56 became an issue. In general there was some feeling that the availability of CMHC resources was not entirely predictable (precipitated by several long-term budget freezes), and that the Center involvement was costly in staff time. CONCLUSIONS AND IMPLICATIONS Despite the special organizational problems associated with integra- tion and linkage of health and mental health services, the potential to resolve them was facilitated by the physical proximity of health and mental health providers who were by necessity forced to work together around clinical, training and research matters within a relatively small integrated organization. The pressures created by bringing the health and mental health service delivery systems closer together require continuous attention in order to derive benefits for patients. After more than ten years the delivery of mental health services has continued to increase and the health/mental health provider interface to diversify creatively. Although the Bunker Hill Health Center may not be typical of community or neighborhood health centers with respect to the extensive integration of its outpatient health and mental health services, or the level and complexity of linkages influencing the health/mental health interface, at least some of the following recommendations which emerged for this experience should be applicable to other settings and populations: A scarcity of mental health resources, a high level of need for mental health services within a general medical population, and the availability of new resources combine to make a health/ mental health effort feasible. As in all organizational or programmatic innovations, the support of strong leadership, and sanction by the boards (of directors or trustees) are a minimum essential. Building intra-agency and interagency linkages is a difficult process which demands great political skill and can best be approached on an incremental basis. 10/ Specification of a limited set of health/mental health service objectives that can be measured or examined can serve to initiate and continue the dialogue between health and mental health forces. The selection of health and mental health staff with a holistic approach to medicine which for both types of providers includes a biopsychosocial orientation can contribute to minimizing

57 initial conflict and contribute to perceiving a joint mission in behalf of patients. A view of mental health as integral to health is an important philosophical basis. The development of active and systematic communication at the board, administrative, program, and provider levels can serve to cement the joint endeavor and facilitate cooperation. Shared authority between systems and among sub-systems requires con- siderable good will, education and effort to reduce conflict and keep the focus on service to patients instead of process. Initial clarity of the agreements for fiscal support of the mental health services can provide some "assurance" that as the health center mission evolves, mechanisms for sup- porting mental health services in a health care system will also continue to evolve. Physical proximity alone (co-location) is not sufficient to change attitudes or practice; diligent and responsive attention to the critical health/mental health interface activities around clinical, training and evaluation/research - functions is necessary to sustain healthy interaction and to promote growth.

58 REFERENCES Jacobson A.M., Regier, D.A., Burns, B.J.: Factors relating to the use of mental health services in a neighborhood health center. Public Health Reports 93:232-238, 1978. . 2. 4. 5. Orso, C.L.: Delivering ambulatory health care: The successful experience of an urban neighborhood health center. Medical Care 17:111-126, 1979. Borus, J.F.: Neighborhood health centers as providers of primary mental health care. New England Journal of Medicine 295:140- 145, 1976. Burns, B.J., Orso, C.L., Jacobson, A.M., et al: Utilization of health and mental health outpatient services in organized medical care settings: Bunker Hill Health Center of the Massachusetts General Hospital. Final report submitted in fulfillment of NIMH Contract No. 278-76-0027 (DB). 1977. Coleman, J.V., Patrick, D.L.: Psychiatry and general health care. American Journal of Public Health 68:451-457, 1978. 6. Hankin, J.R., Burns, B.J., Goldberg, I.D., et al: The utilization of medical care services by persons with mental disorder. Pre- sented at the annual meeting of the American Psychiatric Associa- tion, Atlanta, Georgia, May, 1978. Eastwood, M.R., Trevelyan, M.H.: Relationships between physical and psychiatric disorder. Psychological Medicine 2:363-370, 1972. 8. Matarazzo, R.G., Matarazzo, J.D., Saslow, G.: The relationship between medical and psychiatric symptoms. Journal of Abnormal and Social Psychology 62:55-61, 1961. 9. Goldman, H.: The integration of health and mental health services: a sociohistorical perspective. An Invited Background Paper for the Institute of Medicine Conference, The Provision of Mental Health Services in Primary Care Settings, April 2-3, 1979. 10. DeWitt, J.: Managing the human service "system": What have we learned from services integration? Human Services Monograph Series 4, 1977.

59 DELIVERY OF MENTAL HEALTH SERVICES IN A FEE-FOR-SERVICE PEDIATRIC PRACTICE Lawrence Pakula, M.D. Timoni~,m, Maryland I am a board certified pediatrician, the senior pediatrician in a group of three who practice as a fee-for-service group. We have a strong orientation toward behavioral pediatrics. The practice pro- vides well child care, acute illness care--both in the office and in the hospital-- and what we call consultation care. Consultation care involves spending from one-half to two hours with a family, child, school, or other pertinent organization. Charges for the service are based primarily on time spent and services rendered. The requirements necessary for the provision of adequate mental health services in a primary care setting are adequate training for the provider, allotment of significant time to provide the service, and~appropriate remuneration for the time spent and the quality of service offered. In my case, the subsequent training after pediatrics residency and my Air Force tour of duty consisted of a one-year fellowship, entitled "The Psychiatric Aspects of Pediatrics," with Dr. Leon Eisenberg at the Children's Psychiatric Service at Johns Hopkins. It was literally the only place that I could find in 1962 that would allow a pediatrician to work fulltime without two years of prior adult psychiatry training and at least an implied commitment to complete child psychiatry training. It should be remembered that 15 years ago pediatricians, general practitioners, and internists were not supposed to need this training and were even discouraged from pursuing it. Each of my associates has been a fellow for one year in the Child Psychiatry program, receiving similar training following pediatric residency. In addition, I have been involved actively in teaching programs in the Department of Pediatrics and the Children's Psychiatric Ser- vice at Johns Hopkins and as a consultant in Special Schools of the Board of Education of Baltimore County. In each of these situations I have been identified as a provider of primary care who is interested in and capable of providing mental health services in addition to pediatric services. This has further served to make our patients aware of our complete range of services and interests. We are not participants in any insurance plan, including Blue Shield of Maryland.

60 To help control costs, we each now generally schedule consultations at a time when at least one other physician is in the office. This serves to relieve some of the economic pressures on the office. It also serves to eliminate the pressures of answering the tele- phone and handling acute problems that many primary care physicians complain interfere with their delivering mental health services. Initial screening of the consultation request may be done by either the nurse or the physician. All requests for consultations, however, are discussed with the family by the physician prior to the arrangement of the appointment. This allows for more appropriate scheduling, and increases the likelihood that all pertinent parties and material will be available at the time of the meeting. The entire session may be devoted, as is appropriate, to history taking, discus- sion and interpretation, and even physical and neurological examination in some cases. In addition, when appropriate, relatively simple screening tools such as the Peabody Picture Vocabulary IQ Test, Goodenough Draw-A-Man Test, and Gray Oral Reading Test are utilized. Often a disposition can be made after the initial interview. In my case, typically the initial interview lasts an hour-and-a-half. Appro- priate follow-up by us also usually is arranged at that time. We do not employ or have formal arrangements or association with any psychiatrist, psychologist, social workers, or other mental health specialists. We have done this to avoid the appearance of conflict of interest and to take advantage of the variety of both public and private services available in the community. For example, for school age children, the services of the Division of Psychological Services of the Board of Education of Baltimore County might be utilized. In the near future, we may consider hiring a psychiatric social worker who has considerable experience in medical and school settings as an addition to our staff. An active approach to mental health delivery in the primary care setting begins with the first well child visits. Interpreta- tion of parental concerns and observation by the physician is an integral part of the program. When we feel it is appropriate, we do ask families to spend additional time and money in consultation with us. Acceptance of this approach, we believe, is based both on the willingness of the physicians to discuss this type of infor- mation openly and on the parents' belief in the competence of the medical care being offered. Because of the way the practice is structured, we undoubtedly have attracted and lost some patients primarily because they are sympathetic or unsympathetic to this approach. It should also be noted that each physician has a calling hour in the morning at home beginning at 7:15 a.m., during which any subject may be discussed.

61 Some typical areas of concern brought to us in our consultation times and other settings have included: (1) Adjustment of family members to pregnancy and birth, including sibling rivalry and postpartum depression; (2) effects of children on marital relationships; (3) divorce and other marital problems as they affect children emotionally and physically; (4) school problems; (5) parental problems such as illness, death, alcoholism, separation, and unemployment; and (6) anorexia nervosa as a medical and psychiatric problem. We typically have referred patients to other professionals under a limited variety of circumstances. The primary reason for referral has been unusual severity and/or chronicity of the problem. On some occasions, however, it has been deemed advisable to refer to a mental health professional in order to maintain effectively our role as primary care physicians. We often have done rather extensive workups prior to referral sometimes out of necessity, when the family was unwilling to accept such a workup in any other setting. The feedback to us is that, in comparison with the typical referrals from primary care settings where such screening is omitted, the patients we refer appear to be more appropriate cases, are better motivated, and are more likely to follow through on recommended therapies. Our impression is that this system of pre-evaluation represents an improved investment in time and money, with results that usually are satisfactory from all viewpoints. The patient with significant underlying medical problems usually seems to need referral, if warranted, either to a psychiatrist with pediatric and/or adolescent training or to someone with ready access to specialized medical support or an unusual willingness and ability to maintain a close follow-up relationship with us. Failure to find such qualified referral resources generally leads to early termination of treatment, unsatisfactory outcome, and decay in the relationship between the primary care physician and the patient. We have found that non-medically oriented or related health agencies find it difficult to handle patients with underlying medical problems. The inability to understand the possible interaction of the disease pro- cess and mental health concerns, combined with their fear of having unfavorable effects on the medical aspects of such cases, leads to impotence or at least to lowered effectiveness. Community mental health clinics have not provided a consistent, reasonable or practi- cal referral source for patients from private practice fee-for-service

62 settings in our experience. Despite the screening of patients, family motivation, and physician intervention, there tend to be long delays and a lack of ongoing professional, personal involvement. The limited treatment schedules apparently often are predetermined by administrative concerns. Financing care for these services does present a problem for the less well motivated, less interested, and less financially able families. However, we find that those with some of the -most severe financial limitations are at least willing to spend both the time and money for the initial evaluation, if they believe that it is beneficial to them and their children and that real help will be offered in overcoming their problems. This is true even in cases where families feel that they could not continue with more extensive consultations. Most parents, when presented with the need for this type of care and the possibility that their health insurance will pay for at least part of it, report ignorance of the coverage and even ignorance about how to obtain appropriate information. Unfortunately, these results seem to vary only slightly when the family itself or an employer is responsible for the premiums. Again, insurance coverage does not seem to be the determinant to us in initial use of consultation time but does seem to affect long-term use of such services . The Task Force on Pediatric Education reports that insurance coverage on children has improved, and that 90% of families have some hospital insurance for their children. But 65% of families still have no insurance covering office visits to a physician. Since policies are generally procedure oriented, counseling and psychological services are usually excluded. Eighty percent of pediatricians report that they offer services somewhat similar to ours, to at least some extent; however, less than half of that 80% report reimbursement for the time spent, and only 5% were always reimbursed in proportion to their time. The parent or guardian was the most common source of the payment. We unfortunately have not carefully kept statistics on the patients for whom we have provided consultative services. However, we believe that a least 80 to 90% of our patients complete the need for consultative services with us in a formal way, during 3 to 5 visits at most. The majority, in fact, require only one extensive visit. Problems requiring more time probably would include persistent school management problems, particularly those involving the use of medications; divorce and the ongoing effects of visitation, custody and other disputes, and their role in physical and mental health; and psychosomatic illness or physical illness with major emotional components.

63 Although there are several problems associated with this type of practice, I must reemphasize that it is possible for even the solo practitioner to adapt this type of service to his practice. However, unless time is allowed by the primary care provider and he or she acknowledges the value of and the reponsibility for such services and also receives adequate remuneration, consultation care will not be offered. Even in a group situation, it may not be possible to offer these services. Unless there are other primary care providers either offering similar services or spending time in some other relatively time consuming aspect of care, i.e., specialty care of some type, the atmosphere may be hostile to offering this service to patients. The cry often is: "I am seeing all of the patients while you see 'those people'." Usually the offer to share patients is passed up. Young pediatricians who have spent an extra period of time in training in the behavioral aspects of pediatrics and wish to add this option to their practice indicate that they have found themselves un- welcome in groups where the idea of setting aside extra time is not practiced. I have had an opportunity to compare notes with practitioners in a variety of circumstances, including EMOs; this is a common experi- ence even there. We have had only limited experience competing with an HMO. There are several now in our areas and we have had some families join them. However, some of these families for whom we provided consulta- tion care in the past have returned for such care from us, with fees paid directly by the families. We do not have any statistics concerning this situation. We often find out about families joining HMOs through requests for transfer of records. It has been our impression that mental health services are most appropriately offered in a primary care setting. The require- ments to make this a viable option are attitude and training of the provider, appropriate scheduling, and adequate remuneration. Aware- ness of these needs and requirements should extend down to the levels of medical students and house officers. There are other added advantages for the primary care provider. This system allows for the delivery of more complete, appropriate care and makes primary care a more rewarding and fulfilling profession. Mental health care cannot be solely the responsibility of psychiatrists, psychologists, and mental health counselors. It is an integral part of reasonable, even minimal, primary care. It must be recognized as such.

65 TREATMENT OF EMOTIONAL PROBLEMS BY PRIMARY PHYSICIANS IN AN HMO Jules V. Coleman, M.D. Community Health Care Center Plan of Greater New Haven The Community Health Care Center Plan of Greater New Haven (CHCP) provides the mental health services required of a federally qualified HMO to its enrolled membership of about 25,000. These include up to 20 visits to a mental health clinician and unlimited visits to a primary physician each year for the treatment of emotional complaints. The mental health services are incorporated in the primary care delivery system which is without limits on the number of total visits available to the enrollee and which can draw on the secondary or tertiary specialty services by referral. The provision of primary health care at CHCP is organized around small primary care teams, each including two physicians, a physician assistant or nurse practi- tioner, two medical assistants, a non-medical mental health clinician and a receptionist. 1/ Team mental health workers are psychiatric social workers, a clinical psychologist or a psychiatric nurse practitioner. They are trained psychotherapists. With the Plan psychiatrists, who provide essential back-up support to primary care and mental health clini- cians, they constitute the mental health service at CHCP. There is no separate psychiatric clinic. One way to visualize the program is to think of it as analogous to a community clinic, with the enrol- lees constituting the community, and the clinic functioning simul- taneously in two interdigitated fields of operation, primary care and mental health, each providing necessary contributions to the other for optimum patient care. We have found that 72% of our patients with emotional problems are cared for by primary physicians alone, and are not seen by mental health personnel. The organizational model for mental health services at CHCP recognizes and supports the key role of primary physicians as mental health providers by including mental health clinicians as members of the primary care team. 2/ Psychiatric services have traditionally developed in separate care systems, such as the public and private mental hospitals, general hospital psychiatric services, child guidance clinics and community mental health centers. HMO's have followed this tradition and set up

66 separate psychiatric clinics as referral resources within their organi- zations, or purchased specialized psychiatric services out of plan. The program at CHOP breaks with this tradition. It follows a basic principle of community psychiatry, that mental health services be integrated within an organization in such a way that they further its functional goals. It subscribes to a concept of psychiatric collaboration with health, welfare or education organizations which themselves serve ego-integrative and illness-preventive functions in addition to their specific human services. The CHOP program in psychiatry is thus oriented to the human problems of patients under the care of primary physicians, as well as to patients referred to mental health clinicians for specialized care. Mental health problems require the resources of many professional groups within the health field as well as in other social support agencies. However, most non-medical resources are utilized in a haphazard way by selected population groups. The medical establish- ment is the only social resource potentially available for the care of health needs of all segments of society, including all age groups, all social classes, and all ethnic groups. In our view, the inclusion of mental health services as an integral component of primary health care offers the only feasible means of substantially improving the distribution of mental health services for the population as a whole. Separation of psychiatry from the mainstream of primary medicine has resulted in the exclusion of large sectors of the population at risk and in need of mental health care. 3/ In our national health care program, primary care physicians occupy a strategic and central position in the delivery of all health services, including psychiatric. Psychiatric liaison services in general hospitals are now firmly established; similar services for patients in ambulatory health settings, although less developed, are rapidly expanding. The present challenge is to explore methods of psychiatric liaison with primary physicians which will be helpful to ambulatory patients with emotional disorders. Given the accessibility of primary physicians to patients with a broad range of emotional disturbances and life problems, it is inevitable that the largest share of responsibility for their care now rests with these physicians, and must continue to do so. For most patients, the decision to seek help for emotional problems in a primary care setting is acceptable when referral to a psychiatric facility is not. All health delivery systems rest on a primary care foundation. Despite the great emphasis on specialization in this country since World War II and the decline in the number of general practitioners, a primary care system has been retained in the practice of general internists and pediatricians -- with or without general practitioners as family physicians. In the further development of our national

67 health program, a first principle must be integration of all medical services in relation to primary care, since the fragmentation of medical care can be avoided in no other way. The development of prepaid, group medical practices creates a large number of natural laboratories for the study of approaches using mental health clinicians as extenders of primary health care. At CHCP, most of the non-medical mental health clinicians are psychia- tric social workers, but we have included a clinical psychologist on one team in medicine and a psychiatric nurse practitioner on another. We have found that professional differences become less important than the functional similarities shaped by similar job demands. The enlarged manpower pool thus provided is important if the collaborative model becomes widely accepted. CHCP studies of the role of primary physicians in mental health care. Prevalence studies of patients with emotional problems at CHCP have paralleled the findings of other studies elsewhere. We found that in a recent period 15.7% of patients presented such prob- lems. However, we also found that patients with emotional problems made 27.6Z of all visits of CHCP for all medical and mental health reasons. We found that patients with emotional problems made three to four times as many visits to the Plan as patients without emotional problems. 5/ It was also noted that 9% of all prescriptions filled by the CHCP pharmacy were for psychoactive drugs, which 25% of all patients received. One might assume that mental health clinicians saw patients with more serious psychiatric diagnoses or more severe symptoms. Overall this was not the case. Primary physicians alone saw a greater per- centage of patients in almost every diagnostic category, including the psychoses. Whether a patient was referred to a mental health clinician was not dependent on the diagnosis but on special circum- stances such as unusual pressure or time demands patients put on the primary physician; complicated personal, marital and family situ- ations requiring the psychosocial skills of the mental health clini- cian; the failure of a patient to improve; legal requirements that a patient be seen by a psychiatrist, and so on. It might be noted that mental health clinicians saw 4.4% of all patients who visited the Plan, and 287 of patients with emotional problems. We should note that in a separate study of patients with chronic emotional problems, we found that mental health clinicians saw a greater number, 53.8% compared to the 46.2% seen by primary physicians alone. _/ Very few patients with emotional problems of any kind were seen by mental health clinicians alone. They continued to be carried by their primary doctors for physical complaints, and in general made twice as many visits to the Plan for medical as for mental health reasons. An important aspect of the mental health work

68 of an HMO is the lack of limitation on long-term general health care. At CHCP, primary and mental health clinicians are partners in such care. The opportunity to consult with mental health clinicians formally, informally, day or night, with continuity of care assured the patient and his family as long as they are members of the Plan, provides the primary physician with a built-in structure of institutional support in dealing with the emotional problems of his patients. We regard the treatment of emotional problems as a process which unfolds over a period of years, with multiple care-givers providing patients with a gradually secured base of personal stability. Some problems and issues in the integrated program. We see a number of organizational and management problems. Whatever mental health program is adopted at the start of a health care plan tends to become institutionally entrenched, with resistance to change. If change is to occur, it would require incentives offered to administra- tors and practitioners. In this connection, small demonstration programs might be helpful. Another issue is how much responsibility the primary physician takes for patients with severe or demanding problems. This gets worked on over time with the development of mutual understanding between primary care and mental health providers of each other's capabilities and limitations, and understanding which may get sorely tried in unusual cases, and which is constantly retested and renewed. The ratio between primary care and mental health caseloads is important. The number of mental health clinicians and the amount of time they spent in seeing patients affected how much primary care treatment of troubled patients was undertaken. Continuing health- mental health collaboration was also a factor. We conclude that staf- fing patterns, caseloads, physician interest, and overall program arrangements were important for the demonstrated increases in primary care treatment of emotional problems. Staffing patterns and staff-patient ratios have remained fairly constant at CHOP over the years. This is reflected in expenditures for mental health personnel per patient per month. The in-Center cost for mental health services for the fiscal year 1974-1975, exclu- sive of the cost associated with primary care clinician treatment, was $.55 per enrollee per month. 2/ The cost in the current fiscal year is $.71 per enrollee per month, but this includes costs for all out-of-Plan psychiatric services, such as hospital emergency room charges and physician charges to hospitalized patients, but not direct hospital charges. The latter were $.55 in 1974-1975 and $.51 in this fiscal year. Expenditures have thus not changed appre- ciably. There is an artifical separation between the mental health practice of primary physicians and psychiatrists. It is assumed that a patient may be regarded as a psychiatric case only if he

69 sees a psychiatrist. We have demonstrated that the diagnostic categories of patients seen by primary physicians do not differ from those seen by psychiatrists. The mental health program at CHCP attempts to bridge the gap between the two approaches to the care of patients with emotional problems.

70 References Coleman, JV, Patrick, DL and Eagle, J. THE PRIMUS PROJECT. Primary Care and Mental Health Services. A description of the service and research programs of Mental Health and Psychiatry. Community Health Care Center Plan, 150 Sargent Dr., New Haven, CT 06511, 1976. 2. Patrick, DL, Eagle, J and Coleman, JV. emotional problems in an HMO. 3. Primary care treatment of Medical Care, 16:47-60, 1978. Coleman, JV and Patrick, DL. Integrating mental health services with primary medical care. Medical Care, 14:654-661, 1976. 4. Glasser, MA, Duggan, TJ and Hoffman, MA. Obstacles to utilization of prepaid mental health care. Am. J. Psychiatry, 132:710-715, 1975. 5. Patrick, DL, Coleman, JV, Eagle, J and Nelson, E. Chronic emotional problem patients and their families In an HMO. Inquiry, 15:166-180, 1978. 6. Coleman, JV. Chronic mental illnesses in an HMO. Clinical findings. Submitted for publication. 7. Coleman, JV, Lebowitz, ML and Anderson, FP. ~ . . . ~ ~ Social work in a pediatric primary health care team In a group practice program. Social Work in Health Care. 1:489-497, 1976. 8. Coleman, JV, Patrick, DL and Baker, SL. The mental health of children in an HMO program. J. of Pediatrics, 91:150-153, 1977.

71 QUESTIONS AND COMMENTS ON TREATMENT OF EMOTIONAL PROBLEMS BY PRIMARY PHYSICIANS IN AN Jerry M. Wiener, M.D. Department of Psychiatry and Behavioral Sciences The George Washington University Medical Center Washington, D.C. My questions and comments will deal not only with Dr. Coleman's presentation, but also with issues he raises-in his full paper which time did not allow him to cover. My response is in the nature of concerns I have about first some of the priorities which influence HMO practice and second about some of the assumptions that define or justify the HMO organization we heard described. In the first category is the extent to which the HMO incentive system implicitly rewards the false negative, in contrast to the fee- for-service system which has been criticized for rewarding the false positive. General medical practice has begun to take active steps to control the extent of false positives, e.g., PSRO, second opinions before surgery, etc. But at the same time, in fact, medicine as a healing art and science prefers the false positive to the false nega- tive in terms of not missing a treatable illness. How then, does the HMO system, particularly emphasizing an early detection and prevention model, protect against the false negative diagnosis and treatment? In regard to the second, these assumptions need to be identified so that they may be examined, as follows: 1. Mental health professionals are largely interchangeable, even as backup to the primary care physician, so you provide the lowest common denominator in the mental health disciplines. Are they really interchangeable? When given a choice, patients prefer not to be referred to specialized mental health and specifically psychiatric ser- vices, even within the HMO system. The danger of this assumption is that it may be a self-fulfilling prophecy and a money-saving advantage. How are patients presented with the choice, and which choice is in their best interest? It seems a curious finding about HMO patients, at a time when the national experience is that the population as a whole is increasingly accepting of emotional illness as psychiatric care.

72 3. The assumption that children have individual resources by which they grow out of their problems, and that the one to two year follow-up data support benign neglect as an effective if not the best intervention. Dr. Coleman's own data in support of this assertion are both inadequate and also para- doxical, in that, e.g., with intervention children diagnosed as having moderate to severe disturbance did better on follow-up than those with mild to moderate disturbance. I am concerned here that the assumptions and the usage of the data justify the commitment of less than adequate resources to meet the special needs of children, in which case the HMO would be duplicating the problems children have in receiving adequate care in other delivery systems. The HMO seems quite satisfied with the organization of mental health services described by Dr. Coleman, but I am concerned that the assumptions that underlie that organization have not been tested, per- haps because they satisfy economic considerations. The economic con- siderations are necessary, but alone are not sufficient to establish the guidelines by which the most efficient and efficacious mental health services may be provided to subscribers.

73 TRAINING OF FAMILY PHYSICIANS IN MENTAL HEALTH SKILLS: IMPLICATIONS OF RECENT RESEARCH David Goldberg, M.D Visiting Professor . Medical University of South Carolina University of Manchester England Dr. Goldberg's paper reported the data from his recent study based on a psychiatrist's observations of over 2000 interviews by 56 family physicians cited three important variables accounting for much of the variance: interest and concern; psychiatric focus, similar to psychosomatic score; and age -- older doctors tending to have higher rates of identification. A number of other studies examined correlation coefficients of patient symptom levels, measured by screening tests, relative to physician ratings of degree of disturbance. While not accounting for the threshhold scores of either physicians or screening tests, the studies did suggest that over 2/3 of the coefficient variance could be accounted for by: interest and concern; and conservatism, a broad personality dimension reflecting inflexibility, resistance to change, and authoritarianism -- less conservative physicians making more accurate assessments. Thus, the ability to make accurate psycho- logical assessments is related both to attitudinal and personality factors and to interviewing skills. The latter clearly provides more promise in terms of improving family practice training ~ , , _ ~ O (The full text of Dr. Goldberg's paper appears in Section VII.)

75 OBSERVATIONS ON THE IMPACT OF PSYCHIATRI' DISORDER UPON PRIMARY MEDICAL CARE Edwin W. Hoeper, M.D.* Psychiatrist, Marshfield Clinic and Chief of Staff, St. Joseph's Hospital, Marshfield, Wisconsin INTRODUCTION Houpt, et al., 1/ recommended to the Institute of Medicine that "general medical services cannot be adequately delivered in isolation from mental health services." They point out that behavorial special- ists need to be concerned with: the prevalence of psychiatric disorder in general medical practice, life-style patterns of behavior which place people at greater risk, and medical disorders complicated by psychiatric disorder or life-style patterns of behavior. This report on a recent study conducted in Marshfield, Wisconsin documents the prevalence of psychiatric disorder in an adult Primary Care Provider (PCP) patient population, and the extent to which the POP records recognition of mental disorder. In addition, psychiatric illness is related to patient distress, functional level, and overall medical utilization. SETTING The study was conducted in Marshfield, a central Wisconsin town of 17,000. The Marshfield Clinic, a 170 physician multispecialty group practice, provides primary care to the residents of Marshfield and the surrounding rural area as well as providing secondary and tertiary referral care to a large segment of central and northern Wisconsin. The responsibility for primary care at the Clinic is divided between the general internal medicine section (8 M.D.'s) and the immediate care section which consists of 4 family practitioners and 3 immediate care physicians. The next closest physician practice is approximately 20 miles from Marshfield. Prepared in Collaboration with Gregory R. Nycz, B.S., Researcher, Marshfield Medical Foundation, 510 North St. Joseph Avenue, Marshfield, Wisconsin and Paul D. Cleary, M.S., Researcher, Center for Medical Sociology and Health Services, University of Wisconsin, Madison, Wisconsin.

76 The base population for the study consisted of the adult resi- dents of the Marshfield zip code area who utilized services of the primary care providers from January through March, 1978. METHODS Previous methodological problems in prior studies dealing with these issues have been addressed in this study by: (l) Measuring prevalence of psychiatric disorder utilizing Research Diagnostic Criteria (RDC) 2/ in a standardized psychiatric interview, The Schedule for Affective Disorder -- Schizophrenia (SADS-L) 3/ to control for reliability and validity of psychiatric diagnosis; (2) Weighting to control for biases due to oversampling of high PCP utilizers and seasonality of the sample; 4/ (3) Comparing sensi- tivity and specificity of PCP diagnosis with RDC psychiatric diag- noses; (4) Conducting a detailed nonparticipation analysis, 4/ and (5) Obtaining an overall functional level with the Global Assessment Scale (GAS). _/ Further details of the study design are available. 4/ RESULTS Table 1 displays the prevalence of psychiatric disorder as measured by three different criteria. Utilizing the SADS-L as the criteria for mental disorder, the weighted estimate of the prevalence of RDC psychiatric disorder in the adult primary care population was 26.7%. The RDC lifetime prevalence in this group was 49.5%. The prevalence of PCP point diagnosis (within one month of study entry) of psychiatric disorder was 1.8%. When the determination of the presence of psychiatric disorder was broadened to include PCP nondiagnostic recorded recognition (psychiatric symptoma- tology, psychotropic drug abuse, referral to a mental health specialist, counseling), the resulting point prevalence was estimated to be 5.1%. Table 2 presents the weighted estimates of the prevalence of the more frequent specific RDC diagnoses in the adult primary care patient population. These are not mutually exclusive categories; in fact, 25Z of those with RDC diagnoses had two or more such diagnoses. Major depres- sion and phobic disorder were the most prevalent (5.8%~. The GHQ 6/ score was derived from a 30 item symptom screening device which seems to be a measure of distress. 7/ Persons with RDC diagnoses have mean GHQ scores that are greater than the general adult primary care population (mean =4~. The data on mean GAS scores by RDC diagnoses demonstrated that these psychiatric syndromes also adversely affect functional level. The schizophrenic disorders, not shown here, demonstrate even greater distress and dysfunction.

77 TABLE 1. RECOGNITION OF PSYCHIATRIC DISORDER BY RDC CURRENT, RDC EVER, PCP DIAGNOSIS, AND POP RECORDED RECOGNITION IN WEIGHTED SAMPLE Type of Recognition Percent of Adult Primary Care Provider Users RDC Current Diagnosis Present RDC Diagnosis Ever Present Primary Care Provider Psychiatric Diagnosis Present Within One Month of Study Entry Primary Care Provider Recorded Recognition of Psychiatric Di Border Present Within One Month of Study Entry 26.7 49.5 1.8 5.1 TABLE 2 . FREQUENCY, AVERAGI: (~IQ, AND AVERAGE GAS BY THE MOS T PREVALENT RDC DIAGNOSIS IN THE WEIGHTED SAMPLE RDC Diagno si s Frequency* Percent Mean (;HQ Mean GAS Major Depression 14 5.8 11.9 62.3 Minor Depression 8 3.4 6.0 73.4 Intermittent Depression 12 5.0 5.1 71. 8 Generalized Anxiety Di Border 4 1.6 12. 3 67 . 9 Cyclothymic Personality 5 2.0 9.9 70.0 Labile Personality 9 3.7 9.4 70.8 Phobic Di Border 13 5. ~ 6. 5 76 . 4 *Number rounded to integers

78 Table 3 presents the overall medical utilization of patients with and without RDC psychiatric illness for 12 months before study entry and six months after study entry. Patients with a RDC diagnosis have a higher level of medical utilization (4.05 vs. 3.25 for the 12 month period before study entry). For the six months after study entry the comparable rates were 7.77 and 5.49. TABLE 3. AVERAGE ANNUAL VISIT RATES BY PRESENCE OR ABSENCE OF A RDC DIAGNOSIS AT STUDY ENTRY FOR THE PERIODS 12 MONTHS BEFORE AND 6 MONTHS AFTER STUDY ENTRY Weighted Sample . . Annualized Visit Annualized Visit Rate for 62 RDC Rate for 171 RDC Time Period Positive Patients Negative Patients 12 Months Before Study Entry 4.05 3.25 6 Months After Study Entry 7.77 5.49 DISCUSSION The results indicate that there are three problem areas: 1. A significant number of patients seek primary medical care who have a RDC psychiatric disorder which is dis- tressing and affecting functional level. PCP's identify very few of these patients (3 out of 100 true cases). 3. There is an overall increase of medical utilization in the adult primary care population who have a RDC psychiatric illness. With approximately 25% of the adults seeking care in the primary care sector having a dysfunctional and distressing RDC psychiatric disorder, and only approximately 2% currently having a recorded psychiatric diagnosis, better case identification would appear to be a valuable goal to improve quality of medical and mental health care. Two studies 8/, 9/ have shown that if mental health and primary care services are integrated professionally, administratively and structually, there is an improved case identification, in that a

79 greater proportion of the population served is seen by mental health care providers than in non-integrated mental health settings. One of these studies 9/ demonstrated that when mental health workers are physically located in the primary care setting, this identifi- cation improves even more. Goldberg 10/ has shown that utilizing a screening device in a primary care setting can also improve this case identification. The hospital formal consultation-liaison service model applied to the ambulatory care setting would in all likelihood improve case identification. Review of the charts of the RDC identified patients in our study indicated that as Houpt, et al, 1/ described, some had psychiatric disorder only, some had a coexistent medical and psychiatric illness, and some had life- style patterns of behavior which either related to the onset of the illness or complicated treatment; that is, non-compliance with therapeutic regimens. Many were also experiencing social and environmental stressors _/ at study entry. Increased detection per se is of limited utility if problems exist in obtaining adequate treatment for detected cases. Thera- peutic behavioral and psychopharmacological techniques have proven value, but both depend on high levels of patient compliance and therapeutic expertise. As one physician colleague has stated, "I don't want to detect something I can't do much about." The physi- cians' past psychiatric education has been didactic or caring for severe psychiatric patients; therefore, they have little observation- al knowledge of how psychotherapists treat or detect the ambulatory psychiatrically ill. Fran our findings it appears that the PCP is more likely to identify the chronically high utilizing, more dysfunctional, psychiatrically ill patient correctly. 4/ This would reinforce their attitudes about the chronicity of the psychiatrically ill. Conversely, they are less likely to detect the less disturbed acutely psychiatrically ill who are more likely to get better. The PCP may also be aware of a psychiatric problem without making a notation of that problem because of the stigma associated with the label, and may not consider such information appropriate for inclusion in the medical record; a behavior modeled after psychotherapists. Psychophysiologic complaints and mild symptoms of dysphoria viewed by the PCP as problematic are not adequately described by RDC or other DSM-III 11/ criteria. This accounted for most of the PCP misidentification In this study. RDC psychiatric illness is based on combinations along three dimensions: (1) symptom complexes or syndromes, (2) functional status, and (3) socially unacceptable behavior. Although some behaviors and feelings may fit a particular RDC diagnosis, they may not be perceived as problematic by either patients or their physicians. Such discrepancies emphasize the differences among the PCP's orientation (primarily symptomatic and big-medical), the patients' orientation (subjective discomfort or

80 dysfunction), and the RDC (detection of syndromes, dysfunction and deviant behaviors). Future studies are needed to understand the complex interactions of these three dimensions so that providers can offer an integrated therapeutic approach. Densen et al. 12/ in 1959 found that 4% of prepaid plan members of the Health Insurance Plan of New York utilized 25% of ambulatory care visits. In Marshfield, Wisconsin in 1977, these findings were replicated in a prepaid plan population with widely divergent socio- demographic characteristics. 13/ In the Marshfield population, persons diagnosed psychiatrically ill were disproportionately represented among the highest medical utilizers, especially in the chronically high utilizing population during a three year period. 14/ Kogan's 15/ results supported this finding. Our present study indicates that physician psychiatric diagnoses when made are accurate in the Marsh- field setting, suggesting that the psychiatric labelling in the previous study 13/,14/ is valid. Since 1965 numerous English 16/,17/ and United States _/,15/,17/,18/,19/ studies have demonstrated the_ _ _ consistency of the finding of greater ambulatory medical utilization by persons with diagnosed psychiatric disorder than persons without diagnosed psychiatric disorder. In this study, when identification by RDC for psychiatric illness is utilized, this differential in medical utilization holds true. The evidence remains strong that the more chronic, dysfunctional psychiatrically ill population place a heavy burden on the primary health care system. A high priority should be placed on developing screening and evaluation programs in primary care settings for psychiatric illness. To accomplish improved POP case identification, present PCP's need to be reeducated to appreciate the value of the psychiatric diagnostic process coupled with the effect these syndromes have on the three dimensions of dis- tress, dysfunction, and deviant behavior. Caution must be used since identification of high medical care utilizers who are experiencing psychological distress may only encourage further utilization by reinforcing help seeking behavior. One way of accomplishing these goals would be to establish for- malized ambulatory psychiatric consultation-liaison programs in integrated health and mental health care settings. This might en- courage the establishment of linkages between PCP's and specialized mental health providers. Such programs should be temporary since PCP's would probably learn through consultation appropriate techni- ques for case identification, therapeutic intervention, and specialty referral. The PCP would also learn how to establish and maintain linkages with the specialty mental health care providers and no longer need a consultation-liaison person to do this for him. These programs would need an evaluative component to answer questions about the effectiveness of case finding of psychiatric illness in primary care. Should acute psychiatric illness be identified since it appears most patients improve without inter-

81 vention and there may be a risk of actually reinforcing medical utilization by focusing on psychological distress? Which providers are most effective and efficient in the treatment of various chronic and acute psychiatric illnesses? Should providers focus their thera- peutic efforts on distress or dysfunction or a combination of the two? Finally, evaluation procedures need to use scales which measure functional level, distress, and deviant behavior so that therapeutic outcomes can be measured.

82 References Houpt, JL, et al: The relationship of mental health services to general health care. Report to National Academy of Sciences, Institute of Medicine, December, 1977. 2. Spitzer, RL, Endicott J. Robins E: Research Diagnostic Criteria: Rationale and reliability. Arch Gen Psychiatry 35:773-782, 1978. Endicott, J. Spitzer RL: A diagnostic interview: The Schedule for Affective Disorders and Schizophrenia. Arch Gen Psychiatry 35:837-844, 1978. 4. Hoeper, EW, Nycz OR, Cleary PD: Final Report NIMH Contract Number: DBE-77-0071, The quality of mental health services in an organized primary health care setting. March, 1979. Endicott, J. Spetzer RL, Fleiss JL, Cohen J: The Gloval Assessment Scale. Arch Gen Psychiatry 33:6 766-771, 1976. Goldberg, DP, et al: A comparison of two psychiatric screening tests. Br J Psychiatry (London) 129:61-67, 1976. 7. Hoeper, EW, et al: Prevalence of emotional disorder in primary care. To be presented at the American Psychiatric Association Meeting, Chicago, Illinois, May, 1979. 8. Hoeper, EW, et al: Should health care and mental health care services be integrated? Presented at the American Psychiatric Association Annual Meeting, Atlanta, Georgia, 1978. 9. Coleman, JV, Patrick DH: Psychiatry and general health care. Am J Public Health 68:5 451-456, 1978. 10. Goldberg, D, Kay C, Thompson L: Psychiatric morbidity in general practice and the community. Psychol Med 6:565-569, 1976. Diagnostic and Statistical Manual of Mental Disorders. Third Edition DSM-III. Prepared by the Task Force on Nomenclature and Statistics of the American Psychiatric Association. April, 1977 (Draft). 12. Densen, PM, Shapiro S. Einhorn M: Concerning high and low utilizers of service in a medical care plan, and the persistence of utilization levels a three year period. Milbank Memorial Fund Quarterly 37:217-250, July, 1959. .,

83 13. Wenzel, FJ, Hoeper, EW, Mycz, GR: Ambulatory utilization by frequency and consistency of use in a prepaid health plan. Presented at the Group Health Association Meeting, New York, New York, May, 1978. 14. Hoeper, EW, Nycz, GR, Wenzel, FJ: Impact of mental disorder on medical utilization. To be presented at the American Psychiatric Association Meeting, Chicago, Illinois, May, 1979. Kogan, WA, et al: Impact of integration of mental health service and comprehensive medical care. Med Care XIII:11 934-942, 1975. 16. Shepherd, MD, et al: Psychiatric illness in general practice. London, Oxford University Press, 1966. 17. Anderson, R. Francis A, et al: Psychologically related illness and health services utilization. Med. Care XV:5 59-73, 1977 (Supplement) 18. Tessler, R. Mechanic D, Diamond M: distress on physician utilization. 1976. The effect of psychological J Health Soc Behav 17:353-364, . . 19. Hoeper, EW, et al: Mental disorder diagnosis and increased health services utilization in four outpatient settings. Am J Psychiatry (in press).

85 LITERATURE REVIEW ON MANAGEMENT OF EMOTIONALLY DISTURBED PATIENTS IN PRIMARY CARE SETTINGS Janet Hankin, Ph.D. Johns Hopkins Health Services Research and Development Center To examine the management of emotionally-disturbed patients in primary care settings, Dr. Hankin and her colleague Dr. Julianne Oktay reviewed the literature published from 1958 through 1978. The review focused only on patients actually identified by primary care pro- viders in North American settings, using the broadest definition of mental disorder. While referral to a mental health specialist is a treatment option, this review was limited to consideration of psycho- active drug therapy and psychotherapy administered by primary care providers. (The full text of Dr. Hankin's paper appears in Section VII.)

87 SUICIDE PREVENTION BY FAMILY PHYSICIANS' OPPORTUNITIE S AND LIMITATIONS Thomas A. Madden, M.D. Department of Preventive Medicine Rush-Presbyterian -- St. Lukes Medical Center Chicago, Illinois I think that I have, in fact, very little to tell you that will be new to you about the epidemiology of suicide or the need for a preventive role on the part of the primary care physician. Our interest was not that of research per se but arose from the need to inform ourselves about the problem and how better to tackle it. About the context of our subject: First, in terms of the size of the problem, it is said that at a conservative estimate, in the United States there are 225,000 attempts a year, actions intended to or capable of resulting in death by the individual's own hand. Rather more than one in 10 of these are successful. Suicide is 13th among the leading causes of death. The figures are for 1977. Having said that, for the primary care doctor, the problem will always remain an individual one. Large studies may show differences in profile between the failed and the essentially never intended death, but I do not think that these will greatly help the primary care doctor. Second, there is a recognized connection between the problem of death by one's own hand and that of personal isolation. Looking at these SMRs for the year 1965 to 1967 (Registrar General statistics), those on the extreme right may cause you to reflect how lucky you are if you have been successful in pairing for life at an appropriate age. They also reflect other very complex factors, and the implications of the table extend far beyond this particular topic: to the planning services, to the primary care work load. Looking at suicide, there was a great deal of literature from socio- logical and psychiatric sources over a century or more but the paper which most attracted us was Peter Sainsbury's first Maudsley monograph (1955), which he described as 'an ecological study.' He looked at 409 consectuve suicides over a three-year period in a north London Coroner's Office. I would like you to take note of Sainebury's defini- tion of bereavement, not simply that of a death in the family, but a wider definition, which has implications for our own work. I am also indebted to him for this quotation from Bacon: "For little do men perceive what solitude is and how far it extendeth. The Latin adage

88 Table 1: Married Single 1965-1967 Widowed Divorced Males: (Ages 15-84) Resp. Tuberculosis 79 203 147 212 Cirrhosis of liver 88 142 138 220 Accidental poisoning 64 157 225 346 Suicide 68 152 125 184 Hospital "utilization" shows similar variation. General, 1967 Commentary - Registrar meeteth with it a little, magna civitas, magna solitude, because in a great town friends are scattered, so that there is not that fellowship, for the most part, which is in less neighborhoods." The association between size of community and suicide is well-documented in the United States: the smaller the community, the rarer the occurrence. Other important associations are brought out in Sainsbury's work: with poverty, with lack of work, with alcoholism, with broken marriage or divorce, with a broken home in childhood, and with inner city residence. If you were to ask where all such things come together, they do so precisely in such an area as that where I practiced, an old borough in the center of the metropolis at a time of unsettlement in the population, a process which is called 'urban renewal.' With the consent of the local coroner, we looked at 250 consecutive suicides presenting in the records of the Coroner's Court, Southwark, covering the area served by Guy's, King's College and St. Thomas' Hospitals, essentially that of our own practice area. The principal author, Dr. W.C. Lettington, was my partner. The third author, Ian Cross, was a medical student then studying in the practice. We examined marital status at the time of death. Note the figures for the widowed, the only column in which there is a female over male preponderance. I have said nothing of age and class, although we looked at these. Recently bereaved females had a mean age of 54, but there was an interesting bimodal distribution with an older group in the 60 to 75 year age range. We classified those of "stable marital status," whether single or married, and those

89 TABLE 2. Marital Status (at the time of death) Married Separated Divorced Single Unknown status Widowed 98 (55 M, 43 F) 30 (20 M, 10 F) 10 (6 M, 4 F) 74 (47 M, 27 F) 3 35 (12 M, 23 F) of "altered marital status" to see whether this rave any help. In tact, it did not. We could not determine from the records how long previously the second group had been separated, divorced or widowed. |., We went on therefore to look at a group across both the above who were "recently bereaved." Nineteen of the first group, those of stable status, and 33 of the those of altered marital status, 52 deaths in all, were recently bereaved; out of our total of 250, one- fifth of all. If you look at the causes given, less than half were associated with bereavement by death. In terms of Sainsbury's wider definition, deaths were associated with marital breakdown, a lovers' quarrel, the admission of a loved one to hospital, an adolescent leaving home, and so on. TABLE 3. GROUP C 19 of group A (stable marital status) 33 of group B (altered marital status) 52 deaths, one fifth of the total were recently bereaved. (44% of Group C, 9% of all deaths) had experienced the death of a loved one within the preceding 12 months. 29 (56% of Group C, 12% of all deaths) had experienced marital breakdown, a lovers' quarrel, the admission of a loved one to a hospital, an adolescent leaving home. Of these, marital breakdown affected 16 (M : F - 3 : 1). (13 M, 6 F) (19 M, 14 F)

9o As to the manner of death, there emerged one profile of a person compliant, biddable, responsive to family, to friends' or neighbors' suggestions to see a doctor, willing to see the doctor frequently. These persons died from self-administered drugs. There were others whose deaths display anger, protest, violence, perhaps punishment of another. We looked at events leading up to death as to whether these might help to predict outcome. The recently bereaved did not differ from other groups in relation either to suicidal threat or suicidal attempt, although suicidal threat occurred three times more frequently. Lastly, we looked at the existence of contact with the psychiatric services in this group. When we looked at the first two groups, we had found that 40 percent overall had formed some psychiatric contact; in the recently bereaved group only about half that number had formed such a contact. Figure 1. Relationship between duration of bereavement and suicide among 52 suicides (Group C) 26- 20 — an ._ ~ 15 a E of 10 5 — O .—..... : :. , ---,; .............. ...................... ,...,.:. ...'..'...2..::'. ........... .. ,. :., .;........... :,......... .. : 2 ....... ........ '. ......... . . . . . .... . . ......... ' ........... ........... _ _. _..,,.,-., ...... . .: : --: _ ~ -. . . . ....,,,, ..,- . it, 17_ 3 4 5 6 7 8 9 10 11 12 Number of months after bereavement This stresses the vital role of the primary care physician, the only person likely to be able to make a contact at that time; a resume of something that you all know, pointing to the crisis situation that may exist after recent bereavement. What can we say about the role of the primary care doctor in prevention? Cer- tainly young doctors and nurses in emergency rooms do a great deal to prevent, the change to the utilization of natural gas from carbon monoxide prevents, the change in the prescribing of drugs from barbiturates to benzodiazepines prevents. What it comes down to is always a question of measuring risk against resources, and people are very prone to score risks without being aware of the

91 importance we should attach to resources. You can tell me exactly what the Holmes & Rahe Scale says of somebody who is bereaved, who has also been ill, but you cannot score for all the part played by the patient's coping resources in avoiding this fate. Such coping resources may be money in the bank, may be previously learned skills in dealing with hardships; but most often they are people, members of the family, friends, neighbors, the local community, local authority services. Those life events which convey risk are much less than half the story. (1) I think the family doctor needs to be aware of the patient's situation as a crisis. In practice, nothing will distinguish the failed overdose from the accomplished suicide, but if life continues, there is chance of adjustment. (2) There is the duty to inquire directly into feeling of self- destruction, self-blame, unworthiness, and to take the appropriate steps, including, if necessary, admission to a psychiatric unit. (3) There is a need to provide continuous and committed support, if necessary, daily. This is very demanding of the primary care physician. (4) The doctor must have the ability to deal with the patient's or relatives' expectation of drug therapy. (5) We need to know the patient's place in the social network, of other resources that could be invoked on the patient's behalf e For example, we matched two recently bereaved men who supported each other until their need of company was no longer acute, and although they had enjoyed their association, they did not continue with it. In this context, too, I want to mention the work of the primary care team, from secretaries through practice nurses, social workers, and so on. (6) We need to learn skillfully to deal with grief and anger, with despair, and to encourage mourning. In this instance, we were only able to help a mother who had a stillbirth when we insisted with the local hospital that she be allowed to see and to touch the stillborn child, which had been denied her. (7) Finally, we need to know more of the dynamics of family life. I would like to give two illustrations. One is that classic of the daughter who has stayed on through middle life to look after elderly parents, a tower of strength to everybody, including the family doctor, right through the funeral and who then, with intense deprivation of role, becomes not the person who makes the most noise, but the person in the greatest need.

92 In another direction, expectation of role is a major problem: a recently bereaved father is typical. He is widely invited, he is well supported, he is fine everywhere except with his daughter. He believed, until it was made clear to him that this was not so, that, to display to her how much he had loved her mother, he needed constantly to show himself in the grieving position that he occupies exclusively with her. Thus, there is much in family dynamics that we have to be aware of. Lastly, I would like to say that it is always a human sum, the youngest person in our study being a girl of 15, the oldest a man of 86. Thus, if you asked me to sum up the role of the family doctor, I would refer to Polonius, who telling Hamlet about the actors, said they were the best in the world, either for . . . . pastoral-comical, historical-pastoral, tragical-historical, tragical-comical-historical pastoral, scene individual or scene unlimited. In the case of the family doctor, I think there is no doubt but that the role is diagnostical-therapeutical-pastoral, and this not in the sense of Shakespeare, but in the sense of "feed my lambs" and "feed my sheep."

~3 THE PRIMARY CARE PHYSICIAN: MENTAL HEALTH SKILLS Robert S. Lawrence, M.D. Division of Primary Care and Family Medicine Harvard Medical School Boston, Massachusetts Primary care practice provides special opportunities for the prevention, diagnosis, and treatment of mental disorders, through decreased stigma associated with the patient's mental illness, greater likelihood of familiarity with the patient's social con- text, and greater potential for long-term followup. However, studies traditionally have shown that most general practitioners, while they may recognize the emotional situations of their patients, are not qualified to treat mental health problems -- despite characteristics of the primary care therapeutic relation- ship that should enable physicians to treat many mental disorders more effectively than can specialists. Even when physicians are adept at helping their patients through normal life crisis, they generally lack skills to deal adequately with mental problems such as neurotic disorders, psychophysiological disorders, behavioral disturbances of childhood, character disorders, and more severe depressive and schizophrenic psychotic reactions. Dr. Lawrence suggests integration of mental health skills into primary care training programs as a mechanism for eliminating these deficien- cies and enhancing the primary care physician's ability to ameliorate both physical and mental health problems. (The full text of Dr. Lawrence's paper appears in Section VII.)

5 9l THE ROLE OF THE PSYCHIATRIST IN PRIMARY CARE SETTINGS: ISSUES AND PROBLEMS Alan Jacobson, M.D. Joslin Clinic Boston, Massachusetts My task today is to describe some of the challenges and problems for the psychiatrist working in primary care settings. This discus- sion is based upon my experience in two very different settings, a neighborhood health center and a large multispecialty group practice. The focus of this presentation will be on problems rather than on the benefits of this form of practice which have been so well described by other speakers. The work of the psychiatrist in a primary care setting combines central features of practice from three other settings: (1) the out- patient or office practice; (2) the psychiatric liaison service; and (3) the community mental health center. The primary care psychiatrist must be able to use the kinds of diagnostic and therapeutic skills of the outpatient therapist, for this will be an important feature of what the clinical service offers the patient and the referring physi- cian. Secondly, like the liaison psychiatrist he is called upon to help manage patients with medical illness and help staff cope with their responses to medically ill and chronically disabled patients. The primary care psychiatrist will also require the skills of a translator of psychiatric concepts and views for the non-psychiatric practitioners in the general practice. The ability to move across and between disciplines requires an ability to explain the complex and sometimes confusing concepts of psychiatry in non-technical language. Being at the interface of psychiatry and medicine means the psychiatrist must also have some interest in the medical pro- blems of his colleagues' patients in order to experience the setting as relevant to his professional interest. This task of translator is a difficult one, for psychiatry is frequently misunderstood by outsiders. The recent Time magazine cover story on "Psychiatry's Depression" reflects this, for this story, while probably inaccurate in its presentation of the way psychiatrists view themselves, reflects accurately the way others view us. Those others include physicians and nurses. They find our concepts fuzzy, our suggestions confusing and imprecise and our jargon impenetrable. Yet they need us for they have a large group of patients who have major psychosocial issues interfering with

96 their lives and leading them to these general practitioners. Thus, the practitioners keep returning for advice only to find again, as exemplified by one physician who was unusually sensitive to the psycho- logic problems of his patients, that when we discuss it often sounds more like "gossip" than clinical case discussion. Third, the primary care psychiatrist must recognize that the patients of a general practice are a population at risk for mental illness and thus provide an opportunity for the establishment of programs for early intervention or at least alternative interven- tion, using the relatively acceptable medical setting. Examples of this are the development of the adolescent clinic with psychiatric consultation regarding the psychosocial problems of these patients. A population-based focus may also have an impact on direct patient care in that recognition of patients who may be able to use psychiatric ser- vices but are ambivalent may lead to use of outreach techniques like home visits. In addition, the psychiatrist in primary care who sees all the patients of the medical practice as potential clients must be flexible in making therapeutic contracts which reflect the patient's ambivalence. This may lead to extensive contact over years but with only a few visits at a time. This type of intervention, in fact, closely imitates the model of the general practitioner's lifetime involvement with patients where actual interventions occur with long lapses of inactivity. For the generalist and the patient, however, these lapses do not mean termination or dropping out; they are simply an inactive phase of the relationship. The application of this approach, which I term extensive therapy, is a direct reflection of a population-based focus in which the psychiatrist attempts to engage all patients with psychosocial disturbance -- not just those who are willing to call their disturbance psychiatric or those who are "appropriate" for psychotherapy. Thus the primary care psychia- trist is also a community psychiatrist with the community defined by the medical practice. The problems for the psychiatrist are multiple. He must be able to carry out very different types of tasks; he must fight the temptation to interpret the problems of translation as reflec- tions of personal ineptitude or a demonstration that psychiatry must be reduced to a few hard facts. It would be nice if we had lab tests, but we do not; a return to our medical roots by giving up the complex, sometimes fuzzy concepts which we are still building will be a temptation for the psychiatrist practicing in a medical facility. Ironically, the generalist will be an ally in this parti- cular fight, for as much as he'd like us to be precise he also recognizes that he wants us to join him in managing chronically disabled patients. Besides the danger of loss of identity by being in a non-psychiatric setting, the psychiatrist faces the problem of losing funding. The separation of psychiatric funding for research, training and services since World War II has lead to major growth in the field of psychiatry. By linking psychiatry to medicine the

97 loss of identity at a personal level may lead to a loss of identity at an institutional and national level. Psychiatric studies often seem soft and services are difficult to measure in terms of clear outcome; thus, the push towards linkage must be tempered by a realization that the unique identity and goals of psychiatry should not be lost by a simple reductionism.

99 REIMBURSEMENT FOR MENTAL HEALTH SERVICES IN PRIMARY CARE SETTINGS Alexander Richman, M.D. in collaboration with Murray G. Brown, Ph.D. and Vernon Hicks, M.D. Training and Research Unit in Psychiatric Epidemiology Dalhousie University Halifax, Nova Scotia It has been generally acknowledged that the experiences of other industrialized countries who have had some form of universal health insurance coverage should be studied as the U.S. seeks to initiate such a program. Dr. Richman describes elements of the Canadian health insurance system that have implications for the kinds of benefits that could lead to provision of truly comprehensive health/mental health services for the U.S. population. The Royal Commission of Health Services recommended in 1964 medical services benefits covering the full range of diagnosis and treatment for all physical and psychiatric conditions. The Canadian National Health Plan -- "medicate" -- with this benefit package went into effect in 1969. Under the Canadian medicare program, reimburse- ment covers 50% of medical costs for those provinces in which health systems feature comprehensive and universally available services, accessibility to users to all income levels, and portability of bene- fits between provinces. The fee-for-services plan allows patients to choose their physicians, who in turn submit reimbursement claims to provincial agencies of the national system. Throughout Canada during 1977-78, 3.8 million psychiatric services (170 services per 1000 population) were reimbursed at $91 million ($4042 per 1000 population); 43% of those services were provided by general practitioners. Data for Saskatchewan during 1977-78 show the following trends: 1) Overall, 13.3% of the total Saskatchewan population received psychiatric services -- 18,400 from specialized mental health hospitals or clinics, 10,100 from general hospitals, and 94,600 from private practitioners. 2) Of the total population diagnosed with mental disorder, functional psychoses accounted for 30% of inpatient psychiatric services, 9% of mental health clinic services, 7% of general hospital ward services, and 4% of services from private practice physicians. Neuroses and psychosomatic disorders accounted for 25Z of psychiatric inpatients and 73% of private practice patients. 3) Of all patients who saw physicians, 88%

100 were seen by general practitioners and 12% were seen by psychia- trists; 6% were seen by both. 4) In terms of utilization, 77% of the patients who saw physicians received 28.1% of all medical services, while 4% utilized 39% of all services. 5) About 10% of patients seen by general practitioners used 50Z of the mental health services provided, and 20% of patients seen by psychiatrists used 50% of those mental health services. Thus, the dispropor- tionate use of mental health services was greatest among patients of general practitioners. The data suggest a number of issues to be considered as more physicians become providers of reimbursable psychiatric services: 1) How can epidemiologic methods and classification systems be standardized to decrease variability among practitioners? 2) How can quality of care be guaranteed? Quality assurance in ambulatory mental health care will depend on more reliable definitions of diagnostic and therapeutic criteria and goals, distinguishing mental disorders from psychological reactions and emotional responses. In addition, it should be made clear that diagnosis of mental disorder does not necessarily justify treatment. 3) Does reimbursement for psychotherapy under national health insurance represent economic subsidy to the rich from the poor? and 4) How can psychiatric skills in clinical care appropriately be matched with mental health problems? (The full text of Dr. Richman's paper appears in Section VII.)

101 PROBLEMS OF ANALYZ ING THE COST OFFSET OF INCLUDING A MENTAL HEALTH COMPONENT IN PRIMARY CARE* Emily Mumford, Ph.D. Herbert J. Schlesinger, Ph.D** Gene V. Glass, Ph.D. University of Colorado Medical School Denver, Colorado To assess whether there is cost-offset attributable to including a mental health component in primary health care would seem to be a straightforward research problem. Instead it is encumbered with difficulties from several sources, some conceptual and methodological, others practical and statistical. First, one of the hopes of those who propose including signifi- cant mental health coverage in primary care settings is that it will serve to reduce medical utilization and hence reduce costs. But the very assumption that mental health services should reduce the utili- zation of other medical services needs to be examined closely. Certainly mental health care should be required to demonstrate that it is effective and even that it is cost-effective, which is to say efficient. But should it have to justify itself by displacing other medical services? There are certain circumstances when such an offset might be expected, chiefly when physical health services are misutilized and overutilized, that is, substituted for unavailable, inaccessible, or inconceivable mental health care. The reasoning underlying this expectation is that people will seek help for emotional problems whether or not a proper channel is provided. Emotional problems frequently have somatic components and it is easier for many people to ask for medical attention for sleep- lessness, "tension," low back pain or G.I. disturbance rather than for "depression." Reducing misutilization must be part of any attempt to improve health service systems. But misutilization may coincide with either underutilization or overutilization. Self-neglect or abuse is *The work reported was supported in part by the National Institute of Mental Health, Division of Mental Health Service Programs, under Contract NIMH 278-77-0049(MH) and 278-78-0037(MH). **Paper was presented at Conference by Dr. Schlesinger.

102 common among the emotionally ill, a factor which may contribute to the excess morbidity and mortality from physical disease noted among psychiatric patients (Goldberg, et al., 1979~. One effect of successful mental health intervention for such persons should be to increase medi- cal utilization (Bows, et al., 1979~. For example, services are mis- utilized and underutilized by parents who fail to bring a child to medical care even though a reversible disease process has been identified through an expensive screening program. Services are misutilized and eventually overutilized when preventive services are not used and avoidable illness results in hospitalization. Misutilization also occurs when neglect of prenatal care contributes to unneccessary com- plications at delivery or to the birth of a defective child. Recog- nizing the paradox that failure to use medical services can result in overuse and misuse of health services is central to many preventive efforts and health education campaigns. The time, money and health potential wasted when medical advice is sought but ignored is the subject of a burgeoning literature on '"compliance" with medical advice (Barofsky, 1977; Sackett and Haynes, 1976~. Excessive utilization directly raises costs of medical service. But it is also of great concern to physicians since it may reflect inappropriate utilization and hence lower quality of medical care, ironically at higher expense. Physicians, patients and third-party payers share an interest in realizing an optimal level of medical utilization that results in improved functioning with a minimum of unnecessary procedures, visits and hospitalizations. A second conceptual problem is the supposed distinction between mental health care and physical health care. While conceptually distinguishable, emotional and physical distress are so inextricably intertwined that it can be counterproductive to encourage treatment of physical malfunction while discouraging treatment of emotional malfunction. Furthermore, the strategies that would be required to assign patients to services treating only psychological disorders or only physical disorders boggle the mind in light of the many studies that demonstrate the coexistence of both kinds of problems. A third conceptual problem has to do with the very nature of the major variables of interest. Health itself, which is the prime outcome variable, is largely represented in research by medical utilization, a presumed negative indicator. Often the definition of health and whether an emotional problem is "a problem in living," a religious problem, an ethical problem, a social problem, or a medical problem seems to depend on whose purse is being gored. Medical utilization is often chosen as the indicator of convenience. It is not a simple variable, but rather a congeries of hospital days, doctor visits, procedures done and medicine prescribed. Since it is expressible in dollars spent, it is a convenient way to refer to the output of a health services system.

103 But there are other "hard," or easy to measure, indicators, and any number of "soft" ones that could be considered. "Soft" indi- cators generally refer to subjective issues, feeling well or ill, satisfied with medical services or not. While more difficult to measure, they may be equally or even more important as are the "hard" ones to the persons who are the patients of these studies, as well as to society. Not withstanding all these caveats, it is important for planning health care systems to know if providing mental health services has any effect on the utilization of other medical services. While we will focus on the influence of psychotherapeutic intervention, medi- cal utilization is also influenced by a wide range of psychosocial, economic, geographic and financial factors (Engel, 1977; Lipowski, 1977, 1975; Glass, 1977; Fabrega, 1975; Kaplan, 1975; Mechanic, 1976~. We list some of these in order to warn the reader about the complexity of the field. Among the factors to which medical utilization is sensitive are the following: 1. The tendency in our society to "medicalize" life problems; the "selling of medicine" as holding the key to health, happiness, sexual success and social popularity (Aday and Andersen, 1978; Eisenberg, 1977~. 2. Perceptions of health--emotional and physical--and the place of health in hierarchies of values (Kaminsky and Slavney, 1976; Lesse, 1974~. 3. Perceptions of "rights"--to care, concern, treatment equality, attention--all expressible as "the right to treatment" (M''mford, 1977~. Secondary gain--the use of the experience of complaint of illness and medical attention to legitimatize a wish to be cared for, to be excused from fulfilling obliga- tions and responsibilities, to fulfill a desire for "compensation" either for actual disability or for a sense of being treated unfairly by life (Mechanic, 1976; Dohrenwend and Dohrenwend, 1969~. Economic conditions--it has been observed that visits to physicians and rates of hospitalization increase in times of economic decline (Brenner, 1973~. 6. Availability of health services--including geographic factors, clinic and office hours, presence of emergency services or walk-in services, financial conditions, and social and cultural expectations and patterns of help- seeking of the "patient" population (Mechanic, 1976; Clancy and Gove, 1974; Webb, et al., 1977~.

104 Stress and "life events" may lead to emotional distress presenting as physical disease. Emotional distress, etc., may also affect existing disease, exacerbating symptoms or affecting the body directly or indirectly. Emotional factors, etc., may also trigger visits to physicians for a previously ignored physical illness (Mechanic, 1972; Clayton, 1973; Fontana, et al, 1976; Roghmann and Haggerty, 1972~. 8. Physical illness may itself lead to social and/or psychological problems, including family disruption, which in turn may lead to emotional distress further aggravating the physical illness and leaving the patient dependent on medical facilities (Rappaport, 1975; Engel, 1977; Gersten, et al., 1977~. 9. Patterns of response to medical advice resulting from the nature of care given and the setting in which it is given as well as cultural factors and personal idiosyncrasies (Barofsky, 1977; Becker and Maiman, 1975; Sacket and Haynes, 1976; Rosenberg and Raynes, 1976; Stimson and Webb, 1975~. 10. Professional convictions of health care providers about the appropriate "treatment" of complaints presented to them (e.g., surgeons tend to recommend surgery ) (Lawson and Jick , 1976; Sedgewick , 1974; Vayda, 1973; Bunker, 1970~. Economic motives of health care providers (Fuchs, 1974) . 12. Iatrogenic illness (Bercel, 1968; Fuchs, 1974; Reidenberg, 1968; Ogilvie and Reudy, 1967 ~ . 13. Risk-taking behavior--smoking, overeating, lack of exercise, careless driving, etc. (Pomerleau, et al., 1975; Singer, et al., 1976~. 14. Environmental conditons--pollution, unsafe industrial conditions, traffic hazards, etc. (Fuchs, 1974~. 15. The nature and effectiveness of social controls over availability of health care and its utilization (Goshen, 1963; McCarthy and Widmer, 1974~. This list is not exhaustive and many of the factors can be further sub-divided. Since medical utilization is sensitive to so many factors it would be all the more noteworthy if the availability and use of

105 psychotherapy or other forms of ''mental health," or psychologically informed, intervention can be shown to influence it. Psychotherapy is also a complex variable subject to multiple influences. To summarize, one must consider: 1. The kind of psychotherapy offered. Psychotherapies are classifiable in terms of the unit treated, e.g., individual or group, couples or family, in terms of the theory that inspires the effort, defines the presumed mechanisms or process and explains the results. 2. The time dimension of psychotherapy. How frequently is the patient seen for sessions of what length and over what period, intermittently or continuously? Long duration of treatment and high frequency of ses- sions are sometimes considered to guarantee "intensi- vity" while a brief episode of infrequent visits of psychotherapy may be dismissed as "superficial." These appellations may not be justified. 3. Therapies are not always conducted in a "pure" form but may be combined with each other as well as with other means of providing social support, and with medication. 4. The quality of the psychotherapy given. Even more difficult than knowing precisely what occurs between therapist and patient is knowing how representative the therapy is of what it was supposed to be, and how well it was done. In addition to the barrier of patient privacy, the conceptual tools for evaluating diverse therapies directly are rudimentary. Again, since psychotherapy and medical utilization are each complex and messy variables, these complexities, if they could be assumed to vary randomly, might be expected to wash out any true relationships that more refined studies might reveal. Thus, if studies conducted at the present state of the art do reliably show a relationship, they would argue for remarkably robust effects. In addition to conceptual and philosophical issues there are methodological problems besetting utilization research that make it difficult to estimate cost offsets precisely. First, most of the research is clinical. Patients are seen in naturalistic settings where treatment conditions cannot generally be manipulated to achieve experimental cleanliness. In clinical research we know that "the intervention" is not the only variable at work. A multifactorial approach is needed but rarely are there sufficient patients and suf- ficiently controlled conditions. More importantly, all the factors

106 that operate in the field have not yet clearly been identified. Second, much of the research in the field has been archival, or retrospective, making use of existing data gathered for purposes other than research. A third issue, the problem of selecting appropriate control groups, has flawed many studies in the field, in ways we shall indicate. A fourth issue is the time when one should assess the outcome of psycho- logical intervention. Should it be directly at the end of treatment, or a month or a year or a decade later? While these multiple difficulties could lead one to conclude that research in this field is all but impossible, much work has actually been done. The following is a summary of findings presented elsewhere in greater detail (Mumford, Schlesinger and Glass, 1978~. REVIEW OF STUDIES OF THE EFFECTS OF PSYCHOTHERAPY ON MEDICAL UTILIZATION We will discuss two types of research evidence on the effects of psychotherapy on medical utilization: 1) archival, time-series studies; and 2) controlled clinical experiments. For all of the shortcomings of the archival approach, it has the advantage of "naturalness" and some major sources of bias from intrusive research interventions are avoided. Although controlled clinical experiments may suffer in vary- ing degrees from ''reactive" (Campbell and Stanley, 1966) influences resulting from patient and doctor awareness of the research, they may compensate with superior experimental controls. Archival Time-Series Studies of the Impact of Psychotherapy on Medical Utilization - Sixteen studies that examined patients' medical records for a period of time before and after entry into psychotherapy were located in the published and unpublished literature through April, 1979. They differ in whether they use control groups, in duration of observation and in frequency of pre- and post-therapy observation points. In general, the use of well matched control groups, longer duration of observations and higher frequency of observation imply higher quality of experimental design. One important feature of design, however, critically affects interpretation of findings. The distinction between relative and absolute time base is crucial in assessing the validity of the evidence. Observations on a "relative" time base are accumulated at points before or after the date when each patient entered therapy. Thus, "one month pre-therapy" might be June 1975 for one person but August 1976 for another. "Absolute time," in contrast, is the same

107 for all persons in the group under observation; e.g., in absolute time, one month pre-therapy is April 1977 for each person in the group. The importance of this distinction derives from the susceptibility of relative time series to invalidation from the effect of regression to the mean. This "regression effect" causes mischief in time-series quasi-experiments in the following manner: in reaction to the extreme deviation of the curve, the experimenter is prompted to take some action to bring the curve back to some more typical value. In these circumstances, the experimenter is likely to mistake the expected regression effect, i.e., the tendency in a series for extreme values to be followed by values closer to the mean, for the effect of his own actions. Hence, the fallacy of "reactive intervention" (Glass, Willson, Gottman, 1975~. In regard to the present analysis, suppose an individual monitors his states across time. When discomfort reaches an extremely high level, he might be prompted to seek psychotherapy. The start of therapy might coincide with the expected decline of the discomfort curve and would likely be interpreted by patient and therapist alike as an ameliorative effect of treatment. An alternative explanation of this quick "improvement" is statistical regression; a movement toward the typical level of the series will reliably follow the high or low points in a time-series. If a control group is not matched for the "discomfort" variable with observations accumulated on a relative time base, there would be no comparable expectation for its values to regress toward the mean. The methodological issue of relative versus absolute time looms especially large in view of possible spurts of high utilization around personal crises such as a death in the family (Parkes, 1970~. "Statistical regression" is of course not in itself an explana- tion. It is an all-purpose, mathematical name for the movement of data values. There is nothing logically inconsistent with the claim that a person's scores regressed statistically and the claim that all the reasons for the change in those scores are known. Typically, however, the many and varied causes of the probabilistic fluctuations of a time-series are unknown and can be regarded as "error." Table I summarizes the finding of all 15 studies in comparable terms: Three studies (Goldberg, McHugh, and Kessler) did not employ control groups e Their entries in the final column are the difference between pretest and post-test observations, the best estimate of effect in the absence of a control group. Since the most of the pre/ post differences for the psychotherapy groups are confounded by the regression (or "reactive intervention") problem, it might be wise to forebear drawing conclusions. However, the hope of salvaging even a tentative finding from all this effort is strong and justi- fies risks when methodological scruples would permit no conclusions at all.

108 ~ ~ o ~ ~ ~ ~ ~ . . . . . . ~ . o ~ o o U. o i . . _ ~ c 0 a, 0 _, ' O ~ ~ ~ ' ~ , ~ _ ID 1 v. ~ $ , · ~ al o ~ ° ~ ~ ~ 5 ~ I ~ . ~ . — ; 8 N _ ID ~ 0 0 0 ~ O ~ . . e In ~ ~ U} 8 ~ ~ ~ ~ ~ ~ ~ · · _t -I V ~ V vat ~ ID - ~ ~ ~ ~ ~ ~ ~ ~ I U o C~ O O O ^~ ~ ~ V ~ G & ~~. ~ 36 ~ ~ o U ~ V o ~ c 0 ~ ~ 43 ~ ~ ~ 8 ° ' ~ ~ ° ~ ~ ' ' > ~ ~ A' ' . · · · ~ ~ _ 1 ~ ~ ,`~' I 60 ~ 8 e ~ I I l^- -~-1- c 1 ~ ~o ~ ~ ~ ~ . ~ ~ ~ ~ ~ ~ ~ ~ ~ u | E. V 3 ° ° ° ~ ~ ;' ~e ;2 ° ~ ° ° ° | ~ oO ~ ~D ,,~ ~ a° D ~ _ C, ,, ~ I ~ O ~ ' I ~ S ~ · · · ~ · ~ ~ =. ~ ~ - . ~ O ~ O O . I ~ —1 _ O tI-1 O _1 iD ~ l~ O ~D l~ r~! ~ r ~J Q 4J ~ ~ O 10 ~ ~ ~ t~ ~ ~ ~ ~ ~ 4e :- 0 0 ,$ ·o ~ 0 4~ ~ te~ ~~~ r~ O -~ {~~ de _ ~ · ~ ~ ~ ~ ~ ~ a~ C ~ t~ ~r ' U~ a°~ r~ (D a~ ~~~ \o \D 0 i4 S4 ~`, ~ AS u' ~ —0 a' ~ ~ ~ ~ ~ ~ ~ `° ~i ~ ~ ~ ~ . ~ ~ ~ ~ ~ ~ ~ (D o~ e~ 110 4.~ - — ~ ~ 01 ~ 0 ~ ~ ~ ~ ~ EE ~ ~ O ~ O ~D O dJ ~ 0 _ 4J _1 ~ ° S4 0 0 113 ~ ~ (D 1~ ~ ~ ~J ~ tO _~ ~C O4 o4 54 0 ad ~ ~ O X ~ 04 O4 3~ D. tO3. Q. °= 54 8. ~ ~ ~ ~ 4~ _1 Ut ~ ~ SJ ~ S~ h ~ S4 ~ ~ U] m o' P. a. D. ~ ¢~. ~ e 0 ~ 0 ~ ~ ~d ~ · · ~ ~ ~ · ° ~ ~ ~ ~ ~ ~ es ~ al ~ )4 14 s ~ ~ ~ ~ ~ ~ 8 8 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ 8 ~ ~ 8 ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ - -— ~ ~ ~ ~ ~ ~ ~ ~ - - ~ ~ ~ ~ ~ ~ - - o ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~4 ~ E ~ ~ ~ ~ C C O Id ~ ~ a ~ ~ 0 a os m ~ ~ 0 ~ ~ ~ &! ~ ~ ;~ ~L2 o! a! d! 2 , ~ dt ~o . · ~ ~ , ~ 3 o ~ ~ ~ ,~, ° o ~ ~ . j O c m ~ v v v v v v a, ~ v a v ~ a ~ ~ 0 ~ ~4 ~D O , , D ~ c ~ ~ ,' ~ i U N u c u c ~ u v a ~ v S D U RI Q S S S S S S S ~ C S O eY . t ~ g = I v a - .1 ·1 ~I c a g II a 2 0 - , o, a, ~ o _ C 0 g c _ · _ . _ . , ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ =m ~ ~ —~ ~ ~ ~ ~ ~ ~ ~ ~ ~ ~ a -4 ° ~ ~ ~ ~ :~ O ~ ~ ,`: ~ ,, ~ ~ ~ ', ~ 'd ~ ~ O a.— ~ ~

109 In most studies inadequacies in design probably favor the psycho- therapy group over the control group. Perhaps the only exception is is the McHugh study with its apparent long-term, upward trends in utili- zation. Recognizing then, that the figures are probably biased in favor of psychotherapy. the average benefit of therapy is about a 13% decrease in medical utilization. The upward trend among the Mexican- Americans studied by McHugh may represent increased attention to health as well as the expectable and even hoped for rise in utilization when new services are provided to a previously underserved population. If the McHugh study, which was based on new services in a previously underserved area, is eliminated, then the average benefit calculated from the remaining differences in the last column becomes a 19% decrease in medical utilization as a result of psychotherapy. Because of the problems in design of most of these studies, the hypothesis is quite plausible that they reflect no beneficial effect of psychotherapy on medical utilization. It seems highly unlikely that the true benefits exceed the 19% estimated decrease in medical utilization obtained by eliminating the McHugh study from Table I. The conclusion can be drawn with confidence that psychotherapy reduces medical utili- zation by between 0% to 19%. More careful research is needed to narrow this band of uncertainty. Controlled, Experimental Studies of the Effects of Psychotherapy on Medical Utilization Controlled experiments entail the assignment of patients to two or more different conditions (e.g., psychotherapy vs. a wait-list control) and the subsequent measurement of outcomes from the different conditions. Sufficient controlled, clinical experimental studies suitable for meta-analys~s (Glass, 1977) on the effects of psycho- therapy on medical utilization were found in three areas: alcoholism, asthma, and recovery from medical crisis (surgery and heart attack). The details of quantitative reviews of these experiments are presented elsewhere (Mumford, Schlesinger and Glass, 1978~. In brief, psycho- therapy (primarily behavioral therapies and hypnotherapy) shows impressively large effects on ameliorating the effects of asthma. The effects are even substantial on the reduction of utilization of direct medical services; only 23% of the therapy subjects used as many medical services as half the control subjects. Averaging results across studies of psychological treatment of alcoholism yields success rates of 51% and 33% for psychotherapy and control conditions, respectively. To put these findings into cost- offset terms, we could conclude that on the average 20 hours of psycho- therapy produces 18 "successes" (sobriety 6 months after therapy) out of every 100 persons treated. The relapse rates suggest that if the benefits of the therapy are to be sustained, it must be readminister- ed at periodic intervals. Also, the correlation across the 15 studies

110 between the number of hours of therapy and the differential "success" rate was positive and reasonably large: +.49. Review of Controlled Experimental Studies on the Effects of Psycho- logically Informed Intervention on Patients in Medical Crisis We have been able to locate 23 experimental studies that test the effect of providing emotional support and/or understanding as an ad- junct to medically required care for patients undergoing surgery and recovering from heart attack. The course of recovery was compared with that of a control group of patients not provided the special attention. The circumstances and findings of each study and the problems in analyzing them as a group have been summarized elsewhere (Schlesinger, Mumford and Glass, 1979~. In order to compare and pool results from different studies, an "effect size" was computed. The effect size is a standardized measure of average difference between the treatment and control group on an outcome variable. The effect sizes for all 117 outcome indicators in the 23 studies average +.43 implying that the intervention groups do better than the control groups by nearly one-half standard deviation. These findings are consistent across studies; fewer than 18% of the 117 out- come comparisons were negative. Among the 117 outcome measures 66 are highly relevant to the physical recovery process ("anesthesia time," "units of blood," "degree of hypothermia," and "days in hospital") while 51 have more to do with patient comfort ("self-report of sadness". When effect sizes are calculated separately for these two types of outcome, the compari- son slightly favors larger effects for the more medically relevant indices, at +.45 versus +.40. A subset of the outcome indicators is particularly important for its cost offset implications. Ten studies reported the amount of time spent in the hospital by the treatment and the control groups. The average difference in days hospitalized for these ten studies weighted equally or weighted according to the number of patients studied is slightly more than two days in favor of the intervention group. Is this difference statistically reliable? The estimate is based on data from approximately 2,000 intervention and control patients across the ten studies. Seven of the ten studies gave infor- mation on the standard deviation of duration of hospitalization. The average standard deviation is about 4.75 days and t = 8.53, significant at any reasonable level. If we analyze the findings using the study as the unit of analysis, a significant t of at least 3.07 results.

111 These effects occur even though the interventions are mostly modest and not tailored to the needs of the individual patient, i.e., all patients in the experimental groups received the intervention under study. Two studies that attempted to match the intervention to the patient show that when the coping style of patient is compatible with the type of support provided, the intervention appeared to be more effective (DeLong, 1971; Kennedy, 1966~. SUGARY A review of the problems in determining whether a cost offset can be expected from introducing a mental health component in primary care shows that the problems are conceptual and methodological as well as practical and statistical. The problems are difficult to solve but not insurmountable. Most of the archival studies of the effects of psychotherapy on medical utilization are flawed by problems of experimental design. A critical, quantitative review of 15 such studies that takes account of these flaws indicates a likely reduction of between O and 19% in medical utilization and costs. Further work is needed to narrow the band of uncertainty. A critical, quantitative review of 15 controlled, experimental studies assessing the effects of various kinds of psychotherapy on alcoholism and 13 such studies on asthma show positive effects on outcome indicators with clear implications for a significant and sizeable cost offset. A similar review of studies of the effects of "psychologically informed" intervention on patients recovering from heart attack or surgery shows a clear cost-offset resulting from a more than two-day shorter hospital stay for the intervention group. Quite aside from the intrinsic value of offering specific care for patient's emotional problems and humane and considerate care for their medical and surgical problems, the evidence is that providing psychotherapy and psychologically informed care can be cost effective and that a cost offset may result from the inclusion of a mental health component in primary care systems.

112 References Aday, L. (study director) and Andersen, R. America's health care health care system: A comprehensive portrait. Robert Wood Johnson Foundation Special Report, No. 1, 1978, pp. 4-15. Barofsky, I., (Ed.) Medication Compliance. Thorofare, New Jersey: Charles B. Slack, Inc., 1977. Becker, M.N. and Maiman, L.A. Sociobehavioral determinants of compliance with health and medical care recommendations. Medical Care, 13~1~:109- 24 (January) 1975. Bercel, N.A. Concluding remarks. Diseases of the Nervous System, 29 Supplement (3~:77-78 (March) 1968. Borus, J.F., Burns, B.J., Jacobson, A.M., Macht, L.B., Morrill, R.G. and Wilson, E.M. Neighborhood health centers as providers of coordinated mental health care. Background paper for Invitational Conference on the Provision of Mental Health Services in Primary Care Settings, April 2 - 3, 1979. Brenner, H. Mental Illness and the Economy. Cambridge, Massachusetts: Harvard University Press, 1973. Budman, S.H., Wertlieb, D., Budman, S. and Demby, A. Maximizing the offset of medical utilization via psychological services: A strategy for intervention. Paper presented at the National Institute of Mental Health, April 5, 1979. Bunker, J.P. Surgical manpower, a comparison of operations and surgeons in the United States and in England and Wales. New England Journal of Medicine, 282~3~:135-144 (January 15) 1970. Cambell, D.T. and Stanley, J.C. Experimental and Quasi-experimental Designs for Research. Chicago, Illinois: Rand McNalley, 1966. Clancy, K. and Gove, W. Sex differences in mental illness: An analysis of response bias in self reports. American Journal <'f Sociology, 80(1):205-216 (July) 1974. Cummings, N.A. and Follette, W.T. Psychiatric services and medical utilization in a prepaid health plan setting, (Part II). Medical Care, 6~1~:31-41 (January/February) 1968.

113 DeLong, R.D. Individual differences in patterns of anxiety arousal, stress-relevant information and recovery from surgery. Dissertation Abstracts International, 32: 554B-555B, 1971. Dohrenwend, B.P. and Dohrenwend, B.S. Social Status and Psychological Disorder. New York: John Wiley & Co., 1969. Duehrssen, A. and Jorswiek, E. An empirical and statistical inquiry into the therapeutic potential of psychoanalytic treatment. Der Nervenarzt, 36(4):166-169, 1965. Eisenberg, L. Disease and illness. Culture Medicine and Psychiatry, 1(1):9-23 (April) 1977. Engel, G. The need for a new medical model: A challenge for bio- medicine. Science, 196(4286):129-136 (April 8) 1977. Engel, G. Emotional stress and sudden death. Psychology Today, 11(6): 114-118; 153-154 (November) 1977. Fabrega, H. The position of psychiatry in the understanding of human disease. Archives of General Psychiatry, 32~12~:1500-1512 (December) 1975. Fink, R., et al. Psychiatric treatment and patterns of medical care. Unpublished report to NIMH, (7169), pp. 33-51. Follette, W. and C~'mrnings, N. Psychiatric services and medical utilization in a prepaid health plan setting. Medical Care 5~1~:25-35 (January-February) 1967. Fontana, A.F., Dowds, B.N., Marcus, J.D., and Rakusin, J.M. Coping with interpersonal conflicts through life events and hospitalization. The Journal of Nervous and Mental Disease, 162~2~:88-98 (February) . . 1976. Fuchs, V.R. Who Shall Live? New York: Basic Books, Inc., 1974. _ Gersten, J.C., Langer, T.S., Eisenberg, J.G. and Simcha-Fagan, 0. An evaluation of the etiologic role of stressful life-change events in psychological disorder. Journal of Health and Social Behavior, 18~3~:228-244 (September) 1977. Glass, D.C. Behavior Patterns, Stress and Coronary Disease. Hillsdale, New Jersey: Lawrence Erlbaum Associates, 1977. Glass, G.V., Willson, V.L. and Gottman, J.M. Design and Analysis of Time-Series Experiments. Boulder. Colorado Cr~1 or~-~1n Acc~r;=t-H University Press, 1975.

114 Goldberg, E.L., Comstock, G.W. and Hornstra, R.K. Depressed mood and subsequent physical illness. American Journal of Psychiatry, 136(4B):530-534 (April) 1979. Goldberg, I.D., Krantz, G. and Locke, B.Z. Effects of a short-term outpatient psychiatric therapy benefit on the utilization of medical services in a prepaid group practice medical program. Medical Care, 8~5~:419-428 (September/October) 1970. Goldensohn, S.S. and Fink, R. Mental health services for medicaid enrollees in a prepaid group practice, (EMO). Presented at the 131st annual meeting of the American Paychiatric Association, Atlanta, Georgia, May 11, 1978. Goshen, C.E. The high cost of nonpsychiatric care. General Practi- tioner, 27(4):227-235 (April) 1963. Graves, R. and Hastrup, J. Effects of psychological treatment on medical utilization in a multidisciplinary health clinic for low income minority children. Paper presented Southwestern Psycho- logical Association Meeting, New Orleans, Louisiana, April, 1978. Jameson, J., Shuman, L.J. and Young, W.W. The effects of outpatient psychiatric utilization on the costs of providing third-party coverage. Research Series 18, Blue Cross of Western Pennsylvania, December, 1976, pp. 1-38. Kaminsky, M.J. and Slavney, P.R. Methodology and personality in Briquet's Syndrome: A reappraisal. American Journal of Psychiatry, 133(1):85-88 (January) 1976. Kaplan, H.B. Understanding the social and social-psychological antecedents and consequences of psychopathology: A review of reports of invitational conferences. Journal of Health and Social Behavior, 16~2~:135-151 (June) 1975. Kennecott Copper Corporation. INSIGHT, a program for troubled people. P.O. Box 11299, Salt Lake City, Utah, (undated draft). Kennedy, J.A. and Bakst, H. The influence of emotions on the outcome of cardiac surgery: A predictive study. Bulletin of the New York Academy of Medicine, 42~10~: 811-849 (October) 1966. Kessler, L. Episodes of psychiatric care and medical utilization in a prepaid group practice. Doctor of Science Dissertation, Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland, May, 1978. Kogan, W.S., Thompson, D.J., Brown, J.R. and Newman, H.F. Impact of integration of mental health service and comprehensive medical care. Medical Care, 13~11~:934-943 (November) 1975.

115 Lawson, D.H. and Jick, H. Drug prescribing in hospitals: An inter- national comparison. American Journal of Public Health, 66~7~: 644-648 (July) 1976. Lesse, S. Masked Depression. New York: Jason Aronson, 1974. Lipowski, Z.J. Psychiatry of somatic diseases: Epidemiology, patho- genesis, classification. Comprehensive Psychiatry, 16~2~:105-124 (March/April) 1975. Lipowski, Z.J. Psychosomatic medicine in the seventies: An overview. The American Journal of Psychiatry, 134~3~: 233-244 (March) 1977. McCarthy, E.G. and Widmer, G.W. Effects of screening by consultants on recommended elective surgical procedures. New England Journal of Medicine, 291~25~:1331-1335 (December 19) 1974. McHugh, J.P., Kahn, M.W. and Heiman, E. Relationships between mental health treatment and medical utilization among low-income Mexican- American patients: Some preliminary findings. Medical Care, 15(5):439-444 (May) 1977. Mechanic, D. Social psychologic factors affecting the presentation of bodily complaints. New England Journal of Medicine, 286~20~: 1132-1139 (May 18) 1972. Mechanic, D. Sociocultural and socio-psychological factors affecting personal responses to psychological disorder. Journal of Health and Social Behavior, 16~4~:393-404 (December) 1975. . Mechanic, D. The Growth of Bureaucratic Medicine. New York: John Wiley & Sons, 1976. Mumford, E. Culture: Life perspectives and the social meanings of illness. In: Simons, R. and Pardes, H., (Eds.), Understanding Human Behavior in Health and Illness. Baltimore: Williams and Wilkins, 1977, p. 173-183. Mumford, E., Schlesinger, H.J. and Glass, G.V. A critical review and indexed bibliography of the literature up to 1978 on the effects of psychotherapy on medical utilization. 1978 (unpublished report to NIMH contract No. 278-77-0049 (MH)~. Olbrisch, M.E. Evaluation of a stress management program for high utilizers of a prepaid university health service. Dissertation . submitted to Department of Psychology, The Florida State University, August, 1978. Ogilvie, R.I. and Ruedy, J. Adverse drug reactions during hospitali- zation. Canadian Medical Association Journal, 97~24~:1450-1457 (December 9) 1967.

116 Parkes, C.M. The first year of bereavement: A longitudinal study of the reaction of London widows to the death of their husbands. Psychiatry, 33~4~:444-467 (November) 1970. Patterson, D. and Bise, B. Report pursuant to NIMH contract number 282- 77-0219-MS. January, 1978. Pomerleau, O., Bass, F. and Crown, V. Role of behavior modification in preventive medicine. New England Journal of Medicine, 292~24~: 1277-1282 (June 12) 1975. Rappaport, M. Medically oriented psychiatry: An approach to improving the quality of mental health care. Hospital and Community Psychiatry, 26~12~:811-815 (December) 1975. Regier, D.A., et al. Epidemiological and health services research findings in four organized health/mental health service settings. Paper presented at the ADAMHA Health Maintenance Organization Conference (November 30) 1977. Reidenberg, M.M. Registry of adverse drug reactions. Journal of the American Medical Association, 203~1~:85-88 (January) 1968. Roghmann, K.J. and Haggerty, R.J. Family stress and the use of health services. International Journal of Epidemiology, 1~3~:279-286, 1972. Rosen, J.C. and Wiens, A.N. Changes in medical problems and utiliza- tion of medical services following psychological intervention. In press, American Psychologist, 1979 (prepublication ms). Rosenberg, C.M. and Raynes, A.E. Keeping Patients in Psychiatric Treatment. Cambridge, Massachusetts: Ballinger Publishing Co., 1976. Sackett, D.L. and Haynes, R.B. Compliance with Therapeutic Regimens. Baltimore: Johns Hopkins University Press, 1976. Schlesinger, H.J., Mumford, E. and Glass, G.V. The effects of psychologically-informed intervention on recovery from medical crisis. In Aldrete, J.A. and Guerra, F. Emotional Responses to Anesthesia and Surgery. Grune and Stratton, in press.

117 Sedgewick, P. Medical individualism. (September) 1974. Hastings Center Studies, 2~3~:69-80 Singer, E., Garfinkel, R., Cohen, S.M. and Srole, L. Mortality and mental health: evidence from the Midtown Manhattan restudy. Social Science and Medicine, 10~11 & 12~:517-525 (November/ December) 1976 Stimson, G. and Webb, G. Going to See the Doctor. Routledge and Kegan Paul, 1975. London/Boston: Uris, J.S. Effects of medical utilization and diagnosis on general medical care utilization in a prepaid clinic setting. Report by Western Interstate Commission on Higher Education (WICHE) Intern, Boulder, Colorado, 1974. Vayda, E. A comparison of surgical rates in Canada and in England and Wales. The New England Journal of Medicine, 289~23~: 1224-1229 (December) 1973. Webb, S.B., Jr., Thompson, J.D. and Whitt, I.B. in emergency department utilization. Inquiry (December) 1977. . ~ Statewide trends , 14:402-408

119 PLENARY PRESENTATION APRIL 3, 1979 MENTAL HEALTH AS AN INTEGRANT OF PRIMARY CARE Michael Shepherd, M.D. Professor of Epidemiological Psychiatry Institute of Psychiatry University of London Epidemiological investigations have demonstrated the high pre- valence of psychiatric morbidity presented at the level of primary health care. Further investigation shows that many of these conditions are closely associated with physical ill-health and/or social pathology. The implications of these findings for practice and research at the health/mental health interface are discussed. (The full text of Dr. Shepherd's paper appears in Section VII.)