Frontier Mental Health Services Resource Network

How Little We Know. The impact of mental health policy on those who live in the most isolated rural areas can easily be ignored. Only 2.8 million Americans or 1% of the total population live in "frontier" rural areas (counties with less than 7 persons/sq.mi., 1990 Census). The extremely small relative size of these numbers is conducive to benign neglect of frontier areas — not only within political and policy-making arenas, but also within the research arena. The challenge of small numbers of people spread across large geographic areas permeates virtually every aspect of rural population inquiry. Definition of frontier rural areas, design of analytic methodologies, development of information systems, development of policies that drive allocation of resources, and configuration of services to meet need and demand are all impacted. In an effort to aid development of strategies for surveillance of demand for mental health services in frontier areas, this letter discusses the differentiation of and mutual interrelationships between need, demand, and use and surveillance strategies as a model for dynamic action-oriented policy and research.

Need, Demand, and Use: Differentiation and Mutual Interrelationships

Need for Services. Much has been written about need for services and needs assessment (for recent examples, see Bell, Goldsmith, Lin, Hirzel, & Sobel, 1982; Goldsmith, Lin, Jackson, Bell, & Manderscheid, 1988; Cleary, 1989). Cleary's statement that, "need for services is a relative concept that is a function of the perspective of the person using the term" (1989, p. 161), potentially describes not only that author's perspective of need, but also the variability of definitions within the literature. Two fundamental conceptual frameworks, however, have tended to underlie the development of specific definitions. They are a public health framework (most familiar to mental health and related fields) and an economic framework (particularly germane to the management of health and mental health care). Within the public health framework, need for mental health services has often been defined as having a diagnosable disorder or condition that could profit from receiving service or treatment for that condition (Landerman, Burns, Swartz, Wagner, & George, 1994). Diagnostic or dysfunction criteria usually define need within the public health framework.

Warheit, Bell, and Schwab (1977) identified five general approaches to the assessment of needs: key informants, community forums, rates under treatment, social indicators, and field surveys. The three most often discussed are direct assessment through epidemiologic field surveys; indirect assessment using social indicators; and inferences from rates under treatment. Each procedure has its costs, strengths, and weaknesses. Most public health oriented authors conclude that direct assessment through field surveys is more accurate, though much more expensive, than indirect methods.

Demand for Services. Demand involves the explicit or latent request for a service. Latent demand would occur if an individual requested service for one problem, such as indigestion, when the underlying problem was something else requiring a different intervention strategy.

Demand is affected by need, by economic structure, and finally, by service availability. To use an analogy, if at 10:30 p.m. I look in my refrigerator and discover I have no milk, my need for milk may result in my going to the nearest store to demand milk. If I have insufficient money or if the store is 50 miles away, I may decide not to demand or use milk at this time. If I know before looking in the refrigerator that the nearest milk is 100 miles away, and my car won't start, I may not only not demand milk, I may not even perceive that I need it, and may not exhibit any symptoms of milk need.

Use of Services. Service use is the easiest to define. It is represented by the simple direct measurement of observed empirical variables such as: number of admissions, number of hospital days, or number of outpatient visits. The only difficulties involved are defining what services to enumerate and how to code, scale, and combine enumerations into meaningful indices.

Relationship between Need, Demand, and Use. There is generally a weak association between need and demand. Need and demand differ based on factors such as the nature and type of the identifier of need and variations in help seeking behavior of the person in need. Was diagnosis, impairment, distress, or some combination used as the identifier of need? Was the need determined by professional assessment or by the person in need? Need and demand would tend to be similar if distress or impairment was used as the identifier and if the person in need self-identified. Need and demand may also differ based on the procedures and perspectives used to describe each concept. For example, if a rates-under-treatment variable were used to assess need in an area, then need and demand would show a relatively high association, since both need and demand would be estimated by the same set of variables. Other combinations, such as an epidemiologic field survey assessment of need in combination with an assessment of latent demand might result in low associations, with widely different estimates of need and demand.

There is generally a stronger association between demand and use. These concepts are usually estimated by similar or identical measures. The concepts and estimates differ depending on:

In an economic framework discussed by Landerman et al. (1994), need, demand, and use can be described in terms of the concept of "full price" (the "equilibrium" intersection of demand- and supply-price curves for a given commodity or service at a particular point in time) as follows:

In other words, usage volume of a service at a given "full price" simultaneously defines need, demand, and use--that is, all persons needing the service both demand it and use it, paying the going price. If insurance, discounts, or subsidies with respect to the service increase this equilibrium usage level, such increase is viewed as "not needed" and/or "overuse". On the other hand, if circumstances prevent usage of the service from reaching the "full price" level, and persons exist who would pay the full price to use it if it were available, this decrease represents "residual" or "unmet need". This latter situation might well be found in isolated or frontier areas, while the former might be seen in areas of high service availability (cities, suburbs, etc.) where public subsidies and insurance programs provide greater than "normal" access to the service. The key dynamic variables here are the level of demand and the supply (or availability) of the service at different price levels; "need" is directly expressed as demand, and "unmet need" is also "unmet demand". This economic model is thus different from the behavioral or public-health model, where need (i.e., diagnosis or dysfunction) is defined independently of either demand for, or use of, a service.

Service Availability and the Economic Model. Much of the discussion of demand for mental health services is in the context of economic models (Wells, Manning, Duan, Ware, & Newhouse, 1982; Keeler, Welles, Manning, Rumpel, & Hanley, 1986; Landerman, et al., 1994; Scheffler & Miller, 1989; Frank & McGuire, 1986). In general there are two competing models, demand-side and supply-side. In the current market, a rapid shift from demand-side to supply-side models is occurring. Demand-side cost sharing insurance models (fee for service) manipulate costs, copayment amounts, and annual maximum out-of-pocket expenditures to affect and test elasticity of demand. These models offer little or no incentive for the provider to resist demand (e.g. Keeler, et al., 1986; Landerman, et al., 1994; Frank & McGuire, 1986). Supply-side models, including cost sharing or managed care models (capitated, pre-paid), offer incentives for providers to resist demand, particularly for high cost services. These models are based on a prospective payment for a pool of eligible persons (see McGuire, 1989; Mechanic, Shlesinger, & McAlpine, 1995; Freeman & Trabin, 1994; Hornbrook & Berki, 1985). The supply-side model clearly sets out different economic forces which impact need, demand, and use of services than in demand side models (McGuire, 1989; Mechanic, et al., 1995; Freeman & Trabin, 1994; Hornbrook & Berki, 1985; Feldman, 1994).

Any surveillance system for mental health services demand must be sensitive to concepts of need, utilization, and economic models of service delivery.

Surveillance Strategies: A Model for Dynamic Action-Oriented Policy and Research

The term surveillance, as it is used here, comes from the field of public health epidemiology (Thacker, Berkelman, & Stroup, 1989; Thacker & Stroup, 1994). As defined by Thacker, et al. (1989, p. 188), in association with their work with the Centers of Disease Control and Prevention (CDC), surveillance is:

The key concepts here, differentiating 'surveillance' from other data-related activities, are underscored by Nobre and Stroup (1994, p. 408), who define public health surveillance as:

Surveillance methodologies, as conceptualized and developed within the public health domain, offer a particularly relevant approach to demand assessment.

Though by the above definitions not essential to surveillance, the integration of multiple data sources within surveillance systems is often assumed or recommended. For example, Thacker and Stroup (1994, p. 389) in their discussion of future health surveillance trends state:

Data sources for a surveillance system are adapted to meet the objectives of the surveillance. They may include: statistically valid surveys, hospital records, emergency records, police logs, insurance claims, eligibility rolls, and vital statistics. Although an ideal surveillance system may theoretically be designed, in practice most systems typically 'make-do' with imperfect and incomplete information, using estimation procedures to attempt to make up for data deficiencies. Surveillance systems need to continuously strive to improve data sources, and by linking to decision procedures, refine what data may be required to monitor the system. An integrated surveillance system spanning the vast geography of western frontier states would be highly desirable for the formation of an integrated policy for frontier mental health issues.

A surveillance system, then, is an information system that is data-based, systematic, and ongoing. It is also requisitely linked to stakeholders and decision-making processes. Clearly implied is the notion that surveillance requires not only defining relevant often multiple-source data sets and appropriate analytic methodologies, but also identifying and describing stakeholders and means of disseminating information to those stakeholders.

Stakeholders and Decision-Making. The conceptual and methodological ties of surveillance to stakeholders and decision-making processes provide an orientation well-suited to guiding demand assessment efforts, as well as to efforts in the broader field of rural and frontier mental health assessment. Beeson (1992, p. 4), in addressing future directions vital to the field, observes:

Beeson's observations as well as those within other well-articulated overviews of the rural/frontier mental health field (e.g., Murray & Keller, 1991; Wagenfeld, Murray, Mohatt, & DeBruyn, 1994; Patton, 1989), confirm the need for research and assessment approaches that support action-oriented decision-making by a broad diversity of stakeholders in multiple and complex arenas. These stakeholders span national, state and local levels, private as well as public spheres, policy makers and practitioners within a broad scope of disciplines, and most clearly, the consumers and potential consumers of mental health services.

Somewhat ironically, responsiveness to these numerous and diverse stakeholders may be equally or more important within the most isolated and remote areas than within the densely populated metropolitan areas. In frontier areas, stakeholders or stakeholder groups have minimal opportunity for face-to-face interactions with each other. There is also little opportunity for personal experience of the larger geographic context within which the mental health service system is located. The phrase 'out of sight is out of mind' would be too simplistic in this context. However, timely provision of information and data that effectively depict mental health need, demand, use, and service delivery systems may be particularly valuable to those making decisions on mental health service delivery in frontier areas. Surveillance, by definition, provides not only a conceptual context that recognizes this need, but also a methodological context tailored to address it.

On-going Dynamic Collection, Analysis and Dissemination of Data. A second aspect of the 'surveillance' orientation that is particularly well-suited to the current demand assessment environment is its emphasis on ongoing collection, analysis and dissemination of data (Thacker & Stroup, 1994). As noted by Kessler (1988), the field of needs assessment has been dominated by static measurement approaches. Thus, for example, direct need and demand assessments have often involved one-time administration of a survey instrument; less frequently, they have involved repeated administrations at various defined intervals, perhaps separated by a period of years. Similarly, indirect need and demand assessments have often relied on one-time or discrete intermittent samplings of data from sources such as Census or administrative records.

Undoubtedly numerous factors converged to support emphasis on such static approaches. As an example, Congressional legislation passed during the 1960s and 1970s specified the "assessment of need for services" by applicants for federal grant funds. This spurred a proliferation of activity in needs assessment that carried, in essence, a demand characteristic of 'single-shot' measurement. Less obvious, but perhaps more important, the oldest conceptualizations of 'need' emphasize 'states of being,' with need defined as an objectively determinable state or behavior pattern of an individual (Ciarlo, 1999). Static measurement approaches were thus viewed as appropriate to assessment of static phenomena. Most apparently, until recent years the limitations imposed by predominantly paper-and-pencil-based manpower-intensive technologies highly restricted the use, if not the conceptualization, of more dynamic approaches. Even the advent of widely dispersed computer technology did not greatly impact the static emphasis, but rather facilitated a heightening of the complexity and efficiency of traditional static measurements.

Developments in these and other domains over recent years have, however, underscored not only the desirability but also the feasibility of moving toward a more dynamic model. Increasingly diverse and complex conceptualization of the factors relevant to policy development supportive of judicious allocation of increasingly scarce funding for health and mental health services have moved theory and practice into realms that may incorporate stability but often emphasize variability within a context of change (Cook & Mizer, 1994). Information and communication technologies have reached far beyond the proliferation of isolated mainframe and desktop personal computers to a rapidly evolving environment of multimedia networking that supports virtually instantaneous communications, analysis, and feedback substantially unlimited by geographic distances or boundaries. Correspondingly, major advances in analytic methodologies — from within disciplines including psychology, sociology, epidemiology, economics, market research, mathematics, information sciences, laboratory sciences, and clinical medicine — have immeasurably enhanced capabilities in such areas as time series analysis. This now facilitates a broad variety of identification, estimation, forecasting ,and diagnostic activities, and geographic information systems that significantly heighten the sophistication and speed with which spatial analyses can be conducted.

The desirability of ongoing data collection, analysis, and dissemination characteristic of surveillance systems is particularly compelling at the juncture of frontier populations and assessment of demand for mental health services. The non-urban, non-metropolitan areas of the United States have historically been dynamically changing areas, not only in terms of variations in population numbers, but also in terms of population demographics, economics, communication patterns, structural interrelationships, and a variety of other dimensions (see for example Wagenfeld, et al.,1994; Murray & Keller, 1991; Butler & Beale, 1994; Patton, 1989). Additionally, many of the very policy issues that are driving heightened interest in securing empirical data regarding demand and need for mental health services in rural/frontier areas are issues that can potentially be expected to bring substantial and evolutionary change to the health/mental health service delivery systems in those environments (see for example Mechanic, 1986; Mechanic & Aiken, 1989).

Finally, demand itself is a phenomenon subject to variations occurring in association with changes in such domains as need, help-seeking characteristics, and service delivery system characteristics (Murray & Keller, 1991; DeLeon, et al., 1989). Non-simplistic longitudinal approaches can be expected to play a valuable role in enhancing understanding of these interrelationships (see for example, Goldsmith, et al., 1988).

In sum, current perspectives of the nature and roles of demand for health/mental health services, particularly in the rural/frontier context, yield a portrait of variability rather than constancy across time. These perspectives substantiate the suitability of the types of ongoing research and measurement approaches that are an integral aspect of surveillance methodologies. Ongoing collection, analysis, and dissemination of results offer a requisite sensitivity to change (Cook & Mizer, 1994). It also offers timeliness of response to the varied needs of diverse stakeholders, while concomitantly establishing foundations requisite to the understanding of underlying principles vital to scientific endeavor.

Diverse Data Sources. The surveillance model, in addition to time-related aspects of study, also offers conceptualizations of data source definition and utilization that are particularly suited to the complex realities of the rural/frontier mental health domains. Within the public health arena, surveillance systems have commonly incorporated health data from a broad diversity of sources. They have utilized data from , ". . . populations (individuals and communities) and health care providers (physicians, hospitals, and laboratories), as well as from local, state, and federal health, mental health, law enforcement, and other administrative agencies" (Thacker & Stroup, 1994, p. 384). This orientation potentially provides a richness of data suited to more basic research endeavors as well as to the policy-related decision-making and evaluation tasks. While it may be readily apparent, it should be noted that utility or potential utility must guide the selection of data and data sources in order to assure that the surveillance system is not reduced simply to warehousing unrelated and trivial information. Utilization of diverse data sources holds particular relevance for assessments of demand within rural and frontier settings. For a variety of reasons, frontier populations tend to seek mental health services from a diversity of sources, local and distant, that may or may not be identified as mental health providers. Therefore, many observers have noted the need for a multi-disciplinary involvement in frontier studies using indirect assessment.

An Example: Surveillance of Demand for Emergency Public Mental Health Services in Rural and Frontier Colorado. For the last several years Colorado's ten mental health centers with large rural catchment areas participated with the state’s Division of Mental Health in a National Institute of Mental Health (NIMH) fundedresearch demonstration project called the Colorado Rural Crisis Study (CRCS) (Wackwitz & Wilson, 1992). One aspect of this study involved screening for potentially serious mental health crises that came to the attention of the centers' emergency systems. Of Colorado's 63 counties, 16 were not part of the study, and an additional 12 were eliminated due to partial or incomplete reporting. For the remaining eight community mental health centers (serving 35 counties), nine continuous months of data were selected from a time window of June 1992 through October 1993. The windows were not identical due to differential phase-in and phase-out of the centers. Thus the data set is a typically imperfect surveillance system containing information about client contacts (demand for service) and usage of 24-hour specialty mental health service systems.

We differentiated and then labeled the counties participating in this study on the basis of three characteristics that together appeared to capture their degrees of "frontier-ness" and "rurality" (versus "urbanicity"):

Nonmetropolitan counties with population densities of less than 7 persons per square mile:

Nonmetropolitan counties with population densities of 7 or more persons per square mile:

Using this schema, this Colorado database contained four of the five types of counties:

County Type	No. of Counties	Combined 1990 population
Frontier	15	54,931
"Urbanized" Frontier	5	46,141
"Urbanized" Rural	14	268,125
Urban	1	131,821

To estimate relative demand, the aggregated mental health emergency contacts for the four county types were first multiplied by 4/3 to convert 9 months data to a base of 12 months. These numbers were then divided by the appropriate 1990 population figures to obtain an emergency contacts or "demand" rate for the several county types. As can be seen from Figure 1, demand seems to increase as one moves from frontier to urban types, with the greatest increase between frontier and "urbanized" frontier. However, this analysis is incomplete, and using additional data leads to a more refined and somewhat different conclusion.

Along with indicators of county of residence, the database included town and address information. With this additional information each type can be subdivided by a variable indicating if the contact was from an "urban" county sub-area (population greater than or equal to 2,500) or a "rural" county sub-area. Figure 2 presents this breakdown along with a replication of the total group analysis from Figure 1. From Figure 2, it is apparent that for these data, the partitioning into urban and rural components results in a remarkable similarity across all county sub-area types. The urban demand rate is 4.3 to 5.5 times the rural (non-urban) rate, while the rural demand rate varies by only 0.07%. Thus, averaging demand for rural and urban sub-areas within each county type masked differences based on the level of "rurality" (here defined by the US Census as residence outside of a town or city of at least 2,500 people). The above finding brings us back to the intersection of need, demand, use, and availability. Without further study of these other variables we are unlikely to know if need, and therefore demand is greater in urban areas, or if demand is less simply because services are less available.

Final Observations

This leads to a final observation regarding the suitability of surveillance models. Rural and frontier areas can be found virtually all across the country, and the characteristics of these populations are diverse. Further, many of the policy decisions regarding mental health services will impact a large variety of rural/frontier areas or populations. Kessler (1988) provides an information–related perspective, " . . . In turn, the noted need for research and assessment approaches that hold the potential to support action-oriented decision-making by a broad diversity of stakeholders in multiple and complex arenas implies the desirability of systems responsive to varying information requirements." Thus, while as Cleary (1989) observes, ". . . the statement that need, demand and service use are three distinct variables affected by different factors is now almost a truism," it is also true that demand has come to be considered one element of a tightly integrated package of need, demand, use, and help-seeking.

To meet the information requirements of decision-making entities, surveillance studies must first clearly define a set of goals, objectives, and purposes of the proposed surveillance. Data should be collected from sources other than just specialty mental health providers, including primary care, insurance, emergency systems, and police sources. It is important to know the area of service as well as the area of residence of the client served, and to use the smallest definable area (such as the census tract rather than the county) so that differences are not masked. Sociodemographics and social indicators can help to assess differences between types and differences in need. Finally, surveillance of need should be linked with surveillance of demand, surveillance of services, and surveillance of economics.

In sum, the study of demand for health/mental health services in rural and frontier areas cannot reasonably be considered to be the study of a phenomenon at rest, nor even of forces in equilibrium. More appropriately, it is the study of processes marked by continuous activity and change. That study, it would seem, requires not static measurement approaches, but approaches as dynamic as the processes that inspire it.

References

Beeson, P. (1992). Rural mental health research: The next generation. Outlook, 2, 2-5.

Bell, R.A., Goldsmith, H.F., Lin, E., Hirzel, R.K., & Sobel, S. (Eds.). (1982). Social indicators for human service systems. Louisville, KY: Department of Psychiatry, University of Louisville.

Butler, M., & Beale, C. (1994). Rural-Urban Continuum Codes for Metro and Nonmetro Counties, 1993 (Staff Report No. AGES 9425). Washington, D.C.: U.S. Department of Agriculture, Economic Research Service.

Ciarlo, J. (1999). Estimating and monitoring need for mental health services in frontier rural areas. University of Denver. Manuscript in preparation.

Cleary, P.D. (1989). The need and demand for mental health services. In C. A. Taube, D. Mechanic, & A. A. Hohmann (Eds.), The future of mental health services research. Washington, D.C.: National Institute of Mental Health.

Cook, P.J., & Mizer K.L. (1994, December). The revised ERS county typology: An overview (Rural Development Research Report 89). Washington, D.C.: Rural Economy Division, Economic Research Service, U.S. Department of Agriculture.

DeLeon, P.H., Wakefield, M., Schultz, A.J., Williams, J., & VandenBos, G.R. (1989). Rural America. Unique opportunities for health care delivery and health services research. American Psychologist, 44, 1298-1306.

Feldman, R. (1994). The cost of rationing medical care by insurance coverage and by waiting. Health Economics, 3, 361-372.

Frank, R.G., & McGuire, T.G. (1986). A review of studies of the impact of insurance on the demand and utilization of speciality mental health services. Health Services Research, 21, 241-265.

Freeman, M.A., & Trabin, T. (1994). Managed behavioral health care: History, models, key issues, and future course. Washington, D.C.: Department of Health & Human Services.

Goldsmith, H.F., Lin, E., Jackson, D.J., Bell, R.A., & Manderscheid, R.M. (1988). The future of mental health needs assessment. In H.F. Goldsmith, E. Lin, R.A. Bell, & D.J. Jackson (Eds.), Needs assessment: Its future (Publication No. (ADM)88-1550). Washington DC: Superintendent of Documents, US Government Printing Office.

Hornbrook, M.C., & Berki, S.E. (1985). Practice mode and payment method. Effects on use, costs, quality, and access. Medical Care, 23, 484-511.

Keeler, E.B., Welles, K.B., Manning, W.G., Rumpel, J.D., & Hanley, J.M. (1986). The demand for episodes of mental health services. Rand Corporation.

Kessler, R.C. (1988). A critical evaluation of mental health needs assessment. In H.F. Goldsmith, E. Lin, R.A. Bell, & D.J. Jackson (Eds.), Needs assessment: Its future (Publication No. (ADM)88-1550). Washington DC: Superintendent of Documents, US Government Printing Office.

Landerman, L.R., Burns, B.J., Swartz, M.S., Wagner, H.R., & George, L.K. (1994). The relationship between insurance coverage and psychiatric disorder in predicting use of mental health services. American Journal of Psychiatry, 151, 1785-1790.

McGuire, T.G. (1989). Combining demand- and supply-side cost sharing: The case of inpatient mental health care. Inquiry, 26, 292-303.

Mechanic, D. (1986). The concept of illness behavior: Culture, situation and personal predisposition. Psychology in Medicine, 16, 1-7.

Mechanic, D., & Aiken, L.H. (1989, Fall). Capitation in mental health: Potentials and cautions. New Directives in Mental Health Services, 43, 5-18.

Mechanic, D., Shlesinger, M., & McAlpine, D.D. (1995). Management of mental health and substance abuse services: State of the art and early results. The Milbank Quarterly, 73, 19-55.

Murray, J.D., & Keller, P.A. (1991). Psychology and rural America. Current status and future directions. American Psychologist, 46, 220-231.

Nobre, F.F., & Stroup, D.F. (1994). A monitoring system to detect changes in public health surveillance data. International Journal of Epidemiology, 23, 408-416.

Patton, L. (1989). Setting the rural health services research agenda: The congressional perspective. Health Services Research, 23, 1005-1051.

Scheffler, R.M. & Miller, A.B. (1989). Demand analysis of mental health service use among ethnic populations. Inquiry, 26, 202-215.

Thacker, S.B., Berkelman, R.L., & Stroup, D.F. (1989) The science of public health surveillance. Journal of Public Health Policy, 10, 187-203.

Thacker, S.B., & Stroup, D.F. (1994). Future directions for comprehensive public health surveillance and health information systems in the United States. American Journal of Epidemiology, 140, 383-397.

Wackwitz, J.H. & Wilson, N.Z. (1992). The Colorado rural crisis study: Comparing methods for delivering rural crisis services. Outlook, 2, 21-23.

Wagenfeld, M.O., Murray, J.D., Mohatt, D.F., & DeBruyn, J.C. (1994). Mental health and rural America: 1980-1993. Washington, D.C.: Office of Rural Health Policy & Office of Rural Mental Health Research.

Warheit, G., Bell, R.A., & Schwab, J. (1977). Needs assessment approaches: Concepts and methods (Dept. of Health, Education, and Welfare Publication No. (ADM)77-472). Washington DC: Superintendent of Documents, US Government Printing Office.

Wells, K.B., Manning, W.G., Jr., Duan, N., Ware, J.E., & Newhouse, J.P. (1982). Cost sharing and the demand for ambulatory mental health services. Rand Corporation.

Surveillance of Demand for Mental Health Services in Frontier Areas

Letter to the Field No. 15 - DRAFT

Introduction

Need, Demand, and Use: Differentiation and Mutual Interrelationships

Final Observations

References