Frontier Mental Health Services Resource Network

The often-limited availability of mental health services to residents of rural America has been an issue of importance to those residents, to states with large rural populations, and to the federal government (Human & Wasem, 1991). A closely related issue is that of utilization of such services by rural persons needing mental health (MH) services — that is, to what degree are these people willing and able to use what services may be available in their area? This Letter discusses MH service utilization primarily in terms of comparisons of rural service use with that of non-rural areas, and points to some potentially important implications for governments hoping to provide greater assistance with mental and emotional problems than has been typical in the past.

Recent epidemiologic surveys have established that, on the whole, the need for MH services is only slightly less prevalent for Americans living in rural areas than it is for urban or suburban residents. For example, a national survey by Kessler, et al. (1994) found that the prevalence rate of any formally diagnosable mental disorder was only 1.1 times higher in major metropolitan areas than in rural (i.e., nonmetropolitan) areas. Also, Ciarlo and Tweed (1992) found that while relatively isolated "rural non-towns" in Colorado had somewhat lower prevalence of need for services (diagnoses and everyday dysfunction) than urbanized areas, the more populous rural "towns" (that is, urban areas of at least 2,500 people) had very nearly the same need prevalence as major cities in that state. In sharp contrast, however, there is considerable evidence that the availability of services for mental disorders is substantially lower in rural areas than in urban ones — and especially so in the lowest-density, more isolated rural areas often termed "frontier" (Goldsmith, Wagenfeld, Manderscheid, & Stiles, 1996; Holzer, Mohatt, Goldsmith, & Ciarlo, in press). Hence, it is important to learn whether this relative scarcity of MH services in rural areas may reduce the number and proportion of needy residents able to access and receive or "utilize" MH services. The purpose of this Letter is to examine the extent to which there may be differential utilization of available services in rural and non-rural areas, through reanalysis of the 1989 Mental Health Supplement to the National Health Interview Survey (NHIS) (National Center for Health Statistics, 1993).

Use of the Mental Health Supplement of the NHIS provides both advantages and disadvantages for the present purpose. Its greatest strengths are its large sample size (n = 84,572) and national sampling frame. It provides enough nonmetropolitan respondents for accomplishing meaningful analysis of at least part of the "rural" component of MH service usage (i.e., rural towns). Further, the data are relatively recent (1989) compared to survey data from the Epidemiologic Catchment Area Study (ECA) (Robins & Regier (1991), which was conducted between 1980 and 1983. The data are not much older than those of the National Comorbidity Survey (NCS) by Kessler, et al. (1994), which was conducted between 1990 and 1992. One disadvantage of this NHIS Supplement is that it assessed mental disorders by a simple self-report or informant report of whether a disorder is present. Unlike the ECA and NCS, it does not employ a formal diagnostic questionnaire and thus reports lower prevalence rates. It has been speculated that this is because respondents are unable or unwilling to identify illnesses without the extended probing of a structured diagnostic instrument, or without the feedback and labeling received in treatment of mental disorders. On the other hand, the Supplement asks simple direct questions about the use of MH services that are roughly equivalent to those asked in the ECA and NCS. The relationship between prevalence and utilization will be explored further below.

Defining "Rural" and Other Areas

Unfortunately, data to examine MH issues in rural America are often limited by the particular definitions of "rural" used in epidemiologic studies of geographic areas, and/or the specific ways in which geographic data are categorized or "geo-coded"; the NHIS Supplement is no exception. Therefore, in addressing the above mental health issues, it is important to revisit definitions of "rural" and "urban" areas at least briefly, because they can be relatively complex and cut across a variety of social, cultural and economic issues. For a more thorough discussion of these definitional issues, the reader is referred to FMHSRN's Letters to the Field No. 2 (Ciarlo, Wackwitz, Wagenfeld, & Mohatt, 1996) and No. 21 (Zelarney & Ciarlo, 1999) of this same series.

"Urban" generally denotes cities, and implies cities of some size. A commonly used designation for urban is "metropolitan" or "metro", defined by the Office of Budget Management as an area having either a city with more than 50,000 residents or a Census Bureau defined urbanized area and a total population of at least 100,000. As a designation, however, being identified as metro and therefore urban can fail to differentiate the vast organizational differences between central cities of a large metropolitan area, the smaller cities of the same metropolitan area, and even the suburbs and exurbs which are part of the overall area. For many reasons, health care organizations, such as hospitals and clinics, are found concentrated in central cities; persons living in adjacent areas are expected to "go downtown" for care. Satellite clinics may be available outside the central city, but these are frequently not well documented in available services data sets.

"Rural" areas are perhaps even more difficult to define and characterize than urban areas. Most are defined by either being outside a metro areas (nonmetropolitan or nonmetro) or by the lack of a city or town greater than 2,500 persons. However, some rural areas are adjacent to urban areas and others are not, thus greatly influencing access to some kinds of health care. Some rural areas have farm-based economies while others do not. The Frontier Mental Health Services Resource Network has focused on isolated "frontier" areas having less than seven persons per square mile. Other approaches such as the "Rural-Urban" typology of the U.S. Dept. of Agriculture (Hines, Brown, & Zimmer, 1975; Butler & Beale, 1994) deal more with the size and proximity of cities and towns to a county area. However, these definitional schemes often do not characterize the cultural or economic diversity of rural areas, either in general or as related to mental health. For that reason, any conclusions about the mental health of U.S. "rural" and "urban" areas must be carefully limited to just the specific meanings of these terms as used in a given study.

A variant of the metropolitan vs. non-metropolitan definition of "urban" and "rural" was used in the NHIS study discussed here. It provides four levels of urban-rural coding: residence within a metropolitan area or outside one (nonmetropolitan); if within a metro area, whether in a central city or not; and if outside, whether the residence is a farm or nonfarm.

Comparative Prevalence of Disorders in Rural and Urban Areas

The ECA project has reported some limited "rural/urban" comparisons for a number of disorders, employing the common non-metropolitan vs. metropolitan county definitions of these terms. The urban lifetime prevalence for any disorder is 34%, which is only slightly higher than 32% for rural areas (Robins & Regier, 1991), and for the past one-year prevalence the comparison is 21% vs. 20%. It should be noted that the ECA had relatively small rural samples, primarily in North Carolina and Missouri, which were compared to cities such as New Haven, Baltimore, St. Louis, and Los Angeles. In the NCS study, which was based on a national sample, comparisons were made that differentiated major urban, other urban, and rural, with the latter being equivalent to nonmetropolitan. These comparisons showed only slightly higher prevalence of several disorders for the major metropolitan and other urban as compared to rural areas, but none of the comparisons was significant. The prevalence rate of any formally diagnosable mental disorder was only 1.1 times higher in major metropolitan areas than in rural (i.e., nonmetropolitan) areas, which was not statistically significant.

In considering these small differences in ECA and NCS prevalence rates surveyed across areas, one might expect there to be at most a small differential between urban and rural utilization of services. However, both the ECA and the NCS reported rates of services utilization which were far smaller than the reported disorder prevalence rates. Hence, for the NHIS with its sharply lower prevalence rates, any utilization rates are expected to be still smaller.

Methods

The NHIS is a large national survey of a variety of health conditions initiated in 1957 and conducted continuously since that time by the U.S. Bureau of the Census, under specifications from the National Center for Health Statistics. The sampling design is closely linked to the Current Population Surveys of the U.S. Census and consists of a multistage sampling design of the noninstitutionalized civilian population of the U.S. The sampling design is intended to result in approximately 49,000 housing units and 132,000 persons per year.

The core questions of the NHIS survey include basic health and demographic items, which are asked of a person in a household or of a household informant about each eligible member of the household. These items include disability days, physician visits, acute and chronic conditions, limitations of activity, and hospitalizations. In addition, representative subsamples of households are asked to respond to questions on special health topics, and in 1989 mental health was one of those topics. The MH questions included items asking whether specific disorders were present, and whether particular types of MH services were used and when. Interviews were completed with 96 percent of all eligible households.

The identification of mental disorders in the 1989 mental health supplement was based on the question: "During the past 12 months did this [person] have [..any of the ..] following mental and/or emotional disorders?" The listed items included: schizophrenia, paranoid or delusional disorder, manic episodes, manic depression, major depression, personality disorder, senility, alcohol abuse, drug abuse, and/or mental retardation. Then the respondent/informant is asked whether this person "had any other mental disorder" and "what the other disorder is called". Finally, a summary "mental disorder reported" was coded which includes the mental disorders and senility. Substance abuse or mental retardation were coded separately.

Utilization of services was identified by the question "When [this person] last saw a mental health professional about [any] disorder(s)." The answers were coded into the categories: less than two weeks, less than one month, less than three months, less than one year, less than five years, five years or more, and never. Those with no known mental disorder received a blank code.

Results

Table 1 presents the distribution of the sample and the population estimates by the urban-rural categorization. The sample column is the number of persons who responded or who had informant provided information. The population is the weighted population estimate based on the sample, and the percentages are calculated on the weighted population estimates.

Category	Sample	US Population	Percent*
Metropolitan, central city	27,122	55,200,859	30.75
Metropolitan, noncentral city	37,905	85,071,068	47.39
Nonmetropolitan, nonfarm	18,287	36,850,642	20.53
Nonmetropolitan, farm	1,258	2,406,353	1.34
Total	84,572	179,528,922	100.01

* The population figures and percentages are based on population weights (NWTFA).

Table 2 presents the number and percentage of specific disorders reported by respondents or household informants broken down by the type of urban-rural residence. It is immediately apparent that the rates presented for mental disorders are substantially lower than those presented in epidemiologic surveys based on diagnostic interviews, and by an order of magnitude or more (for example, the rates for the ECA or the NCS noted above were 21%-urban and 20%-rural).

	Metropolitan				Nonmetropolitan
	Central city		Noncentral city		Nonfarm		Farm
	N	Percent	N	Percent	N	Percent	N	Percent
Schizophrenia, paranoid or delusional disorder	106	0.34	69	0.17	44	0.24	4	0.35
Manic episodes, manic depression, and major depression	307	1.16	388	1.08	198	1.05	10	0.75
Personality disorder, senility, other mental disorder	156	0.55	152	0.39	109	0.62	9	0.64
Any mental disorder	569	2.04	609	1.64	351	1.91	23	1.73
Substance abuse (alcohol abuse, drug abuse)	106	0.37	131	0.35	52	0.32	1	0.08
Any mental disorder or substance abuse	675	2.40	740	1.99	403	2.23	24	1.81
Mental retardation	45	0.16	53	0.14	33	0.16	4	0.36
Any of the above	720	2.56	793	2.13	436	2.39	28	2.17
No disorder reported	26402	97.44	37112	97.87	17851	97.61	1230	97.83

Table 3 presents a typology of reported utilization of MH services by persons with a mental disorder, excluding substance abuse and mental retardation. The four levels are persons with no reported mental disorder; disorder but no use of services; past use; and use in the last 12 months (current utilitzation). Percentages are weighted for the adult population. As can be seen, reported current utilization is highest for those in central cities (1.22%), followed by 1.06% for nonfarm, 0.92% for noncentral-city areas, and 0.91% for farm areas. Rates of utilization in the past are quite low, and are again lowest for the farm (0.16%) and noncentral-city metropolitan areas (0.25%). Combining past and current utilization yields rates of lifetime utilization not much higher than for the last year.

	Metropolitan				Nonmetropolitan
	Central city		Noncentral city		Nonfarm		Farm
	N	Percent	N	Percent	N	Percent	N	Percent
No disorder reported – no utilization	26553	97.96	37296	98.36	17936	98.09	1235	98.27
Disorder reported - never used services	141	0.50	172	0.46	98	0.54	9	0.67
Disorder reported –used in the past only	81	0.32	94	0.25	59	0.31	2	0.16
Disorder reported -used in last 12 mos.	347	1.22	343	0.92	194	1.06	12	0.91

Table 4 presents utilization in the last 12 months as a percentage of persons with a disorder reported for the same time period. Those with no reported utilization but a current disorder can be considered cases of unmet need for services. Those in the central cities reported the highest utilization relative to disorder, with 59.75% reporting current use. Noncentral-city, metropolitan area residents reported the next highest level of utilization —56.45%. The nonfarm areas were next with 55.43% utilization, and the farm areas had the lowest utilization (52.53%) and the highest unmet need (47.47%).

	Metropolitan				Nonmetropolitan
	Central city		Noncentral city		Nonfarm		Farm
	N	Percent	N	Percent	N	Percent	N	Percent
Any mental disorder	569	100.00	609	100.00	351	100.00	23	100.00
Current use	347	59.75	343	56.45	194	55.43	12	52.53
No current use	222	40.25	266	43.55	157	44.57	11	47.47

Table 5 presents the type of mental health professional last seen by persons who received specialty mental health services. In each of the urban-rural areas the mental health professional identified is predominantly a psychiatrist. Psychologists are listed as a distant second, which is unexpected given the role of psychologists in mental health facilities and the broad definition of psychologist used in this study. Only 14 persons reported contact with a doctoral level psychologist. Social workers and nurses are rarely identified as the last MH contact. The larger number of "other mental health workers" is exclusive of the categories designated above, but may reflect the work of case workers or specialists whose formal designation is unknown. Only a few nonpsychiatrist physicians were identified, most likely because the question used the term "mental health professional".

	Metropolitan				Nonmetropolitan
	Central city		Noncentral city		Nonfarm		Farm
	N	Percent	N	Percent	N	Percent	N	Percent
Psychiatrist	258	62.10	245	56.94	145	60.64	10	76.46
Psychologist*	67	17.50	77	20.25	51	19.70	2	14.64
Social Worker**	19	4.33	23	6.25	7	2.87	-	-
Other MD	8	1.71	16	3.38	5	1.84	-	-
RN	-	-	3	0.65	1	0.44	-	-
Other MH worker	57	14.37	55	12.53	38	14.52	1	8.89

Discussion

Results from the above analyses are largely consistent with our expectations regarding urban-rural differences, given the initial finding that the reported prevalence of MH disorders is sharply lower than in standard mental health epidemiologic surveys such as the ECA, NCS, and CSHS. Our expectation was that urban-rural differences in the prevalence of mental disorder would be relatively small, paralleling other surveys. The present analyses show a somewhat lower reported prevalence in nonmetropolitan areas, but the large sample size of the present survey makes it clear that a simple "urban-rural" classification is inadequate to characterize the differences found. The highest rates of Any Mental Disorder are reported in the central cities of metropolitan areas; however, the noncentral-city metropolitan areas show a much lower prevalence of disorder—lower even than the nonmetropolitan farm areas. Further, the nonmetropolitan nonfarm areas show higher prevalence than farm areas, and very nearly reach the level of central cities. Interestingly, this same pattern of disorder prevalence was found in Colorado subareas (Ciarlo & Tweed, 1992), where the non-central city metropolitan areas were labeled "exurban", and the nonmetropolitan nonfarm areas were called "rural towns". Finding the same patterning of prevalence of MH service needs on a national-scale survey thus makes it clear that the debate over urban-rural differences must move beyond the use of a simple urban-rural dichotomy. While it is unfortunate that the present data set does not include the greater levels of differentiation afforded by the identification of "frontier" areas or the use of the USDA rural-urban typology (Butler & Beale, 1994), it does indicate clearly that greater refinement in our classifications are essential.

An important issue is the lower prevalence rates generated from the present survey in comparison to the estimates generated by the diagnostic surveys of the ECA and NCS. The most likely explanation for these differences is the simple self-report of "disorder" used in the NHIS, which inquires about the disorders in terms of their labels or names. In contrast, the diagnostic surveys ask about patterns of symptoms found in DSM-III or IV diagnostic systems, and subsequently identify all disorders matching the presenting symptomtalogy (some or all of which disorders may not have been identified by the respondents, their families, or their doctors). The latter is a far more sensitive diagnostic approach than simple self-report of disorders, because even persons who have psychiatric symptoms may not recognize them as such, and/or may not have the knowledge necessary to identify and label them as disorder. Further, knowledge of mental illness and the elements of making a self-diagnosis vary widely throughout U.S. culture, with some expectations that younger, better educated, and more cosmopolitan persons are more likely to have the knowledge base to do so than persons who grew up in a less psychologically-oriented culture. Such cultural differences are likely to be related to the urban-rural categorization, but not necessarily in a tidy manner.

The NHIS Mental Health Supplement also addresses issues of service utilization, with some clear urban-rural differences. Again, however, we note that the reported rates of services utilization are much lower than reported in the ECA and NCS surveys. This is perhaps to be expected, given the greater focus on mental health issues in the latter surveys; it certainly raises the issue of how important it is to establish a specifically mental health symptoms/problems context when performing interviews related to use of MH services.

We have also considered the question of whether extensive use of "household informants" in the NHIS survey would make a difference; such differences were observed when informant interviews were conducted in the ECA project. In the present study, analysis of informant responses regarding another household member showed lower rates of reported disorder and service utilization than did self-reports. The overall rate of disorder for completely self-reported information is 2.33%, for partial self-report 2.21% and for a proxy interview 1.81%. Similarly, the reported utilization rate for complete self-report is 1.15%, for partial self-report it is 1.05%, and for proxy interviews it is 0.80%. This may have significantly lowered over-all results, as about one-third of the data come from proxy interviews.

Notwithstanding these concerns, we do see trends for lower MH services utilization in the nonmetropolitan areas as compared to metropolitan central cities, though again the nonmetropolitan nonfarm areas are fairly close to the latter. This appears to be true both as a percentage of the total population and as a percentage of only those persons in need. And again it is apparent that utilization rates are lower in the noncentral-city metropolitan areas than in the nonfarm nonmetropolitan areas. Finally, farm areas show the lowest levels of utilization in both absolute and relative terms. However, the tiny number of western-U.S. farm areas included (51) make this conclusion hard to generalize to "frontier"-area farms, the vast majority of which lie west of the 100th meridian (which runs from North Dakota to Texas).

It was surprising that the highest rates of MH service utilization were reported for psychiatrists, followed by psychologists of all levels and unspecified mental health workers. This pattern was unexpected because the staffing of most mental health facilities is weighted in favor of nonpsychiatrists. Further, we know that much of the care nationally for mental health problems is provided in primary care settings, yet nonpsychiatrist M.D.s were barely represented in the

provider list (perhaps because the utilization questions asked when a person "last saw a mental health professional"). This may also be partly because many persons treated in primary-care settings do not receive psychiatric diagnostic labels, even if they receive psychoactive medications; hence, such utilization may have gone unreported in this survey.

In conclusion, we must note that the expected pattern of lower rates of utilization in nonmetropolitan areas was confirmed by this survey, despite the unexpectedly low rates of disorder and utilization reported. Finally, it should also be noted that ongoing changes in the entire health care system, especially the advent of managed mental health care (or "managed behavioral health care" as it is often termed), are likely to have brought about changes in the provision and utilization of such services over the decade since these data were collected.

References

Butler, M.A., & Beale, C.L. (1994) Rural-urban continuum codes for metro and nonmetro counties, 1993. Washington, D.C.: U.S. Department of Agriculture, Economic Research Service.

Ciarlo, J. A., & Tweed D.L. (1992). Exploring rural Colorado’s need for mental health services: Some preliminary findings. Outlook , 2(3), 29-31.

Ciarlo, J.A., Wackwitz, J.H., M.O. Wagenfeld, & Mohatt, D.F. (1996). Focusing on "frontier": Isolated rural America (Letter to the Field No. 2). Denver, CO: Frontier Mental Health Services Resource Network.

Goldsmith, H.F., Wagenfeld, M.O., Manderscheid, R.W., & Stiles, D.J. (1996). Geographical distribution of organized mental health services (Chapter 8). In Mental health, United States, 1996 (pp. 154-167). Rockville, MD: U.S. Department of Health and Human Services.

Hines, F.L., Brown, D.L., & Zimmer, J.M. (1975). Social and Economic Characteristics of the Population in Metro and Nonmetro Counties. 1970. (Report #AER-272). Washington D.C.: Economic Research Service, U.S. Department of Agriculture.

Holzer, C.E. III, Mohatt, D.F., Goldsmith, H.F., & Ciarlo, J. (in press). The Availability of Health and Mental Health Providers by Urban-Rural County Type. In Mental Health, United States, 1998. Rockville, MD: U.S. Department of Health and Human Services.

Human, J. & Wasem, C. (1991). Rural mental health in America. American Psychologist, 46(3), 323-339.

Kessler, R.C., McGonagle, K.A., Zhao, S., Nelson, C.B., Hughs, M., Eshleman, S., Wittchen, H.U., & Kendler, K.S. (1994). Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Archives of General Psychiatry, 51, 8.

National Center for Health Statistics. (1993). 1989 National Health Interview Survey (CD-ROM). Washington D.C.: U.S. Department of Health and Human Services, Public Health Service, Centers for Disease Control and Prevention.

Robins, L.N. & Regier, D.A. (Eds.). (1991) Psychiatric disorders in America: TheEpidemiologic Catchment Area Study. New York: The Free Press.

Zelarney, P., & Ciarlo, J. (1999). Defining and describing frontier areas in the United States: An update (Letter to the Field #21). Denver, CO: Frontier Mental Health Services Resource Network.

Mental Health Service Utilization in Rural and Non-Rural Areas