| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
RESEARCH AND PRACTICE |
Gopal K. Singh is with the Division of Cancer Control and Population Sciences, National Cancer Institute, National Institutes of Health, Bethesda, Md. Mohammad Siahpush is with the VicHealth Centre for Tobacco Control, Cancer Control Research Institute, Anti-Cancer Council of Victoria, Carlton, Australia.
Correspondence: Requests for reprints should be sent to Gopal K. Singh, PhD, MS, MSc, National Cancer Institute, Division of Cancer Control and Population Sciences, 6116 Executive Blvd, Suite 504, MSC 8316, Bethesda, MD 20892-8316 (e-mail: gopal_singh{at}nih.gov).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Objectives. This study examined rural–urban gradients in US suicide mortality and the extent to which such gradients varied across time, sex, and age.
Methods. Using a 10-category rural–urban continuum measure and 1970–1997 county mortality data, we estimated rural–urban differentials in suicide mortality over time by multiple regression and Poisson regression models.
Results. Significant rural–urban gradients in age-adjusted male suicide mortality were found in each time period, indicating rising suicide rates with increasing levels of rurality. The gradient increased consistently, suggesting widening rural–urban differentials in male suicides over time. When controlled for geographic variation in divorce rate and ethnic composition, rural men, in each age cohort, had about twice the suicide rate of their most urban counterparts. Observed rural–urban differentials for women diminished over time. In 1995 to 1997, the adjusted suicide rates for young and working-age women were 85% and 22% higher, respectively, in rural than in the most urban areas.
Conclusions. The slope of the relationship between rural–urban continuum and suicide mortality varied substantially by time, sex, and age. Widening rural–urban disparities in suicide may reflect differential changes over time in key social integration indicators. (Am J Public Health. 2002;92:1161–1167)
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
A number of studies have shown that rural areas tend to have higher suicide mortality rates than urban areas.12,14–16 Besides physical and social isolation and limited opportunities for social interaction and networks in rural areas, a number of other societal factors, such as unfavorable changes in the demographic structure and socioeconomic and industrial activity that causes social instability and disruption, have been cited as possible reasons for comparatively higher suicide rates in rural areas.12,14–16 The extent to which suicide rates vary in response to the degree of rurality or urbanization has not been well studied. Furthermore, the degree to which rural–urban disparities in US suicide mortality among men and women have changed over time has not been examined. This report empirically examines rural–urban patterns in US suicide mortality and the extent to which rural–urban gradients in male and female suicide mortality have changed, both overall and for those aged 15 to 24, 25 to 64, and 65 years and older, during the period 1970 to 1997.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
Using national mortality data files maintained by the National Center for Health Statistics, we obtained age-, sex-, and county-specific annual suicide deaths from 1970 through 1997.1,2 Age-, sex-, and county-specific population estimates for 1970 to 1997 prepared by the US Bureau of the Census were used as denominators.18,19 Each of the 3103 counties on the mortality data set was assigned 1 of the 10 rural–urban continuum codes. Alaska was not assigned a code and was therefore excluded from our analysis. Age-, sex-, and county-specific deaths and populations were summed within the 10 categories of the rural–urban variable and the following 6 time periods: 1970 to 1974, 1975 to 1979, 1980 to 1984, 1985 to 1989, 1990 to 1994, and 1995 to 1997. The adjustment by age of suicide mortality rates was performed by the direct method, with the age composition of the 1990 US population used as the standard. We calculated ageadjusted death rates and standard errors across each rural–urban category and time period by using 5-year age-specific death rates for the following 18 age groups: 0 to 4, 5 to 9, 10 to 14, 15 to 19, 20 to 24, 25 to 29, 30 to 34, 35 to 39, 40 to 44, 45 to 49, 50 to 54, 55 to 59, 60 to 64, 65 to 69, 70 to 74, 75 to 79, 80 to 84, and 85 years and older.1,20
The rural–urban differentials in suicide mortality were estimated separately for men and women. To examine the extent to which the rural–urban gradient varied across time, we modeled, using multiple linear regression, county-specific age-adjusted suicide rates (the dependent variable) as a function of rural–urban continuum (treated as a continuous independent variable) separately for each of the 6 time periods. The yearly suicide mortality trends for the most rural and most urban county groups were summarized by annual exponential rates of change.21 To estimate relative risks of suicide for each rural–urban continuum category, we fitted Poisson regression models to the age-, sex-, and county-specific suicide death data with a log link function and the corresponding stratum-specific log population as an offset variable for the time periods 1980 to 1984, 1985 to 1989, 1990 to 1994, and 1995 to 1997.22 The age- and sex-specific Poisson models were also fitted for 1970 to 1974 and for 1975 to 1979, but these models are not presented or discussed here for the sake of brevity. In all Poisson models, the most urban county group was selected as the reference category. All models, fitted by the SAS GENMOD procedure,23 showed reasonable fit as determined by the likelihood ratio statistic or deviance.
We also estimated the effect of the rural–urban continuum on suicide mortality after adjusting for county-level variations in ethnic composition and for divorce rate, an important indicator of social disintegration.3,4 High divorce rates are associated with rising suicide rates,3,4,11,12 and Whites and American Indians have substantially higher suicide rates than other racial/ethnic groups in the United States.24 Thus, we included county-level divorce rates (number of divorces per 1000 people aged 15 years and older), the percentage of Whites, and the percentage of American Indians in 1970, 1980, and 1990 as control variables along with rural–urban continuum (coded as a continuous variable) in the sex- and time-specific regression models of county suicide rates. Other important social integration indicators, such as unemployment rate, percentage of population living alone, percentage employed in agricultural or farming occupations, and income inequality, were considered but were excluded from the county-level regression models because of their high collinearity with divorce rate.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
shows age-adjusted male and female suicide rates for the 10 rural–urban county groups over each of the 6 study time periods. For men, suicide rates were generally higher in more rural areas than in urban areas, especially for the more recent time periods. For the periods 1980 to 1984, 1985 to 1989, 1990 to 1994, and 1995 to 1997, the male suicide rate for the most rural county group was respectively 21%, 26%, 37%, and 54% higher than the rate for the most urban county group. The rural–urban gradient for males, measured by the regression slope of county-level age-adjusted suicide rates on the rural–urban continuum, was positive and statistically significant for each time period except 1970 to 1974, suggesting rising suicide rates with increasing levels of rurality. Moreover, the size of the rural–urban gradient increased consistently across time, indicating widening rural–urban differentials in male suicide mortality.
|
Figure 1 shows yearly trends in ageadjusted suicide rates for the most rural and most urban county groups. Consistent with the aforementioned results in Table 1
, the widening of rural–urban differentials for men and the narrowing of differentials over time for women is clearly evident. During the period 1970 to 1997, while the suicide rate for men in the most urban areas decreased at an average annual rate of 0.46% (95% confidence interval [CI] = 0.30%, 0.61%), the rate for men in the rural areas grew at an average annual rate of 1.08% (95% CI = 0.85%, 1.31%). The observed suicide rate for women in the most urban areas decreased at an average annual rate of 3.19% (95% CI = 3.03%, 3.36%), whereas the rate for women in the rural areas declined only slightly or remained fairly stable from 1970 to 1997.
|
, not only did rural–urban patterns in suicide mortality vary by sex, but the crude rural–urban gradients in suicide were generally much steeper for men than for women. Men had substantially higher suicide rates than women in each county group and time period. However, sex differentials in suicide mortality appear to have increased over time in almost all county groups. For example, the suicide rate was 5 times greater in 1970 to 1974 and 6.7 times greater in 1995 to 1997 for men than for women in the most rural county group. The suicide rate was 2.3 times greater in 1970 to 1974 and 4.3 times greater in 1995 to 1997 for men than for women in the most urban county group.
Table 1 also shows the effect of rural–urban continuum on suicide mortality after adjustment for ethnic composition and divorce rate. Adjustment for ethnic composition and divorce rates widened the rural–urban differentials in suicide for men across time, with the adjusted gradient varying from 0.34 in 1970 to 1974 to 1.30 in 1995 to 1997. However, adjustment for ethnic composition and divorce rates to a large extent accounted for rural–urban differences in suicide mortality among women between 1970 and 1989 and widened the differentials for 1990 to 1994 and 1995 to 1997 such that rural areas had significantly higher suicide rates than urban areas—a pattern consistent with that for men.
Analysis of time trends in age-adjusted rates may mask important differences in trends in age-specific suicide rates. Therefore, in Table 2 we present relative risks of suicide for each county group across time, estimated separately for the young (15–24 years), those at working age (25–64 years), and the elderly (
65 years). As seen in Table 2, the higher the degree of rurality, the higher the suicide rate among young men, especially in the 1990s. Compared with young men in the most urban areas, young men living in the most rural areas had 17% higher suicide rates in 1980 to 1984 and 60% higher suicide rates in 1995 to 1997. Rural–urban differentials in suicide mortality were somewhat less pronounced for men aged 25 to 64 years, with the relative rate ratios between the most rural and the most urban areas varying from 1.16 in 1980 to 1984 to 1.48 in 1995 to 1997. For men aged 65 years and older, rural–urban differentials in suicide were substantial in each of the time periods, although they did not appear to increase across time as sharply as those for young men.
|
. Controlling for ethnic composition and divorce rate substantially increased the male suicide differential between the most rural and most urban county groups for each age cohort, and the differential grew across time for young and working-age men. For the period 1995 to 1997, men in each age cohort in the most rural areas had almost twice the suicide rate of their counterparts in the most urban areas.
The data in Table 2 indicate that rural–urban patterns in female suicide mortality vary by age. While there are no consistent rural–urban patterns in female youth suicide from 1980 to 1994, young women living in the most rural areas had a 55% higher suicide rate in 1995 to 1997 than their most urban counterparts. Among the working-age women, the suicide rate tended to be 20% to 25% lower in the most rural areas during 1980 to 1984 and 1985 to 1989; however, by 1995 to 1997, the suicide rate appeared to have increased with decreasing levels of urbanization. For the elderly women, the decrease in suicide rate was associated with increasing levels of rurality in all time periods, and the magnitude of the rural–urban differentials did not appear to change appreciably across the 4 time periods. Compared with women aged 65 and older living in the most urban areas, women of the same age group living in the most rural areas had a 42% lower suicide rate in 1980 to 1984 and a 31% lower suicide rate in 1995 to 1997. Adjustment for ethnic composition and divorce rate explained most of the observed rural–urban suicide differentials among elderly women, but the suicide differential between the most rural and most urban county groups increased over time for young women aged 15 to 24. The 1995–1997 suicide rate for young women in the most rural areas was 85% higher than the rate for those in the most urban areas.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
The results of the study are highly consistent with the Durkheimian theory of social integration and suicide.3–5,8,9,11,12 According to Durkheim, low levels of social integration are associated with high suicide rates. Durkheim defined social integration as being attached to social groups, maintaining interpersonal ties, and feeling allegiance to social groups. Low levels or an absence of social integration, as measured by living alone or experiencing marital disruption, represent social isolation and the atomization of individuals in a community.3 The increasing rural–urban gradients in suicide observed in this study might to some degree reflect differential changes over time in the social integration indicators for rural and urban areas. Changes in the socioeconomic and demographic structure of a community can have both short- and long-term implications for the health of that population, including suicide rates. Declines in traditional farm activity, the change from an agriculture-based economy to a more service- and manufacturing-oriented economy, and the substantial population loss due to birth deficits and out-migration in most rural communities can lead to a deemphasizing of traditional institutions such as the family and religion, a progressive weakening of social and community ties, and a loss of people's sense of community.
Both rural and urban areas have experienced profound social and demographic changes during the past 3 decades. However, these changes have affected rural areas much more adversely than urban areas. Our own analysis of the 1970 to 1990 census and vital statistics data (not presented here for the sake of brevity) indicates that a variety of social integration and demographic factors are substantially correlated with county-level suicide rates in each of the 6 time periods.25 For example, between 1970 and 1990 the population increase was almost 5 times greater in the most urban areas than in the most rural areas. While the most rural areas of the country as a whole experienced a substantial drop (20%) in the crude birth rate, the most urban areas saw their birth rate increase by 4%. The birth deficit (excess of deaths over births) was particularly dramatic in the rural areas, which experienced a 61% increase in deaths over births. In the most urban areas, there was a 25% increase in births over deaths. Although the rate of agriculture-sector employment was considerably greater in the rural than the most urban areas, the rate declined by 32% for the rural areas and increased by 95% for the most urban areas. Divorce rates increased twice as rapidly, and the percentage of those living alone rose 3 times more rapidly in the most rural areas than in the most urban areas. Household crowding, as indexed by the percentage of housing units with more than 1 person per room, declined by 67% in the most rural areas and by 12% in the most urban areas. The rate of female participation in the labor force and income inequality grew at a much higher rate in the most rural areas than in the most urban areas.
Changing societal attitudes toward suicide in terms of its recognition and labeling may also have contributed to rural–urban disparities over time.26 Furthermore, to the extent that those migrating to rural areas or those returning from urban areas had higher suicide rates than native rural residents, they might have had an important effect on trends.12,15 Despite the substantial birth deficit, the rural population during the period 1970 to 1990 grew owing to net migration. Those migrating to rural and nonmetropolitan areas during the 1970s and 1990s tended to be elderly retirees, blue-collar workers, and disenchanted city residents.27 The higher suicide rates observed among rural migrants have been attributed to social disruption, lack of social support and networks, change in lifestyle, and the increased alienation that accompanies migration.12,15
Geographic variations in access to firearms may also have contributed to the observed rural–urban disparities in suicide. Access to firearms is more common in rural than in urban areas.15,16,28 Studies have shown a strong association between increased firearm availability and a steadily upward trend in firearm suicide rates from 1946 to 1982.29–31 Our own analysis of the mortality data (not shown) indicates consistently growing rural–urban disparities in firearm suicide rates from 1979 to 1997, with rural areas having rates more than twice those of the most urban areas in 1997. Firearm suicides accounted for over 75% of all rural suicides during the period 1979 to 1997, whereas firearm suicides represented half of all suicides in the most urban areas.2
To help reduce rural–urban disparities in suicide, public health researchers and policymakers need to closely monitor geographic and temporal trends in social integration measures. Social and public health policies that emphasize investment in social integration or social capital through job creation, provision of gainful employment and social services, and improved social support and networks through community organization and involvement, especially for the rural young and elderly, may lower suicide rates.3 Restricting access to firearms may also reduce suicide rates, particularly in rural areas.
Although suicide is one of the most individualistic acts, we examined variations in county-level suicide rates rather than individual suicide risks. Our study design was therefore ecological, and the present analysis is not likely to be characterized by ecological fallacy or bias. We modeled ecological variations in suicide rates primarily as a function of an ecological variable, degree of rurality or urbanization, which is not quite reducible to the level of the individual. This study has certain limitations, however. The rural–urban continuum variable we used to stratify all of the nation's counties for the period 1970 to 1997 was developed in 1993 by using population and commuting criteria from the 1990 census.17,25 It is possible that these criteria may not apply as well to the earlier time periods, especially the 1970s. In other words, the classification of counties into specific rural–urban county groups, particularly into the small and medium-sized urban county groups, may have been somewhat different during the 1970s and the early 1980s. This may have introduced some inconsistency in the estimation of rural–urban gradients in suicide across time. Second, the accuracy with which suicide is recorded as a cause of death on the death certificate may differ between rural and urban areas. Because of social stigma, suicide deaths may be more likely to be underreported in rural areas than in urban areas. As a result, the observed rural–urban suicide differentials reported here may be underestimated for men and overestimated for women. Trends in suicide rates could also be affected if the registration of suicide and errors in classifying suicide deaths varied systematically between rural and urban areas and over time.
|
Footnotes |
|---|
Accepted for publication March 23, 2001.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONReferences |
|---|
2. National Center for Health Statistics. Vital Statistics of the United States, 1970–1997, Volume II: Mortality, Parts A and B. Washington, DC: Public Health Service; 1999.
3. Siahpush S, Singh GK. Social integration and mortality in Australia. Aust N Z J Public Health. 1999;23:571–577.[Medline]
4. Singh GK, Wilkinson AV, Song FF, et al. Health and Social Factors in Kansas: A Data and Chartbook, 1997–98. Lawrence, Kan: Allen Press; 1998.
5. Kposowa AJ, Breault KD, Singh GK. White male suicide in the United States: a multivariate individual-level analysis. Soc Forces. 1995;74:315–323.
6. Trout DL. The role of social isolation in suicide. Suicide Life Threat Behav. 1980;10:10–23.[Medline]
7. Seeman TE. Social ties and health: the benefits of social integration. Ann Epidemiol. 1996;6:442–451.[Medline]
8. Trovato F, Jarvis GK. Immigrant suicide in Canada: 1971 and 1981. Soc Forces. 1986;65:433–457.
9. Travis R. Halbwachs and Durkheim: a test of two theories of suicide. Br J Sociol. 1990;41:225–243.[Medline]
10. Fernquist RM, Cutright P. Societal integration and age-standardized suicide rates in 21 developed countries, 1955–1989. Soc Sci Res. 1998;27:109–127.[Medline]
11. Leenaars AA, Lester D. Domestic integration and suicide in the provinces of Canada. Crisis. 1999;20:59–63.[Medline]
12. Wilkinson KP, Israel GD. Suicide and rurality in urban society. Suicide Life Threat Behav. 1984;14:187–200.[Medline]
13. Dewey R. The rural–urban continuum: real but relatively unimportant. Am J Sociol. 1963;66:60–66.
14.
Saunderson T, Haynes R, Langford IH. Urban–rural variations in suicides and undetermined deaths in England and Wales. J Public Health Med. 1998;20:261–267.
15. Morrell S, Taylor R, Slaytor E, Ford P. Urban and rural suicide differentials in migrants and the Australian-born, New South Wales, Australia 1985–1994. Soc Sci Med. 1999;49:81–91.
16. Dudley MJ, Kelk NJ, Florio TM, Howard JP, Waters BG. Suicide among young Australians: an interstate comparison of metropolitan and rural trends. Med J Aust. 1998;169:77–80.[Medline]
17. Butler MA, Beale CL. Rural–Urban Continuum Codes for Metro and Nonmetro Counties, 1993. Washington, DC: Economic Research Service, US Dept of Agriculture; 1994. Staff report 9425.
18. Sink L. Estimates of the Population of Counties by Age, Sex, Race and Hispanic Origin: 1990 to 1998. Washington, DC: US Bureau of the Census; 1999.
19. Hollmann FW. United States Population Estimates, by Age, Sex, Race and Hispanic Origin: 1980 to 1988. Washington, DC: US Bureau of the Census; 1990. Current Population Reports, Series P-25, No. 1045.
20. Singh GK, Kochanek KD, MacDorman MF. Advance report of final mortality statistics, 1994. Month Vital Stat Rep. 1996;45(3)(suppl):1–80.
21.
Singh GK, Yu SM. Infant mortality in the United States: trends, differentials, and projections, 1950 through 2010. Am J Public Health. 1995;85:957–964.
22. Agresti A. An Introduction to Categorical Data Analysis. New York, NY: John Wiley & Sons Inc; 1996.
23. SAS/STAT Software: Changes and Enhancements Through Release 6.12: The GENMOD Procedure. Cary, NC: SAS Institute Inc; 1997.
24. Health, United States, 1999. Hyattsville, Md: National Center for Health Statistics; 1999.
25. The Area Resource File (ARF): Public Use File Technical Documentation. Rockville, Md: Health Resources and Services Administration; 1996.
26. Morrell S, Taylor R, Quine S, Kerr C. Suicide and unemployment in Australia 1907–1990. Soc Sci Med. 1993;36:749–756.
27. Johnson KM. The rural rebound. Reports on America. 1999;1(3):1–19.
28. Wiktor SZ, Gallaher MM, Baron RC, Watson ME, Sewell CM. Firearms in New Mexico. West J Med. 1994;161:137–139.[Medline]
29. Zwerling C, McMillan D, Cook PJ, et al. Firearm injuries: public health recommendations. Am J Prev Med. 1993;9(suppl 1):52–55.
30. Baker SP, O'Neill B, Ginsburg MJ, Li G. The Injury Fact Book. New York, NY: Oxford University Press; 1987.
31. Lester D, Clarke RV. Note on "suicide and increased availability of handguns in the United States": the influence of firearm ownership on accidental deaths. Soc Sci Med. 1991;32:1311–1313.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
