International Journal of Psychiatry in Clinical Practice, 2009; 13: 259264

ORIGINAL ARTICLE

The relationship between elderly suicides rates, household size and family structure: A cross-national study

AJIT SHAH
University of Central Lancashire, Preston and West London Mental Health NHS Trust, London, UK

Abstract Background. Cultural factors may influence cross-national variations in elderly suicide rates. Methods. A cross-national study examining the relationship between elderly suicide rates and (i) mean household size and (ii) family structure was conducted with the a priori unidirectional hypothesis that larger mean household size and greater proportion of extended families may imply a greater number of people being potentially available within the household to provide support and respect to the elderly and to hold them in high esteem, and ultimately leading to reduction in elderly suicide rates. Data on elderly suicide rates was ascertained from the World Health Organisation website. Data on mean household size and family structure was ascertained from a report from the Inter-American Development Bank. Results. The main findings were elderly suicide rates in both sexes were significantly correlated with the mean household size (negative), percentage of extended households (negative; except in males aged 6574 years where this approached significance) and percentage of single person households (positive). Conclusions. The impact of mean household size and family structure on elderly suicide rates may interact with and be modified and mediated through cultural factors. The contribution of cross-national differences in cultural factors on elderly suicide rates requires further study by formally measuring cultural factors with validated instruments.

Key Words: Cultural factors, suicides, elderly suicides

Introduction Cross-national comparisons of elderly suicide rates [13] can generate testable aetiological hypotheses [4]. Elderly suicide rates for both sexes, ascertained from the World Health Organisation (WHO) data bank for 1992, were low in Malta, Greece, Albania, Armenia, Tajikistan, Uzbekistan, Mexico and Columbia and high in Austria, Belarus, Denmark, Estonia, Finland, France, Hungary, Kazakhstan, Latvia, Lithuania, Russia, Slovenia, Switzerland and rural China [5]. Similarly, elderly suicide rate in both sexes for the age-bands 6574 years and 75 years, ascertained from the latest available WHO data, were the lowest in Caribbean, central American and Arabic countries, and the highest in central and eastern European, some oriental and some western European countries [3]. Potential explanations for such regional and cross-national variation in elderly suicide rates include cross-national differences in cultural factors, genetic factors, prevalence of mental

illness in the elderly, life expectancy, socio-economic status, the availability of healthcare services, and public health initiatives to improve the detection and treatment of mental illness, mental health and suicide prevention [3]. The potential impact of cultural factors on crossnational differences in elderly suicide rates requires further consideration. Suicide rates among nonwhite Americans [6,7], Indians [8,9], Arabs in Jordan [10], men in Kuwait [11], Malays in Singapore [12], Indian immigrants to the UK [13,14], and in some east European countries [2] decline with increasing age. Traditionally, in these societies the elderly are respected, held in high esteem and live in closely knit families, and this offers protection against loneliness and despair, which otherwise may lead to suicide [5]. These factors may also explain the low elderly suicide rates in Thailand [15] and among Malays in Singapore [16]. A similar hypothesis may also explain high suicide rates among

Correspondence: Professor Ajit Shah, West London Mental Health NHS Trust, Uxbridge Road, Southall, Middlesex UB1 3EU, UK. Tel: 44 208 354 8140. Fax: 44 208 354 8898. E-mail: ajit.shah@wlmht.nhs.uk

(Received 25 November 2008; accepted 11 March 2009)

ISSN 1365-1501 print/ISSN 1471-1788 online # 2009 Informa UK Ltd. DOI: 10.3109/13651500902887656

260

A. Shah household. This report did not clarify what classification for the family structure would be used if one child was under 20 years and the other over 20 years in the same household. Moreover, this report did not provide separate data for nuclear households with children over the age of 20 years and under the age of 20 years and thus separate analysis for nuclear households with children under the age of 20 years could not be conducted. However, this is unlikely to be important in this study as very few people over the age of 65 years will have children under the age of 20 years. The relationship between elderly suicide rates, in both sexes in both the age-bands, and (i) the mean household size, and the percentage of (ii) single person, (iii) nuclear and (iv) extended households was examined by using Spearmans rank correlation test (r). The independent effect of each of these variables on elderly suicide rates independent of other known correlates of elderly suicide rates [4] was examined using multiple regression analysis with the Enter method because variables measuring mean household size and family structure were intercorrelated. Elderly suicide rates was the dependent variable. The mean household size, the percentage of single person households, the percentage of nuclear households, the percentage of extended households, gross national domestic product (GDP) and life expectancy were the independent variables. Data on GDP and life expectancy were obtained from the WHO website (www.who.int/countries/en/). General population suicide rate was not included as an independent variable in the multiple regression analysis because general population suicide rates include elderly suicide rates and therefore cannot be considered to be independent. Results Full data set was available for 27 countries. This constituted 31% (27/87) of the countries on which we previously ascertained data on elderly suicide rates [3,4]. A list of these 27 countries is given in Table I. Table I also provides the values for general population suicide rates, elderly suicide rates, mean household size and family structure for each country. The mean [standard deviation) and median (range) of general population male suicide rates across the countries were 18.3 (13.9) and 13.5 (3.469.3) per 100,000 population respectively. The mean (standard deviation) and median (range) of general population female suicide rates across the countries were 4.9 (3.2) and 4 (0.812.2) per 100,000 population respectively. Table II illustrates the relationship between suicide rates, in both sexes in both the age-bands, and

the elderly in Japan [17,18], Hong Kong [19], China [20,21] and Taiwan [22] because they have lost their traditional role in the family. Additional evidence supporting this, from Hong Kong, Taiwan and China, includes: mismatch between the traditional dependence of elderly on their children for emotional and financial support and their childrens ability to provide this [20,2224]; the traditional expectation of the elderly to live with their children or grandchildren not being met [20,22]; the greater effect on the elderly of their childrens negative attitudes [20]; and the migration of children to urban areas or to other countries [19,20]. Lower suicide rates among elderly Malays in Singapore were, in part, attributed to larger household size [16]. Higher suicide rates among elderly Chinese in Singapore and the elderly in China were, in part, attributed to the reduction in the number of caregivers [16,24] and family size [16]. However, there is a paucity of validated instruments to formally measure many of the cultural factors described above. Therefore, a cross-national study examining the relationship between elderly suicide rates and (i) mean household size and (ii) family structure was conducted with the a priori unidirectional hypothesis that larger mean household size and greater proportion of extended families may imply a greater number of people being potentially available within the household to provide support and respect to the elderly and to hold them in high esteem, and ultimately lead to reduction in elderly suicide rates.

Methods Data collection The methodology for ascertaining suicide rates is described elsewhere [3,4] and summarised here. Data on elderly suicide rates for males and females in the age-bands 6574 years and 75 years were ascertained from the WHO website (www.3.who.int/ whosis/mort/table1.cfm). The median (range) of the latest available year for this data from different countries was 2000 (19852003). Data on the mean household size (defined as the average number of people in the household in a given country), and the percentage of single person, nuclear and extended households was ascertained from a report from the Inter-American Development Bank [25]; these were calculated using data from national household surveys between the years 1994 and 1998 [25]. Szekely and Hilgert [25] defined a nuclear household to be composed of a couple and children under the age of 20 years, whilst a similar household with children over the age of 20 years was defined as an extended

Table I. A list of countries and data on suicide rates, mean household size and family structure. General General Female suicide Female suicide population rate population rate rate 6574 rate 75 males females 5.5 4.8 2.6 4.9 1.5 0.7 0.0 1.0 0.9 11.10 5.1 10.8 0.7 15.1 5.7 1.1 9.0 4.6 1.9 7.8 18.2 9.4 3.8 3.4 4.0 8.3 2.6 4.1 5.8 6.4 2.8 2.2 0.7 0.0 0.8 1.6 7.5 17.7 18.2 0 34.6 5.9 1.2 10.9 3.2 0.0 7.6 28.3 12.7 7.8 3.7 4.0 9.6 3.8 13.4 21.10 6.4 18.4 18.2 8.1 11.6 6.0 11.6 32.3 27.9 20.4 3.4 45.5 11.1 6.3 12.7 18.4 8.4 26.6 63.9 18.9 13.5 10.8 17.1 29.0 8.8 3.5 5.3 1.6 5.2 3 2.4 2 2.7 5.4 10.2 9.5 7 0.8 12.2 3.3 1.3 5.9 6.0 1.3 5.0 11.9 8.1 3.7 3.1 4.0 5.5 1.5

Country Argentina Australia Brazil Canada Chile Columbia Costa Rica Equador El Salvador Finland France Germany Guatamala Hungary Italy Mexico Netherlands Norway Panama Poland Russia Sweden Thailand United Kingdom United States of America Uraguay Venezuela

Mean household size 3.49 1.95 3.86 1.96 3.95 4.26 4.08 4.6 4.65 1.78 1.97 1.84 5.24 2.16 2.45 4.55 1.89 1.76 4.67 2.3 2.25 1.48 3.67 1.89 1.87 3.2 4.58

Single person family (%) 15 34 8 34 7 7 7 7 6 39 31 39 4 27 18 6 73 44 8 22 22 57 9 32 38 16 6

Nuclear family (%) 42 43 50 43 43 42 51 47 40 48 51 45 46 44 40 49 16 42 45 45 45 43 42 52 39 39 42

Extended family (%) 42 19 41 19 49 51 42 46 54 12 18 16 50 30 42 45 11 13 47 32 30 0 49 15 18 44 52

Male suicide rate 6574 26.5 19.2 13.0 16.9 31.3 17.0 18.4 4.4 8.8 39.9 42.6 29.0 6.0 73.9 17.9 9.7 15.0 23.7 12.0 34.3 94.5 29.5 14.0 8.7 22.7 70.4 23.8

Male suicide rate 75 31.1 26.3 17.9 22.7 31.9 17.4 19.4 8.6 19.2 50.3 86.6 60.9 9.2 121.1 32.4 20.7 21.8 30 34.4 28.7 80.0 42.2 18.0 10.4 42.4 95.1 27.7

Suicide and household size 261

*Suicide rates are per 100,000 of the relevant age-group population.

262

A. Shah

Table II. The relationship between elderly suicide rates and mean household size, and percentage of single person, nuclear and extended households. Male 6574 years suicide rates *(4.494.5) Mean household size *(1.485.24) Percentage single Person households *(473) Percentage of Nuclear households *(1652) Percentage of Extended households *(054) r 0.48 P 0.011 r 0.41 P 0.034 NS r 0.35 P 0.075 (NS) Male 75 years suicide rates *(8.614.1) r 0.48 P 0.012 r 0.48 P 0.012 NS r 0.39 P 0.042 Female 6574 years suicide Female 75 years suicide rates *(018.2) rates *(034.6) r 0.73 P B0.0001 r 0.75 P B0.0001 NS r 0.67 P B0.0001 r 0.66 P B0.0001 r 0.65 P B0.0001 NS r 0.56 P B0.0001

NS, not significant. *Ranges are given in brackets and suicide rates are per 100,000 of the relevant age-band general population.

(i) the mean household size, and the percentage of (ii) single person, (iii) nuclear and (iv) extended households. There were significant negative correlations between suicide rates, in both sexes in both the elderly age-bands, and the mean household size and percentage of extended households (except in males in the age-bands 6574 years the correlation with mean household size only approached significance). There were significant positive correlations between suicide rates, in both sexes in both the elderly agebands, and percentage of single person households. There was no significant correlation between elderly suicide rates and percentage of nuclear households. On multiple regression analysis mean household size was independently associated with suicide rates in males aged 6574 years (P B0.00001), males aged 75 years (P 0.006), females aged 6574 years (P B 0.0001) and females aged 75 years (P B0.0001). GDP was independently associated with suicide rates in males aged 6574 years (P 0.042), females aged 6574 years (P B0.00001) and females aged 75 years (P 0.011). The percentage of single person households, the percentage of nuclear households, the percentage of extended households and life expectancy were not independently associated with suicide rates in both sexes in both the elderly age-bands. Discussion Some methodological issues need consideration. Cross-national data on suicide rates should be viewed with caution because: data were available for only 27 countries and this small sample size may be open to both type 1 and type 2 statistical errors [26,27]; the validity of this data was unclear [1,28]; the legal criteria for the proof of suicide vary between countries and in different regions within a country [27,28]; some countries, may have poor death registration

facilities [28]; and cultural and religious factors and stigma attached to suicide may lead to underreporting of suicides [27,29]. Some countries are large and heterogeneous and a single national figure for suicide rates may mask regional variations in suicide rates [27,28]. This makes it difficult to prove the study hypothesis because of uneven reliability of suicide data in large countries with regional variations. The findings should be considered as a simple comparative observation across countries. Data on mean household size, and the percentage of single person, nuclear and extended households were ascertained from a report from Inter-American Development Bank; Szekely and Hilgert [25] calculated these variables using data from national household surveys. The validity of data from national household surveys with varying designs and the utility of data from these sources to calculate household size and family structure is unclear. However, there were no significant differences in suicide rates in both sexes in both the elderly age-bands between the 27 countries with data on household size and family structure and the 60 countries without this data. Moreover, despite a wide ranging search, including the WHO and United Nations database, Inter-American Development Bank was the only available source of the required crossnational data. The significant correlations between suicide rates, in both sexes in both the elderly age-bands, and mean household size (negative), percentage of extended families (negative) and percentage of single person households (positive) may have several explanations. First, the findings may be an artefact of the methodological issues described above. It is possible that the observed absence of a correlation between suicide rates and the percentage of nuclear families have been because very few elderly people are likely to have children under the age of 20 years

Suicide and household size 263 (the definition used for nuclear families) and so may be subject to type 1 statistical error. Second, the findings may simply reflect variations in general population suicide rates; however, the a priori study hypothesis specifically for elderly suicides was confirmed. Third, the findings were consistent with previous observations of positive correlations between elderly suicide rates and the proportion of elderly in the total population [4,30], and the ratio of elderly people to younger people [31] in large cross-national studies. Fourth, the findings were consistent with reports from China and Singapore, where greater number of caregivers and larger family size afforded protection against elderly suicides [16,24]. Thus, the findings may be genuine and an alternative hypothesis pertaining to cultural factors may be important and is explored below. The observed negative correlations between elderly suicide rates and mean household size and percentage of extended households supports the a priori unidirectional hypothesis that larger household size and greater proportion of extended families may imply a greater number of people being potentially available within the household to provide support and respect to the elderly and to hold them in high esteem, and ultimately lead to reduction in elderly suicide rates. The observed positive correlation between elderly suicide rates and percentage of single person households is also consistent with this hypothesis. The impact of mean household size and the percentage of extended households on elderly suicide rates may interact with and be modified and/or mediated through cultural factors. These cultural factors were described in the Introduction and include: the degree of respect and esteem given by younger generations to the elderly; the traditional dependence of elderly on their children for emotional and financial support and their childrens ability to provide this; the traditional expectation of the elderly to live with their children or grand-children; the greater effect on the elderly of their childrens negative attitudes; and the number of available caregivers, household size and family size. Countries with larger mean household sizes and greater number of extended households potentially have a greater number of people available within the household and within close geographical proximity to positively contribute to these cultural issues, and this may ultimately lead to reduction in elderly suicide rates. However, this study merely examined the relationship between elderly suicide rates and mean household size and family structure, and cultural factors were not formally measured. Therefore, considerable caution should also be exercised in assuming that mean household size and family structure can act as proxy measures for cultural factors. Moreover, mean household size and family structure may also interact with, modify and mediate the effect of other factors on elderly suicide rates. An important factor in this context is the contribution of mean household size and extended household on the socio-economic status and income inequality within the household because elderly suicide rates are also influenced by the socioeconomic status and income inequality [4,30]. The findings appeared to be more prominent for suicide rates in females than males. This may be due to the methodological issues discussed above. It may also be because women may be more sensitive to cultural factors described in the last paragraph including: the degree of respect and esteem given by younger generations to the elderly; the traditional dependence of elderly on their children for emotional and financial support and their childrens ability to provide this; the traditional expectation of the elderly to live with their children or grand-children; and the greater effect on the elderly of their childrens negative attitudes. There is little formal evidence to support this speculative explanation. There may be other explanations also given that household size was the only independent predictor of suicide rates in both sexes in both the elderly agebands. More people in the household implies that there are more people to identify suicidal ideation and support the suicidal individual to seek appropriate help. Also, in larger households elderly people are likely to be alone for lesser time periods and this would reduce the opportunity to implement any suicidal plans without being discovered. Both these possibilities may also help reduce elderly suicide rates. The cross-sectional design of the study does not allow definitive conclusions about the aetiological relationship. The contribution of cross-national differences in cultural factors on elderly suicide rates requires further study by formally measuring: (i) the cultural views and attitudes of young people towards the elderly [15]; and (ii) the perception of the elderly on their traditional and changing role in society and their relationship with younger generations. Key points . Cultural Factors can influence elderly suicide rates. . Elderly suicide rates were negatively correlated with the mean household size and the percentage of extended families. . Elderly suicide rates were positively correlated with the percentage of single parent families. . The impact of mean household size and family structure on elderly suicides may be mediated and modified through cultural factors.

264

A. Shah
[15] Lotrakul M. Suicide in Thailand during the period 19982003. Psychiatry Neurosci 2003;60:905. [16] Kua EH, Ko SM, Ng TP. Recent trends in elderly suicide rates in a multi-ethnic Asian city. Int J Geriatr Psychiatry 2003;18:5336. [17] Watanabe N, Hasegawa K, Yoshinaga Y. Suicide in later life in Japan: urban and rural differences. Int Psychogeriatr 1995;7:25361. [18] Shimuzu M. Depression and suicide in late life. In: Hasegawa K, Homma A, editors. Psychogeriatrics: Biomedical and social advances. Amesterdam: Excerpta Medica; 1990. p 3304. [19] Yip PSF, Tan RCE. Suicides in Hong Kong and Singapore: a tale of two cities. Int J Soc Psychiatry 1998;44:26779. [20] Yip PSF, Callanan C, Yuen HP. Urban/rural and gender differentials in suicide rates: East & West. J Affect Disord 2000;57:99106. [21] Yip PSF, Liu KY, Ku J, Song XM. Suicide rates in China during a decade of rapid social change. Soc Psychiatry Psychiatr Epidemiol 2005;40:7928. [22] Liu H, Wang H, Yang M. Factors associated with an unsual increase in elderly suicide rate in Taiwan. Int J Geriatr Psychiatry 2006;21:121921. [23] Yip PSF, Chi I, Yu KK. An epidemiological prole of elderly suicides in Hong Kong. Int J Geriatr Psychiatry 1998;13: 6317. [24] Yip PSF. An epidemiological prole of suicides in Beijing, China. Suicide Life Threatening Behav 2001;31:6270. [25] Szekely M Hilgert M. What drives differences in inequality across countries? Inter-American Development Bank; 2000 www.Vm.udg.mx/DocSem/What%20is%20behind%20ine quality.pdf [26] Moscicki EK. North American perspectives: epidemiology of suicide. Int Psychogeriatr 1995;7:13748. [27] Wasserman D, Cheng Q, Jiang GX. Global suicide rates among young people aged 1519. World Psychiatry 2005;4: 11420. [28] Shah AK, Ganesvaran T. Suicide in the elderly. In: Chiu E, Ames D, editors. Functional psychiatric disorders of the elderly. Cambridge: Cambridge University Press; 1994. p 22144. [29] Abrahams VJ, Abrahams S, Jacob KS. Suicide in the elderly in Kanyambadi block, Tamil Nadu, South India. Int J Geriatr Psychiatry 2005;20:9535. [30] Kiemo K. Towards a socio-economic and demographic theory of elderly suicides: a comparison of 49 countries at various stages of development. 2004. Viewed 2 February 2007 www.soc.ou.se/publications/fulltext/diss2003-3pdf [31] Shah AK, Padayatchi M, Das K. The relationship between elderly suicide rates and the elderly dependency ratios: a cross-national study using data from the WHO data bank. Int Psychogeriatr 2008;20:594604.
/ / / / / / / / / / / / / / / / / / / / / / / / / /

Acknowledgements The helpful comments of the two anonymous reviewers were greatly appreciated. Statement of Interest As author, I have no conflict of interest with any commercial or other associations in connection with the submitted article. References
[1] Diekstra RFW. Suicide and attemptede suicide: an international perspective. Acta Psychiatr Scand 1989;80(Suppl 354):124. [2] Sartorius N. Recent changes in suicide rates in selected eastern European and other European countries. Int Psychogeriatr 1995;7:3018. [3] Shah AK, Bhat R, MacKenzie S, Koen C. Elderly suicide rates: cross-national comparisons and association with sex and elderly age-bands. Med Sci Law 2007;47:2445. [4] Shah AK, Bhatt R, MacKenzie S, Koen C. A cross-national study of the relationship between elderly suicde rates and life expectancy and markers of socioeconomic status and healthcare. Int Psychogeriatr 2008;20:34760. [5] Shah AK, De T. Suicide and the elderly. Int J Psychiatry Clin Pract 1998;2:317. [6] Seiden RH. Mellowing with age: factors affecting the nonwhite suicide rate. Int J Ageing Hum Dev 1981;13:26584. [7] Woodbury MA, Manton KG, Blazer D. Trends in US suicide mortality rates 19681982: race and sex differences in age, period and cohort components. Int J Epidemiol 1987;17: 3562. [8] Adityanjee DR. Suicide attempts and suicide in India: cross-cultural aspects. International J Soc Psychiatry 1986; 32:6473. [9] Bhatia SC, Kahn MH, Medirrata RP. High risk suicide factors across cultures. Int J Soc Psychiatry 1987;33:22636. [10] Daradekh TK. Suicide in Jordan 19801985. Acta Psychiatr Scand 1989;79:2414. [11] Shah AK. The relationship between suicde rates and age: an analysis of multinational data from the World Health Organisation. Int Psychogeriatr 2007;19:114152. [12] Ko SM, Kua EH. Ethnicity and elderly suicide rates in Singapore. Int Psychogeriatr 1995;7:30917. [13] Raleigh VS, Bulusu L, Balarajan R. Suicides among immigrants from the Indian subcontinent. Br J Psychiatry 1990;156:4650. [14] Neeleman J, Mak V, Wessely S. Suicide by age, ethnic group, coroners verdict and country of birth. A three-year survey in inner London. Br J Psychiatry 1997;181:4637.
/ / / / / / / / / / / / / / / / / / / / / / / / / / / /

Copyright of International Journal of Psychiatry in Clinical Practice is the property of Taylor & Francis Ltd and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.