A recent study reported the relationship between suicide rates in both sexes in the age bands 65–74 and 75+ years and the Gross National Domestic Product (GDP), a measure of socio-economic status, as being curvilinear (inverted U-shaped curve) and fitting the quadratic equation y = a + bx–cx2 (where y is the suicide rate, x is the Gini coefficient and a, b and c are constants) (Shah, Reference Shah2010). This relationship was explained using the epidemiological transition hypothesis (Shah and Bhat, Reference Shah and Bhat2010) and the explanatory model included the following sequence of events (Shah, Reference Shah2010): (i) countries with low socio-economic status have poorly developed healthcare system; (ii) a poorly developed healthcare system is associated with increased child mortality; (iii) increased child mortality rates result in reduced life expectancy; and (iv) reduced life expectancy leads to fewer people reaching old age, the age at which the risk of suicide is high. In order to provide support for this hypothesis other data sources related to socio-economic status, quality and quantity of healthcare services and life expectancy were examined.
Data on suicide rates in both sexes in the age bands 65–74 and 75+ years were ascertained from the World Health Organization website (http://www.who.int/whosis/database/mort/table1.cfm) for the latest five consecutive years. In order to minimize the effect of random annual variations in suicides rates, an annual average of the five years was used for data analysis (Shah and Coupe, Reference Shah and Coupe2009). The methodology for ascertaining and calculating these suicide rates has been previously reported (Shah, Reference Shah2009). The median (range) for the year of the suicide rate data was 2005 (1970–2007).
Two measures, other than GDP, of socio-economic status were considered: the Human Poverty Index (HPI) and the Gini Index/coefficient (a measure of income inequality). They were ascertained from the United Nations Development Program website (http://hdr.undp.org/en/media/HDI_2008_EN_Tables.pdf). The median (range) for the year of the Gini coefficient was 2000 (1990–2003). The probability at birth of surviving to the age of 65 years for both sexes for the years 2000–2005 was also ascertained from the same website.
Rates of immunization for tuberculosis (for 2004) and measles (for 2004), the percentage of births attended by skilled health personnel (for 1996–2004) and the number of physicians (per 100,000 population for 1990–2004) were used as measures of the quality and quantity of available healthcare services. These data were also ascertained from the United Nations Development Program website above.
The relationship between suicide rates and the Gini coefficient, a measure of income inequality, was examined with a curve estimation regression model in order to test the hypothesis that a curvilinear relationship (inverted U-shaped curve) fits the quadratic equation y = a + bx–cx2 (where y is the suicide rate, x is the Gini coefficient and a, b and c are constants). The same excercise was repeated for the HPI.
The relationship between suicide rates and the probability at birth of surviving to the age of 65 years, rates of immunization for tuberculosis and measles, the percentage of births attended by skilled health personnel and the number of physicians (per 100,000 population) was examined by Spearman's correlation coefficient (ρ).
Data on suicide rates and the Gini coefficent were available for 66 countries. There was a statistically (at least at 0.05 level) curvilinear (inverted U-shaped) relationship between suicide rates in both sexes and both elderly age bands and the Gini coefficient fitting the quadratic equation y = a + bx–cx2. However, there was no curvilinear relationship with HPI where a complete data set was only available for 39 countries.
There was a positive correlation between suicides rates in females aged 65–74 years (ρ = +0.34, P = 0.002) and 75+ years (ρ = +0.22, P = 0.055) and the probability at birth of surviving to the age of 65 years; there was no significant correlation with males in both age bands. Here a complete data set was available for 80 countries.
There was a positive correlation between suicides rates in males aged 65–74 years (ρ = +0.51, P< 0.0001), males aged 75+ years (ρ = +0.56, P< 0.0001), females age 65–74 years (ρ = +0.51, P<0.0001) and females aged 75+ years (ρ = +0.45, P<0.0001) and the percentage of births attended by skilled health personnel; here a complete data set was available for 77 countries. There was a positive correlation between suicides rates in males aged 65–74 years (ρ = +0.46, P < 0.0001), males aged 75+ years (ρ = +0.46, P < 0.0001), females age 65–74 years (ρ = +0.51, P < 0.0001) and females aged 75+ years (ρ = +0.57, P<0.0001) and the number of physicians; here a complete data set was available for 82 countries. However, there were no significant correlations between suicide rates in both sexes in both elderly age bands and rates of immunization either for tuberculosis or measles. Here a complete data set was available for 59 countries for tuberculosis and 82 countries for measles.
The curvilinear relationship between suicide rates in both sexes in both age bands and the Gini coefficient is consistent with the previously proposed epidemiological transition hypothesis of elderly suicides, but the absence of this relation with HPI is at variance with this hypothesis, although the latter was based on a significantly smaller sample. The positive correlations between suicide rates in both sexes in both elderly age bands and the percentage of births attended by skilled health personnel and the number of physicians is also consistent with the epidemiological transition hypothesis of elderly suicides. However, the absence of this relationship with rates of immunization for tuberculosis and measles is at variance with the epidemiological transition hypothesis. The positive correlation between suicide rates only in females in both age bands and the probability at birth of surviving to the age of 65 years is partly consistent with the epidemiological transition hypothesis.
Overall, the findings provide mixed evidence in support of the epidemiological transition hypothesis of elderly suicides.
