There have been important improvements in the life expectancy of birth cohorts across time in developed countries at around 3 years per decade. Morbidity also fell at a rate of 50% among the elderly between 1984 and 2000 (Fogel 2005). Finch and Crimmins (2004) highlight the fact that declines in mortality among both the young and elderly generally begin in the same cohort – individuals experiencing improving early-life conditions were also the individuals who experienced declining rates of mortality at later ages. In the economics literature, Case et al (2005) have demonstrated that poor childhood health is a predictor of ill health as an adult. A consequence of these findings is that inequalities in adult health may have their origins in early life (Case et al 2002).
It is clear from this body of research that a person’s current health is influenced not only by current behaviour and circumstance, but also by their cumulative experience since birth. For a variety of reasons, it is not easy to quantify such relationships. Although we know that substantial inequalities exist in both child and adult health, causally linking these two facts is difficult. By itself, finding that individuals in poor health now are more likely to have suffered from poor health in early life is not sufficient. This is because poor childhood health is associated with many other factors that influence adult health, including socioeconomic status.
There are a number of attempts to deal with the potential for omitted variables to bias research findings. Black et al (2007) adopt a twin study, which allows them to control for all factors common to each pair (including attributes which are not possible to measure), such as family background and genetics. They find significant impacts of birth weight on height, education and income. Another approach is to use natural experiments, ideally an event which impacts on an individual’s early-life conditions but leaves all other factors constant. The 1918 influenza outbreak in the US is used by Almond (2006) to provide exogenous variation in early-life conditions. He finds significant negative effects on a number of outcomes for the in utero cohorts.
The Irish experience of the 1940s provides an opportunity to examine the causal role of early-life conditions in determining adult health. We use the fact that we have data on individuals who were born very close together (in terms of both time and space), but faced dramatically different early-life conditions. As evidenced from the infant mortality rate, during the early part of the 20th century Ireland lagged behind the rest of the developed world, with infant death rates of up to 12% not uncommon in some of the cities. There was an urban penalty of around 50% preceding 1947. Following 1947, the national infant mortality rate halved within the space of a decade, largely driven by convergence between urban and rural rates (see Figure 1).
Figure 1. Irish urban and rural infant mortality rates 1930-1961
Source: Irish Infant Mortality Database. Note: The urban areas include Dublin, Cork, Limerick, Waterford and Galway county boroughs, with the rural areas comprising the other county areas.
We argue that this change in the public health of the Irish population was an exogenous shock driven by a series of interventions centred on the 1947 Health Act. The available evidence is consistent with the idea that improvements in sanitation, clean water, and overcrowding in Irish cities were the main factors in these changes. At the county level, there is a strong relationship between these variables and the infant mortality rate in 1946 (shown in Figure 2). However, following these public health investments the relationship disappears. The fact that some counties were affected much more than others (i.e. the cities that experienced a much more dramatic fall in infant mortality), means that those born in rural areas offer a natural control group for comparison. As this was an external change driven by government action, we are therefore able to argue that this reduction in the infant mortality rate only reflected changes in early environment, and none of the other possible confounding factors which could mediate the relationship between initial health and later outcomes. Comparing the affected Irish cohorts should give us an accurate estimate of the effects of initial health. An advantage of this event is that it relates to a change in cohort health induced by government intervention as opposed to some extreme event which could have had very specific effects.
Figure 2. Irish county-level infant mortality and shared sanitation (1946)
Source: Irish Census 1946 and Irish Infant Mortality Database.
In order to track the individuals who were affected, we use data from the Irish Census in 2002 and 2006. These micro files are a 5% representative sample of the Irish population which contain information on county of birth. We combine this census data with information from a database of historical infant mortality rates covering every Irish county and urban district which we collected from various sources. The outcome we examine is self-reported disability status, as this is the only health-related variable available. However, we show that this matches up closely with self-reported health, which is known to be a good marker for objective health status.
We conduct a regression analysis where we model disability as a function of the infant mortality rate in the individual’s county of birth at their time of birth (we are restricted to 5-year birth cohorts, as this is what is available in the data). If early life conditions are an important determinant of later health, we would expect that those who were born into places and times of high infant deaths to be more at risk of suffering from a disability now. Along with county-of-birth fixed effects, county-of-residence fixed effects and controls for survey year, we control for county-specific age trends to capture general improvements in disability rates. This means that we are using non-linear breaks in the infant mortality series within counties to identify the effects of early-life conditions.
We use a probit analysis, although the results are robust to different specifications including more experimental approaches such as instrumental variables or difference-in-differences. The results from this analysis are presented in Table 1, which shows the estimated effects of early life conditions on current disability status.
Table 1. Summary of regression results for early life conditions and disability
|Disability||Disability (>40)||Disability (Lags and Leads)|
|County infant mortality (five-year average)||0.506***||0.478***||0.307**|
Notes: *** p<0.01, ** p<0.05, * p<0.1. Standard errors in parentheses, clustered by county age group. Note: Marginal effects at the mean of the independent variables from a heteroskedastic probit model. The standard probit model is rejected on the basis of a Wald test. The county infant mortality variable is the average mortality rate (in deaths per 1,000 live births) for the individual’s five-year birth cohort in county of birth. The first and third columns restrict analysis to those aged over 25 and under 75. The second column restricts analysis to individuals aged over 40 and under 75.
Source: Irish Infant Mortality Database and Irish Census 2002 and 2006. Coefficients are scaled by 1,000. Other variables are age, age squared, gender, county of residence and county of birth fixed effects, county of birth trends, and controls for survey year.
Marginal effects for disability are shown scaled by 1,000. The first column uses those in the sample born in the Republic of Ireland aged between 25 and 75 (this age restriction is due to concerns about selection), while the second restricts the analysis to those aged over 40. In the final column, we add lags and leads of infant mortality (+/- 5 and 10 years) as an additional check on whether infant mortality is correlated with some other variable which manifests itself at an earlier or later stage. In fact, results are similar for all specifications. In each case, we find significant and large effects – having been born into higher infant mortality increases your risk of suffering from a disability now. A 1 unit increase in the infant mortality rate (for your 5-year birth cohort in your county of birth) increases the risk of disability by between 0.3 and 0.5 percentage points. Given that the average disability rate in Ireland according to the 2006 census is around 10%, these are large effects.
Using the estimates from this analysis, the improvement in national early-life conditions in the 1940s (a fall of around 35 points in the infant death rate), was associated with an approximate 12–18% decrease in the probability of disability for those cohorts who benefited. Improvements over this period were even more dramatic in some counties, particularly the urban areas. For example, in Dublin the infant mortality rate fell from an average of 103 in the period 1938-42 to an average of 43 in the period 1948-52.
When we examine the issue of inequality and early-life conditions, we find that those who benefited most from the improving early-life conditions were at the lower end of the education distribution. Table 2 presents marginal effects of the county infant mortality variable for the model presented in column 1 of Table 1 by education level. Education is the proxy we use for socioeconomic status in childhood, as this is the only variable available in the data. We find that the effects of early-life conditions are largest for those with only primary education, while the effects for those with secondary and tertiary education are not significant. This provides further evidence that the roots of health inequalities may lie in early childhood.
Table 2. Marginal effects of early life conditions by education level
|Education Level||Marginal Effect||Standard Error|
|Upper secondary and tertiary||−0.0528||0.114|
Notes: *** p<0.01, ** p<0.05, * p<0.1. Marginal effects of infant mortality on disability status from the heteroskedastic probit model in column 1 of Table 1 evaluated at different levels of education. All coefficients are scaled by 1,000.
In addition to any immediate benefits, public-health interventions can also have important long-run effects by improving the health of the adults who were affected as children. In the Irish case, improving early-life conditions brought about by public health initiatives centred on the 1947 Health Act appear to have benefited the affected birth cohorts by reducing disability rates, particularly for those of lower socioeconomic status. To validate the assumptions behind our model, and address some potential limitations with respect to our outcome measure, we are proceeding with a project to digitise Irish birth and death records from the period.
Almond, Douglas (2006), “Is the 1918 Influenza Pandemic Over? Long-Term Effects of In Utero Influenza Exposure in the Post-1940 US Population,” Journal of Political Economy, 114(4):672-712.
Black, Sandra E., Paul J. Devereux, and Kjell G. Salvanes (2007), “From the Cradle to the Labour Market? The Effect of Birthweight on Adult Outcomes,” Quarterly Journal of Economics, 122(1):409-439.
Case, Anne, Darren Lubotsky, and Christina Paxson (2002), “Economic Status and Health in Childhood: The Origins of the Gradient,” American Economic Association, 95(2):1308-1334.
Case, Anne, Angela Fertig, and Christina Paxson (2005), "The Lasting Impact of Childhood Health and Circumstance," Journal of Health Economics, 24(2):365-389.
Delaney, Liam, McGovern, Mark and Smith, James P. (2011) "From Angela's ashes to the Celtic tiger: Early life conditions and adult health in Ireland," Journal of Health Economics, Elsevier, vol. 30(1), 1-10, January.
Finch, Caleb, and Eileen Crimmins (2004), “Inflammatory Exposure and Historical Changes in Human Life-Spans,” Science, 305:1736-1739.
Fogel, Robert W. (2005), “Changes in the Physiology of Aging During the Twentieth Century,” National Bureau of Economic Research Working Paper 11233.