Editor's Choice |
Lifecourse epidemiology of disease: a tractable problem?
On British television there are many opportunities for time-wasting, but few are as abject as ‘fantasy football’. On the diminishing occasions when it is not possible to track down a televised game to watch, or at least a soap opera about footballers and their wives, teams can be invented and, through a process of weighted averages, compared using the scores from actual games. The sense that humanity is undergoing de-evolution generated by watching this programme is not redeemed by the blather of the presenter, a man whose lack of talent was already clear in 1977 when he was rejected at an audition for the seminal proto-punk band, The Prefects.1
Sometimes not even fantasy football is available to occupy the long summer evenings, but epidemiologists could fill the void by envisaging their fantasy cohort. Imagine you could magic up the perfect study to answer the epidemiological questions you were interested in. Since this issue of the IJE is on lifecourse epidemiology, consider the perfect lifecourse cohort, one up to the task identified by Yoav Ben-Shlomo and Diana Kuh in their definition of lifecourse epidemiology as ‘the study of long-term effects on chronic disease risk of physical and social exposures during gestation, childhood, adolescence, young adulthood and later adult life. It includes studies of the biological, behavioural and psychosocial pathways that operate across an individual's life course, as well as across generations, to influence the development of chronic diseases’.2
The promise of the lifecourse approach is that trajectories to health and disease can be understood at a level of detail that allows for the identification of how and when optimal outcomes can be promoted. The departing British Prime Minister, Tony Blair, certainly believes in the model. He has unveiled his legacy plan to reduce antisocial behaviour, criminality and social exclusion among future generations through the state intervening (or, more colourfully, launching ‘a crackdown’) during pregnancies deemed to be at risk of producing such future undesirables.3
Can lifecourse epidemiologists help in the optimal timing or formation of such a crackdown? What sort of fantasy cohort would allow the lifecourse processes identified by Ben-Shlomo and Kuh, from before conception to death (or imprisonment in the criminality case), to be understood, and the complexity of confounding, reverse causation, measurement error (including correlated errors), selection bias and missing data to be adequately disciplined? Statisticians have proposed ways in which lifecourse data can be better analysed.4 Are improved statistical methods the solution to the complexity of such data? In this issue of the IJE, Andrew Pickles muses on the currently limited statistical frameworks, and suggests we may benefit from using methods that have been applied to studies of criminality and psychology.5 However, as Pickles points out, ‘to the uninitiated [these methods] seem to involve many assumptions’, and it remains unclear whether initiation really reduces this problem. Both Anne-Marie Andersen6 and Yoav Ben-Shlomo7 discuss two papers from the latest follow-up of the 1958 British Birth Cohort,8 which provide somewhat contrasting pictures of early-life influences on health in middle-age. Would more sophisticated statistical methods greatly clarify things? Somehow I doubt it; however we should clearly try and evaluate whether they would. Discussing the paper by Ramsay and colleagues9 concerning the influence of childhood social circumstances on coronary heart disease risk in late adulthood, Maria Glymour identifies the problems of selection bias and measurement error in such lifecourse studies,10 and touches on other problems, but considers the effort to appropriately model these processes worth it, as it promises to strengthen our ability to draw causal inferences in the lifecourse field.
Other differences in emphasis also emerge in this issue. Corvalán and colleagues attempt to identify the period during growth when body mass index changes most strongly predict adulthood adiposity, and growth between age 3 and 7 appears most important.11 Nicolas Stettler considers some problems in how this conclusion is reached.12 Continuing a long-running debate Backlund et al. conclude that their detailed multilevel analysis of income inequality and mortality within a large US cohort does not suggest that a large effect exists,13 whereas Subramanian and Kawachi14 think the data are supportive. Here the issue is not a disagreement on data analysis or data quality, but on the interpretation of the finding of no meaningful influence of income inequality on mortality above the age of 65—the period of life during which the large majority of deaths now occur in rich countries.
Looking at the big picture—what determines population health?—a series of commentators reflect on 30 years of research following on from Samuel Preston's important paper on mortality and economic development.15–22 Although we may have difficulty in reliably identifying exact processes occurring during individual life trajectories that shape later health, at the population level clear and strong patterns emerge. For example, the history of cigarette smoking levels in populations strongly predict differences in lung cancer rates between these populations. Yet even with such a strong risk factor as cigarette smoking is for lung cancer, prediction of individual risk is poor.23 This distinction in our ability to understand population-level phenomena (generally reasonably good) and individual-level processes (generally rather poor) brings us back to the fact that epidemiology is about collective rather than individual experiences. As Jerry Morris wrote in his foundational book ‘The Uses of Epidemiology’ 50 years ago, ‘the unit of study in epidemiology is the population or group, not the individual’,24 an insight developed by Geoffrey Rose in his recognition that the determinants of incidence rates within populations may be different to the determinants of individual incidents of disease.25
Much of lifecourse epidemiology is focused on trajectories of the lives of individuals, in which stochastic processes that we may never be able to measure or understand play an important role. Epidemiologists may be faced with a similar ‘gloomy prospect’ to the one identified by Eric Turkheimer in the context of the inability of behavioural genetic studies to identify important environmental factors shared between siblings within families.26 Lifecourse trajectory-influencing events may often be of chance or idiosyncratic origin, and thus not tractable by current methods.
So perhaps even a fantasy lifecourse cohort would not be up to the task of greatly improving our understanding of the conception to death processes discussed by Ben-Shlomo and Kuh.2 Anne-Marie Andersen refers to epidemiologists donning a ‘biological fig-leaf’,6 but with some biological processes—as opposed to lifecourse trajectory analysis—we can apply node-by-node perturbation analysis to understand causation,27 sometimes through the use of ‘Nature's randomised trials’,28 as Yoav Ben-Shlomo mentions.7 Otherwise, as the Prefects sang, we may just be ‘Going through the motions’.29
References
1 The Prefects. The Prefects are Amateur Wankers. In: Acute Records (2004).
2 Ben-Shlomo Y, Kuh D. Editorial: A life course approach to chronic disease epidemiology: conceptual models, empirical challenges and interdisciplinary perspectives. Int J Epidemiol (2002) 31:285–93.
3 Unborn babies targeted in crackdown on criminality. In: Guardian (2007) May 16. 1–2.
4 De Stavola BL, Nitsch D, dos Santos Silva I, et al. Statistical.issues in lifecourse epidemiology. Am J Epidemiol (2006) 163:84–96.
5 Pickles A. Commentary: Trajectories, Selection and Cumulative Causation. Int J Epidemiology (2007) 36:549–50.
6 Nybo Andersen A-M. Commentary: Life-course and social epidemiology, the biological fig leaf, and Bob Dylan. Int J Epidemiol (2007) 36:540–41.
7 Ben-Shlomo Y. Editorial: Rising to the challenges and opportunities of life course epidemiology. Int J Epidemiol (2007) 36:481–83.
8 Strachan DP, Rudnicka AR, Power C, et al. Lifecourse influences on health among British adults: effects of region of residence in childhood and adulthood. Int J Epidemiol (2007) 36:532–39.
9 Ramsay SE, Whincup PH, Morris RW, Lennon LT, Wannamethee SG. Are childhood socio-economic circumstances related to coronary heart disease risk? Findings from a population-based study of older men. Int J Epidemiol (2007) 36:561–67.
10 Glymour MM. Commentary: Selected samples and nebulous measures: some methodological difficulties in life-course epidemiology. Int J Epidemiol (2007) 36:568–69.
11 Corvalán C, Gregory CO, Ramirez-Zea M, Martorell R, Stein AD. Size at birth, infant, early and later childhood growth and adult body composition: a prospective study in a stunted population. Int J Epidemiol (2007) 36:551–58.
12 Stettler N. Commentary: Growing up optimally in societies undergoing the nutritional transition, public health and research challenges. Int J Epidemiol (2007) 36:559–60.
13 Backlund E, Rowe G, Lynch J, Wolfson MC, Kaplan GA, Sorlie PD. Income inequality and mortality: a multilevel prospective study of 521 248 individuals in 50 US states. Int J Epidemiol (2007) 36:591–97.
14 Subramanian SV, Kawachi I. Commentary: Chasing the elusive null: the story of income inequality and health. Int J Epidemiol (2007) 36:598–601.
15 Preston SH. The changing relation between mortality and level of economic development. Popul. Stud (Camb.) (1975) 29:231–48. Reprinted Int J Epidemiol 2007;36:484–90.[Medline]
16 Kunitz SJ. Commentary: Samuels Preston's ‘The changing relation between mortality and level of economic development’. Int J Epidemiol (2007) 36:484–90.
17 Wilkinson RG. Commentary: The changing relation between mortality and income. Int J Epidemiol (2007) 36:492–93.
18 Riley JC. Commentary. Int J Epidemiol (2007) 36:494–95.
19 Mackenbach JP. Commentary: Did Preston underestimate the effect of economic development on mortality? Int J Epidemiol (2007) 36:496–97.
20 Bloom DE, Canning D. Commentary: The Preston Curve 30 years on: still sparking fires. Int J Epidemiol (2007) 36:498–99.
21 Leon DA. Commentary: Preston and mortality trends since the mid-1970s. Int J Epidemiol (2007) 36:500–1.
22 Preston SH. Response: On ‘The changing relation between mortality and level of economic development’. Int J Epidemiol (2007) 36:502–3.
23 Hemminki K, Försti A, Lorenzo Bermejo J. Gene-environment interactions in cancer: do they exist? Ann N Y Acad Sci (2006) 1076:137–48.[CrossRef][Web of Science][Medline]
24 Morris JN. Uses of Epidemiology (1957) Edinburgh: Livingstone.
25 Rose G. Sick individuals and sick populations. Int J Epidemiol (1984) 14:32–38. Reprinted Int J Epidemiol 2001;30:427–32.[Web of Science]
26 Turkenheimer E. Spinach and Ice Cream: Why social science is so difficult. In: Behavior Genetics Principles—Di Lalla LF, ed. (2004) Washington: APA.
27 Tegner J, Bjorkegren J. Perturbations to uncover gene networks. Trends Genet (2007) 23:34–41.[CrossRef][Web of Science][Medline]
28 Hingorani A, Humphries S. Nature's randomised trials. Lancet (2005) 366:1906–8.[CrossRef][Web of Science][Medline]
29 The Prefects. Going through the motions. In: London: Rough Trade Records (1979).
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  G. D. Smith 'Something funny seems to happen': J.B.S. Haldane and our chaotic, complex but understandable world Int. J. Epidemiol., June 1, 2008; 37(3): 423 - 426. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||