PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of intjepidLink to Publisher's site
 
Int J Epidemiol. 2009 April; 38(2): 393–402.
Published online 2008 December 17. doi:  10.1093/ije/dyn268
PMCID: PMC2734074

The triumph of the null hypothesis: epidemiology in an age of change

Abstract

Summary The recent confusion concerning the relation between hormone replacement therapy and cardiovascular disease has stirred a new wave of debate about the value and future of epidemiology. Opponents of epidemiology suggest an ever-diminishing role in an age of small risks and complex diseases, yet proponents are not in consensus about how to adapt their discipline to the challenges associated with ageing societies and changing disease patterns. While epidemiology is likely to be increasingly called upon to make sense of the risks involved with these changes, wading into this era with a mindset and tools that were derived from epidemiology's ‘golden era’ of tackling major risk factors, has created more confusion than understanding. Increasingly, we need to downsize epidemiology to what is testable, measurable, and relevant, based on robust methodology and public health rationale. Applying an evolutionary perspective, that views health problems of modernity as a manifestation of the mismatch between our ancient genes and hi-tech lifestyles, can provide guidance for a 21st century research agenda.

Keywords: Epidemiology, chronic disease, small risk, complex disease, evolutionary epidemiology

Introduction

The recent uncertainty surrounding the relation between hormone replacement therapy and cardiovascular disease (HRT-CVD) has again ignited the debate about the value and future of epidemiology.1,2 The issue this time is more serious than the optimal amount of fruits and vegetables we need to eat daily, as it involves the devastating irony that millions of healthy women have been encouraged to take a medication that may put them at risk of the same ailment they were trying to ward off.3 Underlying this dilemma is a credibility crisis brought about by inconsistencies in the results of various epidemiological studies.4–6 Increasingly, voices within and outside the discipline of epidemiology are calling for a total re-evaluation of its tools and paradigms, some going as far as to suggesting abandoning the field entirely.1,7–11 One can argue whether epidemiology is to blame for this state of affairs by adopting the results of cohort studies to formulate treatment guidelines,12 or has been the voice of reason via arguing caution about the ‘protective’ relation between HRT-CVD,13–15 or is an innocent bystander or even pawn at the hands of mass media and corporate interests that manipulate public opinion about medical treatments.1 Regardless, the unavoidable issue is the legitimate concern about the role of epidemiology in an era of small effect, lifestyle-related risks of chronic diseases. This concern has in recent years stirred calls for major methodological and conceptual reevaluation of observational studies (e.g. case control and cohort),1,5,8–11 as their propensity for subtle forms of bias and confounding can influence their value for the study of small risks of chronic disease. Yet a more suitable starting point would be to restore some of the fundamentals of epidemiological practice based on strong theoretical guidance, proper assessment tools and clear public health rationale. As these elements are usually within researchers’ control, addressing them in the context of new directions to improve the prospects of chronic disease epidemiology is warranted.

A historical snapshot

For some time now, epidemiologists have been debating the future ability of their discipline to accommodate emerging disease patterns resulting from the ageing and lifestyle changes of modern societies.7–9,16–22 Mervyn and Ezra Susser identified three main historical stages of epidemiology reflecting the main health threats of the times and the level of knowledge about them. Starting from the sanitary era with its Miasma paradigm, to the infectious disease era accompanying the germ theory, to our chronic disease/risk factors era with its so called ‘black box’ paradigm, to quote Petr Skrabanek's famous metaphor.7,23 Perhaps, it is ‘black box’ epidemiology, referring to ‘the pursuit of exposure-outcome relations without much attention to biological understanding or inference’, that has been most problematic.22–24 The willingness of epidemiologists to run ahead of biology to influence the societal burden of disease is a longstanding tradition of the discipline with some impressive successes.25,26 But while mechanistic associations can lead to hypothesis formulation in the area of major risk factors,26 they are unlikely to be as successful with small risks, given the complexity of the causal grid. This inadequacy has paved the way for a new phase in epidemiology,9,27,28 called ecoepidemiology by the Sussers.27 The concept of ecoepidemiology is based on a multilevel paradigm called the ‘Chinese boxes’ to reinforce ‘the importance of distal (societal), individual and microbiological interactions in disease development’.27 The ecoepidemiology concept also is an attempt to reclaim the public health edge of epidemiology, thought by many to have been lost amidst an overemphasis on individual-level risk factors.18–21,29

Risk factor epidemiology and the importance of guiding hypotheses

Observational studies have been instrumental for the identification of major risk factors to health (e.g. smoking, hypertension, hypercholesterolemia, malnutrition). Yet the HRT-CVD debate has drawn attention to the potentially high price of making unwarranted claims about small and interconnected associations. Epidemiology's doubters argue that the success stories of epidemiology were all easy hits; the magnitude of the association between cigarette smoking and lung cancer was so large that it could be reliably observed even with flawed study designs.30 However, when we move to the realm of complex diseases and smaller effect sizes, bias and confounding start to creep into cohort and case control studies in a variety of unpredictable ways leading to their derailment in any direction.8,22,30,31 But, if we could establish major risk factors with crude tools, why cannot we be able to assess small risks with better ones?

So far, the uncertainty about epidemiological evidence has led to the ‘here-today-gone-tomorrow nature of medical wisdom’,30 and perhaps the confusion we ourselves have about how to lead a healthy lifestyle.30–32 Notwithstanding our position on the ever-changing nature of scientific knowledge and the processes involved, epidemiology has been hurt most by ill conceived and conducted studies with rushed conclusions. As we witnessed, an excess of such studies, combined with vested interests in certain results and media hunger for the newsworthy, has become a formula for trouble.

Epidemiologists have always been vigilant about the danger of claiming associations that do not exist in reality by adopting the null/alternative hypothesis approach, which emphasizes lower tolerance for such error (i.e. type I error) than for missing a real link (i.e. type II error).33 It is an approach similar to the judicial system, which considers convicting an innocent a greater mistake than letting a criminal go free. This approach emphasizes as well the need for research to be driven at the outset by a sound and fully articulated hypothesis. The wisdom of this safeguard seems to be lost on many researchers nowadays, who like to formulate and interpret their studies by what comes out of the logistic regression grinder. Given researchers’ ingeniousness in explaining exotic associations, and the ever-expanding volume of knowledge, it is not hard to find biological explanations for contradictory findings. For example, studies showing positive associations between exposure to pets and childhood asthma attribute this association to animal allergens (compatible with the allergy paradigm), while studies showing negative associations attribute it to pet-related microbial products (compatible with the hygiene paradigm).34,35 There are ample examples of tailor-made post hoc hypotheses, transforming epidemiology from a rational to a ridiculous endeavour, and highlighting the growing importance of epidemiological studies being guided by well-grounded a priori hypotheses.

The asthma epidemic and the importance of measurement

In the current age of small risk factors, the need for epidemiology to be grounded in strong inference is matched only by the need for sensitive tools to assess exposures and outcomes. As an example, childhood asthma is a recent global epidemic, whose secular and spatial patterns favour environmental causes.36,37 Yet more than two decades of intensive research has yet to yield a single target for intervention.8,38 To get out of this deadlock, researchers have been calling for large prospective studies, studies starting as early in life as possible and to differentiate between different asthma phenotypes.39–41 Others have argued, invoking Geoffrey Rose's seminal work,42 that the etiological factors involved in the asthma epidemic are likely to be homogenously distributed within populations, indicating the need for international comparisons.43 These attempts to shore up asthma research reflect sound epidemiological reasoning, but have yet to advance our understanding of the etiology of the asthma epidemic, despite a big price tag.44–46

So, is this impasse in asthma research due to an inadequate conceptual framework or faulty study designs and tools of epidemiology? Research into the asthma epidemic was guided for the most part by the hygiene hypothesis, which relied on early observations of the ‘protective’ effect of sibship size and birth order to suggest that the asthma epidemic could be the side effect of our increased hygiene and success in curbing infectious diseases.47 Biological underpinnings of the hygiene hypothesis were soon provided based on the preferential development of immune-allergic responses in the absence of infectious stimuli.48,49 Certainly, the hygiene hypothesis provided a timely model for an environmentally driven epidemic incorporating changes in our lifestyles with a good dose of evolutionary wisdom. While evidence for and against the hygiene model are accumulating, it has become apparent that it is an over-simplification of a more complex picture.45,50–52

Part of the reason why we still do not have a verdict on the hygiene hypothesis despite countless studies, lies in the uncertainty surrounding the measurement tools we are applying in asthma research. For the most part, this research relied on self-reported or parental-reported questionnaires to assess asthma symptoms and exposures. As a result, ‘asthma’ was transformed gradually into ‘wheeze,’ as the symptom and the disease became synonymous in epidemiological studies. Even in English, this substitution results in crude assessment of a disease that has several phenotypic and age-related manifestations53 but in other languages, translations of ‘wheeze’ can relate poorly to asthma.54,55 Video questionnaires or objective markers (e.g. bronchial hyperresponsiveness, BHR) also have been used for outcome assessment, but none has come close to being a gold standard for the assessment of asthma in epidemiological studies.56–64 Exposure assessment in asthma studies has not fared better, as questionnaires again have been the main tool to assess exposure to infection, animals and their products, air pollutants, certain foods and medicines, and secondhand smoke.65,66 For example, exposure to air pollution in the International Study of Asthma and Allergies in Children (ISAAC) was assessed based on questions asking about (i) the frequency of trucks passing on the residential street on weekdays; and (ii) the severity of traffic noise that forced participants to close the window.67 One cannot expect such assessment to provide credible information about this exposure, let alone study its relative contribution among a variety of other exposures measured with similarly crude instruments. In cases where more objective markers of exposure were sought (e.g. endotoxin, antibodies, skin tests) they were either non-specific/sensitive, were poor markers of chronic exposure relevant to the development of asthma, or were perhaps offset by the crude measurement of the outcome.68–73 The resultant confusion was even greater than researchers’ ingenuity to explain it, to the extent that it is not uncommon to see asthma studies conclude that factor X was associated with past year wheeze, but not ever-wheeze or asthma diagnosis, or with sensitization but not asthma symptoms or BHR, or for that matter any combination of a myriad of outcomes that arguably represent the same clinical entity.74–86 Such inconsistencies have yet to invoke a major re-examination of research paradigms and tools used to study asthma, perhaps signifying ‘black box’ epidemiology at its darkest hour.

From static to dynamic and relevant epidemiology

The transformation needed in small risk epidemiology involves not only applying better guiding models and sensitive markers, but perhaps as importantly, a departure from a static perception of the relation between exposure and disease, reflecting the era of questionnaires and categorical variables, to a more dynamic, context-relevant and life-course approach.87,88 As such, ‘snapshot’ exposure assessment should be replaced with modelling of exposures based on information from a variety of sources/levels (GIS, mobility patterns, biosensors), the use of sensitive markers of long-term exposure (e.g. hair/nail nicotine for exposure to ETS, or glycosolated haemoglobin for control of blood sugar), and the incorporation of macro-level attributes (e.g. area level characteristics).89–94 This can potentially sharpen our ability to approximate the dynamic nature of our interaction with our ever-changing environments. Similar transformation is needed in our perception of risk to reflect its dynamic nature across people's life roles, past experiences and current behaviors and contexts. For example, many of us will fall under some risk category for several chronic diseases only by virtue of our age, which may lead us to seek controversial treatments that can be themselves risky (e.g. vitamin/mineral supplements).95–97

This inner conceptual dynamism should be supplemented with an outer public health orientation. Without that, epidemiology may dwell in the realm of the ideal rather than deal with the constraints of reality. For example, even if efficacy trials confirm the benefits of a school-based programme or five portions/day of fruits in obesity prevention, this can be useless if such results are not translatable into society-wide sustainable interventions.98 Another form of this reductionist approach is for epidemiological studies to statistically control for socioeconomic deprivation, zooming in on more proximal behavioral determinants, while relegating an apparently major determinant of chronic disease risk to the status of a nuisance variable.20 Both approaches shift focus from the society to the sufferers,17,20 who are arguably blamed for failing to follow the health recommendations we have imparted to them for years, and, in the process, stripping epidemiology of its public health essence. This is not to say that public health cannot be advanced by purely medical interventions (e.g. vaccination), or that we need to structure people's lives around some health ideals, but to desensitize epidemiology to a major determinant of chronic disease, such as socioeconomic deprivation has perhaps contributed to the widening of health inequalities, particularly in the USA.99 Classifications such as social epidemiology, clinical epidemiology, infectious disease or cardiovascular epidemiology, may have facilitated this selectiveness in how we treat different variables, by creating a false perception of compartmentalization of different disease processes within individuals and populations.

The opposite can be true as well, as some studies of distal-level determinants lack clear public health orientation. For example, what are the public health implications of demonstrating that low neighbourhood-level educational attainment is detrimental to cognitive functions of elderly residents,100 as one can argue that intervening on older residents to improve their cognitive ability is more practical than increasing the average education level of whole geographic areas. Finally, many relations sought by epidemiologists are ‘fly-by-night’ endeavours embodying no foreseeable public health rationale or follow-up plan, an example of which is the finding of a positive association between husbands’ participation in housework and wives’ psychosocial health.101 One can argue that such a study proves the obvious, can neither be free of residual confounding (husbands who help with housework are likely to be a different breed from those who do not), nor has a clear public health message (who should use this information and how?). For epidemiology to stay relevant, associations must not only be driven by statistical empiricism, but must have a clear public health rationale.

Epidemiology in the age of complex lifestyle-related diseases

It has been hard for epidemiologists to adapt to changing patterns of diseases and their risks in modern societies, partly because of the pace of such changes. New epidemics, such as obesity, asthma and depressive disorders are evolving rapidly and creating pressure for evidence-based solutions. In response, inspired by past successes with major risk factors, epidemiologists rushed to the scene with their usual tools. As it turned out, the task this time was more difficult, casting some serious doubt on epidemiology's ability to respond to current and future threats to health.8,9,30 While only time can tell the fate of epidemiology, one major lesson to be learned from the HRT-CVD story is that observational epidemiology can never be free of residual confounding when comparisons involve health-oriented behaviours.102,103 Yet, observational studies can be valuable for the study of long-term side effects of drug treatments, as these are mostly unintended and unpredictable, therefore are not ‘confounded by indication;’ meaning that they are usually not associated with the treatment decision.104–106 A similar valid scenario for observational epidemiology is the study of potential risk/protective factors that are unknown to the public, as this reduces the probability of associative-selection bias based on differential health awareness between the comparison groups.107

In reality, as we continue to monitor the health of populations, epidemiology will likely be called upon increasingly to make sense of the risks involved in the dramatic changes in our lifestyles and environments. But to continue doing epidemiological studies using the same tried and failed approaches is not an option; the asthma example is a clear indicator of how epidemiology can turn into an absurd exercise when aims are put ahead of tools and concepts. Realizing the current crisis of credibility, epidemiologists have responded by calling for larger and longer studies, coupling of epidemiology with molecular genetics (e.g. genetic epidemiology and biobanks), strong inference, incorporating multilevel attributes and for greater attention to residual confounding.8,108–114 They have also mobilized to improve the reporting of observational studies (STROBE guidelines) to allow editors, reviewers and consumers of epidemiological data to make informed judgement on the quality of reported studies.115 All these approaches can potentially improve epidemiology's ability to zoom in on small risks, yet the defining feature of the small risk/chronic disease era has so far been the triumph of the null hypothesis (i.e. no effect).8,9,116

Such responses reflect healthy self-criticism on the part of epidemiologists, but a clearer articulation of priorities is needed in the face of mounting criticism. Bigger samples and longer studies, for example, can lead to the magnification of errors, loss to follow up, and increase in costs, while applying multilevel approaches without a sound theoretical framework runs the risk of becoming another form of ‘black box’ epidemiology.117 Genetic epidemiology on the other hand, is not free of the problems of observational epidemiology,8 yet recent years have witnessed some promising advances in this field, especially the application of Mendelian randomization to control for environmental confounders in observational studies.118 Mendelian randomization utilizes the random distribution of genetic alleles at the time of gamete formation to identify genetic variants that robustly predict environmentally modifiable exposures and uses them as un-confounded proxies (instruments) for those exposures. While the promise of this approach is still unfolding,119 there are currently not many conditions that can be studied this way, i.e. in which we have a clear understanding of the genetic basis of suspected etiological exposures.

As epidemiologists relish the promise of ecoepidemiology,28,111 this concept can come with its own caveats. Namely, a comprehensive hypothesis based on this approach can involve ‘many-to-many’ factors to account for,8,120 as the range and possibilities of gene–environment interactions and pathways involved in chronic disease form a vast causal universe. As such, a vision that lies between ‘black box’ and ‘Chinese boxes’ may be needed, whereby certain causal pathways can be identified and worked upon without the need to unravel the whole complexity of the relation between exposures, genes, and chronic diseases. For example, if we can establish an effective way to influence obesity in the society (e.g. through policy that rewards/supports active transportation), we can potentially influence a good deal of CVD, and perhaps a variety of other illnesses, without needing to have all the information about the factors and mechanisms involved. Ecological models of major health risks are emerging, which can harness candidate pathways for study and intervention (Figure 1). In parallel, methodological paradigms and analytical tools are being developed to accommodate this new direction (e.g. complexity theory, multilevel modelling, pathway analysis).98,115,121–124 For example, the use of multilevel analysis for the study of obesity and CVD risk has helped to shift the focus from personal behaviours to encompass wider environmental influences (e.g. neighbourhood adversities) that may be more amenable to sustainable interventions through policy.125–129 The widening of health inequalities in the US, driven mainly by non-linear socio-economic and political dynamics,99 underscores the importance of these new trends in epidemiology.

Figure 1
Simplified multilevel approach to the study of environmental influences on obesity98

An evolutionary perspective for epidemiology

Given the methodological constraints imposed on epidemiological research by the complex nature of problems we are increasingly facing, epidemiology can benefit from some guidance as to what risk factors represent good targets for exploration, and how to interpret inconsistent epidemiological data. An evolutionary perspective can help provide a guiding framework for the epidemiology of chronic diseases, being viewed as a result of the immense adaptive pressures brought about by the mismatch between our genes still lingering at the hunter-gatherer era and our hi-tech lifestyles.130 Accordingly, exposures relevant to our health are those that either underwent a rapid change within a short time, or represent an obvious diversion from the environments that prevailed during most of our evolution. In this sense, our eating, mobility, recreation, socialization and communication patterns, as well as our increasingly indoor existence should be relevant to the development of chronic diseases such as obesity, asthma, CVD and depression.131

As broad as this perspective can be, its application can help to sift through the tides of confusing health information we are bombarded with each day. For example, examining the HRT-CVD relation under the evolutionary lens would have raised a big red flag about a potential for harm; women did not evolve to have lifelong active ovaries, despite the clear reproductive advantage such a trait would have conferred. The same would apply to the long-held protective relation between low fat diet and CVD risk132–136, which was recently refuted in a randomized clinical trial.137 As huntergatherers, humans until 500 generations ago consumed mainly wild and unprocessed food foraged and hunted from their environment and rich in fats, fibre, vitamins and minerals.138–140 So, the low-fat mania perhaps does not tell the whole story about the diet–CVD risk relation, especially when corporate voices join the choir by promoting low-fat options, potentially creating a complacency that can even lead to increased consumption.98

Another benefit of the evolutionary perspective is to help understand the difficulties facing genetic epidemiology, based on the fact that the direction of evolution of traits is from phenotype to genotype and not the opposite.141 This can underlie the existence of multiple genetic pathways to each outcome complicating the genetic study of chronic diseases. To add another layer of complexity is the fact that phenotype fitness is not a simple measure, but represents a complex fitness landscape involving the trade-off of numerous traits, each of which is the result of even greater genetic diversity.142 For complex traits therefore, considerable genetic variance can be maintained within the population even in the face of strong selection forces.143 Such complexity undermines the hopes that genetic epidemiology will provide ready answers to the puzzles of chronic diseases.

Finally, an evolutionary perspective can guide intervention research. Specifically, rather than basing nutritional and physical activity recommendations on observational data that may well reflect a systematic difference in the way we assemble our comparison groups (e.g. those who eat ‘healthy’ are likely to be engaged in other less-measurable health behaviours), we may need to inform such recommendations with knowledge of our ancestral dietary and activity patterns.144,145 In addition, the appreciation that we have a genetically hard-wired taste for energy-dense food can help us understand why most dietary-based obesity interventions fail in the long run.98 So instead of trying to work more forcefully against our instincts by increasing the intensity and length of interventions, we may be better off trying to manipulate food taste and energy content,146 to imitate the Paleolithic diet but without the excess calories.

Concluding remarks

Apparently, the golden era of major risk factor epidemiology is giving way to a less glorious, but certainly more complex and perhaps more important one of studying small and interconnected risk factors related to our ever-changing lifestyles and environments. We perhaps have just scratched the surface of what epidemiology can achieve and how it can help us understand how unfavourable environments are shaping our behaviour and health. Wading into this era with a major risk factor mentality and instruments has created confusion. While endorsing new developments in epidemiological research, the era of chronic disease epidemiology mandates more than ever the need to rely on sound theoretical models as well as accurately measurable outcomes and exposures. So in contrast to the calls for larger, longer and wider-reach epidemiology, what is advocated here is to downsize epidemiological research to what is testable, measurable and relevant. An evolutionary perspective of the dynamic interaction of humans with their environments can help guide such a research agenda. In the age of publish or perish, vested interests, publication bias, scarce funding, media's hunger for hit news and web publishing, epidemiology can best navigate these rough waters by being anchored in a clear sense of its inner methodological constraints and outer public health thrust.

Funding

NIDA (R01DA024876-01 to W.M.).

Acknowledgements

Wasim Maziak thankfully acknowledges Dr Kenneth Ward for critically reading and editing this article.

Conflict of interest: None declared.

KEY MESSAGES

  • There is a crisis of credibility facing epidemiology today, brought about by the barrage of studies with less than optimal methods and conflicting results.
  • As epidemiology enters the era of chronic disease and small risk, it becomes more critical for epidemiological studies to be guided at the inception by well-grounded hypotheses, a dynamic perception of the relation between exposure-outcome and to utilize accurate assessment tools.
  • Novelty or methodological precision should not substitute for public health relevance when evaluating epidemiological studies.
  • New conceptual (e.g. multilevel ecoepidemiology) and methodological (e.g. Mendelian randomization) advances should be embraced in light of the need to downsize epidemiology to what is testable, measurable and relevant.
  • An evolutionary and dynamic understanding of our interactions with our changing environments can provide a guiding context for epidemiological research.

References

1. Lawlor DA, Davey Smith G, Ebrahim S. Commentary: the hormone replacement-coronary heart disease conundrum: is this the death of observational epidemiology? Int J Epidemiol. 2004;33:464–67. [PubMed]
2. Stampfer M. Stampfer. Commentary: Hormones and heart disease: do trials and observational studies address different questions? Int J Epidemiol. 2004;33:454–55. [PubMed]
3. Harrington DM, Howard TD. From presumed benefits to potential harm—hormone therapy and heart disease. N Engl J Med. 2003;349:519–21. [PubMed]
4. Whittmore AS, McGuire V. Observational studies and randomized studies of hormone replacement therapy: what can we learn from them? Epidemiology. 2003;14:8–10. [PubMed]
5. Michels KB. Hormone replacement therapy in epidemiologic studies and randomized clinical trials—are we checkmate? Epidemiology. 2003;14:3–5. [PubMed]
6. Taubes G. Epidemiology faces its limits. Science. 1995;269:164–69. [PubMed]
7. Susser M, Susser E. Choosing a future for epidemiology: I. Eras and paradigms. Am J Public Health. 1996;86:668–73. [PubMed]
8. Buchanan AV, Weiss KM, Fullerton SM. Dissecting complex disease: the quest for the Philosopher's Stone? Int J Epidemiol. 2006;35:562–71. [PubMed]
9. Davey Smith G, Ebrahim S. Epidemiology—is it time to call it a day? Int J Epidemiol. 2001;30:1–11. [PubMed]
10. Pritchard KI. Should observational studies be a thing of the past? J Natl Cancer Inst. 2008;100:451–52. [PubMed]
11. Le Fanu J. Rise and fall of modern medicine. Lancet. 1999;354:518. [PubMed]
12. Barton S. Which clinical studies provide the best evidence? The best RCT still trumps the best observational study. BMJ. 2000;321:255–56. [PMC free article] [PubMed]
13. Petitti DB, Perlman JA, Sidney S. Noncontraceptive estrogens and mortality: long-term follow-up of women in the Walnut Creek Study. Obstet Gynecol. 1987;70(3 Pt 1):289–93. [PubMed]
14. Barrett-Connor E, Stuenkel CA. Hormone replacement therapy (HRT)—risks and benefits. Int J Epidemiol. 2001;30:423–26. [PubMed]
15. Vandenbroucke JP. How much of the cardioprotective effect of postmenopausal estrogens is real? Epidemiology. 1995;6:207–8. [PubMed]
16. Rothman KJ. Commentary: Epidemiology still ascendant. Int J Epidemiol. 2007;36:710–11. [PubMed]
17. Pearce N. Commentary: The rise and rise of corporate epidemiology and the narrowing of epidemiology's vision. Int J Epidemiol. 2007;36:713–17. [PubMed]
18. Pearce N. Traditional epidemiology, modern epidemiology, and public health. Am J Public Health. 1996;86:678–83. [PubMed]
19. Savitz DA, Poole C, Miller WC. Reassessing the role of epidemiology in public health. Am J Public Health. 1999;89:1158–61. [PubMed]
20. McMichael AJ. Prisoners of the proximate: loosening the constraints on epidemiology in an age of change. Am J Epidemiol. 1999;149:887–97. [PubMed]
21. Krieger N. Epidemiology and social sciences: towards a critical reengagement in the 21st century. Epidemiol Rev. 2000;22:155–63. [PubMed]
22. Davey Smith G. Reflections on the limitations to epidemiology. J Clin Epidemiol. 2001;54:325–31. [PubMed]
23. Skrabanek P. The emptiness of the black box. Epidemiology. 1994;5:553–55. [PubMed]
24. Susser M. Does risk factor epidemiology put epidemiology at risk? Peering into the future. J Epidemiol Community Health. 1998;52:608–11. [PMC free article] [PubMed]
25. Savitz DA. In defense of black box epidemiology. Epidemiology. 1994;5:550–52. [PubMed]
26. Greenland S, Gago-Dominguez M, Castelao JE. The value of risk-factor (‘black-box’) epidemiology. Epidemiology. 2004;15:529–35. [PubMed]
27. Susser M, Susser E. Choosing a future for epidemiology: II. From black box to Chinese boxes and eco-epidemiology. Am J Public Health. 1996;86:674–77. [PubMed]
28. Poole C, Rothman KJ. Our conscientious objection to the epidemiology wars. J Epidemiol Community Health. 1998;52:613–14. [PMC free article] [PubMed]
29. Shy CM. The failure of academic epidemiology: witness for the prosecution. Am J Epidemiol. 1997;145:479–84. [PubMed]
30. Taubes G. Do We Really Know What Makes Us Healthy?. The New York Times; 2007. Sep 17,
31. Mayes LC, Horwitz RI, Feinstein AR. A collection of 56 topics with contradictory results in case-control research. Int J Epidemiol. 1988;17:680–85. [PubMed]
32. Taubes G. Nutrition. The soft science of dietary fat. Science. 2001;291:2536–45. [PubMed]
33. Schulz KF, Grimes DA. Bias and causal associations in observational research. Lancet. 2002;359:248–52. [PubMed]
34. Apter AJ. Early exposure to allergen: is this the cat's meow, or are we barking up the wrong tree? J Allergy Clin Immunol. 2003;111:938–46. [PubMed]
35. Heinrich J, Gehring U, Douwes J, et al. INGA-Study Group. Pets and vermin are associated with high endotoxin levels in house dust. Clin Exp Allergy. 2001;31:1839–45. [PubMed]
36. Kay AB. Allergy and allergic diseases. First of two parts. N Engl J Med. 2001;344:30–37. [PubMed]
37. Eder W, Ege MJ, von Mutius E. The asthma epidemic. N Engl J Med. 2006;355:2226–35. [PubMed]
38. von Mutius E. Allergies, infections and the hygiene hypothesis—the epidemiological evidence. Immunobiology. 2007;212:433–39. [PubMed]
39. Sheikh A, Strachan DP. The hygiene theory: fact or fiction? Curr Opin Otolaryngol Head Neck Surg. 2004;12:232–36. [PubMed]
40. Rebordosa C, Kogevinas M, Sørensen HT, Olsen J. Pre-natal exposure to paracetamol and risk of wheezing and asthma in children: A birth cohort study. Int J Epidemiol. 2008;37:583–90. [PubMed]
41. Wright AL. Epidemiology of asthma and recurrent wheeze in childhood. Clin Rev Allergy Immunol. 2002;22:33–44. [PubMed]
42. Rose G. Sick individuals and sick populations. Int J Epidemiol. 1985;14:32–38. [PubMed]
43. Lewis S. ISAAC—a hypothesis generator for asthma? International Study of Asthma and Allergies in Childhood. Lancet. 1998;351:1220–21. [PubMed]
44. Beasley R, Ellwood P, Asher I. International patterns of the prevalence of pediatric asthma the ISAAC program. Pediatr Clin North Am. 2003;50:539–53. [PubMed]
45. Maziak W. The asthma epidemic and our artificial habitats. BMC Pulm Med. 2005;5:5. [PMC free article] [PubMed]
46. Buchanan AV, Weiss KM, Fullerton SM. On stones, wands, and promises. Int J Epidemiol. 2006;35:593–96.
47. Strachan DP: Hay fever, hygiene, and household size. BMJ. 1989;299:1259–60. [PMC free article] [PubMed]
48. Martinez FD, Holt PG. Role of microbial burden in aetiology of allergy and asthma. Lancet. 1999;354(Suppl 2):SII12–15. [PubMed]
49. Cookson WO, Moffatt MF. Asthma: an epidemic in the absence of infection? Science. 1997;275:41–42. [PubMed]
50. Ramsey CD, Celedón JC. The hygiene hypothesis and asthma. Curr Opin Pulm Med. 2005;11:14–20. [PubMed]
51. Effros RM, Nagaraj H. Asthma: new developments concerning immune mechanisms, diagnosis and treatment. Curr Opin Pulm Med. 2007;13:37–43. [PubMed]
52. van Schayck CP, Knottnerus JA. No clinical evidence base to support the hygiene hypothesis. Prim Care Respir J. 2004;13:76–79. [PubMed]
53. Stein RT, Martinez FD. Asthma phenotypes in childhood: lessons from an epidemiological approach. Paediatr Respir Rev. 2004;5:155–61. [PubMed]
54. Pekkanen J, Remes ST, Husman T, et al. Prevalence of asthma symptoms in video and written questionnaires among children in four regions of Finland. Eur Respir J. 1997;10:1787–94. [PubMed]
55. Hong SJ, Kim SW, Oh JW, et al. The validity of the ISAAC written questionnaire and the ISAAC video questionnaire (AVQ 3.0) for predicting asthma associated with bronchial hyperreactivity in a group of 13–14 year old Korean schoolchildren. J Korean Med Sci. 2003;18:48–52. [PMC free article] [PubMed]
56. Remes ST, Pekkanen J, Remes K, Salonen RO, Korppi M. In search of childhood asthma: questionnaire, tests of bronchial hyperresponsiveness, and clinical evaluation. Thorax. 2002;57:120–26. [PMC free article] [PubMed]
57. Giovannangelo M, Gehring U, Nordling E, et al. Childhood cat allergen exposure in three European countries: The AIRALLERG study. Sci Total Environ. 2006;369:82–90. [PubMed]
58. Crane J, Mallol J, Beasley R, Stewart A, Asher MI. International Study of Asthma and Allergies in Childhood Phase I study group. Agreement between written and video questions for comparing asthma symptoms in ISAAC. Eur Respir J. 2003;21:455–61. [PubMed]
59. Pizzichini MM, Rennie D, Senthilselvan A, Taylor B, Habbick BF, Sears MR. Limited agreement between written and video asthma symptom questionnaires. Pediatr Pulmonol. 2000;30:307–12. [PubMed]
60. Michel G, Silverman M, Strippoli MP, et al. Parental understanding of wheeze and its impact on asthma prevalence estimates. Eur Respir J. 2006;28:1124–30. [PubMed]
61. Braun-Fahrländer C, Gassner M, Grize L, et al. Comparison of responses to an asthma symptom questionnaire (ISAAC core questions) completed by adolescents and their parents. SCARPOL-Team. Swiss Study on Childhood Allergy and Respiratory Symptoms with respect to Air Pollution. Pediatr Pulmonol. 1998;25:159–66. [PubMed]
62. Gharagozlou M, Khalili S, Hallajmofrad M, Mohammadzadeh R, Mousavi G, Golkari H. Gender similarity in low agreement between written and video ISAAC asthma questionnaires. Monaldi Arch Chest Dis. 2006;65:184–88. [PubMed]
63. Hederos CA, Hasselgren M, Hedlin G, Bornehag CG. Comparison of clinically diagnosed asthma with parental assessment of children's asthma in a questionnaire. Pediatr Allergy Immunol. 2007;18:135–41. [PubMed]
64. Rahimi Rad MH, Hejazi ME. Agreement between written and video asthma symptoms questionnaires in school children in Urmia, Iran. Iran J Allergy Asthma Immunol. 2007;6:21–25. [PubMed]
65. Asher MI, Keil U, Anderson HR, et al. International Study of Asthma and Allergies in Childhood (ISAAC): rationale and methods. Eur Respir J. 1995;8:483–91. [PubMed]
66. Burney PG, Luczynska C, Chinn S, Jarvis D. The European Community Respiratory Health Survey. Eur Respiratory J. 1994;7:954–60. [PubMed]
67. Behrens T, Taeger D, Maziak W, et al. Self-reported traffic density and atopic disease in children. Results of the ISAAC Phase III survey in Muenster, Germany. Pediatr Allergy Immunol. 2004;15:331–39. [PubMed]
68. Maziak W. Endotoxin and asthma. N Engl J Med. 2003;348:171–74. [PubMed]
69. Rylander R. Endotoxin and asthma. Am J Respir Crit Care Med. 2006;173:1177. [PubMed]
70. Lynch NR, Hagel IA, Palenque ME, et al. Relationship between helminthic infection and IgE response in atopic and nonatopic children in a tropical environment. J Allergy Clin Immunol. 1998;101(2 Pt 1):217–21. [PubMed]
71. Obihara CC, Beyers N, Gie RP, et al. Respiratory atopic disease, Ascaris-immunoglobulin E and tuberculin testing in urban South African children. Clin Exp Allergy. 2006;36:640–48. [PubMed]
72. Ponte EV, Lima F, Araújo MI, Oliveira RR, Cardoso LS, Cruz AA. Skin test reactivity and Der p-induced interleukin 10 production in patients with asthma or rhinitis infected with Ascaris. Ann Allergy Asthma Immunol. 2006;96:713–18. [PubMed]
73. Ota MO, van der Sande MA, Walraven GE, et al. Absence of association between delayed type hypersensitivity to tuberculin and atopy in children in The Gambia. Clin Exp Allergy. 2003;33:731–36. [PubMed]
74. Magnusson LL, Olesen AB, Wennborg H, Olsen J. Wheezing, asthma, hayfever, and atopic eczema in childhood following exposure to tobacco smoke in fetal life. Clin Exp Allergy. 2005;35:1550–56. [PubMed]
75. Schachter LM, Salome CM, Peat JK, Woolcock AJ. Obesity is a risk for asthma and wheeze but not airway hyperresponsiveness. Thorax. 2001;56:4–8. [PMC free article] [PubMed]
76. Yarnell JW, Stevenson MR, MacMahon J, et al. Smoking, atopy and certain furry pets are major determinants of respiratory symptoms in children: the International Study of Asthma and Allergies in Childhood Study (Ireland) Clin Exp Allergy. 2003;33:96–100. [PubMed]
77. Arshad SH, Kurukulaaratchy RJ, Fenn M, Matthews S. Early life risk factors for current wheeze, asthma, and bronchial hyperresponsiveness at 10 years of age. Chest. 2005;127:502–8. [PubMed]
78. Bolte G, Schmidt M, Maziak W, et al. The relation of markers of fetal growth with asthma, allergies and serum immunoglobulin E levels in children at age 5–7 years. Clin Exp Allergy. 2004;34:381–88. [PubMed]
79. Annus T, Björkstén B, Mai XM, et al. Wheezing in relation to atopy and environmental factors in Estonian and Swedish schoolchildren. Clin Exp Allergy. 2001;31:1846–53. [PubMed]
80. Davey G, Venn A, Belete H, Berhane Y, Britton J. Wheeze, allergic sensitization and geohelminth infection in Butajira, Ethiopia. Clin Exp Allergy. 2005;35:301–7. [PubMed]
81. Omenaas E, Fluge O, Buist AS, Vollmer WM, Gulsvik A. Dietary vitamin C intake is inversely related to cough and wheeze in young smokers. Respir Med. 2003;97:134–42. [PubMed]
82. Hermann C, De Fine Olivarius N, Høst A, Begtrup K, Hollnagel H. Prevalence, severity and determinants of asthma in Danish five-year-olds. Acta Paediatr. 2006;95:1182–90. [PubMed]
83. Lau S, Illi S, Platts-Mills TA, et al. Multicentre Allergy Study Group. Longitudinal study on the relationship between cat allergen and endotoxin exposure, sensitization, cat-specific IgG and development of asthma in childhood–report of the German Multicentre Allergy Study (MAS 90) Allergy. 2005;60:766–73. [PubMed]
84. Carter PM, Peterson EL, Ownby DR, Zoratti EM, Johnson CC. Relationship of house-dust mite allergen exposure in children's bedrooms in infancy to bronchial hyperresponsiveness and asthma diagnosis by age 6 to 7. Ann Allergy Asthma Immunol. 2003;90:41–44. [PubMed]
85. Cullinan P, MacNeill SJ, Harris JM, et al. Early allergen exposure, skin prick responses, and atopic wheeze at age 5 in English children: a cohort study. Thorax. 2004;59:855–61. [PMC free article] [PubMed]
86. Harris JM, Williams HC, White C, et al. Early allergen exposure and atopic eczema. Br J Dermatol. 2007;156:698–704. [PubMed]
87. Kuh D, Ben-Shlomo Y, Lynch J, Hallqvist J, Power C. Life course epidemiology. J Epidemiol Community Health. 2003;57:778–83. [PMC free article] [PubMed]
88. Lawlor DA, Davey Smith G, Patel R, Ebrahim S. Life-course socioeconomic position, area deprivation, and coronary heart disease: findings from the British Women's Heart and Health Study. Am J Public Health. 2005;95:91–97. [PubMed]
89. Clougherty JE, Levy JI, Kubzansky LD, et al. Synergistic effects of traffic-related air pollution and exposure to violence on urban asthma etiology. Environ Health Perspect. 2007;115:1140–46. [PMC free article] [PubMed]
90. Buxton DB. Nanotechnology in the diagnosis and management of heart, lung and blood diseases. Expert Rev Mol Diagn. 2007;7:149–60. [PubMed]
91. Sørensen M, Bisgaard H, Stage M, Loft S. Biomarkers of exposure to environmental tobacco smoke in infants. Biomarkers. 2007;12:38–46. [PubMed]
92. Frank LD, Sallis JF, Conway TL, Chapman JE, Saelens BE, Bachman W. Many pathways from land use to health associations between neighborhood walkability and active transportation, body mass index, and air quality. J Am Plann Assoc. 2006;72:75–87.
93. O’ Campo P. Invited commentary: Advancing theory and methods for multilevel models of residential neighborhoods and health. Am J Epidemiol. 2003;157:9–13. [PubMed]
94. Maziak W. Asthma and the exposure-disease tenet. J Clin Epidemiol. 2002;55:737–40. [PubMed]
95. Bolland MJ, Barber PA, Doughty RN, et al. Vascular events in healthy older women receiving calcium supplementation: randomised controlled trial. BMJ. 2008;336:262–66. [PMC free article] [PubMed]
96. Buijsse B, Feskens EJ, Kwape L, Kok FJ, Kromhout D. Both {alpha}- and -Carotene, but Not Tocopherols and Vitamin C, Are Inversely Related to 15-Year Cardiovascular Mortality in Dutch Elderly Men. J Nutr. 2008;138:344–50. [PubMed]
97. Heinen MM, Hughes MC, Ibiebele TI, Marks GC, Green AC, van der Pols JC. Intake of antioxidant nutrients and the risk of skin cancer. Eur J Cancer. 2007;43:2707–16. [PubMed]
98. Maziak W, Ward KD, Stockton MB. Childhood obesity: are we missing the big picture? Obes Rev. 2008;9:35–42. [PubMed]
99. Unnatural causes. Is inequality making us sick. [(Accessed March 12, 2008)]; www.unnaturalcauses.org.
100. Wight RG, Aneshensel CS, Miller-Martinez D, et al. Urban neighborhood context, educational attainment, and cognitive function among older adults. Am J Epidemiol. 2006;163:1071–78. [PubMed]
101. Khawaja M, Habib RR. Husbands’ involvement in housework and women's psychosocial health: findings from a population-based study in Lebanon. Am J Public Health. 2007;97:860–66. [PubMed]
102. Navas-Acien A, Bleys J, Guallar E. Selenium intake and cardiovascular risk: what is new? Curr Opin Lipidol. 2008;19:43–49. [PubMed]
103. Lawlor DA, Davey Smith G, Kundu D, Bruckdorfer KR, Ebrahim S. Those confounded vitamins: what can we learn from the differences between observational versus randomised trial evidence? Lancet. 2004;363:1724–27. [PubMed]
104. Vandenbroucke JP. When are observational studies as credible as randomised trials? Lancet. 2004;363:1728–31. [PubMed]
105. Smeeth L, Douglas I, Hubbard R. Commentary: we still need observational studies of drugs—they just need to be better. Int J Epidemiol. 2006;35:1310–11. [PubMed]
106. Avorn J. In defense of pharmacoepidemiology—embracing the yin and yang of drug research. N Engl J Med. 2007;357:2219–21. [PubMed]
107. Vandenbroucke JP. Observational research, randomised trials, and two views of medical science. PLoS Med. 2008;5:e67. [PMC free article] [PubMed]
108. Weed DL. Commentary: rethinking epidemiology. Int J Epidemiol. 2006;35:583–86. [PubMed]
109. Coggon D. Commentary: Complex disease—responding to the challenge. Int J Epidemiol. 2006;35:581–83. [PubMed]
110. Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations. Int J Epidemiol. 2004;33:30–42. [PubMed]
111. Krieger N. Postmenopausal hormone therapy. N Engl J Med. 2003;348:2363–64. [PubMed]
112. Schwartz S, Susser E. Commentary: What can epidemiology accomplish? Int J Epidemiol. 2006;35:587–90. [PubMed]
113. Millikan RC. Commentary: The human Genome: philosopher's stone or magic wand? Int J Epidemiol. 2006;35:578–81. [PubMed]
114. Merikangas KR, Low NCP, Hardy J. Commentary: understanding sources of complexity in chronic diseases: the importance of integration of genetics and epidemiology. Int J Epidemiol. 2006;35:590–92. [PubMed]
115. von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. STROBE Initiative. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: guidelines for reporting observational studies. Epidemiology. 2007;18:800–4. [PubMed]
116. Ioannidis JP. Why most published research findings are false: author's reply to Goodman and Greenland. PLoS Med. 2007;4:e215. [PMC free article] [PubMed]
117. Galea S, Ahern J. Invited Commentary: considerations about specificity of associations, causal pathways, and heterogeneity in multilevel thinking. Am J Epidemiol. 2006;163:1079–82. [PubMed]
118. Smith GD, Lawlor DA, Harbord R, Timpson N, Day I, Ebrahim S. Clustered environments and randomized genes: a fundamental distinction between conventional and genetic epidemiology. PLoS Med. 2007;4:e352. [PMC free article] [PubMed]
119. Lawlor DA, Timpson NJ, Harbord RM, et al. Exploring the developmental overnutrition hypothesis using parental-offspring associations and FTO as an instrumental variable. PLoS Med. 2008;5:e33. [PMC free article] [PubMed]
120. Poole C, Rothman KJ. Our conscientious objection to the epidemiology wars. J Epidemiol Community Health. 1998;52:613–14. [PMC free article] [PubMed]
121. Egger G, Swinburn B. An ‘ecological’ approach to the obesity pandemic. BMJ. 1997;315:477–80. [PMC free article] [PubMed]
122. Pearce N, Merletti F. Complexity, simplicity, and epidemiology. Int J Epidemiol. 2006;35:515–19. [PubMed]
123. Diez Roux A. Multilevel analysis in public health research. Ann Rev Public Health. 2000;21:171–92. [PubMed]
124. Royston P. A useful monotonic non-linear model with applications in medicine and epidemiology. Stat Med. 2000;19:2053–66. [PubMed]
125. Krull JL, MacKinnon DP. Multilevel modeling of individual and group level mediated effects. Multivariate Behav Res. 2001;36:249–77.
126. Lin G, Spann S, Hyman D, Pavlik V. Climate amenity and BMI. Obesity (Silver Spring) 2007;15:2120–27. [PubMed]
127. Singh GK, Kogan MD, van Dyck PC. A multilevel analysis of state and regional disparities in childhood and adolescent obesity in the United States. J Community Health. 2008;33:90–102. [PubMed]
128. Joshu CE, Boehmer TK, Brownson RC, Ewing R. Personal, neighbourhood and urban factors associated with obesity in the United States. J Epidemiol Community Health. 2008;62:202–08. [PubMed]
129. Diez Roux AV, Merkin SS, Arnett D, et al. Neighborhood of residence and incidence of coronary heart disease. N Engl J Med. 2001;345:99–106. [PubMed]
130. Eaton SB, Cordain L, Lindeberg S. Evolutionary health promotion: a consideration of common counterarguments. Prev Med. 2002;34:119–23. [PubMed]
131. Maziak W. The hygiene hypothesis and the evolutionary perspective of health. Prev Med. 2002;35:415–18. [PubMed]
132. Oh K, Hu FB, Manson JE, Stampfer MJ, Willett WC. Dietary fat intake and risk of coronary heart disease in women: 20 years of follow-up of the Nurses’ Health Study. Am J Epidemiol. 2005;161:672–79. [PubMed]
133. Liu S, Stampfer MJ, Hu FB, et al. Whole grain consumption and risk of coronary heart disease: results from the Nurses’ Health Study. Am J Clin Nutr. 1999;70:412–19. [PubMed]
134. Liu S, Manson JE, Lee I-M, et al. Fruit and vegetable intake and risk of cardiovascular disease: the Women's Health Study. Am J Clin Nutr. 2000;72:922–28. [PubMed]
135. Fung TT, Stampfer MJ, Manson JE, Rexrode KM, Willett WC, Hu FB. Prospective study of major dietary patterns and stroke risk in women. Stroke. 2004;35:2014–19. [PubMed]
136. Kris-Etherton PM, Harris WS, Appel LJ. American Heart Association Nutrition Committee. Fish consumption, fish oil, omega-3 fatty acids, and cardiovascular disease. Circulation. 2002;21:2747–57. [PubMed]
137. Howard BV, Van Horn L, Hsia J, et al. Low-fat dietary pattern and risk of cardiovascular disease: the women's health initiative randomized controlled dietary modification trial. JAMA. 2006;295:655–66. [PubMed]
138. O’Keefe JH Jr, Cordain L. Cardiovascular disease resulting from a diet and lifestyle at odds with our Paleolithic genome: how to become a 21st-century hunter-gatherer. Mayo Clin Proc. 2004;79:101–8. [PubMed]
139. Milton K. The critical role played by animal source foods in human (Homo) evolution. J Nutr. 2003;133(11 Suppl 2):S3886–92. [PubMed]
140. Cordain L, Eaton SB, Sebastian A, et al. Origins and evolution of the Western diet: health implications for the 21st century. Am J Clin Nutr. 2005;81:341–54. [PubMed]
141. Weiss KM, Buchanan AV. Evolution by phenotype: a biomedical perspective. Perspect Biol Med. 2003;46:159–82. [PubMed]
142. Nettle D. The evolution of personality variation in humans and other animals. Am Psychol. 2006;61:622–31. [PubMed]
143. Houle D. How should we explain variation in the genetic variance of traits? Genetica. 1998;102–103:241–53. [PubMed]
144. Eaton SB. The ancestral human diet: what was it and should it be a paradigm for contemporary nutrition? Proc Nutr Soc. 2006;65:1–6. [PubMed]
145. Eaton SB, Eaton SB. An evolutionary perspective on human physical activity: implications for health. Comp Biochem Physiol A Mol Integr Physiol. 2003;136:153–59. [PubMed]
146. Ackroff K, Sclafani A. Energy density and macronutrient composition determine flavor preference conditioned by intragastric infusions of mixed diets. Physiol Behav. 2006;89:250–60. [PubMed]

Articles from International Journal of Epidemiology are provided here courtesy of Oxford University Press