International Epidemiological Association Research Articles

A dictionary of epidemiology

When the first edition of A Dictionary of Epidemiology was launched in 1983,1 its principal aim was to facilitate communication among health professionals. The field was relatively young and comprised...

Cabbages and condoms

Case-cross over studies are used to assess brief exposures (e.g. high-ambient temperature) that may cause a temporary increase in risk of diseases (e.g. death), and have the advantage of controlling for confounding by factors that differ between individuals such as age and co-morbidities. A novel application is reported in this issue in an evaluation of factors determining condom use in female bar and hotel workers in Tanzania. Unsurprisingly condoms were more often used with casual rather than regular sex partners, for hotel sex and when gifts were received— all indicative of a high-risk situation. Condoms were less often used with men 410 years older than the woman; partner alcohol use and knowledge of HIV status were not associated with condom use. The authors focus on the decision-making process—those women reporting that they decided or that it was a mutual decision were 10 times more likely to use condoms—and comment that it is much easier for a woman to negotiate with a stranger than with her regular partner. Interventions attempting to increase use of condoms in long-term relationships, supposedly based on mutual trust, are unlikely to provide a satisfactory solution. So, the authors consider interventions to reduce the number of concurrent partners and promote use of female condoms may be more useful. Chlamydia infection is one of the commonest sexually transmitted diseases contributing substantially to the 340 million new cases of bacterial or protozoan sexually transmitted infections. World Health Organization, in common with the British health service and the US Preventive Services Task Force, believe that Chlamydia screening is a good thing. So, why do Nicola Low and colleagues conclude ‘there is an absence of evidence supporting opportunistic Chlamydia screening in the general population younger than 25 years?’ The previous reviews of the evidence suggested fair to good evidence to support screening. The main issue appears to be the organizational system in which screening is conducted—if you use a population register to invite women for screening rather than chance encounters with women attending health services for various reasons (‘opportunistic’ screening), some low quality evidence of benefit can be found. These authors were not randomized controlled trial (RCT) obsessed in their definition of evidence—they note that Chlamydia rates were falling in Sweden before the opportunistic screening service became operational. Low and colleagues suggest that clinical equipoise on the balance of benefits and harms of screening remains—but the alacrity with which World Health Organization and other influential groups and health services have adopted Chlamydia screening suggests that there is no equipoise. Indeed, the language of our commentators from CDC Atlanta on the review is interesting: ‘. . . it is risky to draw conclusions [given the small number of studies reviewed] about which approach to screening works best’. Presumably, it is not possible to countenance a view that screening maybe just does not E-mail:shah.ebrahim@lshtm.ac.uk Figure 1 A novel use for condoms Published by Oxford University Press on behalf of the International Epidemiological Association

Cohort Profile: The North West Adelaide Health Study (NWAHS)

Published by Oxford University Press on behalf of the International Epidemiological Association © The Author 2008; all rights reserved.

Preventing HIV in resource-limited settings: evidence for action, from cross-sectional surveys

The current issue of the International Journal of Epidemiology features seven papers that cover a range of issues that are central to our understanding of the HIV epidemic and its control in resource-limited settings. One cross-sectional study highlights the importance of education as a protective factor against HIV in young South African women, another illustrates the complex relationship between socioeconomic position and HIV infection in Tanzania, and a further prevalence study examines the importance of lack of circumcision as a risk factor in the South Indian state of Andhra Pradesh. Male circumcision is also the topic of a modelling study and an accompanying commentary, which examine the potential long-term impact of male circumcision on HIV prevalence in sub-Saharan Africa. Another article in this issue systematically reviews the risk of HIV transmission from orogenital sex, and the final piece estimates the HIV prevalence in Dehong Prefectrue of Yunnan Province China, which is close to the ‘Golden Triangle’ and has a serious HIV epidemic, fuelled by injection drug use. Almost three decades (and a Nobel prize) after the discovery of the virus, HIV/AIDS continues to be a massive burden in many resource-limited settings. Young people aged 15–24 years account for almost half of all new HIV infections in adults, and subSaharan Africa remains the region most heavily affected by HIV. What can be done to stem the epidemic? Keep them in school, say Pettifor and colleagues, based on a nationally representative sexual behaviour and HIV testing survey in South Africa, which showed that not completing high school was the most important risk factor for HIV infection in young women. The authors focused on women aged 15–24 years who reported to have only one life-time partner, and therefore seemingly were at low risk of infection. The prevalence of HIV-1 infection was nevertheless high: 15.0% compared with 3.8% in men of the same age reporting one life-time partner. Over threequarters of women had not completed high school, and their HIV prevalence was 16.9% compared with 8.6% among the minority of women completing their education (crude odds ratio 2.15). The survey was cross-sectional and it is therefore unclear when women contracted HIV, before or after dropping out of school. Dropping out of school was clearly related to sociodemographic and behavioural factors, but unfortunately we are not told which of the variables included in the multivariable model were responsible for the substantial increase in the odds of infection associated with dropping out (adjusted odds ratio 3.75). The study of Pettifor and colleagues supports the notion that social policies which keep young women in school will have many benefits, including the prevention of HIV infection. Interestingly, a randomized evaluation comparing three school-based HIV/AIDS interventions in Kenya recently showed that reducing the cost of education by paying for school uniforms, but not training teachers in HIV/AIDS education, reduced dropout rates, teen marriages and teen pregnancies. The importance of education is questioned by the data from the 2003/2004 Tanzania HIV/Aids Indicator Survey (THIS). Mshisha and colleagues focussed on sexually active adults and found that while education was not associated with HIV infection, there was a positive association between standard of living and HIV infection: in women, the prevalence of HIV infection increased from 2.7% in the lowest category of the household standard of living index to 14.1% in * Corresponding author. E-mail: egger@ispm.unibe.ch 1 Institute of Social and Preventive Medicine, Finkenhubelweg 11, University of Bern, CH-3012 Bern, Switzerland. 2 Swiss Tropical Institute, Socinstrasse 57, University of Basel, CH-4002 Basel, Switzerland. Published by Oxford University Press on behalf of the International Epidemiological Association

A new disease entity in man: a report on etiologic and clinical observations

First, a quick review of the background: while working in a malaria control campaign in the northern section of the state of Minas Gerais, we came across a large bug locally called ‘‘bar-beiro,’’ which, like the bedbug, infests homes and attacks man at night, when the lights are out. Since this was a blood sucking insect and such insects are of importance in human and animal pathology as vectors of disease, we examined ‘‘barbeiro’’ specimens and found in their hindguts a flagellate of crithidial morphology. The next step was to ascertain whether this flagellate was just a parasite of the insect or a stage in the life cycle of a parasite of vertebrates. In almost all marmoset (Callithrix penicillata) specimens from the same region, we had found a parasitic trypanosome. We suspected that the ‘‘barbeiro’’ was the intermediate host, and the flagellate forms in its gut were stages in the life cycle of this parasite, Trypanosoma minasense Chagas. Because of this, we sent specimens of the bug to our director and teacher Dr. Oswaldo Cruz, who had them feed on a marmoset. Dr. Cruz succeeded in infecting this marmoset with a hemoflagellate which, being quite different in morphology from Trypanosoma minasense, we named Trypanosoma cruzi. I studied the new parasite and found, through repeated experiments, that the insect was indeed the intermediate host, and that it took at least eight days after a bite for transmission to take place. I then made another trip to Minas Gerais for the purpose of establishing the definitive habitual host of the flagellate. The habits of the ‘‘barbeiro’’ suggested a situation of intradomiciliary infection; we were most impressed by the poor state of health of those who lived in infested houses. After conducting physical examinations, we found that the affected persons, especially the children, showed symptoms of a chronic disease unlike those already known and described. These symptoms included some that occur in trypanosomiasis in man and household animals, such as generalized lymph node enlargement, edema, swollen face, etc. We then recalled many patients in that area who had sought me out earlier, sometimes with and sometimes without fever, and these patients always presented the same intense morbid condition and symptoms for which I was then unable to diagnose the cause. Their fever did not yield to quinine, and no malaria parasites could be found in their blood. Our first breakthrough was finding a cat with parasitic hemoflagellates in a house where ‘‘barbeiros’’ were abundant. The first examinations of fresh blood from chronic sufferers were without result. When we were later called to treat a child in grave condition, febrile, with very pronounced swelling in the face, many enlarged nodes in different regions, and thyroid hypotro-phy, we found numerous flagellates in freshly drawn blood. By staining them I was able to identify them as the same parasite transmitted by the ‘‘barbeiro’’ to laboratory animals. Examination of guinea pigs inoculated with blood from this first patient revealed a development of the same process previously observed, and in the lungs we found a schizogony of eight units identical to those observed in our studies of the life cycle of Schizotrypanum cruzi. We obtained more positive results by inoculating guinea pigs with blood from chronic patients. This procedure established the new human-disease entity produced by Schizotrypanum cruzi. We communicated the results of our work to Dr. Oswaldo Cruz, the director of the Manguinhos Institute, who sent a preliminary report to the National Academy of Medicine. Due to other commitments, we could not begin the detailed clinical study of the disease until ten months after these findings. The transmitting insect is a heteropterous hemipteran, species megistus, genus Conorhinus, family Reduvidae (sic). Dr. Arthur Neiva has recently published a detailed study of its biology in the Memorias do Institute Oswaldo Cruz. Conorhinids invade homes, where they actively multiply, becoming exceedingly numerous and troublesome to the occupants. The many cracks in the mud walls of primitive, grassthatched huts are the preferred habitat of the bug, where it greatly multiplies. Even better-built houses, if they offer suitable hiding places, are infested with the bug. The bug can find shelter in the cellars of houses, and enters occupied rooms at night through cracks in the floor. On small farms in the region, we found Conorhinids in different outbuildings such as carriage-houses, storerooms, stables, etc. The bug Source: Carlos Chagas: Coletânea de Trabalhos Cientificos. Brasilia, Editora Universidade de Brasilia, 1981. English version in The Challenge of Epidemiology: issues and Selected Readings. Pan American Health Organization, 1988. Published by Oxford University Press on behalf of the International Epidemiological Association

Commentary: Health policies in the US: can they increase or decrease the gap between subgroups of the population? The case of folic acid

Effective on January 1, 1998, the Food and Drug Administration (FDA) amended the standards to identify several enriched grain products, bromated flour, vegetable macaroni and vegetable noodle products by requiring folic acid fortication. Specifically, the FDA required that these products be fortified with folic acid levels ranging from 0.43 to 1.4 mg/pound or 95 to 309 mg/100 g of product. Although this requirement affects the entire population, the purpose of this amendment was to ensure that women of childbearing age consume the US Public Health Service recommended daily allowance of at least 0.4 mg (400 mg) of folic acid daily to reduce their risk of having a child with spina bifida or other neural tube defects (NTDs). Spina bifida and anencephaly affect approximately 4000 children in the US prior to the folic acid fortification amendment. However, approximately 1000 of these cases have been prevented with folic acid fortification. As with most conditions in the US, estimates for NTDs vary by race and ethnicity, with Hispanics exhibiting the highest estimates and blacks and Asians the lowest. Moreover, NTDs varies by socioeconomic status defined using income and education with the least educated and those with low income exhibiting the highest estimates. Low folate levels not only affect children in utero, but also evidence suggests that low folate levels are associated with increased prevalence of cancer, cardiovascular disease, physical and mental functioning. Thus, despite the fact that the fortification was targeted to women of childbearing age, the FDA amendment could be a preventive paradox for population health by benefiting everyone. The latter could have implications for health disparities across subgroups of the population if differences exist across the outcomes benefiting from the fortification. Findings from existing studies on folic acid preand post-fortification suggest that there are disparities between groups. For instance, Ford and Bowman showed differences in red blood cell (RBC) folate concentrations between non-Hispanic blacks and nonHispanic whites with the highest differences among those with greater than a high school diploma regardless of gender. Conversely, differences between Mexican Americans and non-Hispanic whites were significant for men with at least 9 years of education and for women at the educational attainment extreme (<9 years of education and more than a high school diploma). Ganji and Kafai showed a decrease in prevalence of low RBC folate between 1988–94 and 1999–2002 across all racial/ ethnic and poverty-income ratio groups. However, nonHispanic blacks continue to exhibit a higher prevalence of low RBC folate through this period. Unlike previous studies examining trends between the preand post-fortification period, Dowd and Aiello focused on investigating the magnitude of the racial/ethnic and socioeconomic status (defined using poverty to income ratio) disparities assessed as absolute and relative differences. Using data from the National Health and Nutrition Examination Surveys (NHANES), Dowd and Aiello examined racial/ethnic and income disparities in folate status in US adults aged 25 years or older before (1991–1994) and after (1999–2002) enactment of the folic acid fortification. They found an increase in relative, but a decrease, in absolute difference before and after the fortification. Non-Hispanic blacks had 1.6 (95% CI: 1.4–1.9) greater odds of having low RBC folate (<362.6 nmol) levels compared with non-Hispanic whites in the pre-fortification period. This estimate was 3.7 (95% CI: 2.8–5.0) in the postfortification period. In the pre-fortification period, individuals in the bottom income quartile had a 1.3 (95% CI: 1.1–1.4) greater odds of having low RBC compared with those in the highest income quartile. Finally, in the post-fortification period, this estimate was 2.1 (95% CI: 1.6–2.7). The corresponding absolute differences were 233 (pre-) and 121 (post-fortification) Department of Health Sciences, Graduate Program in Public Health, Lehman College, CUNY, 250 Bedford Park Boulevard West, Gillet 336, Bronx, NY 10468, USA. E-mail: Luisa.Borrell@lehman.cuny.edu Published by Oxford University Press on behalf of the International Epidemiological Association

Bulimia among female students in Marrakesh

During August 17th–21st, 2014, the University of Alaska Anchorage, along with other local, state, and federal agencies throughout Alaska, will host the 20th International Epidemiological Association's (IEA) World Congress of Epidemiology (WCE 2014). The theme for this Congress is “Global Epidemiology in a Changing Environment: The Circumpolar Perspective.” The changing environment includes the full range of environments that shape population health and health inequities from the physical to the social and economic. Our circumpolar perspective on these environments includes views on how political systems, work, immigration, Indigenous status, and gender relations and sexuality affect the global world and the health of its people. Suggestions and insights from the 3rd North American Congress of Epidemiology (2011) and the first-ever joint regional workshop co-organized by the IEA North American Region and the IEA Latin American and Caribbean Region held at the 19th IEA World Congress of Epidemiology (2011) have helped direct the focus for WCE 2014. Since the Arctic regions are feeling the effects of climate change first, we believe focusing on the emerging data on the health impacts of climate change throughout the world will be an important topic for this Congress. This will include a broad range of more traditional epidemiology areas such as infectious disease epidemiology, environmental epidemiology, health disparities, and surveillance and emergency preparedness. Addressing health inequities and promoting health equity is likewise a key concern of the Congress. This Congress will also host presentations on injury epidemiology, occupational health, infectious diseases, chronic diseases, maternal and child health, surveillance and field epidemiology, mental health, violence (from self-directed, e.g., suicide, to interpersonal to structural), psychoactive substance use (including tobacco), and measures of subjective health. Attention will be given to epidemiology's theoretical frameworks and emphasizing knowledge translation, from epidemiology to health systems, to policy, and to the broader public. We also plan to offer many hands-on workshops including practical uses of epidemiology to improve health systems and reduce health inequities within and between countries; the manner in which epidemiology can inform public health practice; the understanding and use of the Dictionary of Epidemiology; and many others.

Commentary: Smoking and human papillomavirus infection: the pursuit of credibility for an epidemiologic association

As the most important among the modifiable risk factors for many cancer types tobacco smoking is dutifully treated as a potent confounder when epidemiologists explore new candidate relations. The role for smoking changes from ‘confounder’ to ‘confounded’ variable in anogenital cancers. Among the latter, cervical cancer is a case in point. Smoking was first suspected as a risk factor for cervical cancer in the mid 60s but although most studies have replicated the finding the association was always seen with suspicion because of confounding by sexual activity. Number of sexual partners and age at first intercourse are unequivocal key determinants of risk, which has long suggested that an infectious sexually-transmitted agent plays an aetiological role. Controlling for these two measures of sexual behaviour tended to dampen the associations with smoking in cervical cancer studies but in many of them it remained statistically detectable. Concerns about residual confounding by unmeasured sexual behaviours or insufficient control of confounding due to misclassification of sexual activity information led the International Agency for Research on Cancer (IARC) to defer judgment about a carcinogenic role for smoking in cervical cancer in its first monograph focusing on tobacco products in 1986. This intractable situation was nicely captured in 1994 in an insightful article in the Journal which demonstrated that, as long as the putative sexuallytransmitted aetiologic agent remained unmeasured and uncontrolled for in an epidemiological study, the association with smoking would continue to be hopelessly confounded, despite best efforts at adjusting for sexual activity variables. It did not help when the first molecular epidemiologic studies in the late 80s heeded this advice by placing the human papillomavirus (HPV) at centre stage for the first time. Paradoxically, HPV infection measured in these earlier studies was not associated with sexual activity and was only feebly associated with cervical cancer. Epidemiologists learned that misclassification of HPV status could account for the incoherent results and were quick to adopt improved molecular techniques for detecting HPV, which led to the recognition of this virus as a necessary cause of cervical cancer and ushered in a new era in cervical cancer prevention via HPV vaccination and screening with HPV tests. Once measurement error issues related to cervical HPV detection were largely resolved the smoking-cervical cancer association could be verified with reasonable confidence. The IARC revisited its previous conclusions and listed cervical cancer among those causally related to smoking in 2004. A pooled analysis of 23 modern molecular epidemiologic studies that included over 13 000 cases and 23 000 controls and carefully controlled for HPV status found an increased risk of squamous cell carcinomas with a dose–response relation with number of cigarettes. Although the controversy subsided, the role of epidemiology was not yet over; a mechanism for the carcinogenic action of tobacco smoking needed to be documented. The aetiological pathway that begins with sexual activity as distal variable, then includes HPV infection as intermediate endpoint, and ends with cervical cancer is consistent with both ‘upstream’ and ‘downstream’ smoking effects relative to HPV infection. Documenting both still requires careful attention to confounding by sexual activity, which is strongly correlated with smoking habits. In this issue of the Journal, Vaccarella et al. advanced our understanding of the upstream relation by focusing on smoking as a determinant of HPV infection in the large IARC multi-centre study of HPV prevalence, which included over 10 000 women throughout the world. Their state-of-the-art methods for detecting HPV DNA in cervical specimens were conducted in world-class laboratories. The authors were acutely aware of the confounding problem and made clever use of the standardized questionnaire information to control for age, number of sexual * Corresponding author. Department of Epidemiology & Biostalistics and Department of Oncology, McGill University, Montreal, Canada. E-mail: eduardo.franco@mcgill.ca Department of Epidemiology & Biostatistics and Department of Oncology, McGill University, Montreal, Canada. Published by Oxford University Press on behalf of the International Epidemiological Association

Cohort profile: the Centers for AIDS Research Network of Integrated Clinical Systems

Highly active antiretroviral therapy (HAART) delays disease progression and death. However, the treatments incompletely control HIV replication, only partially restore immune function, have significant shortand long-term toxicities, and eventually fail in many patients with consequent development of HIV drug resistance. Thus, there is increasing need for information to guide HIV-infected patients and their providers in making decisions regarding optimal use of antiretroviral therapies. Although clinical trials provide valuable information about efficacy and side effects of antiretroviral treatment, they have limited size, duration and power to detect effects on clinical outcomes, focusing instead on surrogate endpoints such as virologic failure, treatment discontinuation or composite outcome measures. Outside the clinical trial setting, there is tremendous heterogeneity among HIV-infected patients. The prevalence and impact of important health conditions such as hepatitis C virus (HCV) co-infection, mental illness and substance abuse likely contribute to increased toxicity and decreased clinical effectiveness of HAART regimens among the broader spectrum of patients treated in routine care. Cohorts with significant diversity in HIV disease severity, comorbidities and demographic distributions are required to provide information regarding long-term outcomes and complications of HIV infection in the modern HAART era. The Centers for AIDS Research (CFAR) Network of Integrated Clinical Systems (CNICS) was created to better define the relationship between patient and treatment factors and long-term clinical outcomes among HIV-infected patients in the HAART era. The CFARs are a national network of centres of excellence for HIV care and research established by the National Institutes of Health (NIH) whose mission is to support a multi-disciplinary environment for basic, clinical, epidemiologic, behavioural and translational research in the prevention, detection and treatment of HIV infection and AIDS. There are 19 CFARs located at academic and research institutions throughout the United States. The objective of the CNICS project is to integrate clinical data from the large and diverse population of HIV-infected persons receiving care at CFAR sites to investigate questions related to HIV disease management that cannot be readily addressed through traditional randomized controlled clinical trials and other cohort studies. Investigators with expertise in basic, clinical, translational and epidemiologic research, in addition to medical informatics, are collaborating on the CNICS project. The potential to build a comprehensive clinical data repository for HIV disease was greatly advanced by the work of CFAR investigators at the participating CNICS sites who had instituted point-of-care electronic medical record systems (EMRs) with the dual purpose of providing real-time clinical information to facilitate the delivery of HIV care and capturing standardized clinical data to support populationbased HIV research. The initial four CNICS sites were Case Western Reserve University, University of * Corresponding author. Center for AIDS Research, University of Washington, 325 9th Ave, MS 359931, Seattle, WA 98104, USA. E-mail: kitahata@u.washington.edu 1 Department of Medicine, University of Washington, Seattle, 98195, USA. 2 Department of Medicine, Case Western Reserve University, Cleveland, 44106, USA. 3 Department of Medicine, University of California, San Diego, 92110, USA. 4 Department of Medicine, Harvard University, Boston, 02115, USA. 5 Department of Medicine, Johns Hopkins University, Baltimore, 21218, USA. 6 Department of Medicine, University of Alabama, Birmingham, 35209, USA. 7 Department of Medicine, University of California, San Francisco, 94143, USA. Published by Oxford University Press on behalf of the International Epidemiological Association

Corporate influences on epidemiology

Corporate influences on epidemiology have become stronger and more pervasive in the last few decades, particularly in the contentious fields of pharmacoepidemiology and occupational epidemiology. For every independent epidemiologist studying the side effects of medicines and the hazardous effects of industrial chemicals, there are several other epidemiologists hired by industry to attack the research and to debunk it as 'junk science'. In some instances these activities have gone as far as efforts to block publication. In many instances, academics have accepted industry funding which has not been acknowledged, and only the academic affiliations of the company-funded consultants have been listed. These activities are major threats to the integrity of the field, and its survival as a scientific discipline. There is no simple solution to these problems. However, for the last two decades there has been substantial discussion on ethics in epidemiology, partly in response to the unethical conduct of many industry-funded consultants. Professional organizations, such as the International Epidemiological Association, can play a major role in encouraging and supporting epidemiologists to assert positive principles of how science should work, and how it should be applied to public policy decisions, rather than simply having a list of what not to do.

Commentary: On 'Quality in epidemiological research: should we be submitting papers before we have the results and submitting more hypothesis generating research?'

Lawlor’s excellent proposals to mitigate publication bias complements other ideas, such as (truly) blind review of completed studies by blanking-out study results. Nonetheless, despite my general agreement with Lawlor, I think certain conventional presumptions in the editorial are in error and have blinded discussants to important sources of publication bias. Most authors write as if publication bias is chiefly about tendency to not publish null results. Studies linking products with hazards provide examples to the contrary, as in the Vioxx debacle (one that happened to come to prominence). Typically, the sponsor or conductor is the affected industry, with the power to either suspend study activity or else to press for further data collection or analysis before publication. A strategy that extends a study or analysis only when a particular type of result (e.g. a positive finding) emerges at first is biased against reporting that type of result. Of course, some researchers follow similarly biased strategies when their initial findings contravene strongly held beliefs, whether those beliefs are in the null or an alternative. In any event, it is time to recognize that publication bias has many forms, some away from the null, others toward the null. For similar reasons, it is time to stop using the term ‘hypothesis generating’ for studies that screen multiple associations, because the term conceals the true nature and potential bias of such studies. Even in the most speculative fishing expedition, the fact that an association was examined means a hypothesis about it was generated and approved for examination at the outset. These studies do not create hypotheses any more than screening programs create people. What they do is screen a selected set of hypotheses, usually according to dubious statistical criteria (such as multiple-testing procedures that aggravate publication bias by reducing alpha levels at the expense of power). To understand why my complaint concerns more than terminology, consider that the manner in which the hypotheses are selected for examination opens an avenue for publication bias, in the broad sense of bias in what we see in the published literature. As with patient selection for medical screening, there can be considerable (and faulty) judgement in what hypotheses get selected for examination and how they are examined, beginning with data-collection decisions and continuing through the analysis. Certain questions may be avoided for fear of the answer, but most often are ignored due to certainty about the answer. We all do the latter; for example, we fail to ask a person’s favourite movies in the firm belief that such information would provide no useful health-risk information beyond demographics, diet and exercise. Essential as pre-data hypothesis selection may be, these decisions are based on political, academic and personal judgements about pursuit of certain questions, and thus subject to the biases prevalent in their context. The fact that hypotheses are generated and screened before a study is done needs to be acknowledged, instead of concealed by confused terminology. Consider the Los Angeles Heart Study. Begun in the 1940s, it neglected to obtain baseline smoking information. In fairness to the authors, this was at a time when, in the Anglo-American world, a smoking effect on lung cancer was still regarded with scepticism; an effect on heart disease would probably have been lumped by some in the same category as flying saucers. Nonetheless, the narrow range of data collected helped ensure that this study would play only a minor role in the landmark discoveries of cardiovascular risk factors during subsequent decades, and reduced the pool of early evidence regarding smoking and heart disease. As for biased publication of collected data, submission and review innovations cannot substitute for mandatory study registration at the outset of data collection, for otherwise the very existence of data could be missed. Ethical screening bodies such as Institutional Review Boards and Research Ethics Committees may hold the key to preventive measures. Via ethical guidelines, such boards exert extensive control over study approval. Those guidelines can be used to argue that studies collecting data on human subjects should be required to register their approved proposal or protocol in a searchable public database. Mandatory registration is not enough to prevent publication bias, however; study results must be available to the public. Rationales offered for not publishing results have included protection of proprietary information or patient confidentiality (the latter has been invoked when the study has few subjects or few cases). If the data holder is unresponsive to inquiries, other Departments of Epidemiology and Statistics, University of California, Los Angeles, USA. E-mail: lesdomes@ucla.edu Published by Oxford University Press on behalf of the International Epidemiological Association The Author 2007; all rights reserved. International Journal of Epidemiology 2007;36:944–945 doi:10.1093/ije/dym174

Quality in epidemiological research: should we be submitting papers before we have the results and submitting more hypothesis-generating research?

Lawlor’s excellent proposals to mitigate publication bias complements other ideas, such as (truly) blind review of completed studies by blanking-out study results. Nonetheless, despite my general agreement with Lawlor, I think certain conventional presumptions in the editorial are in error and have blinded discussants to important sources of publication bias. Most authors write as if publication bias is chiefly about tendency to not publish null results. Studies linking products with hazards provide examples to the contrary, as in the Vioxx debacle (one that happened to come to prominence). Typically, the sponsor or conductor is the affected industry, with the power to either suspend study activity or else to press for further data collection or analysis before publication. A strategy that extends a study or analysis only when a particular type of result (e.g. a positive finding) emerges at first is biased against reporting that type of result. Of course, some researchers follow similarly biased strategies when their initial findings contravene strongly held beliefs, whether those beliefs are in the null or an alternative. In any event, it is time to recognize that publication bias has many forms, some away from the null, others toward the null. For similar reasons, it is time to stop using the term ‘hypothesis generating’ for studies that screen multiple associations, because the term conceals the true nature and potential bias of such studies. Even in the most speculative fishing expedition, the fact that an association was examined means a hypothesis about it was generated and approved for examination at the outset. These studies do not create hypotheses any more than screening programs create people. What they do is screen a selected set of hypotheses, usually according to dubious statistical criteria (such as multiple-testing procedures that aggravate publication bias by reducing alpha levels at the expense of power). To understand why my complaint concerns more than terminology, consider that the manner in which the hypotheses are selected for examination opens an avenue for publication bias, in the broad sense of bias in what we see in the published literature. As with patient selection for medical screening, there can be considerable (and faulty) judgement in what hypotheses get selected for examination and how they are examined, beginning with data-collection decisions and continuing through the analysis. Certain questions may be avoided for fear of the answer, but most often are ignored due to certainty about the answer. We all do the latter; for example, we fail to ask a person’s favourite movies in the firm belief that such information would provide no useful health-risk information beyond demographics, diet and exercise. Essential as pre-data hypothesis selection may be, these decisions are based on political, academic and personal judgements about pursuit of certain questions, and thus subject to the biases prevalent in their context. The fact that hypotheses are generated and screened before a study is done needs to be acknowledged, instead of concealed by confused terminology. Consider the Los Angeles Heart Study. Begun in the 1940s, it neglected to obtain baseline smoking information. In fairness to the authors, this was at a time when, in the Anglo-American world, a smoking effect on lung cancer was still regarded with scepticism; an effect on heart disease would probably have been lumped by some in the same category as flying saucers. Nonetheless, the narrow range of data collected helped ensure that this study would play only a minor role in the landmark discoveries of cardiovascular risk factors during subsequent decades, and reduced the pool of early evidence regarding smoking and heart disease. As for biased publication of collected data, submission and review innovations cannot substitute for mandatory study registration at the outset of data collection, for otherwise the very existence of data could be missed. Ethical screening bodies such as Institutional Review Boards and Research Ethics Committees may hold the key to preventive measures. Via ethical guidelines, such boards exert extensive control over study approval. Those guidelines can be used to argue that studies collecting data on human subjects should be required to register their approved proposal or protocol in a searchable public database. Mandatory registration is not enough to prevent publication bias, however; study results must be available to the public. Rationales offered for not publishing results have included protection of proprietary information or patient confidentiality (the latter has been invoked when the study has few subjects or few cases). If the data holder is unresponsive to inquiries, other Departments of Epidemiology and Statistics, University of California, Los Angeles, USA. E-mail: lesdomes@ucla.edu Published by Oxford University Press on behalf of the International Epidemiological Association The Author 2007; all rights reserved. International Journal of Epidemiology 2007;36:944–945 doi:10.1093/ije/dym174

Commentary: Epidemiology still ascendant

Commentary: Epidemiology still ascendant

Commentary: Using innovative inequality measures in epidemiology

Commentary: Using innovative inequality measures in epidemiology

Commentary: Observational studies and the art of accurately measuring influenza vaccine benefits

As Orenstein and colleagues point out in an elegant paper in this issue, observational studies that compare the incidence of influenza-like illness in vaccinated and unvaccinated groups are theoretically prone to underestimating the vaccine’s true effectiveness (VE). This is because many other respiratory pathogens cause similar symptoms; these other infections form a sizeable background of influenza-like illness cases in both case and control groups that are not preventable by influenza vaccination. Orenstein et al. reasonably call for laboratory confirmed endpoints, which have rarely been obtained in observational studies of influenza vaccine effectiveness in populations targeted for vaccination. These theoretical effects of low endpoint specificity are perhaps not so novel, but this paper and its careful quantification of the issue is timely because the problems of interpreting results from studies that use low-specificity end-points seem to have been all but forgotten in the contemporary literature. In a perfect world, there would be plenty of ‘Gold standard’ evidence from randomized placebo-controlled clinical trials (RCTs) that measure vaccine efficacy using highly specific laboratory-confirmed influenza endpoints. But these are scarce in the influenza literature. A placebo control group is simply not an option when studying vaccine benefits in populations that are already recommended for vaccine (such as seniors and persons with high-risk conditions). For that reason, observational studies have long made up the largest part of the evidence base, especially for influenza vaccine benefits in seniors. Assuming a near-perfect world, Orenstein et al. theoretically explore the expected performance of cohort and case-control study designs that use laboratory-confirmed endpoints, the latter with two different approaches to control selection. They focused on the consequences of less-than-perfect sensitivity and specificity of the rapid laboratory tests (an increasingly popular choice over culture-confirmation as the price of these kits falls) and on the prevalence of influenza relative to other respiratory pathogens. However, their simulations did not explore the possibility of selection bias leading to various degrees of mismeasurement. Orenstein et al. first explored a base-case scenario of a paediatric population, assuming realistic parameters of attack rates of influenza (15%) and other respiratory pathogens (30%), and rapid tests with 80% sensitivity and 90% specificity. In this scenario, all observational study designs performed about the same, although case-control studies using non-lab-positive controls had a slight advantage. All underestimate the true VE by about 15%; for example, a true VE of 70% would be measured at 56–60%. But as can be seen from results of the sensitivity analyses (Figure 2 in ref.), under a scenario in which the influenza attack rate is low relative to other respiratory illness (5% vs 30%), the VE would be substantially underestimated; for example, a VE of 70% would be measured at 40% or lower. So the good news from Orenstein et al. is that observational studies can generate reasonably accurate VE estimates, as long as laboratory tests are highly (590%) specific and as long as the incidence of influenza is relatively high. But as the authors point out, the bad news is that the use of endpoints with low specificity—such as clinical influenza-like illness without laboratory confirmation—is not appropriate. The consequences can be seen by perusing their Figure 2. If the specificity of the endpoint is 70%, for example, the underestimation is profound (measured VE at 10%). With extrapolation of Orenstein’s findings to a lower level of sensitivity of the outcome, it would naturally produce far worse results. Halloran and Longini previously developed a classy framework for cohort studies to analytically adjust measured VE for less specific outcomes in order to provide true VE estimates; however, these authors also did not consider the possible complications of selection bias. In the end, however, we must come back to the not-evenclose-to-perfect world in which we all live. Here, observational studies often report on a variety of endpoints that include unconfirmed influenza-like illness, pneumonia hospitalization Corresponding author. National Institute of Allergy and Infectious Diseases, National Institutes of Health, 6700B Rockledge Drive, room 1107, Bethesda 20892, Maryland, USA. E-mail: lsimonsen@niaid.nih.gov Laboratory of Infectious Diseases, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland, USA. The online version of this article has been published under an open access model. Users are entitled to use, reproduce, disseminate, or display the open access version of this article for non-commercial purposes provided that: the original authorship is properly and fully attributed; the Journal and Oxford University Press are attributed as the original place of publication with the correct citation details given; if an article is subsequently reproduced or disseminated not in its entirety but only in part or as a derivative work this must be clearly indicated. For commercial re-use, please contact journals.permissions@oxfordjournals.org Published by Oxford University Press on behalf of the International Epidemiological Association. The Author 2007; all rights reserved. International Journal of Epidemiology 2007;1–2 doi:10.1093/ije/dym084

Commentary: Dietary patterns in transition can inform health risk, but detailed assessments are needed to guide recommendations

Commentary: Dietary patterns in transition can inform health risk, but detailed assessments are needed to guide recommendations

Commentary: Growing up optimally in societies undergoing the nutritional transition, public health and research challenges

The question that underlies the study by Corvalan et al., published in this issue of ‘International Journal of Epidemiology’, has far-reaching consequences: what is the growth pattern that will lead to optimal health both in the shortand long-term for children living in countries in nutritional transition? A few decades ago, when obesity, hypertension, diabetes and coronary disease were marginal (or marginally recognized) public health problems in low-income countries, the answer was easy. The more infancy weight gain, the better, in order to prevent undernutrition, related infections, deaths and impaired physical and neurological development. The short-term benefits of sufficient infancy weight gain are unquestionable. But now that chronic adult diseases have become public health priorities in countries in nutritional transition and that the obesity epidemic is taking global proportions, the question of possible negative consequences of excessive infancy or childhood weight gain has become more relevant. Increasing evidence suggests that rapid weight gain during some periods of growth may be associated with the programming or imprinting of long-term obesity and other cardiovascular risk factors. The balance of benefits and risks has become more difficult to find in societies that deal with the ‘dual burden of [. . .] the unfinished agenda of infectious diseases and the emerging agenda of non-communicable diseases’. In this context, the challenge of public health nutrition is to define clear and coherent messages for health workers and parents to promote growth patterns that are optimal for short-term and long-term health. The first condition for designing such messages is to have strong scientific evidences to support them. Unfortunately, when it comes to defining optimal growth for children in societies in nutritional transition, evidence remains very incomplete. The lifecourse approach to chronic disease epidemiology provides a useful framework, but many scientific challenges have limited the interpretation and usefulness of existing research. First, most studies of obesity programming are observational and no matter how many are conducted and how well they are designed, observational studies will not demonstrate causality and therefore will have limited practical or public health implications. For example, unknown or unmeasured confounding factors, such as common genetic predisposition between the exposure and the outcome, cannot be eliminated in observational studies. Unfortunately, in the context of a lifecourse approach to chronic disease epidemiology, by definition, results of prospective randomized intervention studies will take a lifetime to become available. Meanwhile, decisions are made based on the follow-up of a few intervention studies designed for other purposes or based on observational studies. Most observational studies are analyses of cohorts constituted to address questions other than our current concerns. Consequently, analyses are driven by data availability, rather than by a priori hypotheses. For example, birth weight is more widely available than infancy or childhood growth data and therefore many studies have focused on fetal growth. In contrast, in the study by Corvalan et al., growth status was assessed at several time points in a cohort where stunting and adult overweight are frequent, allowing the authors to plan a more hypothesisrather than data-driven analysis. Another strength of this study is that data on infancy length and childhood height, as well as some assessment of adult adiposity, enabled the analysis of separate aspects of growth: changes in length vs changes in body mass index (BMI), as well as adult lean mass vs adiposity. The data show that changes in BMI during infancy and between the ages of 3 and 7 years, but not during fetal life or between the ages of 1 and 3 years, are predictive of adulthood adiposity. Changes in length during any period appear to be predictive of adult lean mass, but also, to some extent, of adult adiposity. Even if ideal data are available, scientists face the challenges of choosing optimal analytical methods to identify periods of growth that are more strongly associated with adult outcomes. Human growth during one period is not independent from the growth in the previous period. Additionally, the magnitude of variance around the average growth pattern depends on the duration of the period considered and on the expected variance from the norm during that period. For example, moving from the 50th to the 95th percentile in the 4 months between birth and age four months represents a more important variance from the norm than moving between the same two percentiles in the 8 months between age 4 months and 1 year. Some periods, such as early infancy or puberty, are E-mail: nstettle@upenn.edu Division of Gastroenterology, Hepatology, and Nutrition, The Children’s Hospital of Philadelphia and Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Philadelphia, PA, USA. Published by Oxford University Press on behalf of the International Epidemiological Association The Author 2007; all rights reserved. International Journal of Epidemiology doi:10.1093/ije/dym058

Commentary: The ubiquity of prostate cancer: echoes of the past, implications for the present: "What has been will be again, what has been done will be done again,; there is nothing new under the sun." ECCLESIASTES 1:9

Commentary: The ubiquity of prostate cancer: echoes of the past, implications for the present: "What has been will be again, what has been done will be done again,; there is nothing new under the sun." ECCLESIASTES 1:9

Commentary: On the clinical prediction of pre-eclampsia and its enigmatic aetiology

Commentary: On the clinical prediction of pre-eclampsia and its enigmatic aetiology

On the frequency of occurrence of occult carcinoma of the prostrate

On the frequency of occurrence of occult carcinoma of the prostrate