IJE Advance Access originally published online on May 20, 2004
International Journal of Epidemiology 2005 34(2):485-487; doi:10.1093/ije/dyh230
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Published by Oxford University Press on behalf of the International Epidemiological Association
Letters to the Editor |
Development of environmental epidemiology in Latin America: scope, methodological issues, priorities
Servizio di Epidemiologia dei Tumori, Dipartimento di Scienze Biomediche e Oncologia Umana, Università di Torino, Via Santena 7, 10126 Torino, Italy. E-mail benedetto.terracini{at}fastwebnet.it
 |
The quarrel around the oil fields |
|---|
 Top The quarrel around the... The mission of epidemiologyWhich priorities for research...To conclude, a historical...References |
|---|
In 2004, a discussion about environmental epidemiology in Latin America cannot ignore the recent debate prompted by a study carried out in Ecuador.1 The debate was focused on the extent to which (environmental) epidemiological research, with its limits and uncertainties, can and must legitimate public health measures. In somewhat crude terms, the conflict has been described as a watershed between the pressure on epidemiologists to either remain ‘on the side of the angels’ (that is, forcing the interpretation of results in order to protect people's health) or preserve the scientific credibility of the discipline by strict adherence to rules of causal inference.2
The study which triggered the debate was an ecological study in four Ecuadorean provinces. It estimated a statistically significant excess of cancer cases (largely, stomach cancer in men and cervical cancer in women) among residents in the 4 counties located around the oil fields compared with the 11 counties with no oil fields. The authors acknowledge that no conclusive causal inference can be drawn from their study, but they believe that their results should support some recommendations. They recall a previous study of theirs, in another oil area in Ecuador,3 in which samples of the water used by the population to drink, wash themselves, and wash the linen were found to contain concentrations of oil-derived hydrocarbons 10–300 times higher than the limit of 0.01 parts per million established by the European Union. In addition to more-appropriate epidemiological studies, the authors recommended implementation of a systematic environmental monitoring (‘in order to estimate, control and eliminate the sources of pollution’) and the production of cancer incidence data. The authors point out that these recommendations are also supported by an element of plausibility: some components of the crude oil are carcinogenic and their mechanism of action is known. In addition, their findings are consistent with observations reported in the literature: elsewhere in the world, studies with a different design have suggested a carcinogenic risk. Finally, the authors argue that the ethical responsibility raised by the precautionary principle cannot be ignored.4
The debate has some semantic connotations, which are not easily perceived by those whose mother tongue is not English. In his critical comments, Jack Siemiatycki2 states that the study only provides a hint (a term for which my Oxford Dictionary provides a spectrum of definitions ranging from ‘very small trace’ to suggestion) whereas the authors' assessment (1) is that their findings are compatible (my emphasis) with an association (which they carefully do not connote as causal) between life around the oil wells and cancer incidence.
Substantially, there are two pivotal points in Siemiatycki's warning against an excessive partisanship of the epidemiologists who choose to stay on the side of the angels.5 On one hand, he questions the correctness of the comparability and integration of findings in Ecuador with results of studies around oil fields carried out elsewhere, on different populations and in different conditions of exposure. On the other hand, he dissents on the weight to be given to the risk estimates from a single epidemiological study in making recommendations for an increased control of the sources of environmental pollution. Fundamentally, he does not dissent on the essence of the recommendations, but he denies that the recomendations can be derived from the results of the epidemiological study.
Interestingly, within the debate, no mention was made of any editorial dilemma regarding the publication of a study which is undoubtedly ‘weak’. Strictly speaking, if the study was unreliable, the journal should have rejected it. However, Siemiatycki5 states that he recommended the publication (presumably as a referee for the journal) because it raises important questions and should be used to focus on the possible weight of chance or bias.
|
The mission of epidemiology |
|---|
|
TopThe quarrel around the...The mission of epidemiologyWhich priorities for research...To conclude, a historical...References |
|---|
At first sight, the debate raises once again the mission of epidemiology, whether ‘scientific’ or ‘action-oriented’. The value of the former is universal: epidemiology ought to be focused on the most adequate inferential method, arguing about causality of associations which—when demonstrated—are valid in any country, period, and circumstance (except, perhaps, for infrequent ethnic differences in vulnerability). On the contrary, the core value of the ‘action-oriented’ perspective is local, and relates to the extent to which causal associations which have already been demonstrated (or at least hypothesized), are relevant to a given local problem.
The tension between these two perspectives ignores the importance of the context in which epidemiology carries out its mission, irrespective of the basic findings. In this particular episode, the context is typical of developing countries and for this reason one must ask: which environmental epidemiology is possible? And, which epidemiology is needed for populations who are poor and exposed to the consequences of a deregulated industrial and agricultural production, largely governed by large corporations (including the multinationals of tobacco: see, for instance, President Menem's veto to the Neri's proposal to stop cigarette publicity in Argentina6)?
How can this context be described and quantified? A crude but numerical indicator is provided by Medline. The Table below shows the results of a very modest citation-finding exercise through some combinations of keywords and the names of some countries in Latin America and in countries where a relatively sophisticated epidemiological expertise is available.
|
A possible interpretation of these figures is that in Latin America there are few environmental circumstances deserving epidemiological attention, but this is hardly tenable (it is worth noticing, nevertheless, that, per unit of population, more studies on childhood lead blood levels have been carried out in Ecuador than in Italy).
More likely, these figures reflect difficulties encountered by environmental epidemiology in Latin American, among which the numerical insufficiency of local expertise is not the only one (this problem has been approached through co-operation with outstanding international centres7). The core problem is the construction of knowledge, at all stages of the process, ranging from the generation of aetiological hypotheses to the estimation of the efficacy of public health interventions. Had such an investigation such as the Hurtig and San Sebastian's study1 been conducted in Europe, it would have included components that were not feasible in the Ecuadorian context. I am alluding, for instance, to: (1) the availability of data from geographic information systems, (2) the collection of routine mortality rates in small areas countrywide, (3) the assessment of the reliability of endpoints typically used in epidemiological studies, and (4) the correspondence to international standards of current statistics, such as hospital discharge diagnoses, mortality data, and nationwide cancer registration (in countries where research strategies are fragile, international collaboration can protect local investigators from petty interferences from local power: this was also the Italian experience, and not so long ago).
I am not supporting the ineluctability of a double standard in environmental epidemiology. What should be acknowledged is that there are differences between countries and between periods. While contemporary epidemiology has been criticized for neglecting the social-historical context of disease, it must be admitted that there are no easy solutions to this problem, as contextual (social, economic, cultural, technical, and environmental) circumstances keep changing.8 Nowadays, some environmental aetiological interpretations of studies in cancer epidemiology carried out in Italy during the 70s sound naive: nevertheless, those studies contributed significantly to shaping the culture of public health workers (also because these studies were subject to international levels of critique). Given the same environmental problems, the spectrum of models for field investigations is far larger in countries where both the epidemiological tradition in the academy and the attention of public health authorities to environmental problems have been more robust. In terms of methods, the major problems in developing countries are the limited availability of quantitative estimates of exposure and the lack of standardized measures of outcome.9
When considered in the light of this web of circumstances, it seems overly simplistic to state that ‘Hurtig and San Sebastian's position, if taken to its logical limit, could be used to justify dishonest science if it provides support for a particular public health interventions’.5
|
Which priorities for research in environmental epidemiology in Latin American countries? |
|---|
|
TopThe quarrel around the...The mission of epidemiologyWhich priorities for research...To conclude, a historical...References |
|---|
Environmental pollution is concentrating in Latin America as a consequence of the economic conditions of those countries and the limited regulation of agricultural and industrial production. Severe risks for the populations' health can be postulated, but direct measures of disease occurrence encounter major difficulties. Also on the basis of the Italian experience during the 1960s and 1970s, investigations such as those of Hurtig and San Sebastian help to develop an epidemiological know how in loco and create awareness of the potential consequences of the environmental decay within the health authorities, the population, and the scientific milieu. It is important for the former two categories to be enlightened on the basic concept of statistical power, in order to avoid hasty misinterpretations of ‘negative’ epidemiological findings.
Admittedly, in all societies, the public opinion and public health authorities are more ‘impressed’ by the demonstration of the occurrence of disease (particularly when these are severe or lethal) than by studies which estimate the risk of loss of health brought about by environmental exposures. The concept of risk, somehow, is perceived as less impending, even when based on valid estimates. Nevertheless, in my mind, properly designed epidemiological studies on the prevalence of exposures are to be encouraged, also in consideration of the long latent period of many environmentally induced diseases. There is nothing new in this message: disturbing findings on biological markers of environmental exposures in children in Ecuador have been reported (see for instance refs 10, 11). Similarly, statistics on working conditions, such as the number of children at work, are non-specific but important indicators of the attention given to the protection of a vulnerable fraction of the population from exogenous exposures.12 Finally, documentation on the export to Latin America of dangerous industrial activities and materials (such as asbestos) is already available and should be expanded. Findings of well-designed studies on these aspects should impact on the decision-making processes of public health authorities as much as conventional epidemiological measures of disease occurrence.
|
To conclude, a historical and literary citation |
|---|
|
TopThe quarrel around the...The mission of epidemiologyWhich priorities for research...To conclude, a historical...References |
|---|
At least until the 1940s, every year, in the town of Oruro in Bolivia, on Carnaval Sunday and Monday, Atahualpa was killed on stage by Pizarro. The piece was bilingual, performed in archaic Bolivian Quechua and Spanish. The bilinguism, expressing the mutual incomprehension of the Spanish and the Indigenous population meant drama for the losers and laughter for the winners.13 The possibility of a similar lack of communication—mutatis mutandis—can be read in Hurtig and San Sebastian's final warning: we should reject the view that the ideal world is the one in which epidemiologists with clean hands remain far away from the practice of public health and from the public they are supposed to serve.
|
References |
|---|
|
TopThe quarrel around the...The mission of epidemiologyWhich priorities for research...To conclude, a historical...References |
|---|
1 Hurtig AK, San Sebastian M. Geographical differences in cancer incidence in the Amazon basin of Ecuador in relation to residence near oil fields. Int J Epidemiol 2002; 31:1021–27.
2 Siemiatycki J. Commentary: Epidemiology on the side of the angels. Int J Epidemiol 2002; 31:1027–28.
3 San Sebastian M, Armstrong B, Cordoba JA, Stephens C. Exposures and cancer incidence near oil fields in the Amazon basin of Ecuador. Occup Environ Med 2001; 58:517–22.
4 Hurtig AK, San Sebastian M. Epidemiology on the side of the angels ... or the people? Int J Epidemiol 2003; 32:658–59.
5 Siemiatycki J Response. Int J Epidemiol 2003; 32:658–59.
6 Barnoya J, Glantz S. Tobacco industry success in preventing regulation of secondhand smoke in Latin America: the ‘Latin project’. Tobacco Control 1002; 11:305–14.
7 Hogstedt C, Ahlbom A, Aragon A et al. Experiences from long-term research cooperation between Costa-Rican, Nicaraguan and Swedish institutions. Int J Occup Environ Health 2001; 7:130–35.[Medline]
8 McMichael AJ. The role of epidemiologists in eradicability of poverty (letter). Lancet 1998; 352:1627.[Web of Science][Medline]
9 Cullen MR, Harari R. Occupational health research in developing countries: The experience in Ecuador. Int J Occup Environ Health 1995; 1:39–46.[Medline]
10 Harari R, Forastiere F, Axelson O. Unacceptable ‘Occupational’ exposure to toxic agents among children in Ecuador. Am J Indust Med 1997; 32:185–87.[Medline]
11 Counter SA, Buchanan LH, Laurell G, Ortega F. Field screening of blood levels in remote Andean villages. Neurotoxicology 1998; 19:871–77.[Web of Science][Medline]
12 Kogevinas M, Boffetta P, Pearce N. Occupational Exposure to Carcinogens in Developing Countries. IARC Sci Publication # 129. Lyon: International Agency for Research on Cancer, 1997.
13 This text has been collected in March 1942 by Miss Ena Dargan and was edited by CH Balmori, University of Tucuman (Argentina). It has been quoted in Terracini L. L'incomprensione linguistica nella conquista spagnola, dramma per i vinti, comicità per i vincitori. In I Codici del Silenzio. Alessandria: Edizioni Dell'Orso, 1988, pp. 197–99.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Hurtig and M S. Sebastian Epidemiology vs epidemiology: the case of oil exploitation in the Amazon basin of Ecuador Int. J. Epidemiol., October 1, 2005; 34(5): 1170 - 1172. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||