Epidemiology and ‘developing countries’: Writing pesticides,poverty and political engagement in Latin America

Geographic framing

The major finding that emerged from analysis of the quantitative sample was that most texts employ representations of Latin America to motivate studies in allegedly representative ‘developing country’ sites. In numerous papers, geographic representations play a central role in motivating studies, and introductory sections are framed as contributing to knowledge on pesticide problems in the developing world through examination of a specific case study in a particular Latin American country. A total of 68 of the 88 identified papers explicitly indicate that the featured research site is in a ‘developing’, ‘less developed’, ‘Third World’, ‘lower- and middle-income’, or, in one case, ‘poor’ country, sometimes combining terms within the same article. ² In many such cases, the study’s importance to development issues is identified in the title through expressions such as ‘subclinical health effects of environmental pesticide contamination in a developing country’ (McConnell et al., 1999). Alternately, authors often begin article introductions by describing a situation of broad importance, such as overall estimates of pesticide poisonings worldwide, and then relate it to ‘developing countries, such as Ecuador’ (Cole et al., 1997: 277). In the following passage, 33 farmers from the vicinity of León, Nicaragua, serve as relevant units of analysis for application of an exposure assessment methodology to ‘working conditions in Nicaragua’, and by extension to all ‘developing countries’:

The fluorescent tracer technique has been occasionally used in developing countries … we built up some educational experience with this technique with the intent of expanding this experience towards a robust but valid alternative method for semi-quantitative exposure assessment in a Third World context … it was necessary to adapt it to the particular working conditions in Nicaragua … Nicaraguan farmers usually do not wear protective clothing and experience more extensive pesticide skin contamination… we scored the dermal exposure of 33 farmers from the surroundings of León, Nicaragua … (Aragón et al., 2004: 602)

In contrast, a smaller number of papers begin with a scientific question of relevance to a particular field, and then introduce a Latin American setting as relevant due to its pesticide exposure profile without categorizing it with respect to development (e.g. Crissman et al., 1994; Keifer et al., 1996b; McConnell et al., 1994). This does not mean that those papers framed as being about development issues do not also engage with scientific literatures; indeed, these often display a ‘recycling’ movement (Swales, 1990, 2004) in which they alternately narrow in on a geographically defined research gap and a universal scientific one. The following passage demonstrates both geographic narrowing from all developing countries to Costa Rica and identification of a gap in universal scientific knowledge about the toxicity of the carbamate class of chemicals:

A majority of the pesticide poisonings in developing countries is caused by neurotoxic organo-phosphates and carbamates … Cholinesterase inhibitors bind to acetylcholinesterase, the enzyme that controls the transmission of the nerve impulses at the cholinergic synapses throughout the nervous system … few carbamates have been submitted to extensive neurotoxicity testing … In Costa Rica, both organophosphate and carbamate nematocides (worm killers) are widely used …. (Wesseling et al., 2002: 27)

These place-linked rhetorical moves, unremarkable at face value, demonstrate important aspects of the geographic framing of epidemiologic and related studies. The researchers involved in producing these studies obviously often have significant personal involvement in the research sites. Nevertheless, in these introductions local specificities are typically reduced to national identity, and that national identity is then taken as a marker of ‘developing’ status. As I discuss in the next section, furthermore, these representations portray Latin American countries in ways that systematically avoid certain types of historical or political explanation for this underdevelopment.

Pesticide exposure, poverty, and (a)historical causes

Further analysis of these representations suggests that, somewhat like the ‘bad luck’ geographies identified by Sparke (2009), they often present abridged and selective histories that justify specific technical interventions. These Latin American settings are variously characterized as lacking adequate technical and human resources for public health activities (Aragón et al., 2004: 602), failing to enforce appropriate occupational and environmental health standards (Corriols et al., 2009: 205; McConnell et al., 1990: 1238), and having ‘primitive working conditions’ (McConnell and Hruska, 1993: 1561) such as a lack of personal protective equipment (PPE). In describing these features of ‘the developing world’, papers seldom discuss causal factors from the pre–World War II period or considerations of political power, focusing instead on ‘proximal’ determinants of pesticide exposure. Most strikingly, 27 papers discuss pesticide poisonings and related impacts as ‘caused’ by specific pesticides. For example, ‘In Nicaragua, the neuropathic pesticides methamidophos and chlorpyrifos were identified as causes of these impairments … the poisonings for which the agent was confirmed were all due to acute organophosphates …’ (Delgado et al., 2004: 363). Consistent with this tendency is use of the evocative term ‘poisoning’ without identification of a perpetrator, as in the case of ‘workers poisoned with organophosphates … [and] organophosphates with which hospitalized workers were poisoned …’ (McConnell et al., 1994: 332).

While various papers in the sample do contain historical context, this typically deals with topics such as trends in pesticide use or health impacts, or the economic prominence of agroindustries such as banana, cotton, and flower production, without exploring political economic reasons for these phenomena (e.g. Grandjean et al., 2006; Handal et al., 2007a; Harari et al., 2010; López et al., 2009; Rosenstock et al., 1991; Valcke et al., 2005; Van Wendel de Joode et al., 1996; Wesseling et al., 1996). Another frequently identified but seldom explained cause of pesticide exposure is poverty:

In Nicaragua, subsistence farmers use backpack sprayers … The use of any other application technique such as tractor-mounted boom sprayer is out of their economical possibilities. (Blanco et al., 2008: 536)

they carry a much heavier burden of adverse health and environmental effects with the uncontrolled use of highly toxic pesticides in conditions of poverty… (Bravo et al., 2011: 258)

The vast majority of studies make no attempt to contextualize poverty in light of colonialism and subsequent political economic transformations in Latin America. The geographic framing of such studies thus presents a selectively historicized landscape in which pesticide exposures are related to unexplained poverty, lack of PPE, inadequate policy measures, or contemporary (again usually unexplained) changes in agricultural practices. As documented by Fairclough (2003), omission of explanations for key premises can function rhetorically to naturalize those premises, even when done unintentionally. Latin Americans thus often function rhetorically in this genre as units of analysis for the application of universal scientific methods to an unexplained and, by extension, constitutionally poor region of the world. Political economic explanations of poverty and underdevelopment in Latin America would be one way to help frame pesticide problems differently; of course, inclusion of such material might make it difficult to concisely make the case for traditional epidemiology and public health studies and interventions in the restrictive genre of the scientific journal article, a point to which I return shortly.

Responses

Prominent epidemiologist Kenneth Rothman (1998: 335) warns against making ‘superficial recommendations’, or indeed any policy recommendations at all, in papers submitted to Epidemiology, the journal he founded. One analysis of 50 medical research articles, a category overlapping epidemiology, however, found that ‘it was remarkable how frequently papers ended with a recommendation which was too imprecise to operationalize, or too grand to be implemented by a decision at much lower than ministerial level’ (Skelton, 1994: 459). This suggests the final sentences of an epidemiology article as one of its few sites – albeit a contested one – for expression of individuality or ‘extra-epidemiologic’ content. As Miller (2000) explains, furthermore, the solutions that are proposed for problematic issues are strongly determined by how those issues are framed.

A minority of papers in the quantitative sample contain no explicit indication as to how pesticide-related health issues should be approached or resolved. Schenker et al. (2004), for example, restrict introductory framing of their study of paraquat exposure and respiratory function in Costa Rica ³ to unanswered scientific questions and end with suggestions for further research. In contrast, however, the majority of included papers are positioned as helping to generate solutions to health problems, either in the introductions or in final paragraphs or sentences of discussion sections. One common solution offered in the articles is the establishment and improvement of systems for monitoring and surveillance of pesticide exposures and the risky practices leading to them, so as to inform public health interventions (e.g. Cole et al., 1988; Corriols et al., 2008; Keifer et al., 1996a; McConnell and Hruska, 1993; Wesseling et al., 1996). In contrast, a sizable number of papers conclude that ‘safe use’ of pesticides in developing country contexts is impossible, and that the elimination of highly toxic pesticides is required to improve health:

One short-term policy option is to substitute less dangerous but similarly effective chemical products for the more highly toxic ones…. Another policy solution approaches the problem from the public health viewpoint. Education on the proper management and handling of these dangerous products with clear discussion of the long-term health effects may be useful. However, unless there is a greater supply of personal protective equipment and incentives for their use, such interventions may be ineffective. A longer-term solution is to change potato production technology. Improvements in host plant resistance and development of new pest or disease management technologies may reduce pesticide use without yield reductions. (Crissman et al., 1994: 596–597)

The above passage illustrates several of the solutions proposed, singly or in various combinations, by most of the authors of studied texts. Furthermore, the passage’s reference to ‘the public health viewpoint’ suggests a particular way of looking at the world characteristic of public health researchers and professionals. From this perspective, “public health” has a legitimate, self-evident right and responsibility to intervene in health problems in society. Subtle evidence of this frame in the majority of included papers includes the ways in which the logical link between health problems and health sector intervention is elided. Often, authors will jump from the existence of pesticide-related health problems to the need for physicians and health researchers and policymakers to respond to them. The following example demonstrates this tendency, and also shows how faith in public health’s legitimacy persists in the face of questionable efficacy:

The Nicaraguan economy is based on intensive, pesticide-dependent agriculture (7). As a consequence, APP [acute pesticide poisoning] is considered a severe health problem, especially for agricultural workers but also in the rural population at large… In 1979, the National Unified Health System in Nicaragua included APP cases in its list of mandatory reportable diseases. Despite this, reporting of APP was rare through the early 1980s. Regulations and measures to enforce compliance did not exist. The reporting of information was not linked to any action on the part of the health sector. (Corriols et al., 2008: 774)

The link between the unexplained pesticide-intensiveness of Nicaraguan agriculture and the role of the health sector hides behind the connector ‘as a consequence’, while the passage clearly implies that the reporting of information should be linked to health sector actions. In addition, several papers urge the production of high-quality epidemiologic data on pesticide-related health problems as a way for health researchers to influence policymakers and the public (Cole et al., 2000; Corriols et al., 2008; McConnell and Hruska, 1993). In these and numerous other papers in the quantitative sample, the health sector plays a central role in responding to pesticide-related health impacts. Epidemiologic studies and public health interventions are thus justified on the basis of frequently truncated histories and geographies of Latin America, in which pesticides and poverty cause illness, but nothing causes poverty – it just exists. This framing naturalizes particular intervention routes and serves to close off debate with respect to others. Political responses such as land reform, income redistribution, macroeconomic policy shifts, and removal of incentives favoring export-based agriculture are, accordingly, not discussed.

Exceptions to the rule: political engagement in epidemiology articles

In contrast to this allegedly apolitical (though in reality deeply political) tendency, several early studies done in Nicaragua provide a muted representation of the rise and fall of the revolutionary Sandinista government with whose health system several authors of these studies had worked as ‘health internationalists’. Cole et al. (1988) and Keifer et al. (1996a) highlight, for example, the development of a National Unified Health System in 1979, while Keifer et al. (1996b) mention the period before ‘Nicaragua’s recent agroeconomic changes’ in explaining the production of pesticide exposures (p. 726). Along with McConnell et al.’s (1990, 1992) descriptions of pesticide-control measures implemented by the Sandinistas in the 1980s, these studies provide historical context to the extent that it helps to explain pesticide exposures and access to the data they have analyzed. McConnell and Hruska (1993) go slightly further, referring to ‘a series of social reforms, including redistribution of agricultural land to small farmers, and universal health coverage’ (p. 1559). This same paper also attributes pesticide poisonings among children during the mid-1980s to the fact that the young men who traditionally applied pesticides were ‘away at war’ (McConnell and Hruska, 1993: 1561). This paper appears in the relatively hospitable American Journal of Public Health, which frequently publishes articles engaging with issues of equity and social justice. Its references to social reforms relate to capacity for pesticide poisoning surveillance by the health system, however, while the veiled reference to the US-backed Contra war helps explain the study’s results, a typical function of biomedical article discussion sections (Skelton and Edwards, 2000). It is therefore especially illustrative in that it describes revolutionary politics and US-backed resistance to them, but in a way that appears highly influenced by a conventional biomedical journal article structure.

In Ecuador, Handal et al. (2007b) similarly engage with political debates over ‘new economic endeavors’ such as floriculture and end their paper in the more restrictive journal Epidemiology with a modest recommendation:

The economic benefits of the flower industry may improve child health in some respects, whereas pesticide exposure and environmental contamination may cause harm in others. Both aspects must be considered when assessing the impact of specific industrial exposures on child development and health. (p. 319)

The influence of Epidemiology’s discouragement of policy recommendations (Rothman, 1998) is suggested by an earlier book chapter published out of the same research program, with Handal as tenth author, asserting that ‘[c]ut flower production in countries like Colombia, Costa Rica, Ecuador and Mexico, illustrates neoliberal mechanisms that have been imposed in rural development’ (Breilh et al., 2005: 66–67). The chapter also highlights ‘surplus value extraction’ (p. 75) in floriculture and its negative ecological and social effects, hinting at political thought that Handal et al. (2007b) compress into their article’s final two sentences.

A more recent Ecuador-focused paper, in contrast, is published in the relatively new International Journal for Equity in Health, which is open to explicitly political approaches. Cole et al. (2011a) employ a typical epidemiology article structure, but also devote space to reviewing relevant Latin American social medicine literature. In contrast to the majority of papers in the quantitative sample, the discussion also explains poverty in highland Ecuador as due to ‘a historical legacy of hierarchy and social stratification; persistent ignoring of farmers’ rights by politicians and farmers; inegalitarian social policy for decades; declining productivity among smallholder farmers; and unequal globalization, a widespread underlying reason for inequities in health’ (Cole et al., 2011a: 8). These Ecuadorian and Nicaraguan papers thus illustrate various ways in which some researchers are attempting to navigate the tension between epidemiology’s genres and methods and the implications of the socially stratified settings in which they are working.

Developing countries and (mild) turf wars

I return to this more politically engaged approach to public health shortly. First, however, I trace a slightly acrimonious exchange between two research groups to show how geographic framing is strategically useful in generating scientific journal articles and advancing academic careers. A paper presenting a methodology for assessing dermal pesticide exposures based on assessment of likely exposure determinants (Blanco et al., 2008) is a particularly illustrative example of scalar framing. This paper identifies a public health need in universal scientific terms, highlights its lack in developing countries, and then introduces a methodology developed to remedy this lack and an example of its application in Nicaragua (which is, as a result, an interchangeable ‘developing country’ in the eyes of science):

In developing countries … methods for assessment of exposure must be inexpensive and easy to use…. Aragón et al. (2006) modified Fenske’s VSS to identify patterns of dermal contamination and to estimate dermal exposure to pesticides in Nicaraguan farmers…. Subsistence farmers in developing countries apply pesticides under high-risk conditions … (van Wendel de Joode et al., 1996; Aragón et al., 2001). Thus, we propose ‘determinants of dermal exposure ranking method (DERM) …. We present here an example of how to use DERM in a group of Nicaraguan subsistence farmers’. (Blanco et al., 2008: 535–536)

In this case, the scalar framing is made especially evident by its somewhat precarious dependence on prior texts. A cited study of farmers in a region of western Nicaragua (Aragón et al., 2006) is used to situate Nicaraguan farmers as being ‘in developing countries’. The habits of subsistence farmers in apparently all developing countries are then described, citing a study of 11 banana plantation employees in Costa Rica (Van Wendel de Joode et al., 1996), and another study in western Nicaragua (Aragón et al., 2001). The paper repeatedly highlights a lack of exposure assessment methods relevant to developing countries, particularly singling out an existing method entitled DREAM (DeRmal Exposure Assessment Method). When the creators of DREAM respond in a letter, their tone is tellingly curt. Beginning with a confrontational title, ‘The accuracy of DERM may be a self-fulfilling DREAM’, the letter outlines several ‘major concerns with the presented study and the manuscript’ (Kromhout et al., 2008: 783) and illustrates the professional currency at stake:

Finally, given the familiarity of the authors with our work, they should have known that DREAM has been applied in a developing setting among pesticide exposed vineyard workers in South Africa (van Wendel de Joode et al., 2005). The authors are therefore wrong in believing that ‘validated semi-quantitative DREAMs for pesticides applicable in developing countries are to our knowledge non-existent’. (p. 783)

The use of quotation is a telltale sign of skepticism toward the content of reported speech in scientific writing (Myers, 1990); furthermore, the bald statement that Blanco et al. are ‘wrong’ goes well beyond the polite tone typical of scientific writing (Myers, 1989), hinting at the degree to which Kromhout et al. feel slighted. The passage also employs the characteristic scalar framing discussed at length above, in that it assumes the equivalence of Nicaragua and South Africa, and presumably all other ‘developing settings’. Thus, while the overall goal of the two research groups seems to be to benefit workers and farmers in the global South, they are also clearly attempting to benefit themselves professionally – as scientists typically must – by generating new and representative knowledge uniquely attributable to their efforts.

Metadiscourse: authorial presence in the texts

Conflicts such as the DERM–DREAM exchange are consistent with canonical work in science and technology studies (STS) documenting how scientists must diplomatically contribute novel results to existing bodies of knowledge but without introducing so much novelty that their work lacks credibility (Myers, 1989). Analysis of metadiscourse – first-person usage, evaluation, modals, and the like – in the mixed sample similarly reveals familiar patterns related by Hyland (1998) to the social dynamics of science, specifically the need to highlight a paper’s specific contribution to knowledge while simultaneously emphasizing the author’s scientific objectivity (Myers, 1992). The geographic representations of Latin America documented above typically serve to justify the importance of the paper’s original contribution. Finally, some papers employing this characteristic scalar framing and familiar pattern of metadiscourse also make space for comments resisting the implications of a narrowly scientific perspective.

Use of first-person active voice, for example, fits with previous studies of scientific writing, but in a way that reflects the challenges of carrying out epidemiologic studies in Latin America. Following a largely third-person literature review, authors often employ first-person to highlight the specific contribution of the study, usually responding to a geographically framed research gap (e.g. Cole et al., 2011a; Monge et al., 2004). This is consistent with the observation that many scientists mark out their specific contribution to science with a transition to first-person in article introductions (Myers, 1992). In numerous methodology sections, furthermore, use of first-person tends to mark out specific nonroutine methodological steps requiring specialized knowledge:

We excluded those workers whose year of birth was missing … Differences in cancer morbidity among banana workers as compared to the general population were assessed by comparing the observed numbers of all cancers and specific sites with the expected numbers … to make the reference population comparable to the exposed population, we only used the identified subset… The expected numbers were derived …. The standardized incidence ratio (SIR) was calculated … For observed numbers below 20 the exact method was used … As main exposure indicator we used categories of number of months of employment. Because of underreporting … by the companies, high job turnover within the region, and intermittent migration patterns … we also categorized the number of years … (Wesseling et al., 1996: 1127)

Those activities for which active, first-person voice is used – deciding who to include in the analytic sample, which populations are comparable, which exposure indicator to use, whether to convert a variable of one type into another for analysis – are all ones requiring the professional judgment of an epidemiologist. Those that are described in passive voice – calculating observed versus expected cancer numbers or a standardized incidence ratio, or using the exact method – are all activities taught in basic epidemiology and biostatistics courses. Of relevance here is the observation that passive voice in scientific writing has historically served to convey objectivity and universality (Atkinson, 1999). In Wesseling et al. (1996) and other epidemiology papers in the mixed sample (e.g. Cole et al., 1997; Corriols et al., 2008; Monge et al., 2007), passive voice is largely used to describe methods that a trained research assistant could do; those in active, first-person voice are typically those requiring epidemiologic judgment. In the discussion section of numerous papers, first-person active voice is also used when highlighting the study’s overall findings and their importance or novelty, appearing to emphasize the personal involvement of the authors in making a contribution to science: ‘In conclusion, we have demonstrated the feasibility of adapting the ICF to a new cultural and socioeconomic setting …’ (Monge et al., 2004: 78). Similarly, discussion sections frequently use first person as demonstrated below in (a) comparing results to those found by other investigators, (b) discussing the strengths and weakness of the study, or (c) speculating on explanations for results:

These patterns of first-person usage are largely consistent with how STS work suggests scientists emphasize their contributions to science but also employ the geographic representations documented above in justifying why those contributions are important.

Such familiar dynamics of science are also demonstrated by a characteristically epidemiologic use of evaluation in papers employing the same scalar framing. A study conducted using data from Costa Rica’s population and disease registries flatly states that

[i]ncreased occurrence of soft tissue sarcomas, various lymphohaematopoietic cancers, and cancer of the brain, testis, stomach and prostate have been observed among pesticide exposed populations … The use of dibromochloropropane (DBCP) in the 1970s caused the sterilization of approximately 1500 banana workers … (Wesseling et al., 1996: 1125)

The authors also describe the ‘potential value for epidemiological research’ of the disease registries (Wesseling et al., 1996: 1125). They then state that ‘[k]nowledge about the functioning and quality of these registries was insufficient,’ (p. 1125) and later point out a ‘severe problem in the identification of the deceased’ (p. 1127). Their study aims to contribute to ‘improvements of the registries’ (p. 1125); indeed, the authors claim that ‘one main virtue of this study is the effort per se which has permitted a better understanding of the registries involved’ (p. 1128). Thus, cancers and sterilizations are presented without evaluation, while those factors related to epidemiological study design – data availability and quality, for example – receive more personal authorial attention. Evaluative language such as ‘value’, ‘virtue’, improvements’, and ‘better’ all suggest a characteristically epidemiologic perspective approving of good data sources permitting proper studies. This tendency is consistent with the authors’ view that Costa Rica presents an opportunity to pursue high-quality epidemiology in developing countries:

Despite constraints to conducting research in the third world … epidemiological cancer studies are increasing, mainly of the case-control type …. Cohort studies, however, are almost nonexistent, because of scarce population-based cancer registries and limited possibilities to follow populations over time …. In Costa Rica, high exposures co-exist with nationwide population and disease registries. (Wesseling et al., 1996: 1125)

In stating that ‘[w]orking conditions in Costa Rica are insufficiently controlled’ (Wesseling et al., 1996: 1125), however, the authors indicate their preference that working conditions in Costa Rica should be controlled. In this manner, the paper complements its approval of high-quality epidemiology with disapproval of uncontrolled working conditions, a recognizable set of priorities in occupational health.

The pattern of metadiscourse in a study on Ecuador published almost simultaneously (Cole et al., 1997) similarly indicates epidemiologic priorities. The introduction points out, without evaluation, that ‘[t]he adverse health impacts of agricultural pesticide use has been an ongoing concern among health personnel over the last two decades,’ and describes ‘wide disparities in levels of education and socioeconomic status between general populations in developing countries and those of northern countries in which neurobehavioral tests were developed’ (Cole et al., 1997: 277). This neutral tone contrasts with the evaluative language and modals used in the ‘Results’ and ‘Discussion’ sections to describe the acquisition of data: ‘[c]onsiderable success was achieved in obtaining similar levels of age and education across the exposure groups’ (p. 280); and,

[i]deally, studies of cohorts of workers and/or community members in high exposure situations in the developing world or among different immigrant groups in developed countries would deal with some of these limitations. Participants should be more judiciously selected … Exposures should be more fully characterized across a wider range of nutritional and occupational factors. (p. 285)

This authorial presence indicates approval of proper epidemiologic methods, consistent with the introduction’s framing of the study as responding to ‘[a]n opportunity to carry out such research [i.e. applying sophisticated neurobehavioral methods to a highly exposed population] in another Latin American country’ (p. 277). While this contrast certainly suggests disciplinary priorities approving of rigorous epidemiologic methods and opportunities to use them in developing countries, the use of metadiscourse also approves of improved working conditions in the study site:

Although such further research may clarify the reasons for the neurotoxic effects we observed, we should not let the call for more research obscure the glaringly lower neurobehavioral scores we observed in our potato farming populations. Reductions in clear overexposure to pesticides should not be stalled pending further exploration of other potential contributors. (Cole et al., 1997: 285)

In this manner, a thoroughly epidemiologic paper ends with an explicit recognition that strict adherence to epidemiology’s disciplinary imperatives may conflict with actually improving health in Ecuador.

Genre effects

The papers of the quantitative sample – even with the selective historical and geographical framing I document above – still provide far more socioeconomic context than might be expected based on descriptions of epidemiologists as ‘prisoners of the proximate’ (McMichael, 1999) who characterize the ‘web of causation’ without paying attention to the spider (Krieger, 1994). The inclusion of such context here is especially unusual in a genre that (a) is typically limited in the amount of text it can devote to historical and political economic explanations and (b) must logically and compellingly arrive at the necessity for a specific epidemiology research project.

I have highlighted a few examples of epidemiologists attempting to engage with political power within this restrictive genre, especially in the final sentences of epidemiology papers. Other strategies allowing engagement with ‘extra-epidemiologic’ content appear to include publishing in less restrictive journals; in addition, several authors have published review papers and in other less restrictive genres, occasionally co-authored by social scientists or employing social science methods (e.g. Barraza et al., 2011; Orozco and Cole, 2011; Sherwood et al., 2007). Analysis of two such papers here further suggests that some framing choices in the quantitative sample are due to the restrictive genre, while others persist more broadly.

A review paper involving numerous epidemiologists prominently represented in the quantitative sample provides a programmatic vision for involvement of health professionals in ‘solving the pesticide problem in Latin America’ (Keifer et al., 1997). Affiliations listed by lead author Matthew Keifer include both the University of Washington and CARE-Nicaragua, a main nongovernmental organization (NGO) partner in Sandinista-era pesticide reforms. With respect to geographic framing, the introduction states that, ‘[d]espite awareness of the problems associated with pesticide use, they have persisted, and are even increasing in some regions of the developing world. Nowhere is this more apparent than in Latin America’ (Keifer et al., 1997: 26). In response to these pesticide-related problems, the authors then introduce ‘[a] recent initiative in Central America … a significant new direction in the effort to resolve the pesticide problem in Latin America’ (Keifer et al., 1997: 27). The rhetorical use of scale in establishing terrain (pesticide problems in all developing countries), focusing on a specific representative region, then zeroing in on an initiative in Central America, clearly reflects the general pattern observed in the quantitative sample. That individual developing countries function as interchangeable units of analysis is further suggested by the assumption that ‘lessons learned in Nicaragua and Costa Rica can be used to reduce per-country costs as the project expands’ (p. 29).

With respect to the causes of pesticide exposure, the article lists agricultural production techniques, different ministerial influences on policymaking, and chemical industry lobbying. Despite this multifaceted discussion, however, ‘health sector empowerment’ is presented as the main solution, bypassing larger political economic power structures driving exposure and vulnerability to pesticides. For example, while the paper later highlights the effects of structural adjustment on health services (p. 30), related effects on agriculture with implications for pesticide-intensiveness are left unexplored. In employing the same scalar framing evident in the quantitative sample, then, the paper thus provides a selective (though not superficial) discussion of political economic causal factors, leading to calls for mobilization of the health sector.

Canadian researcher Donald Cole is lead author of numerous quantitative-sample articles on pesticides and health in Ecuador, as well as an early paper emerging out of the Sandinista-era pesticide surveillance program in Nicaragua (Cole et al., 1988). A more qualitative effort (Cole et al., 2006) combines a summary of the Ecuador research program and interviews with research officers at Canada’s International Development Research Centre (IDRC). The paper does not perform the typical geographic narrowing characteristic of Cole’s quantitative papers, although its appearance in a special issue of the Canadian Journal of Public Health about Canada–Latin America connections perhaps means it does not need to. In addition, the paper contains information-dense tables that are largely not discussed in the body of the paper, conveying information on Canada–Latin American solidarity movements and political refugees fleeing Latin America in the 1970s and 1980s. (In contrast, the tables and text are tightly linked in Cole’s epidemiology papers.) The paper’s conclusion critically comments on the IDRC ‘EcoHealth’ program that funded much of the authors’ work:

Can the holistic, almost transcendent systems’ [sic] frameworks promoted by EcoHealth, reminiscent of other ideals in the ‘new public health’, deal with such heterogeneity? … A further challenge is the pressure of market capitalism, which pervades even remote agricultural communities, often with the face of multinational pesticide companies … Although reduction of toxic pesticide use made sense at a field level for Ecuadorian farmers and at a regional level for communities and health authorities, movement globally is slow… Crucial issues of differential power and knowledge between different stakeholders have to be confronted, including in EcoHealth discourse … EcoHealth’s ability to effectively grapple with complex policy environments may depend on its linkages with other research programs that IDRC sponsors on equity, globalization, and international markets. (Cole et al., 1988: I12-I13)

The paper’s unconventional structure and appearance in a supplement edited by a public health historian suggest that it provides a flexible venue for deeper exploration of themes briefly discussed in the closing sentences of several of Cole’s previous epidemiology papers. In comparison to Keifer et al.’s earlier paper, furthermore, the references to ‘differential power and knowledge’ and ‘equity, globalization, and international markets’ also suggest a more complicated response than just health sector involvement.