Article Text

Download PDFPDF

Original article
Epidemiologic investigation of an occupational illness of tobacco harvesters in southern Brazil, a worldwide leader in tobacco production
  1. Patrícia Bartholomay1,
  2. Betine Pinto Moehlecke Iser1,
  3. Patrícia Pereira Vasconcelos de Oliveira1,2,
  4. Tania Esther Herc Holmer dos Santos3,
  5. Deborah Carvalho Malta2,
  6. Jeremy Sobel4,
  7. Lenildo de Moura1,2
  1. 1Field Epidemiology Training Program EPISUS, Health Surveillance Secretariat, Ministry of Health - Brazil, Brasilia, Distrito Federal, Brazil
  2. 2Non-communicable Diseases Coordination, Health Surveillance Secretariat, Ministry of Health - Brazil, Brasília, Distrito Federal, Brazil
  3. 3Non-communicable Diseases Coordination, State Center for Health Surveillance, State Health Secretariat, Porto Alegre, Rio Grande do Sul, Brazil
  4. 4Center for Global Health Center, Centers for Disease Control and Prevention - CDC, Atlanta, Georgia, USA
  1. Correspondence to Patrícia Bartholomay, Health Surveillance Secretariat, Ministry of Health - Brazil, Setor Comercial Sul, Quadra 4, Bloco A, Edifício Principal, 1 andar, CEP:70304-000, Brasilia, Distrito Federal 70304000, Brazil; patricia.bartholomay{at}


Objectives As part of smoking surveillance, the authors conducted an epidemiologic investigation in southern Brazil to identify the occurrence of Green Tobacco Sickness and risk factors for illness and to recommend control and prevention measures.

Methods A 1:2 case–control study matched by subjects' smoking habits. The study population was residents of Candelária, Rio Grande do Sul state, who farm tobacco and provided a urine sample for cotinine measurement by high-performance liquid chromatography. Confirmed cases were persons with compatible clinical presentation (headache, nausea, vomit, dizziness or weakness) and cotinine level >10 ng/ml. Controls were persons without compatible signs or symptoms. The association measure was the matched OR with 95% CIs and p<0.05.

Results Of 33 confirmed cases, 64% were men, average age was 33 years (SD ±11.8 years) and 57% were landowners. Cases have had similar illness in the past and were likelier to be workers hired by farmers–landowners than controls. Multivariate analysis yielded independent association between these variables and illness, controlled for age and sex. Contact with pesticides and working with wet tobacco leaves were not associated with illness.

Conclusions The authors confirmed Green Tobacco Sickness in southern Brazil; the authors recommend investigation of its prevalence in tobacco-growing regions and monitoring of and education about the disease and its prevention by occupational health authorities.

  • Public health
  • health surveillance
  • nicotine
  • epidemiology

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What this paper adds

GTS was first described in the medical literature in the 1970s in the USA Until 2007, there has been no confirmed occurrence of GTS in Brazil, although it is the largest tobacco grower in the world. To strengthen disease surveillance, the Brazilian Ministry of Health conducted epidemiologic investigations with the purpose of identifying the occurrence of this disease in two tobacco-farming regions, each with a distinct mode of farming. Our investigation took place in one county in southern Brazil, in the region in which 96% of tobacco production in Brazil took place. We confirmed the occurrence of GTS and identified several risk factors, demonstrating it is an occupational condition among tobacco farmers of this region. We recommend studies of its prevalence in tobacco-growing regions and monitoring and education about the disease and its prevention by occupational health authorities.


Green Tobacco Sickness (GTS), an occupational illness of tobacco harvesters, is an acute moderate intoxication caused by dermal absorption of nicotine. Its principal signs and symptoms are headache, nausea, vomiting, abdominal cramps, dizziness and weakness.1 In some cases, the condition presents more severe manifestations, requiring more protracted medical attention and hospitalisation.2 3

Published studies describe contact with moist tobacco leaves and failure to use personal protective equipment (PPE) during the harvest as risk factors for illness.3–6 Other factors such as contact between the tobacco plant and more absorptive body surfaces (forehead, back and axilae), consumption of alcoholic beverages and heat exposure at the time of harvest are considered facilitators of nicotine absorption.5 High levels of nicotine in smokers have been described in the literature as protective.

Diagnosis of GTS rests of the following triad: occurrence of compatible signs and symptoms, exposure to tobacco farming and elevated nicotine levels demonstrated by cotinine level measurement. The differential diagnosis includes pesticide intoxication and heat exhaustion. Cotinine is a nicotine metabolite found in urine, blood and saliva that serves as a useful marker for nicotine exposure because its half-life (20 h) is longer than that of nicotine (2 h).7

GTS was first described in the medical literature in the 1970s in the USA and later in other countries, including India and Malaysia.3 8 There has been no confirmed occurrence of GTS in Brazil, although it is the largest tobacco grower in the world, responsible for 13% of world production.9 10

To strengthen disease surveillance and implementation of the Framework Convention on Tobacco Control in Brazil,11 the Brazilian Ministry of Health conducted epidemiologic studies with the purpose of identifying the occurrence of this disease in two tobacco-farming regions, each with a distinct mode of farming. The first investigation took place in northeastern Brazil, in the township of Arapiraca, where tobacco has reportedly been grown since the end of the 19th century.12

Our epidemiologic investigation took place in one county in southern Brazil, in the region in which 96% of tobacco production in Brazil takes place. The objective was to confirm the occurrence of GTS, identify risk factors for illness and recommend control and prevention measures.


The investigation county was selected because it is located in Rio Grande do Sul state, the principle tobacco-producing part of Brazil, where harvest time coincided with the convenience of the investigators and the ready availability of healthcare to residents.

To identify potential risk factors for illness to be included in the investigation, we observed the tobacco production process in situ. We conducted a 1:2 matched case–control study of potential risk factors during the period 17 November to 31 December 2009; this period included the local tobacco harvest. Cases and controls were matched based on smoking status, to control for the impact of smoking on the occurrence of GTS; smoking is known to reduce GTS signs and symptoms,13 and nicotine-induced vasoconstriction in smokers reduces dermal nicotine absorption.3 A smoker was defined as a person who reported smoking, regardless of the number of cigarettes smoked daily.

Persons eligible to participate in the investigation were residents of the county of Candelária, who had worked in tobacco farming in the 7 days preceding the interview. A suspect case was defined as an individual presented with symptoms of acute intoxication (headache, nausea, vomiting, dizziness or weakness) in the 48 h preceding collection of a urine specimen. A confirmed case was defined as a suspect case with a urinary cotinine level >10 ng/ml. Controls were persons who presented no signs or symptoms of acute intoxication (headache, nausea, vomiting, dizziness or weakness) during the 7 days preceding interview.

Case finding was conducted with the support of the local health services: three family health teams, one basic health unit, one hospital and a team of community health agents. Additionally, we advertised the investigation during the Rural Workers' Unions radio program's broadcast on local AM and FM radio stations. Their main audience is farmers residing in the county. For convenience, controls were systematically sought in homes to the left and then to the right of case homes. When more than one eligible potential control was identified in a household, they were listed by alphabetical order, and the control was selected using a random numbers table. Homes without occupant present at the time of the team visit were substituted.

Data were collected using a standardised questionnaire about socioeconomic, individual and clinical characteristics, as well as questions on tobacco and pesticide exposure. The questionnaire was completed by the interviewer.

Association between biologically plausible exposures and illness was subjected to univariate analysis, and those exposures found to be statistically significantly associated with illness were assessed for independence by conditional logistic regression models. Data analysis was conducted in Epi-info 3.5.1. We used the McNemar χ2 test for categorical variables and the Kruskal–Wallis test for continuous variables in non-matched analyses. We calculated matched ORs and 95% CIs; p<0.05 was considered statistically significant.

Urine samples from suspect cases and controls were tested for cotinine by high-performance liquid chromatography, with ultraviolet detector. The alkaline urine was extracted with an dichloromethane.7 Cut-off levels were <20 ng/ml for non-smokers and ≥20 ng/ml for smokers, as established by the testing laboratory (Laboratório de Toxicologia da Pontifícia Universidade Católica do Rio Grande do Sul, Porto Alegre, Brazil). The lower limit of quantitative determination for the assay was 10 ng/ml and the lower limit of detection was 5 ng/ml.

The Ministry of Health determined in advance that the proposed investigational activities fell within the scope of normal measures of disease monitoring and response to an urgent health event and that therefore they did not constitute human subject research requiring National Ethics Committee oversight. Verbal consent was obtained from all subjects as a condition of enrolment. Confidentiality of personal data was assured under Brazilian law, CNS Decree 196/96.


The investigation was conducted in the county of Candelária (population 29 444), Rio Grande do Sul state.14 It is the fourth most significant tobacco-producing county in the state; 7500 hectares are planted annually, yielding approximately 15 000 tons of tobacco each year. Production takes place in small plots owned and farmed by ∼4000 families, which represent 54% of the county's population. Candelária and 18 other counties produce 60% of the tobacco grown in the state of Rio Grande do Sul.10

The harvest phase lasts about 4 months and entails the greatest exposures of farmers to tobacco leaves through the breaking of the leaves at harvest, carrying them in bundles to storage and preparing leaves for drying in ovens. All persons who met the study case and control definitions completed the questionnaire and provided urine samples.

Forty-six persons met the suspect case definition, of which 33 (71.7%) were confirmed by cotinine-level test results. Of confirmed cases, 17 (51.5%) were men, average age was 33 (SD ±11.8 years), 25 (75.8%) were white complexion/race, six (18.2%) mixed race and two (6.1%) black. Education level ranged from 0 to 11 years (median, four). Twelve (36.4%) were smokers, 19 (57.6%) never-smokers and two (6.0%) were ex-smokers, for at least 1 year (table 1). Of the 21 non-smokers and ex-smokers, 13 (61.9%) reported residing or working with a smoker, thus being passive smokers.15 Regarding alcohol intake, 14 (42.4%) reported alcoholic beverage consumptions, of which nine (64.3%) reported consumption of alcohol at least once per week (table 1).

Table 1

Characteristics of cases and controls, Candelária/Rio Grande do Sul State /Brazil, 2008, N=99

The principal occupation of 81.8% of all subjects was farming, with 57.6% being hired by a landowner, that is, they farmed tobacco on land owned by someone else (table 1). The distribution of cases was homogenous by workplace within the rural tobacco-farming zone of the county. Rural county areas without cases were those where rice was grown (figure 1).

Figure 1

Green Tobacco Sickness cases, Candelária/Rio Grande do Sul State/Brazil, 2008.

Principal signs and symptoms among cases were nausea (93.9%), dizziness and vomiting (72.7%), headache (66.7%), abdominal cramps (57.6%), weakness (54.5%), tremor (18.1%) and diarrhoea (12.1%). The median duration of signs and symptoms was 21 h (range: 3–80 h). Of 24 persons who sought medical treatment, two (8.3%) required observation for 24 h and four (17.7%) required observation for up to 4 h.

Of the 33 cases, 20 (60.6%) reported working with wet tobacco and 24 (72.7%) reported that their clothes became wet from dew, rain or sweat. Rainfall was low during the investigation period, with rain on 4 days of the investigation period.

During the harvest, 97.0% of cases engaged in breaking tobacco leaves, 33.3% stitched the collected leaves, 24.2% carried bundled leaves and 15.1% stored leaves in ovens. Most (93.9%) cases reported no contact with pesticides during the 7 days before they experienced onset of symptoms of acute intoxication. Among the 28 cases who knew the date of the last application of pesticides to the crops on which they worked, 78.6% reported that pesticides had been applied more than 96 h before the onset of signs and symptoms of illness.

Clothing most commonly used by confirmed cases included flip-flop sandals (57.6%), long trousers (54.5%), long shirt (63.6%) and hat (100.0%). Using cotton gloves was reported by 48.5% of cases. No farmer reported using PPE.

Acute intoxication in previous years was reported by 69.7% of cases, 56.5% reported illness every year in which they were engaged in tobacco farming and 95.6% reported experiencing the illness during the harvest.

In the case–control study, 66 controls were matched to cases. On univariate analysis, cases were likelier than controls to have experienced acute intoxication symptoms in previous years during tobacco harvest and to be hired workers. Among those who reported illness in previous years, cases were likelier to have been ill more than once. Among those who were engaged in breaking the tobacco leaves, cases harvested a lesser portion of leaves from the superior part of the plants (table 2). On univariate analysis, illness was not associated with exposure to wet leaves or clothes, using conventional clothes or accessories for personal protection, consuming alcohol, working for the first time in tobacco farming, male sex, age <33 and contact with pesticides (table 2).

Table 2

Univariate analysis of risk factors for Green Tobacco Sickness, cases vs. controls, Candelária/Rio Grande do Sul State/Brazil, 2008, N=99

The following variables were included on conditional logistic regression analysis: having acute intoxication symptoms after contact with tobacco in previous years, being a hired farm worker and harvesting superior tobacco leaves, all adjusted for age and sex. This analysis yielded results consistent with those of univariate analysis (table 3).

Table 3

Conditional logistic regression analysis of risk factors for Green Tobacco Sickness, Candelária/Rio Grande do Sul State/Brazil, 2008, N=99

On non-matched analysis of continuous variables, there was a statistically significant difference between average levels of cotinine of cases and controls. Stratifying by smoking use (smoker vs non-smoker), there was a statistically significant difference between average levels of cotinine in cases and control, who were non-smokers (table 4).

Table 4

Average levels of cotinine in cases and controls, Candelária/Rio Grande do Sul State/Brazil, 2008


GTS is often mistaken for the clinically similar intoxication by pesticides, which are used in several phases of tobacco farming.3 In our investigation, however, acute intoxication cases occurred during the harvest phase, which entails the least application of pesticides. Thus, pesticide exposure was not related to illness.

Cotinine levels in individuals presenting GTS were heterogeneous, even when stratifying smokers versus non-smokers. These differences could be explained by factors that modify cotinine levels that have been described in the scientific literature. These include the time between last cigarette smoked and urine collection, the number of cigarettes smoked per day, nicotine tolerance, time from last exposure to sample collection and body mass index.4 16–18 Even so, a statistically significant difference was noted between cotinine levels of case-patients and controls, resembling the findings of the stuy of Pereira et al, although the type of tobacco to which cases were exposed was different.12

The farm workers interviewed in this investigation considered the use of conventional clothes and accessories during labour as a form of protection against GTS. However, our results demonstrate that these clothes conferred no protection. Use of cotton gloves was reported as a measure to protect hands from physical injury and facilitate their cleaning after work; no farm worker reported spontaneously that they were used as PPE. Use of proper PPE would be difficult for tobacco farmers, given that it must be impermeable and tolerable to users at the high ambient temperatures characteristic of harvest season.4 While PPE use has been recommended by other authors, it must be acceptable to the workers.

Tobacco farming in southern Brazil is a family enterprise, entailing the participation of all family members. This differs from the situation in the USA, where GTS is described in male foreign migrant workers from Latin countries and in the first reported outbreak of GTS in Brazil, from a region where the activity is more common among men than women.3 5 12 All family members are involved in the breaking of leaves at harvest, the activity entailing the greatest exposure to nicotine. Our conditional logistic regression analysis confirmed this finding, controlled for the age and sex of the subjects.

Trape-Cardoso et al19 report that leaves from the upper portion of the plant possess higher concentration of nicotine than the remainder of the plant. We observed that farmers collecting leaves from the upper portion of the plant were less likely to experience illness than those who collected leaves from other parts of the plant. This could be explained by the upper leaves being larger, requiring a lesser number of leaves to fill a drying oven, with the pickers consequently having reduced time of exposure to leaves; this is supported by other reports in the literature. Tobacco farmers relate that the smell of tobacco is stronger during the picking of lower leaves, due to the proximity to the soil, presumably due to reduced air circulation by the ground, facilitating the stimulation of nausea, one of the signs of acute intoxication.20 Furthermore, picking lower leaves requires a greater effort, presumably associated with greater perspiration and consequently increased nicotine absorption.

We observed that worker, who were hired, as opposed to those who owned the land, were likelier to experience illness, similar to findings that Latino workers in the USA were at greatest risk for presenting with GTS.5 21 It is possible that this may be related to the lesser ability that these persons had to choose their working hours. Farmer–owners reported that they avoided working during the hottest hours of the day and at the times that tobacco leaves were wet, circumstances they recognise as associated with the illness.

Although the association between GTS and exposure to wet tobacco leaves or working with wet clothes is well documented,22 we did not demonstrate this type of association.5 6 16 23 This may be due to the low power of the study, especially since the investigators heard many reports by farm workers about the occurrence of illness after contact with wet leaves or in wet clothes. Additionally, the occurrence of rain on only 4 days during the investigation period was considered exceptionally low compared with previous harvest seasons.

We believe that the true magnitude of illness in the region is greater than that shown in our investigation. Local farmers consider the signs and symptoms of GTS as inevitably related to their work and report frequently opting for self-treatment, rather than seeking medical care, in order to quickly return to work. This is likely encouraged by the self-limiting nature of the illness.

Possible limitations of our study include memory and information biases occurring during interviews with tobacco farmers, principally among controls. We believe that selection bias may have occurred because controls were not recruited on the same days as cases, but this was likely minimised by the fact that weather conditions did not vary during the investigation period and that cotinine levels are not cumulative. The number of cases and controls in our study was a limiting factor, not permitting the inclusion of more than three independent variables in conditional logistic regression models. Our study did not control for environmental tobacco smoke. It is possible that this exposure could have increased cotinine levels in some cases and controls.

Our investigation confirmed the occurrence of green leaf tobacco sickness in southern Brazil, demonstrating it is an occupational condition among tobacco farmers of this region.

In many studies of GTS from other countries, variation persists in the clinical elements of the case definition,3 19 limiting comparisons of disease incidence and prevalence between studies that used a cohort design. The combination of clinical presentation with cotinine level measurement in urine, blood or saliva should allow more precise estimates; we recommend including cotinine measurement in similar studies in the future.

Since we used a case–control study design, we cannot estimate incidence or prevalence of disease in southern Brazil; this is true of the only other study on this disease in Brazil.12 Studies permitting the determination of GTS prevalence should be conducted in tobacco-growing populations in Brazil, so that the magnitude of the problem can be known. Furthermore, monitoring of the disease by occupational health authorities is necessary, so that appropriate education of the affected populations with regard to control measures, such as use of PPE appropriate for tobacco farming, may take place.

Signs and symptoms of GTS usually resolve within a day, but we do not know whether there are chronic effects associated with daily transdermal nicotine absorption over years in the course of tobacco farming activities. Studies are needed to investigate potential chronic effects in these workers, who are exposed to high doses of nicotine in the course of their working lives.


We thank the following entities and persons: the State Center for Health Surveillance of Rio Grande do Sul, the Central State Laboratory (LACEN) of Rio Grande do Sul, the Municipal Secretariat of Health of Candelária and the Benevolent Hospital Society of Candelária for essential support; the Union of Rural Workers of Candelária, Rio Grande do Sul for assistance during field work; Eduardo Marques Macário and Wildo Navegantes de Araújo for scientific advice and Carlos Eduardo Leite for help with laboratory testing.



  • Competing interest None declared.

  • Patient consent The work described in this paper was conducted as part of a public health response to an outbreak of acute toxicity in tobacco workers. Oral consent was required for all individuals at the moment of the interviews.

  • Ethics approval A public health response to an outbreak of acute toxicity in tobacco workers. These activities were reviewed in advance by the officials within the Ministry of Health of Brazil charged with determining whether actions by public health workers fall under the jurisdiction of a research ethics committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.