Association Between Diabetes and Pesticides: A Case-Control Study Among Thai Farmers

Juntarawijit and Juntarawijit Environmental Health and Preventive Medicine
(2018) 23:3
Environmental Health and
DOI 10.1186/s12199-018-0692-5 Preventive Medicine
RESEARCHARTICLE Open Access
Association between diabetes and

pesticides: a case-control study among Thai
farmers
1* 2
Chudchawal Juntarawijit and Yuwayong Juntarawijit
Abstract
Background: Pesticides are an agricultural chemical suspected to be a significant contributor to a global
diabetes pandemic. The purpose of this study was to confirm previous findings of the link between
diabetes and some agricultural pesticides and to identify the particular pesticides that are most likely to
pose a risk of diabetes in the community.
Methods: A population-based case-controlled study was conducted among residents in the Bang Rakam
district of Phitsanulok Province in Thailand. Lifetime pesticide exposure and other relevant data were
collected from 866 participating cases with diabetes mellitus and 1021 healthy controls.
Results: After adjusting for gender, age, BMI, cigarette smoking, alcohol consumption, family history of diabetes, and
occupation, it was found that the prevalence of diabetes was positively associated with exposure to all types of pesticides,
including insecticides, herbicides, fungicides, rodenticides, and molluscicides, with exposure to rodenticides being
statistically significant (OR = 1.35; 95%CI 1.04–1.76). Among 35 individual brand-named pesticides investigated, we found
statistically significant ORs with three insecticides, including one organochlorine [endosulfan (OR = 1.40; 95%CI 1.01–1.95)],
one organophosphate [mevinphos (OR = 2.22; 95%CI 1.17–4.19)], and one carbamate [carbaryl/Sevin (OR = 1.50; 95%CI
1.02–2.19)]; and one fungicides [benlate (OR = 2.08; 95%CI 1.03–4.20)].
Conclusions: Our results suggest that the occurrence of diabetes among Thai farmer was associated with
pesticide exposure. This finding is in line with previous epidemiological and animal studies. Further study using
a larger sample size is needed to confirm the relationship and to identify the more toxic compounds.
Keywords: Diabetes, Pesticides, Environmental exposure, Insecticides, Herbicides
Background sleeping patterns [6, 7], and pregnancy status [8]. Recent
Diabetes is an important public health problem with 415 research has also suggested environmental contamin-ation
million adults (aged 20–79 years) worldwide suffering the by agricultural chemicals to be a causal factor [9]. The
disease in 2015. By 2040, the figure will rise to 642 suspected compounds are arsenic, mercury, phthalates,
million [1]. In Thailand, diabetes among adult aged ≥ 18 bisphenol A (BPA), and persistent organic pollutants
years drastically increased from 6.9% in 2009 to 8.9% (POPs), especially PCBs, dioxin, and pesti-cides [10–12].
(4.8 million people) in 2014 [2]. The confirmed causal
factors were gender, genetic, and overweight [3]. In Pesticides have received greater attention as they are the
addition, the disease was also strongly associated with most commonly compounds in use today, with more than
many lifestyle behaviors, including lack of active exercise 2.3 billion kilograms used worldwide in 2001 [13]. In an
[4], cigarette smoking [5] and alcohol consumption [3], Agricultural Health Study (AHS) project in the USA, a
large cohort study of 33,457 pesticide applicators reported
* Correspondence: [email protected]
1
an association between diabetes and seven organochlorine
Department of Natural Resources and Environment, Faculty of
Agriculture, Natural Resources and Environment, Naresuan
pesticides, including aldrin, chlordane, heptachlor,
University, Muang District, Phitsanulok 65000, Thailand dichlorvos, trichlorfon, alachlor, and
Full list of author information is available at the end of the article
© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Juntarawijit and Juntarawijit Environmental Health and Preventive Medicine (2018) 23:3 Page 2 of 10
cyanazine [14]. Further study of this program using data Study subjects
of the wives of 13,637 pesticide applicators identified The participants were medical diagnosed diabetic patients
more diabetic risk compounds, including three organo- from seven local, sub-district hospitals in Bang Rakam
phosphates, including fonofos (OR = 1.56; 95%CI 1.11– district. The 7 hospitals were randomly selected from the
2.19), phorate (OR = 1.57; 95%CI 1.14–2.16), and para- 21 sub-district hospitals in the target area. From the 2832
thion (OR = 1.61; 95%CI 1.05–2.46); one organochlorine, patients who sought out-patient follow-up service from
dieldrin (OR = 1.99; 95%CI 1.12–3.54); and one herbi- the selected 7 hospitals, 1000 cases (approximately 35%)
cide, 2,4,5-T/2,4,5-TP (OR = 1.59; 95%CI 1.00–2.51) were randomly selected in propor-tional manner from the
[15]. Study among women who were involved in mixing hospitals for initial interview. The controls were close
or applying pesticides, or fixing pesticide appliances, also neighbors who were without diabetes, having the same
found diabetes and pesticide exposure with odds ratios of gender and aged within ± 5 years of the cases’ age. Due to
2.2 (95%CI 1.5–3.3) [16]. From these studies, it is evident missing information on key data for some of the initial
that gender plays a significant role, as well as the several 2100 patients, such as pesticides usage, age, weight, and
other factors which influence diabetic occur-rence, behavior, the final data analysis was conducted among
including race, ethnicity and economic status, education 1887 subjects (866 cases and 1021 controls).
level, and nutritional conditions [17]. A study among
Mexican-American farmers found an association with This study was approved by the Ethics Board of
several organochlorine pesticides, including trans- Naresuan University (project number 402/59). Written
nonachlor, oxychlordane, β-hexachlorocyclohexane, p,p informed consent for an interview was obtained from each
′-DDT, and p,p′-DDE [18]. Another study using data from subject.
Native Americans found PCBs, dichlorodiphenyldi-
chloroethylene (DDE), and hexachlorobenzene (HCB) to Questionnaire
be associated with diabetes [19]. Data on pesticide exposure was collected using a struc-
With currently limited information, several mechanisms tured questionnaire. Questions were derived from the
have been suggested to explain the linkage between envir- Agricultural Health Study Questionnaire (https://
onmental compounds and diabetes. In animal and cell aghealth.nih.gov/collaboration/questionnaires.html). In-
studies, it was found that organochlorine pesticide and formation collected included demographic data, occupa-
some other POPs reduce cell metabolic functions and tion, health behaviors, body weight, and height. Body
cause obesity [20]. There has been evidence that pollutant mass index (BMI) was calculated as weight (kg) divided
compounds affect insulin secretion, insulin action, and 2
by height squared (m ). Height and weight of partici-pants
glucose homeostasis [21–23], and elsewhere, were measured by interviewers. For eating behav-ior,
physiological and oxidative stress, and cholinesterase amount and frequency of eating vegetable, fruit, and
inhibition [24]. Recently, Park et al. [25] proposed a sweet were collected. The amount of food consumption
molecular scheme to explain how these chemicals caused was categorized to less than a cup/glass, about a cup/
metabolic disorders by damaging the mitochondria, which glass, and more than a cup/glass. Data on eating fre-
are organelles that act as powerhouses of the cells. quency was categorized to never eating or eat less than
Information on diabetes and pesticide linkage is rela- once a month, once a month, 2–3 times per month, once a
tively new, and a significant data gap still remains [12]. week, twice a week, 3–4 times per week, 5–6 times per
This purpose of this case-control study was to identify week, once a day, and twice a day or more. A
more specific compounds which are, or may be, associ- consumption rate per week (cup/glass per week) was then
ated with diabetes, using data on Thai rural residents. calculated, and the quartile of this rate was used for
With the differences in the health backgrounds, life styles, prediction of diabetes risk. Data on lifestyle (morning-
and agricultural practices and behaviors of this particular ness-eveningness) was also collected. It is a self-reported
group, the results can be compared to those previously behavior. Morning types (larks) referred to those who are
reported in the literature which has mostly been usually active during the day and wake up early in the
conducted in the Western world. morning. Evening types (owls) are those who feel alert in
the night and wake up late in the morning.
Lifetime pesticide use and use type, mixing or apply-
Methods ing, were analyzed. Pesticides were categorized into five
This study was a population-based case-control study. It groups: insecticides (organochlorine, organophosphate,
was conducted from February to May 2016 in Bang carbamate, and pyrethoid), fungicides, herbicides, roden-
Rakam district, located in Phitsanulok Province, 377 km ticides, and molluscicides. For each pesticide, data col-
north of Bangkok, Thailand. The district has a popula-tion lected were numbers of years using pesticides (< 1, 2–5,
of 94,980, most of whom are rice farmers [26]. 6–10, 11–20, and > 20 years) and numbers of days in
each year using pesticides (< 5, 5–9, 10–19, 20–39, 40– (morningness-eveningness), ever exercise during the past
59, 60–150, and > 150 days). Cumulative days of expos- 12 months, and frequency of TV watching during leisure
ure were calculated from duration by frequency data using time; and an amount of vegetable, fruit, and sweet intake.
the midpoints of the reported categories. For each group Association between different types of pesticide and
of pesticides, the exposure duration was divided into diabetes were presented in Table 2. It was found that ever use
quartiles of cumulative days of exposure. Among all types of any pesticides cannot predict risk of diabetes. Experience
of pesticides, exposure data of 35 specific pesti-cides was using any types of pesticides was positively associated with
collected. diabetes but the association were statisti-cally significant
The study subjects were interviewed by 50 village only for rodenticide (adjusted OR = 1.35; 95%CI 1.04–1.76).
health volunteers, 5–10 in each sub-district. Prior to data For insecticides, year of exposure to insecticides and
collection, all the health volunteers were trained on how cumulative days failed to predict risk of diabetes. For
to interview and use the questionnaire. fungicides, diabetic risk seems to correlate with both year of
exposure and exposure days, but none with statistical
Statistical analysis significance. For herbicides and mollusci-cides, both year of
Collected data was analyzed using IBM SPSS Statistics exposure and exposure days were inversely associated with
(Version 19) and OpenEpi (Version 3.01). P values < 0.05 the disease. For rodenticides, only exposure days was
were considered statistically significant. In the asso- positively correlated with diabetes.
ciation between diabetes and pesticides exposure, odds Table 3 presents the ORs of each individual pesticide
ratio (OR) was calculated for each group of pesticides for diabetes. Among 20 compound investigated, 15 were
(insecticides, herbicides, fungicides, rodenticides, and positively associated with diabetes, 3 with statistical sig-
molluscicides) and 35 individual compounds. Both crude nificance [endosulfan (OR = 1.40; 95%CI 1.01–1.95),
and adjusted ORs with 95% confidence intervals (CIs) mevinphos (OR = 2.22; 95%CI 1.17–4.19), and carbaryl/
were presented. Adjusted ORs were analyzed using Sevin (OR = 1.50; 95%CI 1.02–2.19)]. For herbicides, of
multiple logistic regression controlled for gender, age the 6 compounds investigated, 4 showed increased ORs.
(continuous), BMI (continuous), cigarette smoking Study also found 7 out of 9 fungicides to be correlated
(packs/year), alcohol consumption (glass/week), family with diabetes, 1 with statistical significant [benlate (OR =
history of diabetes (yes, no), and occupation (farmer, non- 2.08; 95%CI 1.03–4.20).
farmer, other). The control variables were those with
difference between cases and controls in addition to the Discussion
fundamental confounding factors. In this study, the OR of diabetes was statistically associ-
For exposure duration to a group of pesticides, cumu- ated with three types of insecticides investigated,
lative exposure days were computed and categorized into including one organochlorine (endosulfan), one
quartile of exposure days. The diabetic risk was then organophosphate (mevinphos), and one carbamate
predicted, using quartile 1 as a reference. For each (carbaryl/Sevin), as well as one fungicide (benlate). The
specific pesticide, exposure was categorized as ever vs. results were well sup-ported by previous findings either
never used. The cumulative exposure days were not per- from epidemiological studies or an animal study. A recent
formed because the number of subject reported using each study also suggested that pesticides could affect the
individual pesticide was too small. pancreases and interfere with insulin secretion [27] by
damaging the mitochondria cells [27].
Results Of the six organochlorine pesticides investigated, three
Both case and control groups had a similar demographic [chlordane (OR = 1.35; 95%CI 0.56–3.24), endosulfan
data with a comparable proportion of men, 35% for con- (OR = 1.40; 95%CI 1.01–1.95), and heptachlor (OR =
trol and 33% for case (P = 0.36) (Table 1). Comparing 1.13; 95%CI 0.46–2.80)] were positively associated with
case to control, the mean age (57.68 ± 9.00 vs. 57.64 ± diabetes. This finding was consistent with a study which
9.17) and age in each categories did not differ signifi- used US National Health and Examination Survey
cantly (P = 0.99). However, case had a higher average (NHANES) data that reported a strong association
BMI (25.67 ± 6.99) than control (24.06 ± 6.04). Most of between diabetes and PCB, two dioxin congeners, and
the study subjects are farmer, 78.9% for case and 75.3% three organochlorine pesticides (oxychlordane, DDE, and
for control. An average number of alcohol consumption of trans-nonachlor) [10]. A study in the USA under the
case was 7.16 ± 2.34 glass per year, significantly higher Agricultural Health Study program also found diabetes to
than that of control (6.85 ± 2.60) (P < 0.01). However, be correlated with seven organochlorine pesticides (aldrin,
case smoked cigarette in less amount (1.41 ± 5.80 vs 2.56 chlordane, heptachlor, dichlorvos, tri-chlorfon, alachlor,
± 9.36 pack/year, case to control) (P < 0.01). The two and cyanazine) [14]. Compared with these previous
groups of subjects had no difference in lifestyle results, ORs of chlordane in our study
Table 1 General characteristics and variables between case and Table 1 General characteristics and variables between case and
control groups control groups (Continued)
Characteristic Case Control P value** Characteristic Case Control P value**
(n = 866)* (n = 1021) (n = 866)* (n = 1021)
Men (%) 286 (33.0) 357 (35.0) 0.38 10–12 54 (6.2) 68 (6.7)
Age (years) 0.99 Missing 43 (5.0) 37 (3.6)
≤ 44 57 (6.6) 71 (7.0) Mean ± s.d. 7.92 ± 1.12 7.93 ± 1.13 0.85
45–54 246 (28.4) 293 (28.7) Lifestyle (morningness-eveningness)*** 0.77
55–64 373 (43.1) 435 (42.6) Morning types (larks) 799 (92.3) 948 (92.9)
65–74 162 (18.7) 186 (18.2) Evening types (owls) 53 (6.1) 67 (6.6)
≥ 75 28 (3.2) 36 (3.5) Missing 14 (1.6) 6 (0.6)
Mean age ± s.d. 57.68 ± 9.00 57.64 ± 9.17 0.92 Ever exercise during the past 12 months 0.60
Median (min-max) 58.0 (25–84) 58.0 (15–86) Yes 319 (36.8) 387 (37.9)
BMI (kg/m2) < 0.01 No 535 (61.8) 617 (60.4)
< 18.50 27 (3.1) 66 (6.5) Missing 12 (1.4) 17 (1.7)
18.50–22.90 227 (26.2) 379 (37.1) Watching TV during leisure time 0.43
23.00–24.90 178 (20.6) 182 (17.8) Never 34 (3.9) 37 (3.6)
25.00–29.90 283 (32.7) 279 (27.3) Rarely 147 (17.0) 155 (15.2)
> 30.00 115 (13.3) 76 (7.4) Sometimes 331 (38.2) 368 (36.0)
Missing 36 (4.2) 39 (3.8) Very often 254 (29.3) 337 (33.0)
Mean ± s.d. 25.70 ± 6.94 24.06 ± 6.04 < 0.01 Always 54 (6.2) 58 (5.7)
Median (min-max) 24.8 (13.8–141.7) 23.5 (13.8–166.7) Missing 46 (5.3) 66 (6.5)
Occupation 0.02 Intakes of vegetables (cups per week) 0.72
Farmer 650 (75.1) 805 (78.8) Quartile1 126 (14.5) 164 (16.1)
Non-farming private business 164 (18.9) 145 (14.2) Quartile2 315 (36.4) 358 (35.1)
Other 49 (5.7) 70 (6.9) Quartile3 197 (22.7) 238 (23.3)
Missing 3 (0.3) 1 (0.1) Quartile4 177 (20.4) 196 (19.2)
Cigarette smoke < 0.01 Missing 51 (5.9) 65 (6.4)
Never smoked cigarettes 719 (83.0) 798 (78.2) Mean cups per week ± s.d. 12.40 ± 11.58 12.00 ± 11.21 0.46
Have smoked cigarettes 62 (7.2) 67 (6.6) Intakes of fruits (cups per week) 0.95
Current smoker 77 (8.9) 144 (14.1) Quartile1 254 (29.3) 296 (29.0)
Missing 8 (0.9) 12 (1.2) Quartile2 263 (30.4) 314 (30.8)
Mean pack/year ± s.d. 1.41 ± 5.80 2.56 ± 9.36 < 0.01 Quartile3 148 (17.1) 163 (16.0)
Median pack/year 0.0 (0.0–75.0) 0.0 (0.0–120) Quartile4 149 (17.2) 178 (17.4)
(min-max)
Missing 52 (6.0) 70 (6.9)
Alcohol consumption < 0.01
Mean cups per week ± s.d. 5.44 ± 6.91 5.29 ± 6.76 0.64
Never drank 616 (71.1) 647 (63.4)
Intakes of sweets (cups per week) 0.22
Former drinker 112 (12.9) 112 (11.0)
Quartile1 65 (7.5) 67 (6.6)
Current drinker 128 (14.8) 244 (23.9)
Quartile2 316 (36.5) 352 (34.5)
Missing 10 (1.2) 18 (1.8)
Quartile3 159 (18.4) 223 (21.8)
Mean glasses per week 3.11 ± 11.6 4.8 ± 14.4 < 0.01
± s.d. Quartile4 280 (32.3) 311 (30.5)
Median glasses per week 0.0 (0.0–84.0) 0.0 (0.0–84.0) Missing 46 (5.3) 68 (6.7)
(min-max) Mean cups per week ± s.d. 2.91 ± 4.56 2.98 ± 4.57 0.75
Having family history of diabetes < 0.01 *N was 1021 for control and 866 for case unless otherwise indicated
**χ2 test for categorical data; t test for continuous data
Yes 500 (57.7) 365 (35.7)
***Morning types (larks) were those who wake up early and active during the
No 355 (41.0) 643 (63.0) day; evening types (owls) were usually feel alert in the night and wake up late
in the morning
Missing 11 (1.3) 13 (1.3)
Hours of sleep per day 0.95
4–6 77 (8.9) 90 (8.8) were higher (1.35 vs 1.16 in the previous study) but
7–9 692 (79.9) 826 (80.9) lower for heptachlor (1.13 vs 1.20). Our study is the first
to show that endosulfan is strongly associated with
Table 2 Associations between different types of pesticides and diabetes

Case Control OR (95%CI) Adjusted OR*
(95%CI)
Exposed n % Exposed n %
Pesticides (any) (N = 1867) 468 54.5 556 55.1 0.98 (0.81–1.17) 0.94 (0.75–1.17)
Insecticides (any) (N = 1876) 433 50.2 500 49.4 1.03 (0.86–1.24) 1.04 (0.84–1.29)
Number of years using the insecticides (N = 933)
1–5 50 11.6 53 10.6 Referent Referent
6–10 82 18.9 102 20.4 0.85 (0.53–1.38) 0.83 (0.49–1.41)
11–20 99 22.9 126 25.2 0.83 (0.52–1.33) 0.67 (0.40–1.12)
21–30 105 24.3 110 22.0 1.01 (0.63–1.62) 1.01 (0.61–1.68)
> 30 97 22.4 109 21.8 0.94 (0.59–1.51) 0.80 (0.47–1.34)
P for trend*** 0.78 0.80
Number of days using insecticides (quartile) (N = 920)
Quartile1 119 27.5 128 25.6 Referent Referent
Quartile2 99 22.9 116 23.2 0.92 (0.64–1.32) 1.02 (0.68–1.51)
Quartile3 113 26.1 130 26.0 0.94 (0.66–1.33) 0.83 (0.56–1.23)
Quartile4 97 22.4 118 23.6 0.88 (0.61–1.28) 0.86 (0.58–1.28)
P for trend 0.12 0.18
Fungicides (any) (N = 1881) 336 38.9 387 38.0 1.04 (0.86–1.25) 1.00 (0.80–1.24)
Number of years using the fungicides (N = 723)
6–10 65 19.4 88 22.7 0.88 (0.53–1.47) 0.87 (0.50–1.51)
11–20 83 24.7 105 27.1 0.95 (0.58–1.54) 0.93 (0.55–1.58)
21–30 80 23.8 69 17.8 1.39 (0.84–2.30) 1.55 (0.90–2.69)
> 30 62 18.5 70 18.1 1.06 (0.63–1.78) 1.03 (0.58–1.82)
Number of days using fungicides (quartile) (N = 672)
Quartile2 83 24.7 93 24.0 1.09 (0.73–1.62) 1.06 (0.69–1.64)
Quartile3 72 21.4 70 18.1 1.25 (0.82–1.92) 1.13 (0.70–1.80)
Quartile4 62 18.5 79 20.4 0.96 (0.62–1.47) 1.02 (0.64–1.62)
Herbicides (any) (N = 1882) 418 48.4 491 48.2 1.01 (0.84–1.21) 1.13 (0.76–1.67)
Number of years using the herbicides (N = 909)
6–10 83 19.9 118 24.0 0.72 (0.45–1.14) 0.60 (0.36–0.99)**
11–20 102 24.4 113 23.0 0.92 (0.58–1.45) 0.74 (0.45–1.22)
21–30 99 23.7 104 21.2 0.97 (0.61–1.54) 0.84 (0.51–1.38)
> 30 79 18.9 100 20.4 0.80(0.50–1.29) 0.63(0.37–1.06)
Number of days using herbicides (quartile) (N = 844)
Quartile2 76 18.2 87 17.7 1.00 (0.68–1.48) 0.90 (0.59–1.37)
Quartile3 96 23.0 109 22.2 0.99 (0.69–1.43) 0.93 (0.62–1.38)
Quartile4 92 22.0 118 24.0 0.91 (0.64–1.30) 0.84 (0.57–1.26)
Table 2 Associations between different types of pesticides and diabetes (Continued)

Case Control OR (95%CI) Adjusted OR*
(95%CI)
Exposed n % Exposed n %
Rodenticides (any) (N = 1886) 227 26.3 223 21.8 1.28 (1.04–1.58) 1.35 (1.04–1.76)**
Number of years using the rodenticides (N = 450)
6–10 60 26.4 63 28.3 0.77 (0.48–1.25) 0.81 (0.48–1.37)
11–20 45 19.8 43 19.3 0.85 (0.50–1.44) 1.03 (0.57–1.84)
21–30 19 8.4 24 10.8 0.64 (0.32–1.27) 0.87 (0.40–1.88)
> 30 24 10.6 29 13.0 0.67 (0.36–1.26) 0.83 (0.42–1.66)
Number of days using rodenticides (quartile) (N = 425)
Quartile2 68 30.0 56 25.1 0.99 (0.60–1.64) 1.05 (0.61–1.83)
Quartile3 37 16.3 52 23.3 0.58 (0.33–1.01) 0.75 (0.41–1.38)
Quartile4 46 20.3 48 22.0 0.78 (0.45–1.35) 1.09 (0.60–1.99)
Molluscicides (any) (N = 1879) 243 28.2 255 25.1 1.17 (0.95–1.44) 1.26 (0.96–1.64)
Number of years using the molluscicides (N = 498)
6–10 70 28.8 74 29.0 0.88 (0.56–1.40) 0.91 (0.55–1.51)
11–20 50 20.6 63 24.7 0.74 (0.45–1.21) 0.81 (0.47–1.39)
21–30 23 9.0 26 10.2 0.83 (0.43–1.58) 0.88 (0.43–1.80)
> 30 24 9.9 21 8.2 1.07 (0.55–2.09) 1.20 (0.58–2.48)
Number of days using molluscicides (quartile) (N = 471)
Quartile2 37 15.2 34 13.3 0.93 (0.53–1.61) 1.00 (0.54–1.83)
Quartile3 62 25.5 72 28.2 0.73 (0.47–1.15) 0.76 (0.47–1.25)
Quartile4 38 15.6 54 21.2 0.60 (0.36–1.00)** 0.70 (0.40–1.23)
P for trend 0.01** 0.07
*Adjusted for gender, age (continuous), BMI (continuous), cigarette smoking (packs/year), alcohol consumption (glasses/week), family history of diabetes (yes, no),
and occupation (farmer, non-farmer, other)
**Statistically significant (P < 0.05)
***P values for linear trend were derived using a continuous variable with the midpoint value of each category
diabetes. However, although endosulfan has long been methamidophos (OR = 1.04; 95%CI 0.65–1.65). This is
banned in the Western world a long time ago, being one of consistent with a study in the USA by Montgomery et al.
the most toxic POPs, it was still widely used in Thailand [14], which reported an increased ORs of diabetes among
until quite recently. In an animal study, expos-ure to subject exposure to chlorpyrifos (OR = 1.03; 95%CI
endosulfan was shown to affect the endocrine cells of the 0.91– 1.17), coumaphos (OR = 1.26; 95%CI 1.03–1.55),
pancreas of a rabbit and also that it decreased insulin dichlor-vos (OR = 1.21; 95%CI 0.98–1.49), phorate (OR
secretion [27]. = 1.22; 95%CI 1.06–1.42), terbufos (OR = 1.17; 95%CI
For organophosphates, in our study, specific compounds 1.02–1.35), and trichlorfon (OR = 1.85; 95%CI 1.03–
found to be positively associated with diabetes were chlor- 3.33). The ORs of chlorpyrifos and dichlorvos in our
pyrifos (OR = 1.20; 95%CI 0.83–1.75), dicrotophos (OR = study were higher than those in the previous study. This
2.12; 95%CI 0.97–4.61), dichlorvos (OR = 1.03; 95%CI issue will be discussed later. In an animal study, it was
0.41–2.62), EPN (ethyl-p-nitrophenyl) (OR = 1.29; 95%CI found that organophosphate inhibits acetylcholinester-ase
0.83–2.02), mevinphos (OR = 2.22; 95%CI 1.17–4.19), and causes an accumulation of acetylcholine and reduced
monocrotophos (OR = 1.29; 95%CI 0.72–2.32), and insulin production [22, 28].
Table 3 Associations between individual pesticide and diabetes

Exposed Case n (%) Control n (%) OR (95%CI) Adjusted OR* (95%CI)
Organochlorine insecticides
Aldrin Yes 7 (1.5) 11 (2.0) 0.77 (0.30–1.99) 0.79 (0.28–2.23)
No 445 (98.5) 536 (98.0)
Chlordane Yes 12 (2.7) 12 (2.2) 1.22 (0.54–2.75) 1.35 (0.56–3.24)
No 439 (97.3) 536 (97.8)
Dieldrin Yes 6 (1.3) 11 (2.0) 0.65 (0.24–1.78) 0.67 (0.22–2.02)
No 447 (98.7) 536 (98.0)
DDT Yes 32 (7.1) 48 (8.8) 0.79 (0.50–1.26) 0.83 (0.50–1.37)
No 420 (92.9) 500 (91.2)
Endosulfan Yes 112 (24.6) 105 (19.1) 1.38 (1.02–1.87)** 1.40 (1.01–1.95)**
No 343 (75.4) 444 (80.9)
Heptachlor Yes 9 (2.0) 13 (2.4) 0.83 (0.35–1.97) 1.13 (0.46–2.80)
No 444 (98.0) 535 (97.6)
Organophosphate insecticides
Chlorpyrifos Yes 76 (16.8) 78 (14.3) 1.21 (0.86–1.71) 1.20 (0.83–1.75)
No 376 (83.2) 468 (85.7)
Dicrotophos Yes 18 (4.0) 13 (2.4) 1.70 (0.82–3.51) 2.12 (0.97–4.61)
No 436 (96.0) 535 (97.6)
Dichlorvos Yes 10 (2.2) 12 (2.2) 1.01 (0.43–2.36) 1.03 (0.41–2.62)
No 442 (97.8) 536 (97.8)
EPN (ethyl-p-nitrophenyl) Yes 50 (11.0) 52 (9.5) 1.19 (0.79–1.79) 1.29 (0.83–2.02)
No 403 (89.0) 497 (90.5)
Folidol/parathion Yes 105 (23.2) 135 (24.5) 0.93 (0.70–1.25) 0.83 (0.60–1.15)
No 348 (76.8) 417 (75.5)
Methamidophos/Tamaron Yes 43 (9.5) 49 (9.0) 1.07 (0.69–1.64) 1.04 (0.65–1.65)
No 409 (90.5) 497 (91.0)
Mevinphos Yes 26 (5.7) 20 (3.6) 1.61 (0.89–2.92) 2.22 (1.17–4.19)**
No 427 (94.3) 528 (96.4)
Monocrotophos Yes 26 (5.8) 26 (4.7) 1.23 (0.70–2.15) 1.29 (0.72–2.32)
No 426 (94.2) 523 (95.3)
Carbamate insecticides
Carbaryl/Sevin Yes 81 (17.8) 70 (12.8) 1.48 (1.05–2.10)** 1.50 (1.02–2.19)**
No 373 (82.2) 478 (87.2)
Carbofuran/Furadan Yes 112 (24.7) 129 (23.5) 1.07 (0.80–1.43) 1.23 (0.90–1.70)
No 342 (75.3) 421 (76.5)
Carbosulfan Yes 54 (11.9) 50 (9.1) 1.35 (0.90–2.03) 1.44 (0.94–1.12)
No 398 (88.1) 498 (90.9)
Methomyl Yes 89 (19.6) 85 (15.5) 1.33 (0.96–1.84) 1.23 (0.86–1.76)
No 365 (80.4) 462 (84.5)
Pyrethoid insecticides
Abamectin Yes 220 (48.6) 290 (52.4) 0.86 (0.67–1.10) 0.79 (0.60–1.04)
No 233 (51.4) 263 (47.6)
Permethrin/Ambush Yes 15 (3.3) 15 (2.7) 1.23 (0.59–2.53) 1.39 (0.64–3.04)
No 435 (96.7) 533 (97.3)
Table 3 Associations between individual pesticide and diabetes (Continued)

Exposed Case n (%) Control n (%) OR (95%CI) Adjusted OR* (95%CI)
Herbicides
2,4-D Yes 147 (32.1) 167 (30.1) 1.10 (0.84–1.44) 1.04 (0.78–1.39)
No 311 (67.9) 388 (69.9)
Alachlor Yes 44 (9.7) 57 (10.4) 0.93 (0.61–1.40) 0.94 (0.59–1.47)
No 409 (90.3) 490 (89.6)
Butachlor Yes 115 (25.1) 130 (23.7) 1.08 (0.81–1.44) 1.14 (0.83–1.56)
No 344 (74.9) 418 (76.3)
Glyphosate Yes 65 (13.8) 88 (15.8) 1.17 (0.83–1.65) 1.08 (0.72–1.61)
No 405 (86.2) 470 (84.2)
Paraquat/Gramoxone Yes 115 (24.8) 161 (29.0) 1.23 (0.93–1.63) 1.31 (0.97–1.79)
No 348 (75.2) 395 (71.0)
Propanil Yes 94 (20.6) 129 (23.5) 0.85 (0.63–1.15) 0.81 (0.58–1.13)
No 362 (79.4) 421 (76.5)
Fungicides
Benlate Yes 24 (5.4) 16 (2.9) 1.88 (0.99–3.58) 2.08 (1.03–4.20)**
No 424 (94.6) 531 (97.1)
Bordeaux mixture Yes 6 (1.3) 11 (2.0) 0.66 (0.24–1.80) 0.81 (0.29–2.29)
No 444 (98.7) 538 (98.0)
Carbendazim Yes 45 (10.0) 49 (9.0) 1.13 (0.74–1.73) 1.17 (0.74–1.86)
No 405 (90.0) 498 (91.0)
Copper sulfate Yes 14 (3.2) 15 (2.8) 1.15 (0.55–2.40) 1.16 (0.51–2.63)
No 428 (96.8) 525 (97.2)
Mancozeb Yes 31 (6.9) 39 (7.1) 0.96 (0.59–1.57) 1.09 (0.65–1.84)
No 421 (93.1) 510 (92.9)
Maneb Yes 11 (2.5) 15 (2.7) 0.89 (0.41–1.97) 1.03 (0.44–2.41)
No 437 (97.5) 533 (97.3)
Metalaxyl Yes 39 (8.6) 40 (7.3) 1.20 (0.76–1.91) 1.32 (0.80–2.17)
No 413 (91.4) 510 (92.7)
Thiophanate-methyl Yes 13 (2.9) 13 (2.4) 1.23 (0.56–2.68) 1.37 (0.60–3.14)
No 435 (97.1) 534 (97.6)
Zineb Yes 16 (3.6) 9 (1.6) 2.22 (0.97–5.06) 2.23 (0.89–5.57)
*Adjusted for gender, age (continuous), BMI (continuous), cigarette smoking (packs/year), alcohol consumption (glasses/week), family history of diabetes (yes, no),
and occupation (farmer, non-farmer, other)
**Statistically significant (P < 0.05)
For carbamates, our results are in contrast with that re- 95%CI 0.64–3.04). This finding has never been reported
ported by Montgomery et al. [14] in the USA. Of the four in an epidemiological study before. In molecular study,
carbamate insecticides investigated (carbofuran/Furadan, carbamate and pyrethoid compounds demonstrated a
carbosulfan, methomyl, and carbaryl/Sevin), we found that potential effect on circadian disruption which is linked to
all were positively associated with diabetes. For carbaryl, the an increased diabetic risk [30].
ORs were statistically significant (OR = 1.50; 95%CI 1.02– For herbicides, of the 6 compounds investigated, four of
2.19). The OR in our study was higher than the results them (butachlor, glyphosate, paraquat/gramoxone, and
reported by Montgomery et al. [14] (OR = 1.10; 95%CI 2,4-D) could predict risk of diabetes but not statistically
0.95–1.28). In another animal study, carbaryl affected significant. A similar result was also reported in the Agri-
glucose homeostasis and insulin secretion [29]. cultural Health Study [14]. Evidence of the diabetic effect
We also found permethrin/Ambush, a pyrethoid of herbicides was first recognized among Vietnam vet-
insecticide, to be associated with diabetes (OR = 1.39; erans who had used Agent Orange, which is an herbicide
containing dioxins [31]. Dioxins were strongly associated Conclusion

with diabetes in [10]. A study in Australia compared the Our results suggest that the occurrence of diabetes among
mortality of 1999 outdoor staff who had used insecticides Thai farmer was associated with pesticide expos-ure. The
in the period 1935–1996 to those of the general popula- results supported the existing literature that pesticide
tion [32] and that study reported an association between exposure increases the risk of diabetes. Of the 35
mortality as a result of diabetes and herbicide use, at the individual pesticides investigated, those found to show
standard mortality rate (SMR) of 3.57 (95%CI 1.16– statistically significant increased risk of diabetes were
8.32). This is a high level of association. endosulfan, mevinphos, carbaryl/Sevin, and ben-late. The
Of the nine fungicides investigated, eight (benlate, car- ORs in this study were higher than previous studies in
bendazim, copper sulfate, mancozeb, maneb, metalaxyl, Western populations, and the difference is considered to
thiophanate-methyl, and zineb) were positively associ- be explainable by subject discrepancy in terms of health
ated with diabetes. For benlate, the association is statisti- background, safety practices, and nutri-tional status. It
cally significant with OR of 2.08 (95%CI 1.03–4.20). must be noted that recall and selection bias tend to
Although associations between diabetes and fungicides obscure the true association between diabetes and
have never been reported in any epidemiological study pesticides. It is suggested that more studies employing a
prior to this, a recent tissue-culture study reported that larger sample size and better exposure tracking are still
fungicide exposure can cause insulin resistance in fat cells needed to confirm our results and to identify more
and causes higher blood glucose [33]. individual compounds that may be a causal effect of
Compared with the study in the USA [14], most of ORs diabetes. The issue should certainly receive more
found in our study were higher. The possible explanation attention by public institutions, and practical measures to
for this is that Thai people are more likely to have poorer control those harmful pesticides are urgently needed.
health backgrounds, life styles, and risky agricultural
Acknowledgements
practices and behaviors. Data on PPE use and pesticide We acknowledge with great appreciation the village health volunteers in
use behaviors that were also collected in this study the district of Bang Rakam for their assistance and cooperation in data
showed that the rate of PPE utilized was lower than 30% collection. We must also thank the Health Promoting Hospitals in Bang
Rakam for the coordination and data support. We also thank Mr. Roy
and 41% of users did not change their clothes Morien of the Naresuan University Division of Research Administration for
immediately after spraying the pesticides. A further his assistance with English expression and grammar in this article.
localizing factor is that Thai people, particularly rural
Funding
Thais, tend to eat less protein in their diet with greater This study received grants supported by Naresuan University
amounts of vegetables and have poorer educational and (grant no R2560C031).
socioeconomic status, all of which are confirmed diabetic
Availability of data and materials
risk factors [17]. Not applicable.
The major sources of error in this type of study were
recall and selection bias. Recall bias usually occurs when Authors’ contributions
The work was primarily done by Dr. CJ, who was responsible for
using past exposure data which may be biased due to the development of the proposal, data analysis, and preparation of
differences in recall ability between case and control [34]. the manuscript. Dr. YJ joined the project to assist with data
However, as information on pesticide and diabetic linkage collection and analysis, and preparation for manuscript. Both
authors read and approved the final manuscript.
was not well known in Thailand, the recall prob-lem, if it
did occur, would affect both case and control groups Ethics approval and consent to participate
equally. As for selection bias, it often occurs when This study was approved by the Ethics Board of Naresuan University
exposure or disease information are not accurate [35]. In (project number 402/59). Written informed consent for an interview
was obtained from each subject.
our situation, the error for case selection was expected to
be minimal since each case was a medically diagnosed Consent for publication
diabetes patient. However, by using neighbor-ing Not applicable.
controls, it was very likely that the control subject was Competing interests
also exposed to some type of pesticides. This The authors declare that they have no competing interests.
phenomenon would have the effect of lowering the degree
of association between diabetes and any of the pesticides. Publisher’s Note
Data on exercise and the eating of various types of food, Springer Nature remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
including vegetables, fruit, and sweet were also collected in
this study for analysis (Table 1). However, since most of the Author details
1
study subjects were farmers who had similar lifestyles, exer- Department of Natural Resources and Environment, Faculty of
Agriculture, Natural Resources and Environment, Naresuan
cise and eating factors failed to predict diabetic risk and University, Muang District, Phitsanulok 65000, Thailand. 2Faculty of
therefore they were not included in the analytical model. Nursing, Naresuan University, Phitsanulok, Thailand.
Received: 5 October 2017 Accepted: 19 January 2018 24. Rezg R, Mornagui B, El-Fazaa S, Gharbi N. Organophosphorus
pesticides as food chain contaminants and type 2 diabetes: a
review. Trends Food Sci Technol. 2010;21(7):345–57.
25. Park WH, Kang YC, Piao Y, Pak DH, Pak YK. Causal effects
References of synthetic chemicals on mitochondrial deficits and diabetes
1. International Diabetes Federation. Idf Diabetes Atlas— pandemic. Arch Pharm Res. 2013;36:178–88.
7th edition. International Diabetes Federation. 2015. 26. Somchai Dungsuwan. Bang Rakam history. https://www.gotoknow.org/
2. Thai NCD network. Situation of NCDs (2nd report):“Kick off to the Goals.” posts/388665. Accessed 20 Sept 2017.
Nontaburi: The International Health Policy Program, Thailand (IHPP); 2016. 27. Ozmen O, Sahinduran S, Mor F. Pathological and
3. Baliunas DO, Taylor BJ, Irving H, Roerecke M, Patra J, Mohapatra immunohistochemical examinations of the pancreas in subacute
S, et al. Alcohol as a risk factor for type 2 diabetes: a systematic endosulfan toxicity in rabbits. Pancreas. 2010;39(3):367–70.
28. van Koppen CJ, Kaiser B. Regulation of muscarinic acetylcholine
review and meta-analysis. Diabetes Care. 2009;32(11):2123–32.
4. Rockette-Wagner B, Edelstein S, Venditti EM, Reddy D, Bray GA, Carrion- receptor signaling. Pharmacol Ther. 2003;98(2):197–220.
Petersen ML, et al. The impact of lifestyle intervention on sedentary time in 29. Popovska M, Dubocovich M, Rajnarayanan R. The insecticides
individuals at high risk of diabetes. Diabetologia. 2015;58(6):1198–202. carbaryl and carbofuran show high affinity for hMT2 melatonin
receptors (662.12). FASEB J. 2014;28(1 Supplement 662.12):926.5.
5. Willi C, Bodenmann P, Ghali WA, Faris PD, Cornuz J. Active smoking
30. Van Hall A, Popovska-Gorevski MJA, Rajnarayanan R, Dubocovich M.
and the risk of type 2 diabetes. JAMA. 2007;298(22):2654.
Federation proceedings. FASEB J. 2015;29(1 Supplement 776.4):926.5.
6. Yu JH, Yun C-H, Ahn JH, Suh S, Cho HJ, Lee SK, et al. Evening
chronotype is associated with metabolic disorders and body composition in 31. Kang HK, Dalager NA, Needham LL, Patterson DG, Lees PSJ, Yates K,
middle-aged adults. J Clin Endocrinol Metab. 2015;100(4):1494–502. et al. Health status of Army Chemical Corps Vietnam veterans who
sprayed defoliant in Vietnam. Am J Ind Med. 2006;49(11):875–84.
7. Cappuccio FP, D’Elia L, Strazzullo P, Miller MA. Quantity and
32. Beard J, Sladden T, Morgan G, Berry G, Brooks L, McMichael A.
quality of sleep and incidence of type 2 diabetes: a systematic
review and meta-analysis. Diabetes Care. 2010;33(2):414–20. Health impacts of pesticide exposure in a cohort of outdoor
8. Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type workers. Environ Health Perspect. 2003;111(5):724–30.
2 diabetes: a systematic review. Diabetes Care. 2002;25(10):1862–8. 33. Endocrine Society. Fungicide used on farm crops linked to insulin resistance.
https://www.sciencedaily.com/releases/2012/06/120625100902. htm. Accessed
9. Kuo C-C, Moon K, Thayer KA, Navas-Acien A. Environmental
15 Sept 2017.
chemicals and type 2 diabetes: an updated systematic review of
the epidemiologic evidence. Curr Diab Rep. 2013;13(6):831–49. 34. Coughlin SS. Recall bias in epidemiologic studies. J Clin
10. Lee D-H, Lee I-K, Song K, Steffes M, Toscano W, Baker BA, et al. A strong Epidemiol. 1990; 43(1):87–91.
dose-response relation between serum concentrations of persistent organic 35. Grimes DA, Schulz KF. Bias and causal associations in
pollutants and diabetes: results from the National Health and examination observational research. Lancet. 2002;359(9302):248–52.
survey 1999-2002. Diabetes Care. 2006;29(7):1638–44.
11. Vasiliu O, Cameron L, Gardiner J, DeGuire P, Karmaus W. Polybrominated
biphenyls, polychlorinated biphenyls, body weight, and incidence of adult-onset
diabetes mellitus. Epidemiology. 2006;17(4):352–9.
12. Thayer KA, Heindel JJ, Bucher JR, Gallo MA. Role of environmental
chemicals in diabetes and obesity: a national toxicology program
workshop review. Environ Health Perspect. 2012;120(6):779–89.
13. Lah K. Pesticide Use Statistics - Toxipedia. http://www.toxipedia.org/display/
toxipedia /Pesticide+Use+Statistics. Accessed 20 Sept 2017.
14. Montgomery MP, Kamel F, Saldana TM, Alavanja MCR, Sandler DP. Incident
diabetes and pesticide exposure among licensed pesticide applicators:
agricultural health study, 1993-2003. Am J Epidemiol. 2008;167(10):1235–46.
15. Starling AP, Umbach DM, Kamel F, Long S, Sandler DP, Hoppin JA.
Pesticide use and incident diabetes among wives of farmers in the
agricultural health study. Occup Environ Med. 2014;71(9):629–35.
16. Saldana TM, Basso O, Hoppin JA, Baird DD, Knott C, Blair A, et al.
Pesticide exposure and self-reported gestational diabetes mellitus in
the agricultural health study. Diabetes Care. 2007;30(3):529–34.
17. Benjamin N, Kushwah A, Sharma RK, Katiyar AK. Histopathological
changes in liver, kidney and muscles of pesticides exposed malnourished
and diabetic rats. Indian J Exp Biol. 2006;44(3):228–32.
18. Cox S, Niskar AS, Venkat Narayan KM, Marcus M. Prevalence of self-
reported diabetes and exposure to organochlorine pesticides among
Mexican Americans: Hispanic health and nutrition examination survey,
1982-1984. Environ Health Perspect. 2007;115(12):1747–52.
19. Codru N, Schymura MJ, Negoita S, Rej R, Carpenter DO. Diabetes in relation to
serum levels of polychlorinated biphenyls and chlorinated pesticides in adult Submit your next manuscript to BioMed
native Americans. Environ Health Perspect. 2007;115(10):1442–7.
20. Dirinck E, Jorens PG, Covaci A, Geens T, Roosens L, Neels H, et al. Obesity Central and we will help you at every step:
and persistent organic pollutants: possible obesogenic effect of organochlorine • We accept pre-submission inquiries
pesticides and polychlorinated biphenyls. Obesity. 2011;19(4):709–14.
• Our selector tool helps you to find the most relevant journal
21. Sargis RM. The hijacking of cellular signaling and the diabetes
epidemic: mechanisms of environmental disruption of insulin • We provide round the clock customer support
action and glucose homeostasis. Diabetes Metab. 2014;28:13–24. • Convenient online submission
22. Rahimi R, Abdollahi M. A review on the mechanisms involved
• Thorough peer review
in hyperglycemia induced by organophosphorus pesticides.
Pestic Biochem Physiol. 2007;88(2):115–21. • Inclusion in PubMed and all major indexing services
23. Hectors TLM, Vanparys C, Van Der Ven K, Martens GA, Jorens PG, Van Gaal • Maximum visibility for your research
LF, et al. Environmental pollutants and type 2 diabetes: a review of mechanisms
that can disrupt beta cell function. Diabetologia. 2011;54(6):1273–90. Submit your manuscript at
www.biomedcentral.com/submit

Association Between Diabetes and Pesticides: A Case-Control Study Among Thai Farmers

Uploaded by

Association Between Diabetes and Pesticides: A Case-Control Study Among Thai Farmers

Uploaded by

Juntarawijit and Juntarawijit Environmental Health and Preventive Medicine

RESEARCHARTICLE Open Access

Association between diabetes and

Table 2 Associations between different types of pesticides and diabetes

Table 2 Associations between different types of pesticides and diabetes (Continued)

Table 3 Associations between individual pesticide and diabetes

Table 3 Associations between individual pesticide and diabetes (Continued)

containing dioxins [31]. Dioxins were strongly associated Conclusion

You might also like