- Email: [email protected]
Childhood leukemia mortality and farming exposure in South Korea: A national population-based birth cohort study
Cancer Epidemiology 38 (2014) 401–407
Contents lists available at ScienceDirect
Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention journal homepage: www.cancerepidemiology.net
Childhood leukemia mortality and farming exposure in South Korea: A national population-based birth cohort study Eun Shil Cha a, Seung-sik Hwang b, Won Jin Lee a,* a b
Department of Preventive Medicine, Korea University College of Medicine, South Korea Department of Social and Preventive Medicine, Inha University School of Medicine, South Korea
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 November 2013 Received in revised form 30 April 2014 Accepted 4 May 2014 Available online 10 June 2014
Objectives: The aim of this study was to evaluate the relationship between leukemia mortality and exposure to farming among children in South Korea. Methods: A retrospective cohort study of South Korean children was conducted using data collected by the national birth register between 1995 and 2006; these data were then individually linked to death data. A cohort of 6,479,406 children was followed from birth until either their death or until December 31, 2006. For surrogate measures of pesticide exposure, we used residence at birth, paternal occupation, and month of conception from the birth certificate. Farming and pesticide exposure indexes by county were calculated using information derived from the 2000 agricultural census. Poisson regression analyses were used to calculate rate ratios (RRs) of childhood leukemia deaths according to indices of exposure to agricultural pesticides after adjustment for potential confounders. Results: In total 585 leukemia deaths were observed during the study period. Childhood leukemia mortality was significantly elevated in children born in rural areas (RR = 1.43, 95%CI 1.09–1.86) compared to those in metropolises, and in counties with both the highest farming index (RR = 1.33, 95%CI 1.04–1.69) and pesticide exposure index (RR = 1.30, 95%CI 1.02–1.66) compared to those in the reference group. However, exposure–response associations were significant only in relation to the farming index. When the analyses were limited to rural areas, the risk of death from leukemia among boys conceived between spring and fall increased over those conceived in winter. Conclusions: Our results show an increase in mortality from childhood leukemia in rural areas; however, further studies are warranted to investigate the environmental factors contributing to the excess mortality from childhood leukemia in rural areas. ß 2014 Elsevier Ltd. All rights reserved.
Keywords: Cancer Children Cohort Death Pesticides Rural Seasonal variation
1. Introduction Leukemia is the most common form of childhood cancer in the world, accounting for around 30% of all cancers diagnosed in children younger than 15 years of age [1]. However, risk factors for the disease remain largely unknown, and the established causal factors – such as ionizing radiation and congenital genetic syndromes – together explain less than 10% of the cases. Several other environmental agents have been put forward to explain its etiology; these include parental smoking, electromagnetic fields, socioeconomic status, infection, and pesticides [2].
* Corresponding author at: Department of Preventive Medicine, Korea University College of Medicine, 73, Inchon-ro, Seongbuk-gu, Seoul 136-705, South Korea. Tel.: +82 2 2286 1413; fax: +82 2 927 7220. E-mail address: [email protected] (W.J. Lee). http://dx.doi.org/10.1016/j.canep.2014.05.003 1877-7821/ß 2014 Elsevier Ltd. All rights reserved.
Several reviews of epidemiological studies examining the potential association between childhood leukemia and residential pesticide exposure have been conducted [3,4]. Despite this existing epidemiological literature (suggesting elevated childhood leukemia risks related to household environmental pesticide exposure), the evidence for a causal relationship between exposure to agricultural pesticides and childhood leukemia is still limited. Studies on this topic among Asian populations are also lacking, and the results vary between countries due to differences in exposure and demographic characteristics. Moreover, most findings on childhood leukemia and pesticides were from case–control studies. In South Korea, the incidence of childhood leukemia was 5.0 per 100,000 population in 2010, accounting for 33.2% of all cancer incidence among children aged 0–14 years [5]. South Korea has traditionally been an agricultural nation, and it still retains as of 2011 a farming population of about three million (approximately 6% of the total population). The quantity of pesticides used was
402
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
recorded as 19,131 tons [6], and average pesticide consumption was 10.6 kg per hectare in 2011 – higher than that in other developed countries with intensive agriculture [7]. Despite widespread agricultural pesticide exposure in the Korean population, no study has examined the relationship between childhood leukemia and this exposure in South Korea. Therefore, the aim of this study is to examine the relationship between extent of farming and pesticide use in the county of residence, paternal occupation, and season of conception and mortality from childhood leukemia in South Korea by using national population-based retrospective birth cohort data from 1995 to 2006. 2. Materials and methods 2.1. Data sources Details of the study design and of the population have been described previously [8,9]. In brief, data from the national birth and death registration databases of Statistics Korea were used to establish a retrospective cohort of all children born in South Korea between 1995 and 2006. Records from the national birth registration database contain information on each child born (e.g., birth date, gender, address of residence at birth, personal identification number, and birth characteristics such as birth weight and gestational duration), as well as on its parents at the time of the birth of the child (e.g., age, education, and occupation). Of the 6,813,945 total births between 1995 and 2006, we excluded 5.3% of the records that lacked a complete personal identification number – an omission due mainly to administrative disagreements between the district offices of residence and those of birth registration. A total of 6,479,406 records of live births which occurred between 1995 and 2006 in South Korea were obtained and linked to the national death registration database for the same period via their unique personal identification numbers. These individuals were followed from birth until their death or until the end of the follow-up period on December 31, 2006. The maximum age at death was 11 years old. The causes of death in the registered death data are coded according to the International Classification of Diseases, 10th Revision (ICD-10) [10]. A leukemia death was defined as a case marked with ICD codes between C91 and C95. 2.2. Exposure indices As indicators of potential exposure to risk factors associated with farming, including pesticides, we used five indices. First, residence at birth – based on the 243 administrative districts in the cohort data – was categorized as metropolis, city, or rural area. Metropolises included Seoul and six other metropolitan cities in South Korea. Districts in the nine provinces outside these metropolises were classified into either cities or rural areas, depending on governmental administrative divisions by population size and rural characteristics. Second, we calculated the farming index using following algorithm applied in our previous study [11]. County-specific farming index = (% of full-time farm household + % of part-time farm household 0.3) % of farm households weighted with farming years % of farm households weighted with farm size. The calculation process of farming index by area is presented as an Appendix A. Full-time farm population refers to persons engaged exclusively in farming, while part-time farm population refers to farm workers who also were employed in non-farming jobs to earn money for more than 30 days during the course of a year. The score assigned to part-time farm population (i.e., 0.3) was weighted to reflect the intensity of farming compared with the full-time farm population based on our professional discretion. Farming years and
farm size were categorized into quartiles, i.e., 20, 21–32, 33–45, 46 years for farming years and 0.43, 0.44–0.89, 0.90–1.65, 1.66 ha for farm size. Information on the number of farm households by county, as well as farm size and farming years of each farm household, was derived from the agricultural census conducted by Statistics Korea [12]. Agricultural censuses were taken every 10 years from when the first census was taken in 1960 until 1990, and since 1995 they have been conducted on a 5-year basis. We chose the 2000 census as it was the midpoint of our cohort data. Data on the number of total households and population by county in 2000 was drawn from the database maintained by Statistics Korea (http://kosis.kr). Each term of the numerator of farming index represents the agricultural labor by each county and provides an estimate of total farming level per county. Third, we developed a pesticide exposure index for each county. The index was calculated by taking into account the area of cultivated land (Ai) and the frequency (Fi) and hours (Hi) of pesticide application for the major crops, and the algorithm was modified on the basis of a previous study [13]. The county-specific pesticide exposure index can be described as P4 A F H =population. The i represents the four major types of i i i i¼1 farming (i.e., rice, upland, fruit, and greenhouse) in South Korea. The crop-specific area by county was taken from the agricultural census, and Fi and Hi from a nationwide sampling survey of male farmers in South Korea in 2010 [14]. The numerator of the formula indicates an estimate of the total pesticide load. With total population of each county in the denominator, the pesticide exposure index provides an estimate of pesticides used per inhabitant. Fourth, besides these three regional variables, we classified parental occupations obtained from national birth certificate data. Paternal occupation was grouped into non-manual (i.e., legislators, senior officials and managers, professionals, technicians and associate professionals, clerks, and service and sale workers), manual (i.e., craft and related trades workers, plant, machine operators and assemblers, and elementary occupations), others (i.e., students, the unemployed, homemakers, and armed forces), and skilled agricultural, forestry and fishery workers. Finally, since parental exposure to pesticides may vary by month of conception, we defined month of conception for each child as subtracting the duration of gestation from the date of birth. The month of conception was categorized into four seasons; spring (March–May), summer (June–August), fall (September–November), and winter (December–February). 2.3. Statistical analysis The number of person-years was calculated by summing the numbers of days from the date of birth to the date of death or to the end of the follow-up period (December 31, 2006). Poisson regression was used to calculate mortality rate ratios (RRs) and 95% confidence intervals (95%CIs) of leukemia death according to our surrogates of pesticide exposure. For farming index and pesticide exposure index, we assigned counties into one of four groups based on quartiles of each index based on the distribution of total subjects (i.e., <0.02, 0.02–27.9, 28.0– 20,422.3, >20,422.3 for farming index and <0.006, 0.006– 0.05, 0.06–0.54, >0.54 for pesticide exposure index). We adjusted for gender and birth year of children in the analyses in order to consider any gender difference in leukemia mortality and potential cohort effect due to rapid development of medical technologies or accessibility to high-quality medical facilities with birth year. To control for potential confounders, we also included the variable of maternal education and paternal job in the analyses, which have been reported as a significant risk
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
403
Table 1 Distribution of farming exposure indicators and childhood leukemia deaths by residence at birth in South Korea. Farming exposure indicators
Residence at birth (number of counties) Total (n = 243)
Farming index Q1 Q2 Q3 Q4 Pesticide exposure index Q1 Q2 Q3 Q4 Cultivated area per inhabitant Total Paddy Upland Fruit Greenhouse % of full-time farm households Childhood leukemia (1995–2006) Person-years Leukemia deaths Rate per 100,000
Rural (n = 93)
61 61 61 60
0 4 32 57
(0.0) (4.3) (34.4) (61.3)
61 61 61 60
0 5 36 52
(0.0) (5.4) (38.7) (55.9)
9.6 5.2 3.4 0.7 0.3 14.5
19.9 10.8 7.4 1.2 0.5 30.2
42,862,104 585 1.36
factor for childhood cancer death in a previous study [9]. Detailed analyses of leukemia mortality by gender and age group (<1, 1–4, and 5–11 years) were conducted separately, and subgroup analysis was performed within the subset of rural areas. All statistical analyses were performed using Stata 12.0 (StataCorp, College Station, Texas). 2.4. Ethics statement We used publicly available birth and mortality data without any personal identifiers, and thus ethical approval was unnecessary.
City (n = 82)
Metropolis (n = 68)
Number of counties (%) 11 (13.4) 40 (48.8) 28 (34.1) 3 (3.7) Number of counties (%) 11 (13.4) 39 (47.6) 24 (29.3) 8 (9.8) Mean of area (hectare) 5.5 3.0 1.7 0.6 0.2 8.3
4,478,104 78 1.74
50 17 1 0
(73.5) (25.0) (1.5) (0.0)
50 17 1 0
(73.5) (25.0) (1.5) (0.0)
0.4 0.2 0.1 0.2 0.4 0.6
18,448,336 270 1.46
19,935,664 237 1.19
3. Results Rural areas were assigned higher farming exposure indicators, while metropolises were assigned a lower level on these indices (Table 1). Both the average area per inhabitant of cultivated land for all kinds of crops and the average proportion of full-time farm households were the highest in rural areas, followed by those in cities and metropolises as might be expected. Our cohort contained 42,862,104 person-years of observation and a total of 585 leukemia deaths from 1995 through 2006. The overall rate of mortality from childhood leukemia was 1.36 per 100,000 and the highest rate was shown in rural areas, with 1.74 per 100,000.
Table 2 Risks of childhood leukemia death according to farming exposure indicators and gender among children born in South Korea, 1995–2006. Farming exposure indicators
Residence at birth Metropolis City Rural area Ptrend Farming index Q1 Q2 Q3 Q4 Ptrend Pesticide exposure index Q1 Q2 Q3 Q4 Ptrend Paternal occupation Non-manual Manual Others Agriculture Season of conception Winter Spring Summer Fall a
Total (n = 585)
Boys (n = 342) a
Girls (n = 243)
Cases
RR (95%CI)
Cases
RR (95%CI)
Cases
RR (95%CI)
237 270 78
1.00 1.21 (1.02–1.45) 1.43 (1.09–1.86) 0.003
140 154 48
1.00 1.17 (0.93–1.48) 1.49 (1.06–2.09) 0.020
97 116 30
1.00 1.27 (0.97–1.67) 1.34 (0.87–2.05) 0.073
121 164 141 159
1.00 1.30 (1.03–1.64) 1.26 (0.98–1.60) 1.33 (1.04–1.69) 0.042
70 94 76 102
1.00 1.26 (0.93–1.72) 1.15 (0.83–1.59) 1.47 (1.08–2.01) 0.033
51 70 65 57
1.00 1.35 (0.94–1.94) 1.41(0.97–2.04) 1.13 (0.76–1.67) 0.521
121 157 152 155
1.00 1.31 (1.04–1.67) 1.26 (0.99–1.60) 1.30 (1.02–1.66) 0.065
71 86 89 96
1.00 1.20 (0.88–1.65) 1.23 (0.90–1.69) 1.37 (1.00–1.87) 0.057
50 71 63 59
1.00 1.48 (1.03–2.13) 1.30 (0.89–1.89) 1.20 (0.81–1.77) 0.533
379 140 47 19
1.00 1.28 (1.04–1.57) 1.64 (1.20–2.24) 1.10 (0.69–1.76)
211 94 27 10
1.00 1.53 (1.18–1.98) 1.80 (1.20–2.69) 1.00 (0.53–1.92)
168 46 20 9
1.00 0.97 (0.69–1.36) 1.45 (0.89–2.36) 1.25 (0.63–2.48)
139 153 158 135
1.00 0.95 (0.76–1.20) 0.99 (0.79–1.25) 0.98 (0.78–1.25)
77 94 95 76
1.00 1.05 (0.78–1.42) 1.08 (0.80–1.46) 1.00 (0.73–1.38)
62 59 63 59
1.00 0.83 (0.58–1.18) 0.88 (0.62–1.26) 0.96 (0.67–1.38)
Adjusted for gender, birth year, maternal education, and paternal occupation.
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
404
Table 3 Risks of death from childhood leukemia according to farming exposure indicators and gender among children born in rural areas in South Korea, 1995–2006. Farming exposure indicators
Farming index Q1 Q2 Q3 Q4 Ptrend Pesticide exposure index Q1 Q2 Q3 Q4 Ptrend Paternal occupation Non-manual Manual Others Agriculture Season of conception Winter Spring Summer Fall a
Total (n = 78)
Boys (n = 48) a
Girls (n = 30)
Cases
RR (95%CI)
Cases
RR (95%CI)
15 18 23 22
1.00 1.22 (0.61–2.43) 1.56 (0.81–3.01) 1.51 (0.78–2.96) 0.170
8 10 14 16
1.00 1.31 (0.52–3.33) 1.82 (0.76–4.37) 2.16 (0.91–5.11) 0.055
7 8 9 6
1.00 1.09 (0.39–3.03) 1.25 (0.46–3.40) 0.81 (0.26–2.48) 0.801
12 24 25 17
1.00 2.00 (1.00–4.00) 2.10 (1.05–4.20) 1.39 (0.66–2.95) 0.458
6 14 14 14
1.00 2.40 (0.92–6.26) 2.44 (0.93–6.39) 2.39 (0.91–6.28) 0.116
6 10 11 3
1.00 1.57 (0.57–4.34) 1.71 (0.62–4.69) 0.45 (0.11–1.85) 0.412
35 20 11 12
1.00 1.49 (0.85–2.63) 2.48 (1.26–4.89) 1.25 (0.62–2.51)
23 13 6 6
1.00 1.55 (0.76–3.15) 2.01 (0.81–4.94) 0.92 (0.35–2.40)
12 7 5 6
1.00 1.40 (0.54–3.63) 3.42 (1.20–9.74) 1.91 (0.68–5.33)
11 23 21 23
1.00 1.77 (0.86–3.64) 1.69 (0.81–3.50) 2.10 (1.02–4.30)
6 13 14 15
1.00 1.84 (0.70–4.83) 2.05 (0.79–5.35) 2.51 (0.97–6.47)
5 10 7 8
1.00 1.70 (0.58–4.98) 1.24 (0.39–3.91) 1.59 (0.52–4.86)
Cases
RR (95%CI)
Adjusted for sex, birth year, maternal education, and paternal occupation.
The relative risks of childhood leukemia were statistically significantly elevated among children born in rural areas (RR = 1.43, 95%CI 1.09–1.86) as well as in cities (RR = 1.21, 95%CI 1.02–1.45) compared to those in metropolises (Table 2). The risks of mortality from childhood leukemia among children in the counties of the second, third, and highest levels of regional farming and pesticide exposure indices were significantly elevated compared with those born in the reference group for each index, but exposure–response associations were significant only at the farming index. Mortality from childhood leukemia was not elevated for paternal agricultural work with reference to nonmanual occupations. There was no distinct association between childhood leukemia and season of conception. The risk estimates did not significantly differ in the analysis by age group, and the point estimates of the two major subtypes of leukemia (i.e., acute lymphocytic leukemia, acute myeloid leukemia) were not significantly different (data not shown). The results from the analyses among 93 rural counties showed results that did not differ materially from those for total subjects (Table 3). However, in rural areas a positive association between childhood leukemia and surrogate pesticide exposure indices was more apparent in boys than in girls. For season of conception, the risks of mortality from childhood leukemia were found to be higher for boys conceived between spring and fall compared to those conceived in winter. 4. Discussion We observed statistically significant elevated risks of childhood leukemia in rural areas and in counties with a higher level of farming index and pesticide exposure index. Within rural areas, the risk of childhood leukemia death increased among boys conceived between spring and fall compared to among those conceived in winter. Although these results may support the notion that agricultural pesticide exposure might be associated with risk of childhood leukemia, further investigation is required. Most previous studies have consistently reported increased risks of childhood leukemia as a result of farming exposure. Some ecological studies have found elevated risks of childhood leukemia in relation to proximity to and intensity of agricultural activity
[15,16]. Studies in California – applying a pesticide-use reporting database – have found significant increases in childhood leukemia in groups exposed to pesticides [17,18]. A number of case–control studies have also reported that childhood exposure to household insecticides and parental exposure to pesticides prior to and during pregnancy present the greatest risks [19,20]. Some studies, however, found no significant associations between childhood leukemia and either intense agricultural activity or parental occupational exposure to pesticides [21,22]. Children in rural areas or near treated croplands can be exposed to pesticides through agricultural application drift, or through contaminated soil, surface water, household dust and food [23]. Although specific biochemical mechanisms relating pesticide exposure to childhood cancer have yet to be established, some evidence suggests that pesticides may promote chromosomal aberrations, oxidative stress, and immunotoxicity, which could be linked to increased cancer risk [24]. Several pesticides identified as having a possible human carcinogenic effect by recent epidemiological literature [25,26] and US EPA’s Pesticide Program [27] – such as diazinon, chlorpyrifos, butachlor, alachlor, chlorothalonil, dichlorvos, pendimethalin, thiophanate-methyl, ethoprophos, trichlorfon, and oxadiazon – have been widely applied in South Korea [6]. Moreover, children are believed to be more vulnerable to the effects of environmental pollutants than are adults as a result of different physical dimensions, less mature immune systems, unique diets, and metabolic characteristics [23]. In subgroup analysis within rural areas we observed elevated risks of childhood leukemia among boys conceived between spring and fall (when agricultural pesticides are commonly applied in South Korea) compared to those conceived in winter. The fact that seasonality in childhood leukemia was observed only in rural areas may be explained by more common environmental exposure to pesticides in an agricultural setting than in urban areas. Previous studies have found an elevated risk of adverse birth outcome, such as birth defects, for infants conceived during the spraying season [28,29]. The results of a more pronounced association among boys, particularly in rural areas, observed in this study may be explained by greater susceptibility to toxicants such as pesticides among male fetuses, although the role of child gender as a potential
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
modifier of the effects of environmental toxicants has not been much investigated [30]. Another explanation may be that the risk of accidental childhood exposure linked to farm activities affects boys more often than girls. The increase in childhood leukemia mortality risk in rural areas, but the absence of an exposure–response relationship for pesticide exposure index observed in this study, suggest that this increase could reflect factors other than pesticide exposure that we were unable to examine. Besides pesticides, numerous environmental factors – such as infectious agents, socioeconomic status, air pollution, radiation, and other chemicals – have been suggested as risk factors for childhood leukemia [2,31]. Moreover, the socioeconomic status of parents in childhood leukemia may also complicate the associations [31]. Some studies of childhood leukemia have reported a higher incidence with higher socioeconomic status, while other studies have suggested a potential inverse association between socioeconomic status and incidence of childhood leukemia [32]. There were also significant social disparities in the survival of childhood leukemia in South Korea [33]. Although we adjusted for maternal education and paternal occupation in the analyses, residual confounding factors for socioeconomic status may affect incidence and survival of childhood leukemia. No significant association between childhood leukemia and paternal occupation as an agricultural worker was observed in this study, consistent with a systematic review and meta-analysis [34]. Previously, prenatal maternal exposure was reported to have an increased risk of childhood leukemia [34,35]. However, there were too few children (n = 5) whose mothers were agricultural workers in our analysis. There are several limitations to this study. First, regional agricultural pesticide exposure was measured by means of surrogate indices rather than through direct measurement. However, residence in agricultural communities or in proximity to farmland has been reported to be correlated with higher pesticide concentrations in house dust, air, and urinary metabolites compared with non-farm residence [36,37]. In this study, the distribution of counties for farming index and pesticide exposure index by area was similar, and the associations between childhood leukemia and each index also showed similar patterns. In addition, a lack of information on individual or groups of pesticides by area is an important limitation of this study. Nevertheless, the use of the crude surrogates for pesticide exposure in this study may bias effect estimates toward null. Second, the exposure of interest may also have been misclassified by using only a single address based on birth registration records to characterize exposure during the time period of interest. Among our subjects, the overall migration rate by area was 17.8%, and the rates of movement from rural to urban (9.6%) and from urban to rural (8.2%) areas did not appear to differ. Consequently, the phenomenon of children who had resided in rural areas at birth moving to urban areas may have diluted the exposure gradient. Third, the difference between rural and urban residence in childhood leukemia mortality may be affected by variation in access to well-equipped medical facilities and quality of medical services. These factors would influence the geographic disparities for survival from childhood leukemia [38]. Although advanced medical technology and a relatively high number of health-care facilities are present in South Korea, medical resources are unevenly distributed throughout the country [39]. To reduce the
405
geographic discrepancy in survival for childhood leukemia, we performed a subgroup analysis within the subset of 93 rural counties. The results, however, were generally similar to those observed among overall subjects in combined areas. In addition, the results derived from death at the age of less than 1 year, which could be more related as incidence cases than those of older age groups, also showed a similar pattern to those observed among overall subjects. Fourth, we were unable to examine whether the survival effect of childhood leukemia affected the associations observed in this study, since this study is based on deaths with an underlying cause coded as being due to leukemia. Although mortality data are not ideal for determining risk factors for childhood leukemia, we conducted subgroup analyses by age group in order to minimize the survival effects and also controlled for certain variables of socioeconomic status during the analyses in order to minimize the effect of variable access to medical treatment. Finally, the present analysis did not take into account other routes of pesticide exposure in children. Children may be exposed to pesticides through their use in their homes, and further studies need to incorporate information on pesticide exposure at the individual level with that at the ecological level. Despite these limitations, we have shown an increased risk of childhood leukemia mortality among children born in rural areas and in counties with higher levels of farming index and pesticide exposure index using national retrospective cohort data in South Korea. The associations tended to be most apparent among boys, and seasonality in leukemia mortality was observed in rural areas. However, further studies are warranted to investigate whether pesticide exposure contributes to the excess mortality from childhood leukemia in rural areas. Conflict of interests None declared. Acknowledgements This work was supported by Korea University Grant (K1031681), Republic of Korea.
Appendix A. The formula for county-specific farming index In the formula for farming index, the percentage of full-time farm population was a continuous variable based on the full-time farm population divided by the total population of each area. Fulltime farm population refers to persons engaged exclusively in farming, while part-time farm population refers to farm workers who also were employed in non-farming jobs to earn money for more than 30 days during the course of a year. Farming years and farm size were categorized into quartiles, i.e., 20, 21–32, 33–45, 46 years for farming years and 0.43, 0.44–0.89, 0.90–1.65, 1.66 ha for farm size. The score assigned to part-time farm population (i.e., 0.3) was weighted to reflect the intensity of farming compared with the full-time farm population based on our professional discretion. The term ‘weighted’ means multiplying the median values for each category in order to reflect weights according to farming years and farm size.
406
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
No: of full-time farm household No: of part-time farm household 100 þ 100 0:3a Total household Total household No: of farm households with farming during 120 years 100 Total households No: of farm households with farming during 2132 years 14b þ 100 30b Total households No: of farm households with farming during 3345 years 100 40b þ Total households No: of farm households with farming during 4680 years 100 50b þ Total households No: of farm households with farming during 0:0010:429 hectare farm 100 Total households No: of farm households with farming during 0:4300:891 hectare farm 100 0:66b 0:25b þ Total households No: of farm households with farming during 0:8921:650 hectare farm 100 1:22b þ Total households No: of farm households with farming during 1:651165 hectare farm 100 2:50b þ Total households
County-specific farming index ¼
where a – weight for part-time farm population to reflect intensity of farming compared with full-time farm population based on professional discretion and b – median values for each category in order to reflect weights according to farming years and farm size. References [1] American Cancer Society. Global cancer facts & figures. 2nd ed. Atlanta: American Cancer Society, 2011 , http://www.cancer.org/acs/groups/content/ @epidemiologysurveilance/documents/document/acspc-027766.pdf. [2] Belson M, Kingsley B, Holmes A. Risk factors for acute leukemia in children: a review. Environ Health Perspect 2007;115:138–45. [3] Turner MC, Wigle DT, Krewski D. Residential pesticides and childhood leukemia: a systematic review and meta-analysis. Environ Health Perspect 2010;118:33–41. [4] Van Maele-Fabry G, Lantin AC, Hoet P, Lison D. Residential exposure to pesticides and childhood leukaemia: a systematic review and meta-analysis. Environ Int 2011;37:280–91. [5] National Cancer Center. Annual report of cancer statistics in Korea in 2010. Seoul: Ministry of Health and Welfare, 2012. [6] Korea Crop Protection Association. Agrochemical year book. Seoul: Korea Crop Protection Association, 2012. [7] OECD. OECD Environmental Data Compendium 2008. Agriculture. Paris: OECD, 2008, http://www.oecd.org/statistics. [8] Cha ES, Kong KA, Moon EK, Khang YH, Lee WJ. Childhood cancer mortality and birth characteristics in Korea: a national population-based birth cohort study. J Korean Med Sci 2011;26:339–45. [9] Kong KA, Khang YH, Cha ES, Moon EK, Lee YH, Lee WJ. Childhood cancer mortality and socioeconomic position in South Korea: a national populationbased birth cohort study. Cancer Causes Control 2010;21:1559–67. [10] World Health Organization. International statistical classification of diseases and related health problems-tenth revision. Geneva: World Health Organization, 1992. [11] Lee WJ, Son M, Chun BC, Park ES, Lee HK, Coble J, et al. Cancer mortality and farming in South Korea: an ecologic study. Cancer Causes Control 2008;19:505–13. [12] Statistics Korea. Agricultural Census. Daejeon: Statistics Korea, 2000. [13] Wesseling C, Antich D, Hogstedt C, Rodrı´guez AC, Ahlbom A. Geographical differences of cancer incidence in Costa Rica in relation to environmental and occupational pesticide exposure. Int J Epidemiol 1999;28:365–74. [14] Lee WJ. Exposure assessment. In: Lee WJ, ed. Pesticides and health in public health. Seoul: Korea University Press, 2013: 77–112 [in Korean]. [15] Carozza SE, Li B, Elgethun K, Whitworth R. Risk of childhood cancers associated with residence in agriculturally intense areas in the United States. Environ Health Perspect 2008;116:559–65. [16] Rull RP, Gunier R, Von Behren J, Hertz A, Crouse V, Buffler PA, et al. Residential proximity to agricultural pesticide applications and childhood acute lymphoblastic leukemia. Environ Res 2009;109:891–9.
[17] Reynolds P, Von Behren J, Gunier RB, Goldberg DE, Hertz A, Harnly ME. Childhood cancer and agricultural pesticide use: an ecologic study in California. Environ Health Perspect 2002;110:319–24. [18] Reynolds P, Von Behren J, Gunier RB, Goldberg DE, Harnly M, Hertz A. Agricultural pesticide use and childhood cancer in California. Epidemiology 2005;16:93–100. [19] Bailey HD, Armstrong BK, de Klerk NH, Fritschi L, Attia J, Scott RJ, et al. Exposure to professional pest control treatments and the risk of childhood acute lymphoblastic leukemia. Int J Cancer 2011;129:1678–88. [20] Rudant J, Menegaux F, Leverger G, Baruchel A, Nelken B, Bertrand Y, et al. Household exposure to pesticides and risk of childhood hematopoietic malignancies: the ESCALE study (SFCE). Environ Health Perspect 2007;115:1787–93. [21] Glass DC, Reid A, Bailey HD, Milne E, Fritschi L. Risk of childhood acute lymphoblastic leukaemia following parental occupational exposure to pesticides. Occup Environ Med 2012;69:846–9. [22] Walker KM, Carozza S, Cooper S, Elgethun K. Childhood cancer in Texas counties with moderate to intense agricultural activity. J Agric Saf Health 2007;13:9–24. [23] Roberts JR, Karr CJ. Council on environmental health, pesticide exposure in children. Pediatrics 2012;130:e1765–88. [24] Alavanja MC, Ward MH, Reynolds P. Carcinogenicity of agricultural pesticides in adults and children. J Agromedicine 2007;12:39–56. [25] Alavanja MC, Bonner MR. Occupational pesticide exposures and cancer risk: a review. J Toxicol Environ Health B: Crit Rev 2012;15:238–63. [26] Weichenthal S, Moase C, Chan P. A review of pesticide exposure and cancer incidence in the Agricultural Health Study cohort. Environ Health Perspect 2010;118:1117–25. [27] U.S. EPA. Chemicals evaluated for carcinogenic potential; 2012, http://npic.orst.edu/chemicals_evaluated.pdf. [28] Garry VF, Harkins ME, Erickson LL, Long-Simpson LK, Holland SE, Burroughs BL. Birth defects, season of conception, and sex of children born to pesticide applicators living in the Red River Valley of Minnesota, USA. Environ Health Perspect 2002;110:441–9. [29] Winchester PD, Huskins J, Ying J. Agrichemicals in surface water and birth defects in the United States. Acta Paediatr 2009;98:664–9. [30] Vahter M, Gochfeld M, Casati B, Thiruchelvam M, Falk-Filippson A, Kavlock R, et al. Implications of gender differences for human health risk assessment and toxicology. Environ Res 2007;104:70–84. [31] Buffler PA, Kwan ML, Reynolds P, Urayama KY. Environmental and genetic risk factors for childhood leukemia: appraising the evidence. Cancer Invest 2005;23:60–75. [32] Adam M, Rebholz CE, Egger M, Zwahlen M, Kuehni CE. Childhood leukaemia and socioeconomic status: what is the evidence? Radiat Prot Dosimetry 2008;132:246–54. [33] Son M, Kim J, Oh J, Kawachi I. Inequalities in childhood cancer mortality according to parental socioeconomic position: a birth cohort study in South Korea. Soc Sci Med 2011;72:108–15. [34] Wigle DT, Turner MC, Krewski D. A systematic review and meta-analysis of childhood leukemia and parental occupational pesticide exposure. Environ Health Perspect 2009;117:1505–13. [35] Monge P, Wesseling C, Guardado J, Lundberg I, Ahlbom A, Cantor KP, et al. Parental occupational exposure to pesticides and the risk of childhood leukemia in Costa Rica. Scand J Work Environ Health 2007;33:293–303.
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407 [36] Harnly M, McLaughlin R, Bradman A, Anderson M, Gunier R. Correlating agricultural use of organophosphates with outdoor air concentrations: a particular concern for children. Environ Health Perspect 2005;113: 1184–9. [37] Lu C, Fenske RA, Simcox NJ, Kalman D. Pesticide exposure of children in an agricultural community: evidence of household proximity to farmland and take home exposure pathways. Environ Res 2000;84:290–302.
407
[38] Petridou ET, Dimitrova N, Eser S, Kachanov D, Karakilinc H, Varfolomeeva S, et al. Childhood leukemia and lymphoma: time trends and factors affecting survival in five Southern and Eastern European Cancer Registries. Cancer Causes Control 2013;24:1111–8. [39] Moon YO, Park EC, Shin HR, Won YJ, Jung KW, Lee JH, et al. Regional variation in accessing regional hospitals for cancer patients. Korean J Epidemiol 2006;28:152–61 [in Korean].
Contents lists available at ScienceDirect
Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention journal homepage: www.cancerepidemiology.net
Childhood leukemia mortality and farming exposure in South Korea: A national population-based birth cohort study Eun Shil Cha a, Seung-sik Hwang b, Won Jin Lee a,* a b
Department of Preventive Medicine, Korea University College of Medicine, South Korea Department of Social and Preventive Medicine, Inha University School of Medicine, South Korea
A R T I C L E I N F O
A B S T R A C T
Article history: Received 18 November 2013 Received in revised form 30 April 2014 Accepted 4 May 2014 Available online 10 June 2014
Objectives: The aim of this study was to evaluate the relationship between leukemia mortality and exposure to farming among children in South Korea. Methods: A retrospective cohort study of South Korean children was conducted using data collected by the national birth register between 1995 and 2006; these data were then individually linked to death data. A cohort of 6,479,406 children was followed from birth until either their death or until December 31, 2006. For surrogate measures of pesticide exposure, we used residence at birth, paternal occupation, and month of conception from the birth certificate. Farming and pesticide exposure indexes by county were calculated using information derived from the 2000 agricultural census. Poisson regression analyses were used to calculate rate ratios (RRs) of childhood leukemia deaths according to indices of exposure to agricultural pesticides after adjustment for potential confounders. Results: In total 585 leukemia deaths were observed during the study period. Childhood leukemia mortality was significantly elevated in children born in rural areas (RR = 1.43, 95%CI 1.09–1.86) compared to those in metropolises, and in counties with both the highest farming index (RR = 1.33, 95%CI 1.04–1.69) and pesticide exposure index (RR = 1.30, 95%CI 1.02–1.66) compared to those in the reference group. However, exposure–response associations were significant only in relation to the farming index. When the analyses were limited to rural areas, the risk of death from leukemia among boys conceived between spring and fall increased over those conceived in winter. Conclusions: Our results show an increase in mortality from childhood leukemia in rural areas; however, further studies are warranted to investigate the environmental factors contributing to the excess mortality from childhood leukemia in rural areas. ß 2014 Elsevier Ltd. All rights reserved.
Keywords: Cancer Children Cohort Death Pesticides Rural Seasonal variation
1. Introduction Leukemia is the most common form of childhood cancer in the world, accounting for around 30% of all cancers diagnosed in children younger than 15 years of age [1]. However, risk factors for the disease remain largely unknown, and the established causal factors – such as ionizing radiation and congenital genetic syndromes – together explain less than 10% of the cases. Several other environmental agents have been put forward to explain its etiology; these include parental smoking, electromagnetic fields, socioeconomic status, infection, and pesticides [2].
* Corresponding author at: Department of Preventive Medicine, Korea University College of Medicine, 73, Inchon-ro, Seongbuk-gu, Seoul 136-705, South Korea. Tel.: +82 2 2286 1413; fax: +82 2 927 7220. E-mail address: [email protected] (W.J. Lee). http://dx.doi.org/10.1016/j.canep.2014.05.003 1877-7821/ß 2014 Elsevier Ltd. All rights reserved.
Several reviews of epidemiological studies examining the potential association between childhood leukemia and residential pesticide exposure have been conducted [3,4]. Despite this existing epidemiological literature (suggesting elevated childhood leukemia risks related to household environmental pesticide exposure), the evidence for a causal relationship between exposure to agricultural pesticides and childhood leukemia is still limited. Studies on this topic among Asian populations are also lacking, and the results vary between countries due to differences in exposure and demographic characteristics. Moreover, most findings on childhood leukemia and pesticides were from case–control studies. In South Korea, the incidence of childhood leukemia was 5.0 per 100,000 population in 2010, accounting for 33.2% of all cancer incidence among children aged 0–14 years [5]. South Korea has traditionally been an agricultural nation, and it still retains as of 2011 a farming population of about three million (approximately 6% of the total population). The quantity of pesticides used was
402
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
recorded as 19,131 tons [6], and average pesticide consumption was 10.6 kg per hectare in 2011 – higher than that in other developed countries with intensive agriculture [7]. Despite widespread agricultural pesticide exposure in the Korean population, no study has examined the relationship between childhood leukemia and this exposure in South Korea. Therefore, the aim of this study is to examine the relationship between extent of farming and pesticide use in the county of residence, paternal occupation, and season of conception and mortality from childhood leukemia in South Korea by using national population-based retrospective birth cohort data from 1995 to 2006. 2. Materials and methods 2.1. Data sources Details of the study design and of the population have been described previously [8,9]. In brief, data from the national birth and death registration databases of Statistics Korea were used to establish a retrospective cohort of all children born in South Korea between 1995 and 2006. Records from the national birth registration database contain information on each child born (e.g., birth date, gender, address of residence at birth, personal identification number, and birth characteristics such as birth weight and gestational duration), as well as on its parents at the time of the birth of the child (e.g., age, education, and occupation). Of the 6,813,945 total births between 1995 and 2006, we excluded 5.3% of the records that lacked a complete personal identification number – an omission due mainly to administrative disagreements between the district offices of residence and those of birth registration. A total of 6,479,406 records of live births which occurred between 1995 and 2006 in South Korea were obtained and linked to the national death registration database for the same period via their unique personal identification numbers. These individuals were followed from birth until their death or until the end of the follow-up period on December 31, 2006. The maximum age at death was 11 years old. The causes of death in the registered death data are coded according to the International Classification of Diseases, 10th Revision (ICD-10) [10]. A leukemia death was defined as a case marked with ICD codes between C91 and C95. 2.2. Exposure indices As indicators of potential exposure to risk factors associated with farming, including pesticides, we used five indices. First, residence at birth – based on the 243 administrative districts in the cohort data – was categorized as metropolis, city, or rural area. Metropolises included Seoul and six other metropolitan cities in South Korea. Districts in the nine provinces outside these metropolises were classified into either cities or rural areas, depending on governmental administrative divisions by population size and rural characteristics. Second, we calculated the farming index using following algorithm applied in our previous study [11]. County-specific farming index = (% of full-time farm household + % of part-time farm household 0.3) % of farm households weighted with farming years % of farm households weighted with farm size. The calculation process of farming index by area is presented as an Appendix A. Full-time farm population refers to persons engaged exclusively in farming, while part-time farm population refers to farm workers who also were employed in non-farming jobs to earn money for more than 30 days during the course of a year. The score assigned to part-time farm population (i.e., 0.3) was weighted to reflect the intensity of farming compared with the full-time farm population based on our professional discretion. Farming years and
farm size were categorized into quartiles, i.e., 20, 21–32, 33–45, 46 years for farming years and 0.43, 0.44–0.89, 0.90–1.65, 1.66 ha for farm size. Information on the number of farm households by county, as well as farm size and farming years of each farm household, was derived from the agricultural census conducted by Statistics Korea [12]. Agricultural censuses were taken every 10 years from when the first census was taken in 1960 until 1990, and since 1995 they have been conducted on a 5-year basis. We chose the 2000 census as it was the midpoint of our cohort data. Data on the number of total households and population by county in 2000 was drawn from the database maintained by Statistics Korea (http://kosis.kr). Each term of the numerator of farming index represents the agricultural labor by each county and provides an estimate of total farming level per county. Third, we developed a pesticide exposure index for each county. The index was calculated by taking into account the area of cultivated land (Ai) and the frequency (Fi) and hours (Hi) of pesticide application for the major crops, and the algorithm was modified on the basis of a previous study [13]. The county-specific pesticide exposure index can be described as P4 A F H =population. The i represents the four major types of i i i i¼1 farming (i.e., rice, upland, fruit, and greenhouse) in South Korea. The crop-specific area by county was taken from the agricultural census, and Fi and Hi from a nationwide sampling survey of male farmers in South Korea in 2010 [14]. The numerator of the formula indicates an estimate of the total pesticide load. With total population of each county in the denominator, the pesticide exposure index provides an estimate of pesticides used per inhabitant. Fourth, besides these three regional variables, we classified parental occupations obtained from national birth certificate data. Paternal occupation was grouped into non-manual (i.e., legislators, senior officials and managers, professionals, technicians and associate professionals, clerks, and service and sale workers), manual (i.e., craft and related trades workers, plant, machine operators and assemblers, and elementary occupations), others (i.e., students, the unemployed, homemakers, and armed forces), and skilled agricultural, forestry and fishery workers. Finally, since parental exposure to pesticides may vary by month of conception, we defined month of conception for each child as subtracting the duration of gestation from the date of birth. The month of conception was categorized into four seasons; spring (March–May), summer (June–August), fall (September–November), and winter (December–February). 2.3. Statistical analysis The number of person-years was calculated by summing the numbers of days from the date of birth to the date of death or to the end of the follow-up period (December 31, 2006). Poisson regression was used to calculate mortality rate ratios (RRs) and 95% confidence intervals (95%CIs) of leukemia death according to our surrogates of pesticide exposure. For farming index and pesticide exposure index, we assigned counties into one of four groups based on quartiles of each index based on the distribution of total subjects (i.e., <0.02, 0.02–27.9, 28.0– 20,422.3, >20,422.3 for farming index and <0.006, 0.006– 0.05, 0.06–0.54, >0.54 for pesticide exposure index). We adjusted for gender and birth year of children in the analyses in order to consider any gender difference in leukemia mortality and potential cohort effect due to rapid development of medical technologies or accessibility to high-quality medical facilities with birth year. To control for potential confounders, we also included the variable of maternal education and paternal job in the analyses, which have been reported as a significant risk
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
403
Table 1 Distribution of farming exposure indicators and childhood leukemia deaths by residence at birth in South Korea. Farming exposure indicators
Residence at birth (number of counties) Total (n = 243)
Farming index Q1 Q2 Q3 Q4 Pesticide exposure index Q1 Q2 Q3 Q4 Cultivated area per inhabitant Total Paddy Upland Fruit Greenhouse % of full-time farm households Childhood leukemia (1995–2006) Person-years Leukemia deaths Rate per 100,000
Rural (n = 93)
61 61 61 60
0 4 32 57
(0.0) (4.3) (34.4) (61.3)
61 61 61 60
0 5 36 52
(0.0) (5.4) (38.7) (55.9)
9.6 5.2 3.4 0.7 0.3 14.5
19.9 10.8 7.4 1.2 0.5 30.2
42,862,104 585 1.36
factor for childhood cancer death in a previous study [9]. Detailed analyses of leukemia mortality by gender and age group (<1, 1–4, and 5–11 years) were conducted separately, and subgroup analysis was performed within the subset of rural areas. All statistical analyses were performed using Stata 12.0 (StataCorp, College Station, Texas). 2.4. Ethics statement We used publicly available birth and mortality data without any personal identifiers, and thus ethical approval was unnecessary.
City (n = 82)
Metropolis (n = 68)
Number of counties (%) 11 (13.4) 40 (48.8) 28 (34.1) 3 (3.7) Number of counties (%) 11 (13.4) 39 (47.6) 24 (29.3) 8 (9.8) Mean of area (hectare) 5.5 3.0 1.7 0.6 0.2 8.3
4,478,104 78 1.74
50 17 1 0
(73.5) (25.0) (1.5) (0.0)
50 17 1 0
(73.5) (25.0) (1.5) (0.0)
0.4 0.2 0.1 0.2 0.4 0.6
18,448,336 270 1.46
19,935,664 237 1.19
3. Results Rural areas were assigned higher farming exposure indicators, while metropolises were assigned a lower level on these indices (Table 1). Both the average area per inhabitant of cultivated land for all kinds of crops and the average proportion of full-time farm households were the highest in rural areas, followed by those in cities and metropolises as might be expected. Our cohort contained 42,862,104 person-years of observation and a total of 585 leukemia deaths from 1995 through 2006. The overall rate of mortality from childhood leukemia was 1.36 per 100,000 and the highest rate was shown in rural areas, with 1.74 per 100,000.
Table 2 Risks of childhood leukemia death according to farming exposure indicators and gender among children born in South Korea, 1995–2006. Farming exposure indicators
Residence at birth Metropolis City Rural area Ptrend Farming index Q1 Q2 Q3 Q4 Ptrend Pesticide exposure index Q1 Q2 Q3 Q4 Ptrend Paternal occupation Non-manual Manual Others Agriculture Season of conception Winter Spring Summer Fall a
Total (n = 585)
Boys (n = 342) a
Girls (n = 243)
Cases
RR (95%CI)
Cases
RR (95%CI)
Cases
RR (95%CI)
237 270 78
1.00 1.21 (1.02–1.45) 1.43 (1.09–1.86) 0.003
140 154 48
1.00 1.17 (0.93–1.48) 1.49 (1.06–2.09) 0.020
97 116 30
1.00 1.27 (0.97–1.67) 1.34 (0.87–2.05) 0.073
121 164 141 159
1.00 1.30 (1.03–1.64) 1.26 (0.98–1.60) 1.33 (1.04–1.69) 0.042
70 94 76 102
1.00 1.26 (0.93–1.72) 1.15 (0.83–1.59) 1.47 (1.08–2.01) 0.033
51 70 65 57
1.00 1.35 (0.94–1.94) 1.41(0.97–2.04) 1.13 (0.76–1.67) 0.521
121 157 152 155
1.00 1.31 (1.04–1.67) 1.26 (0.99–1.60) 1.30 (1.02–1.66) 0.065
71 86 89 96
1.00 1.20 (0.88–1.65) 1.23 (0.90–1.69) 1.37 (1.00–1.87) 0.057
50 71 63 59
1.00 1.48 (1.03–2.13) 1.30 (0.89–1.89) 1.20 (0.81–1.77) 0.533
379 140 47 19
1.00 1.28 (1.04–1.57) 1.64 (1.20–2.24) 1.10 (0.69–1.76)
211 94 27 10
1.00 1.53 (1.18–1.98) 1.80 (1.20–2.69) 1.00 (0.53–1.92)
168 46 20 9
1.00 0.97 (0.69–1.36) 1.45 (0.89–2.36) 1.25 (0.63–2.48)
139 153 158 135
1.00 0.95 (0.76–1.20) 0.99 (0.79–1.25) 0.98 (0.78–1.25)
77 94 95 76
1.00 1.05 (0.78–1.42) 1.08 (0.80–1.46) 1.00 (0.73–1.38)
62 59 63 59
1.00 0.83 (0.58–1.18) 0.88 (0.62–1.26) 0.96 (0.67–1.38)
Adjusted for gender, birth year, maternal education, and paternal occupation.
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
404
Table 3 Risks of death from childhood leukemia according to farming exposure indicators and gender among children born in rural areas in South Korea, 1995–2006. Farming exposure indicators
Farming index Q1 Q2 Q3 Q4 Ptrend Pesticide exposure index Q1 Q2 Q3 Q4 Ptrend Paternal occupation Non-manual Manual Others Agriculture Season of conception Winter Spring Summer Fall a
Total (n = 78)
Boys (n = 48) a
Girls (n = 30)
Cases
RR (95%CI)
Cases
RR (95%CI)
15 18 23 22
1.00 1.22 (0.61–2.43) 1.56 (0.81–3.01) 1.51 (0.78–2.96) 0.170
8 10 14 16
1.00 1.31 (0.52–3.33) 1.82 (0.76–4.37) 2.16 (0.91–5.11) 0.055
7 8 9 6
1.00 1.09 (0.39–3.03) 1.25 (0.46–3.40) 0.81 (0.26–2.48) 0.801
12 24 25 17
1.00 2.00 (1.00–4.00) 2.10 (1.05–4.20) 1.39 (0.66–2.95) 0.458
6 14 14 14
1.00 2.40 (0.92–6.26) 2.44 (0.93–6.39) 2.39 (0.91–6.28) 0.116
6 10 11 3
1.00 1.57 (0.57–4.34) 1.71 (0.62–4.69) 0.45 (0.11–1.85) 0.412
35 20 11 12
1.00 1.49 (0.85–2.63) 2.48 (1.26–4.89) 1.25 (0.62–2.51)
23 13 6 6
1.00 1.55 (0.76–3.15) 2.01 (0.81–4.94) 0.92 (0.35–2.40)
12 7 5 6
1.00 1.40 (0.54–3.63) 3.42 (1.20–9.74) 1.91 (0.68–5.33)
11 23 21 23
1.00 1.77 (0.86–3.64) 1.69 (0.81–3.50) 2.10 (1.02–4.30)
6 13 14 15
1.00 1.84 (0.70–4.83) 2.05 (0.79–5.35) 2.51 (0.97–6.47)
5 10 7 8
1.00 1.70 (0.58–4.98) 1.24 (0.39–3.91) 1.59 (0.52–4.86)
Cases
RR (95%CI)
Adjusted for sex, birth year, maternal education, and paternal occupation.
The relative risks of childhood leukemia were statistically significantly elevated among children born in rural areas (RR = 1.43, 95%CI 1.09–1.86) as well as in cities (RR = 1.21, 95%CI 1.02–1.45) compared to those in metropolises (Table 2). The risks of mortality from childhood leukemia among children in the counties of the second, third, and highest levels of regional farming and pesticide exposure indices were significantly elevated compared with those born in the reference group for each index, but exposure–response associations were significant only at the farming index. Mortality from childhood leukemia was not elevated for paternal agricultural work with reference to nonmanual occupations. There was no distinct association between childhood leukemia and season of conception. The risk estimates did not significantly differ in the analysis by age group, and the point estimates of the two major subtypes of leukemia (i.e., acute lymphocytic leukemia, acute myeloid leukemia) were not significantly different (data not shown). The results from the analyses among 93 rural counties showed results that did not differ materially from those for total subjects (Table 3). However, in rural areas a positive association between childhood leukemia and surrogate pesticide exposure indices was more apparent in boys than in girls. For season of conception, the risks of mortality from childhood leukemia were found to be higher for boys conceived between spring and fall compared to those conceived in winter. 4. Discussion We observed statistically significant elevated risks of childhood leukemia in rural areas and in counties with a higher level of farming index and pesticide exposure index. Within rural areas, the risk of childhood leukemia death increased among boys conceived between spring and fall compared to among those conceived in winter. Although these results may support the notion that agricultural pesticide exposure might be associated with risk of childhood leukemia, further investigation is required. Most previous studies have consistently reported increased risks of childhood leukemia as a result of farming exposure. Some ecological studies have found elevated risks of childhood leukemia in relation to proximity to and intensity of agricultural activity
[15,16]. Studies in California – applying a pesticide-use reporting database – have found significant increases in childhood leukemia in groups exposed to pesticides [17,18]. A number of case–control studies have also reported that childhood exposure to household insecticides and parental exposure to pesticides prior to and during pregnancy present the greatest risks [19,20]. Some studies, however, found no significant associations between childhood leukemia and either intense agricultural activity or parental occupational exposure to pesticides [21,22]. Children in rural areas or near treated croplands can be exposed to pesticides through agricultural application drift, or through contaminated soil, surface water, household dust and food [23]. Although specific biochemical mechanisms relating pesticide exposure to childhood cancer have yet to be established, some evidence suggests that pesticides may promote chromosomal aberrations, oxidative stress, and immunotoxicity, which could be linked to increased cancer risk [24]. Several pesticides identified as having a possible human carcinogenic effect by recent epidemiological literature [25,26] and US EPA’s Pesticide Program [27] – such as diazinon, chlorpyrifos, butachlor, alachlor, chlorothalonil, dichlorvos, pendimethalin, thiophanate-methyl, ethoprophos, trichlorfon, and oxadiazon – have been widely applied in South Korea [6]. Moreover, children are believed to be more vulnerable to the effects of environmental pollutants than are adults as a result of different physical dimensions, less mature immune systems, unique diets, and metabolic characteristics [23]. In subgroup analysis within rural areas we observed elevated risks of childhood leukemia among boys conceived between spring and fall (when agricultural pesticides are commonly applied in South Korea) compared to those conceived in winter. The fact that seasonality in childhood leukemia was observed only in rural areas may be explained by more common environmental exposure to pesticides in an agricultural setting than in urban areas. Previous studies have found an elevated risk of adverse birth outcome, such as birth defects, for infants conceived during the spraying season [28,29]. The results of a more pronounced association among boys, particularly in rural areas, observed in this study may be explained by greater susceptibility to toxicants such as pesticides among male fetuses, although the role of child gender as a potential
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
modifier of the effects of environmental toxicants has not been much investigated [30]. Another explanation may be that the risk of accidental childhood exposure linked to farm activities affects boys more often than girls. The increase in childhood leukemia mortality risk in rural areas, but the absence of an exposure–response relationship for pesticide exposure index observed in this study, suggest that this increase could reflect factors other than pesticide exposure that we were unable to examine. Besides pesticides, numerous environmental factors – such as infectious agents, socioeconomic status, air pollution, radiation, and other chemicals – have been suggested as risk factors for childhood leukemia [2,31]. Moreover, the socioeconomic status of parents in childhood leukemia may also complicate the associations [31]. Some studies of childhood leukemia have reported a higher incidence with higher socioeconomic status, while other studies have suggested a potential inverse association between socioeconomic status and incidence of childhood leukemia [32]. There were also significant social disparities in the survival of childhood leukemia in South Korea [33]. Although we adjusted for maternal education and paternal occupation in the analyses, residual confounding factors for socioeconomic status may affect incidence and survival of childhood leukemia. No significant association between childhood leukemia and paternal occupation as an agricultural worker was observed in this study, consistent with a systematic review and meta-analysis [34]. Previously, prenatal maternal exposure was reported to have an increased risk of childhood leukemia [34,35]. However, there were too few children (n = 5) whose mothers were agricultural workers in our analysis. There are several limitations to this study. First, regional agricultural pesticide exposure was measured by means of surrogate indices rather than through direct measurement. However, residence in agricultural communities or in proximity to farmland has been reported to be correlated with higher pesticide concentrations in house dust, air, and urinary metabolites compared with non-farm residence [36,37]. In this study, the distribution of counties for farming index and pesticide exposure index by area was similar, and the associations between childhood leukemia and each index also showed similar patterns. In addition, a lack of information on individual or groups of pesticides by area is an important limitation of this study. Nevertheless, the use of the crude surrogates for pesticide exposure in this study may bias effect estimates toward null. Second, the exposure of interest may also have been misclassified by using only a single address based on birth registration records to characterize exposure during the time period of interest. Among our subjects, the overall migration rate by area was 17.8%, and the rates of movement from rural to urban (9.6%) and from urban to rural (8.2%) areas did not appear to differ. Consequently, the phenomenon of children who had resided in rural areas at birth moving to urban areas may have diluted the exposure gradient. Third, the difference between rural and urban residence in childhood leukemia mortality may be affected by variation in access to well-equipped medical facilities and quality of medical services. These factors would influence the geographic disparities for survival from childhood leukemia [38]. Although advanced medical technology and a relatively high number of health-care facilities are present in South Korea, medical resources are unevenly distributed throughout the country [39]. To reduce the
405
geographic discrepancy in survival for childhood leukemia, we performed a subgroup analysis within the subset of 93 rural counties. The results, however, were generally similar to those observed among overall subjects in combined areas. In addition, the results derived from death at the age of less than 1 year, which could be more related as incidence cases than those of older age groups, also showed a similar pattern to those observed among overall subjects. Fourth, we were unable to examine whether the survival effect of childhood leukemia affected the associations observed in this study, since this study is based on deaths with an underlying cause coded as being due to leukemia. Although mortality data are not ideal for determining risk factors for childhood leukemia, we conducted subgroup analyses by age group in order to minimize the survival effects and also controlled for certain variables of socioeconomic status during the analyses in order to minimize the effect of variable access to medical treatment. Finally, the present analysis did not take into account other routes of pesticide exposure in children. Children may be exposed to pesticides through their use in their homes, and further studies need to incorporate information on pesticide exposure at the individual level with that at the ecological level. Despite these limitations, we have shown an increased risk of childhood leukemia mortality among children born in rural areas and in counties with higher levels of farming index and pesticide exposure index using national retrospective cohort data in South Korea. The associations tended to be most apparent among boys, and seasonality in leukemia mortality was observed in rural areas. However, further studies are warranted to investigate whether pesticide exposure contributes to the excess mortality from childhood leukemia in rural areas. Conflict of interests None declared. Acknowledgements This work was supported by Korea University Grant (K1031681), Republic of Korea.
Appendix A. The formula for county-specific farming index In the formula for farming index, the percentage of full-time farm population was a continuous variable based on the full-time farm population divided by the total population of each area. Fulltime farm population refers to persons engaged exclusively in farming, while part-time farm population refers to farm workers who also were employed in non-farming jobs to earn money for more than 30 days during the course of a year. Farming years and farm size were categorized into quartiles, i.e., 20, 21–32, 33–45, 46 years for farming years and 0.43, 0.44–0.89, 0.90–1.65, 1.66 ha for farm size. The score assigned to part-time farm population (i.e., 0.3) was weighted to reflect the intensity of farming compared with the full-time farm population based on our professional discretion. The term ‘weighted’ means multiplying the median values for each category in order to reflect weights according to farming years and farm size.
406
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407
No: of full-time farm household No: of part-time farm household 100 þ 100 0:3a Total household Total household No: of farm households with farming during 120 years 100 Total households No: of farm households with farming during 2132 years 14b þ 100 30b Total households No: of farm households with farming during 3345 years 100 40b þ Total households No: of farm households with farming during 4680 years 100 50b þ Total households No: of farm households with farming during 0:0010:429 hectare farm 100 Total households No: of farm households with farming during 0:4300:891 hectare farm 100 0:66b 0:25b þ Total households No: of farm households with farming during 0:8921:650 hectare farm 100 1:22b þ Total households No: of farm households with farming during 1:651165 hectare farm 100 2:50b þ Total households
County-specific farming index ¼
where a – weight for part-time farm population to reflect intensity of farming compared with full-time farm population based on professional discretion and b – median values for each category in order to reflect weights according to farming years and farm size. References [1] American Cancer Society. Global cancer facts & figures. 2nd ed. Atlanta: American Cancer Society, 2011 , http://www.cancer.org/acs/groups/content/ @epidemiologysurveilance/documents/document/acspc-027766.pdf. [2] Belson M, Kingsley B, Holmes A. Risk factors for acute leukemia in children: a review. Environ Health Perspect 2007;115:138–45. [3] Turner MC, Wigle DT, Krewski D. Residential pesticides and childhood leukemia: a systematic review and meta-analysis. Environ Health Perspect 2010;118:33–41. [4] Van Maele-Fabry G, Lantin AC, Hoet P, Lison D. Residential exposure to pesticides and childhood leukaemia: a systematic review and meta-analysis. Environ Int 2011;37:280–91. [5] National Cancer Center. Annual report of cancer statistics in Korea in 2010. Seoul: Ministry of Health and Welfare, 2012. [6] Korea Crop Protection Association. Agrochemical year book. Seoul: Korea Crop Protection Association, 2012. [7] OECD. OECD Environmental Data Compendium 2008. Agriculture. Paris: OECD, 2008, http://www.oecd.org/statistics. [8] Cha ES, Kong KA, Moon EK, Khang YH, Lee WJ. Childhood cancer mortality and birth characteristics in Korea: a national population-based birth cohort study. J Korean Med Sci 2011;26:339–45. [9] Kong KA, Khang YH, Cha ES, Moon EK, Lee YH, Lee WJ. Childhood cancer mortality and socioeconomic position in South Korea: a national populationbased birth cohort study. Cancer Causes Control 2010;21:1559–67. [10] World Health Organization. International statistical classification of diseases and related health problems-tenth revision. Geneva: World Health Organization, 1992. [11] Lee WJ, Son M, Chun BC, Park ES, Lee HK, Coble J, et al. Cancer mortality and farming in South Korea: an ecologic study. Cancer Causes Control 2008;19:505–13. [12] Statistics Korea. Agricultural Census. Daejeon: Statistics Korea, 2000. [13] Wesseling C, Antich D, Hogstedt C, Rodrı´guez AC, Ahlbom A. Geographical differences of cancer incidence in Costa Rica in relation to environmental and occupational pesticide exposure. Int J Epidemiol 1999;28:365–74. [14] Lee WJ. Exposure assessment. In: Lee WJ, ed. Pesticides and health in public health. Seoul: Korea University Press, 2013: 77–112 [in Korean]. [15] Carozza SE, Li B, Elgethun K, Whitworth R. Risk of childhood cancers associated with residence in agriculturally intense areas in the United States. Environ Health Perspect 2008;116:559–65. [16] Rull RP, Gunier R, Von Behren J, Hertz A, Crouse V, Buffler PA, et al. Residential proximity to agricultural pesticide applications and childhood acute lymphoblastic leukemia. Environ Res 2009;109:891–9.
[17] Reynolds P, Von Behren J, Gunier RB, Goldberg DE, Hertz A, Harnly ME. Childhood cancer and agricultural pesticide use: an ecologic study in California. Environ Health Perspect 2002;110:319–24. [18] Reynolds P, Von Behren J, Gunier RB, Goldberg DE, Harnly M, Hertz A. Agricultural pesticide use and childhood cancer in California. Epidemiology 2005;16:93–100. [19] Bailey HD, Armstrong BK, de Klerk NH, Fritschi L, Attia J, Scott RJ, et al. Exposure to professional pest control treatments and the risk of childhood acute lymphoblastic leukemia. Int J Cancer 2011;129:1678–88. [20] Rudant J, Menegaux F, Leverger G, Baruchel A, Nelken B, Bertrand Y, et al. Household exposure to pesticides and risk of childhood hematopoietic malignancies: the ESCALE study (SFCE). Environ Health Perspect 2007;115:1787–93. [21] Glass DC, Reid A, Bailey HD, Milne E, Fritschi L. Risk of childhood acute lymphoblastic leukaemia following parental occupational exposure to pesticides. Occup Environ Med 2012;69:846–9. [22] Walker KM, Carozza S, Cooper S, Elgethun K. Childhood cancer in Texas counties with moderate to intense agricultural activity. J Agric Saf Health 2007;13:9–24. [23] Roberts JR, Karr CJ. Council on environmental health, pesticide exposure in children. Pediatrics 2012;130:e1765–88. [24] Alavanja MC, Ward MH, Reynolds P. Carcinogenicity of agricultural pesticides in adults and children. J Agromedicine 2007;12:39–56. [25] Alavanja MC, Bonner MR. Occupational pesticide exposures and cancer risk: a review. J Toxicol Environ Health B: Crit Rev 2012;15:238–63. [26] Weichenthal S, Moase C, Chan P. A review of pesticide exposure and cancer incidence in the Agricultural Health Study cohort. Environ Health Perspect 2010;118:1117–25. [27] U.S. EPA. Chemicals evaluated for carcinogenic potential; 2012, http://npic.orst.edu/chemicals_evaluated.pdf. [28] Garry VF, Harkins ME, Erickson LL, Long-Simpson LK, Holland SE, Burroughs BL. Birth defects, season of conception, and sex of children born to pesticide applicators living in the Red River Valley of Minnesota, USA. Environ Health Perspect 2002;110:441–9. [29] Winchester PD, Huskins J, Ying J. Agrichemicals in surface water and birth defects in the United States. Acta Paediatr 2009;98:664–9. [30] Vahter M, Gochfeld M, Casati B, Thiruchelvam M, Falk-Filippson A, Kavlock R, et al. Implications of gender differences for human health risk assessment and toxicology. Environ Res 2007;104:70–84. [31] Buffler PA, Kwan ML, Reynolds P, Urayama KY. Environmental and genetic risk factors for childhood leukemia: appraising the evidence. Cancer Invest 2005;23:60–75. [32] Adam M, Rebholz CE, Egger M, Zwahlen M, Kuehni CE. Childhood leukaemia and socioeconomic status: what is the evidence? Radiat Prot Dosimetry 2008;132:246–54. [33] Son M, Kim J, Oh J, Kawachi I. Inequalities in childhood cancer mortality according to parental socioeconomic position: a birth cohort study in South Korea. Soc Sci Med 2011;72:108–15. [34] Wigle DT, Turner MC, Krewski D. A systematic review and meta-analysis of childhood leukemia and parental occupational pesticide exposure. Environ Health Perspect 2009;117:1505–13. [35] Monge P, Wesseling C, Guardado J, Lundberg I, Ahlbom A, Cantor KP, et al. Parental occupational exposure to pesticides and the risk of childhood leukemia in Costa Rica. Scand J Work Environ Health 2007;33:293–303.
E.S. Cha et al. / Cancer Epidemiology 38 (2014) 401–407 [36] Harnly M, McLaughlin R, Bradman A, Anderson M, Gunier R. Correlating agricultural use of organophosphates with outdoor air concentrations: a particular concern for children. Environ Health Perspect 2005;113: 1184–9. [37] Lu C, Fenske RA, Simcox NJ, Kalman D. Pesticide exposure of children in an agricultural community: evidence of household proximity to farmland and take home exposure pathways. Environ Res 2000;84:290–302.
407
[38] Petridou ET, Dimitrova N, Eser S, Kachanov D, Karakilinc H, Varfolomeeva S, et al. Childhood leukemia and lymphoma: time trends and factors affecting survival in five Southern and Eastern European Cancer Registries. Cancer Causes Control 2013;24:1111–8. [39] Moon YO, Park EC, Shin HR, Won YJ, Jung KW, Lee JH, et al. Regional variation in accessing regional hospitals for cancer patients. Korean J Epidemiol 2006;28:152–61 [in Korean].