Environmental Pollution 187 (2014) 81e89
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Quantitative assessments of indoor air pollution and the risk of childhood acute leukemia in Shanghai Yu Gao a,1, Yan Zhang a,1, Michihiro Kamijima b, Kiyoshi Sakai c, Md Khalequzzaman d, Tamie Nakajima d, Rong Shi a, Xiaojin Wang e, Didi Chen a, Xiaofan Ji a, Kaiyi Han a, Ying Tian a, f, * a
Department of Environmental Health, School of Public Health, Shanghai Jiao Tong University, 280 South Chongqing Road, Shanghai 200025, China Department of Occupational and Environmental Health, Nagoya City University Graduate School of Medical Sciences, Nagoya 466-8550, Japan Department of Environmental Health, Nagoya City Public Health Research Institute, Nagoya 467-8615, Japan d Department of Occupational and Environmental Health, Nagoya University Graduate School of Medicine, Nagoya 466-8550, Japan e Department of Biostatistics, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China f MOE and Shanghai Key Laboratory of Children’s Environmental Health, Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200092, China b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 October 2013 Received in revised form 24 December 2013 Accepted 27 December 2013
We investigated the association between indoor air pollutants and childhood acute leukemia (AL). A total of 105 newly diagnosed cases and 105 1:1 gender-, age-, and hospital-matched controls were included. Measurements of indoor pollutants (including nitrogen dioxide (NO2) and 17 types of volatile organic compounds (VOCs)) were taken with diffusive samplers for 64 pairs of cases and controls. Higher concentrations of NO2 and almost half of VOCs were observed in the cases than in the controls and were associated with the increased risk of childhood AL. The use of synthetic materials for wall decoration and furniture in bedroom was related to the risk of childhood AL. Renovating the house in the last 5 years, changing furniture in the last 5 years, closing the doors and windows overnight in the winter and/or summer, paternal smoking history and outdoor pollutants affected VOC concentrations. Our results support the association between childhood AL and indoor air pollution. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Acute leukemia Child Indoor air Volatile organic compounds
1. Introduction Childhood acute leukemia (AL) is the most common malignant tumor in children, comprising about one-third of all childhood cancers with an average annual incidence of 40.7 per million in Shanghai (Bao et al., 2009). Acute lymphocytic leukemia (ALL) accounts for approximately 80% of all childhood leukemia diagnoses, with incidence peaks at 25 years of age, indicating that exposure early in life is important (Pui, 2000; Wiemels, 2012). Different etiology mechanisms may be linked to subtypes of childhood leukemia. However, the causes of the childhood AL in the majority of cases are not known (Buffler et al., 2005). It is assumed to result from complex interactions between genetic predispositions and environmental factors, including air pollution (Buffler et al., 2005; Wiemels, 2012).
* Corresponding author. E-mail address: [email protected] (Y. Tian). 1 These authors contributed equally to this work. 0269-7491/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envpol.2013.12.029
The exposure to air pollutants can induce many health effects, such as sensory irritation, nervous system impairment, asthma and cancer (Billionnet et al., 2011; Prazad et al., 2008). Indoor air pollution may have a greater impact than outdoor pollution because children spend 90% of their time indoors and they have a greater physiological susceptibility to indoor air pollutants than adults do (Pickett and Bell, 2011; Whitworth et al., 2008). The potential sources of indoor air pollution are numerous, such as building materials and equipment, cleaning products, tobacco smoke, bio-contaminants, outdoor pollution, and combustion processes (Khalequzzaman et al., 2007; Pickett and Bell, 2011). In China, particularly in urban areas, with the living conditions of the residents improving, purchasing a new house and renovating the home has become increasingly popular. Home renovation as a source of indoor pollution and its impact on the health of children is an increasing concern in China (Zhou et al., 2011). Many indoor materials and utilities, such as paints, furnishings, carpets, and household cleaning products, contain volatile organic compounds (VOCs) (Guo et al., 2009; Pickett and Bell, 2011). The renovation of a house (painting, flooring, and new furniture) may
82
Y. Gao et al. / Environmental Pollution 187 (2014) 81e89
significantly affect indoor VOCs concentrations (Kim et al., 2002). For example, the most significant sources of formaldehyde (HCHO) at home are wood products using adhesives that contain ureaHCHO resins (Tang et al., 2009). Among the indoor VOCs, some are carcinogenic such as HCHO, benzene and 1,3-butadiene (Amigou et al., 2011; Whitworth et al., 2008). The previous studies of childhood AL and exposure to air pollutants mainly focused on traffic-related air pollution (Amigou et al., 2011; Reynolds et al., 2004; Whitworth et al., 2008), air pollution from stationary sources (Bailey et al., 2011; Brosselin et al., 2009), and environmental tobacco smoke (Brondum et al., 1999; Metayer et al., 2013). Most of those studies suggested that living close to heavy-traffic roads, living next to a gas station, and exposure to tobacco smoking were associated with increased risks of both childhood ALL and acute myeloid leukemia (AML) (Amigou et al., 2011; Whitworth et al., 2008; Metayer et al., 2013). However, no firm conclusion can be made at present for the association between air pollution and childhood AL given the limited number of studies, variations in study designs, differences in analytic and exposure assessment methodology among study results (Buka et al., 2007; Whitworth et al., 2008). Moreover, most previous studies were depended on representative modeled air pollutant levels rather than the actual measured concentrations, even less research has been conducted on the exposure to indoor air pollution (Pickett and Bell, 2011; Reynolds et al., 2004; Whitworth et al., 2008). There is a compelling need to evaluate the relationship between indoor air pollutants and the risk of childhood AL. We conducted a caseecontrol study of childhood AL in Shanghai. In the present study, we hypothesized that indoor pollutants such as HCHO, benzene, and other types of VOCs may be a risk factor for childhood leukemia. Our purpose was to explore the association between exposure to indoor air pollutants and childhood AL with both questionnaire-based responses and quantitative measurements of nitrogen dioxide (NO2) and VOCs using diffusive samplers. 2. Methods 2.1. Study population Between January 2008 and June 2011, we identified newly diagnosed childhood AL cases (less than 15 years old) in Shanghai. A 1:1 matched caseecontrol study design was used. The selection criteria are published elsewhere (Ding et al., 2012). Briefly, newly diagnosed (less than 2 wk) childhood AL according to the FrenchAmerican-British classification and admitted to four children’s hospitals in Shanghai were recruited in the study. The control subjects were healthy children who received physical check-up from the department of children’s healthcare or who visited the clinic of developmental pediatrics or orthopedics for reasons other than blood diseases and malignant tumors. To ensure that cases and controls were comparable in terms of socioeconomic category, the controls were selected from the same hospital as cases and were matched with cases in gender and age (1 y). Both cases and controls had resided in the same home before diagnosis/reference date for at least one year. Children who had been adopted were not eligible as either cases or controls. A total of 120 cases of childhood AL identified during the study period, and we excluded those who were not newly diagnosed when investigated (n ¼ 1), who did not meet the diagnostic criteria (n ¼ 1) and who refused to participate in this study and did not finish the questionnaire (n ¼ 13). Finally, a total of 105 cases and 105 matched controls were included in the study, and the case participation fraction was 105/118 (89%). Of the 105 pairs of cases and controls, 64 pairs agreed to participate in the indoor environment assessment (participation rate was 64/105 (61%)). We have compared the list of newly diagnosed Al cases included in our study with those registered in the Shanghai cancer registry, and the overlap ratio is more than 80%. All participating parents signed the informed consent form approved by the Shanghai Jiao Tong University School of Medicine Institutional Review Board. 2.2. Questionnaire data The mothers of the cases and controls were interviewed face to face by specifically trained interviewers using a structured questionnaire. Mothers of the children were asked to sign informed consent forms and to fill out questionnaires as soon as the cases were newly diagnosed and the corresponding controls were matched in
hospitals. The interview administered to the mother of a case usually preceded the interview of the mother of a corresponding control, with a lag time of 1e3 months. The questionnaire covered the period from pregnancy to diagnosis (cases) or reference date (controls). It included information on demographics and personal characteristics. The questions related to renovations and building materials included: When did you renovate the house or change furniture? What types of materials did you use for the wall decoration (such as natural wood, plywood, high density wood, water-borne coating, wall paper and paint), for floor construction (such as natural wood materials, high density wood, marble/granite, ceramic tile and carpet), and for furniture (such as natural wood, plywood and high density wood) in the child’s bedroom and in the parents’ bedroom? Several personal behavioral factors were also included, such as the frequency of cleaning the air conditioner, the type of fuel that was predominantly used in the house (such as natural gas, liquefied petroleum or biomass fuel), and whether the doors and windows were left open overnight in the winter and/or summer. In addition, questions about the daily household use of pesticides and whether the home is within 500 m of a gas station, industrial waste, heavy-traffic roads or a chemical factory were included. 2.3. Sampling and analytical methods The indoor air sampling was performed upon the child’s parents’ agreement to participate, and the sampling time for the case and the matched control households was within the same month. The air samples for the analysis of the concentrations of HCHO and NO2 were collected using a diffusion sampler packed with silica gel containing triethanolamine (passive gas tube for HCHO and NO2, Sibata Scientific Technology, Japan). The air samples for the analysis of the sixteen VOCs were collected using a diffusion sampler packed with activated charcoal (passive gas tube for organic solvents, Sibata Scientific Technology, Japan). The duration of sampling was approximately 24 h. Sampling was carried out in the child’s bedroom where the children spend the most time. The samples were taken at a height of 60e70 cm above the floor, corresponding to the height at which the children were positioned when sleeping. All the samples collected in Shanghai were transported to Japan by air and analyzed by the same researchers (Kiyoshi Sakai and Md Khalequzzaman). The VOCs and NO2 were analyzed by Sakai et al. (2004, 2006). In brief, HCHO and NO2 were extracted with distilled water and analyzed on the spectrophotometer with the 4amino-3-hydrazino-5-mercapto-1,2,4-triazole and sulfonamide methods, respectively. The VOCs were extracted with carbon disulfide and analyzed with gas chromatography - mass spectrometry (GCeMS). The GCeMS (5980 Series II/5971A, Hewlett Packard, USA) was equipped with a 60 m 0.25 mm i.d. capillary column coated with a 1.5 mm film of NB-1 (GL Sciences, Japan). The GC oven temperature was first maintained at 45 C for 5 min then programmed to increase to 300 C at 10 C/ min intervals and held at 300 C for 7 min. The analysis was performed with a helium flow rate of 0.9 ml/min under a selected-ion monitoring mode targeting 16 VOCs. For some samples, the analysis was performed under a total-ion monitoring mode in order to examine all of the major peaks followed by selected-ion monitoring mode targeting the 16 VOCs. The detection limit for HCHO and NO2 was defined as the concentration equivalent to the absorbance of 0.03. The detection limit for the other VOCs was defined as 10 times the standard deviation of the response of the lowest standard. This limit was determined each day that analysis occurred, and the worst (highest) limit was considered to be the overall detection limit for this study. The field and media blanks were tested each year of the analysis. There was no contamination of the samples by VOCs. 2.4. Statistical analysis The SPSS software package (version 15.0, SPSS, USA) was used for all of the analyses. The descriptive statistics of the questionnaire data were calculated with a conditional logistic regression model. When the concentration of an indoor chemical substance was below the limit of detection (LOD), the concentration was set at half of the LOD. Because the data remained skewed after lognormal transformation, median and quartiles were used for the descriptive statistics. The Wilcoxon test was used to compare the concentrations of each chemical between the cases and controls, and the Spearman correlation test was employed to determine the correlation among the personal VOCs. The total VOCs were calculated by two methods: a) the sum of the concentrations of the sixteen individual VOCs; b) the sum of the concentrations of the sixteen individual VOCs and HCHO. In addition to analyzing air pollutants individually, we categorized the VOCs into families as alkanes (hexane), aromatics (benzene, toluene, xylene and styrene), chlorinated hydrocarbons (trichloroethylene, tetrachloroethylene, chloroform, 1,1,1trichloroethane, 1,2-dichloroethane, carbon tetrachloride, and p-dichlorobenzene), esters (butyl acetate), aldehydes (HCHO), ketones (methyl ethyl ketone and methyl isobutyl ketone) and alcohols (butyl alcohol) to avoid the problem of multiple comparisons. The distribution of concentrations within each VOC family in the controls was stratified into quartiles. The exposure to each pollutant was categorized as a binary variable by classifying the exposure level as low or high, using the 3rd quartile value of the controls’ exposure as a threshold value. A conditional multiple logistic regression model was used to determine the possible risk factors for childhood AL. We calculated adjusted odds ratios (ORs) and
Y. Gao et al. / Environmental Pollution 187 (2014) 81e89
83
3. Results
their 95% confidence intervals (CIs) in the model, and the potential effects of the covariates were assessed using a forward step-wise regression process. Each possible risk factor was entered separately into the model, and the covariates were selected for the analyses if they were related to childhood leukemia in the literature or were significantly associated (p < 0.10) with the risk of childhood AL in our study (e.g. parental education levels, parental occupations, parental smoking histories, annual household income, household use of pesticides, season of indoor air collection, and outdoor pollution). We repeated the above logistic regression analysis, this time including other potential covariates which were not significantly different between the cases and controls in our study (p > 0.10) (i.e. parental age at delivery and childhood/maternal passive smoking history). As these covariates did not markedly alter the observed results, we did not include them in the final model. The test for linear trends was evaluated by modeling the quartile values as continuous variables in the logistic regression model, again adjusted for the covariates mentioned above. The VOC concentrations were categorized by quartile, and the spearman correlation coefficients were used to test for associations between potential risk factors from questionnaire and indoor air pollutants. Two-tailed tests of significance were used, and p < 0.05 indicated statistical significance.
Of the 105 patients included in the study, 85 were diagnosed with ALL and 20 were diagnosed with AML. The general demographics for overall childhood AL and ALL yielded similar results. The analyses of the questionnaires of all 105 pairs of cases and controls also yielded similar results to those of the 64 pairs for whom the indoor air pollution was measured, so we reported the questionnaire results for the 105 pairs of cases and controls. Table 1 shows the demographic characteristics of the participants. Nearly forty-five percent of the cases and controls were males, and the average age was 5.8 years (standard deviation (SD) ¼ 4.0). Cases and controls had similar distributions of age, gender, birth weight, parental age at delivery, paternal education level, parental occupations, smoking (for parents), passive smoking (for both parents and children), alcohol history (for parents), family
Table 1 Description of characteristics for childhood AL cases and controls. Characteristic
Values
Cases (n ¼ 105) No. (%) or mean SD
Controls (n ¼ 105) No. (%) or mean SD
c2
p
Gender
Male Female
47 (45) 58 (55) 3349.3 487.3 9 (8) 88 (84) 8 (8) 5.8 4.1 3 (3) 55 (52) 47 (45) 27.7 4.0 28 (27) 44 (42) 33 (31) 31.6 6.0 7 (7) 38 (36) 60 (57)
e
e
0.000 4.547
0.989 0.103
e e
e e
0.353 1.519
0.552 0.468
20w 25w 30w
47 (45) 58 (55) 3348.6 411.9 3 (3) 98 (93) 4 (4) 5.8 4.0 5 (5) 52 (49) 48 (46) 27.4 3.6 25 (24) 52 (49) 28 (27) 30.5 5.4 12 (11) 40 (38) 53 (51)
1.972 1.613
0.160 0.446
>High school degree High school diploma High school degree High school diploma
Contents lists available at ScienceDirect
Environmental Pollution journal homepage: www.elsevier.com/locate/envpol
Quantitative assessments of indoor air pollution and the risk of childhood acute leukemia in Shanghai Yu Gao a,1, Yan Zhang a,1, Michihiro Kamijima b, Kiyoshi Sakai c, Md Khalequzzaman d, Tamie Nakajima d, Rong Shi a, Xiaojin Wang e, Didi Chen a, Xiaofan Ji a, Kaiyi Han a, Ying Tian a, f, * a
Department of Environmental Health, School of Public Health, Shanghai Jiao Tong University, 280 South Chongqing Road, Shanghai 200025, China Department of Occupational and Environmental Health, Nagoya City University Graduate School of Medical Sciences, Nagoya 466-8550, Japan Department of Environmental Health, Nagoya City Public Health Research Institute, Nagoya 467-8615, Japan d Department of Occupational and Environmental Health, Nagoya University Graduate School of Medicine, Nagoya 466-8550, Japan e Department of Biostatistics, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China f MOE and Shanghai Key Laboratory of Children’s Environmental Health, Xinhua Hospital, Shanghai Jiao Tong University School of Medicine, Shanghai 200092, China b c
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 October 2013 Received in revised form 24 December 2013 Accepted 27 December 2013
We investigated the association between indoor air pollutants and childhood acute leukemia (AL). A total of 105 newly diagnosed cases and 105 1:1 gender-, age-, and hospital-matched controls were included. Measurements of indoor pollutants (including nitrogen dioxide (NO2) and 17 types of volatile organic compounds (VOCs)) were taken with diffusive samplers for 64 pairs of cases and controls. Higher concentrations of NO2 and almost half of VOCs were observed in the cases than in the controls and were associated with the increased risk of childhood AL. The use of synthetic materials for wall decoration and furniture in bedroom was related to the risk of childhood AL. Renovating the house in the last 5 years, changing furniture in the last 5 years, closing the doors and windows overnight in the winter and/or summer, paternal smoking history and outdoor pollutants affected VOC concentrations. Our results support the association between childhood AL and indoor air pollution. Ó 2014 Elsevier Ltd. All rights reserved.
Keywords: Acute leukemia Child Indoor air Volatile organic compounds
1. Introduction Childhood acute leukemia (AL) is the most common malignant tumor in children, comprising about one-third of all childhood cancers with an average annual incidence of 40.7 per million in Shanghai (Bao et al., 2009). Acute lymphocytic leukemia (ALL) accounts for approximately 80% of all childhood leukemia diagnoses, with incidence peaks at 25 years of age, indicating that exposure early in life is important (Pui, 2000; Wiemels, 2012). Different etiology mechanisms may be linked to subtypes of childhood leukemia. However, the causes of the childhood AL in the majority of cases are not known (Buffler et al., 2005). It is assumed to result from complex interactions between genetic predispositions and environmental factors, including air pollution (Buffler et al., 2005; Wiemels, 2012).
* Corresponding author. E-mail address: [email protected] (Y. Tian). 1 These authors contributed equally to this work. 0269-7491/$ e see front matter Ó 2014 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.envpol.2013.12.029
The exposure to air pollutants can induce many health effects, such as sensory irritation, nervous system impairment, asthma and cancer (Billionnet et al., 2011; Prazad et al., 2008). Indoor air pollution may have a greater impact than outdoor pollution because children spend 90% of their time indoors and they have a greater physiological susceptibility to indoor air pollutants than adults do (Pickett and Bell, 2011; Whitworth et al., 2008). The potential sources of indoor air pollution are numerous, such as building materials and equipment, cleaning products, tobacco smoke, bio-contaminants, outdoor pollution, and combustion processes (Khalequzzaman et al., 2007; Pickett and Bell, 2011). In China, particularly in urban areas, with the living conditions of the residents improving, purchasing a new house and renovating the home has become increasingly popular. Home renovation as a source of indoor pollution and its impact on the health of children is an increasing concern in China (Zhou et al., 2011). Many indoor materials and utilities, such as paints, furnishings, carpets, and household cleaning products, contain volatile organic compounds (VOCs) (Guo et al., 2009; Pickett and Bell, 2011). The renovation of a house (painting, flooring, and new furniture) may
82
Y. Gao et al. / Environmental Pollution 187 (2014) 81e89
significantly affect indoor VOCs concentrations (Kim et al., 2002). For example, the most significant sources of formaldehyde (HCHO) at home are wood products using adhesives that contain ureaHCHO resins (Tang et al., 2009). Among the indoor VOCs, some are carcinogenic such as HCHO, benzene and 1,3-butadiene (Amigou et al., 2011; Whitworth et al., 2008). The previous studies of childhood AL and exposure to air pollutants mainly focused on traffic-related air pollution (Amigou et al., 2011; Reynolds et al., 2004; Whitworth et al., 2008), air pollution from stationary sources (Bailey et al., 2011; Brosselin et al., 2009), and environmental tobacco smoke (Brondum et al., 1999; Metayer et al., 2013). Most of those studies suggested that living close to heavy-traffic roads, living next to a gas station, and exposure to tobacco smoking were associated with increased risks of both childhood ALL and acute myeloid leukemia (AML) (Amigou et al., 2011; Whitworth et al., 2008; Metayer et al., 2013). However, no firm conclusion can be made at present for the association between air pollution and childhood AL given the limited number of studies, variations in study designs, differences in analytic and exposure assessment methodology among study results (Buka et al., 2007; Whitworth et al., 2008). Moreover, most previous studies were depended on representative modeled air pollutant levels rather than the actual measured concentrations, even less research has been conducted on the exposure to indoor air pollution (Pickett and Bell, 2011; Reynolds et al., 2004; Whitworth et al., 2008). There is a compelling need to evaluate the relationship between indoor air pollutants and the risk of childhood AL. We conducted a caseecontrol study of childhood AL in Shanghai. In the present study, we hypothesized that indoor pollutants such as HCHO, benzene, and other types of VOCs may be a risk factor for childhood leukemia. Our purpose was to explore the association between exposure to indoor air pollutants and childhood AL with both questionnaire-based responses and quantitative measurements of nitrogen dioxide (NO2) and VOCs using diffusive samplers. 2. Methods 2.1. Study population Between January 2008 and June 2011, we identified newly diagnosed childhood AL cases (less than 15 years old) in Shanghai. A 1:1 matched caseecontrol study design was used. The selection criteria are published elsewhere (Ding et al., 2012). Briefly, newly diagnosed (less than 2 wk) childhood AL according to the FrenchAmerican-British classification and admitted to four children’s hospitals in Shanghai were recruited in the study. The control subjects were healthy children who received physical check-up from the department of children’s healthcare or who visited the clinic of developmental pediatrics or orthopedics for reasons other than blood diseases and malignant tumors. To ensure that cases and controls were comparable in terms of socioeconomic category, the controls were selected from the same hospital as cases and were matched with cases in gender and age (1 y). Both cases and controls had resided in the same home before diagnosis/reference date for at least one year. Children who had been adopted were not eligible as either cases or controls. A total of 120 cases of childhood AL identified during the study period, and we excluded those who were not newly diagnosed when investigated (n ¼ 1), who did not meet the diagnostic criteria (n ¼ 1) and who refused to participate in this study and did not finish the questionnaire (n ¼ 13). Finally, a total of 105 cases and 105 matched controls were included in the study, and the case participation fraction was 105/118 (89%). Of the 105 pairs of cases and controls, 64 pairs agreed to participate in the indoor environment assessment (participation rate was 64/105 (61%)). We have compared the list of newly diagnosed Al cases included in our study with those registered in the Shanghai cancer registry, and the overlap ratio is more than 80%. All participating parents signed the informed consent form approved by the Shanghai Jiao Tong University School of Medicine Institutional Review Board. 2.2. Questionnaire data The mothers of the cases and controls were interviewed face to face by specifically trained interviewers using a structured questionnaire. Mothers of the children were asked to sign informed consent forms and to fill out questionnaires as soon as the cases were newly diagnosed and the corresponding controls were matched in
hospitals. The interview administered to the mother of a case usually preceded the interview of the mother of a corresponding control, with a lag time of 1e3 months. The questionnaire covered the period from pregnancy to diagnosis (cases) or reference date (controls). It included information on demographics and personal characteristics. The questions related to renovations and building materials included: When did you renovate the house or change furniture? What types of materials did you use for the wall decoration (such as natural wood, plywood, high density wood, water-borne coating, wall paper and paint), for floor construction (such as natural wood materials, high density wood, marble/granite, ceramic tile and carpet), and for furniture (such as natural wood, plywood and high density wood) in the child’s bedroom and in the parents’ bedroom? Several personal behavioral factors were also included, such as the frequency of cleaning the air conditioner, the type of fuel that was predominantly used in the house (such as natural gas, liquefied petroleum or biomass fuel), and whether the doors and windows were left open overnight in the winter and/or summer. In addition, questions about the daily household use of pesticides and whether the home is within 500 m of a gas station, industrial waste, heavy-traffic roads or a chemical factory were included. 2.3. Sampling and analytical methods The indoor air sampling was performed upon the child’s parents’ agreement to participate, and the sampling time for the case and the matched control households was within the same month. The air samples for the analysis of the concentrations of HCHO and NO2 were collected using a diffusion sampler packed with silica gel containing triethanolamine (passive gas tube for HCHO and NO2, Sibata Scientific Technology, Japan). The air samples for the analysis of the sixteen VOCs were collected using a diffusion sampler packed with activated charcoal (passive gas tube for organic solvents, Sibata Scientific Technology, Japan). The duration of sampling was approximately 24 h. Sampling was carried out in the child’s bedroom where the children spend the most time. The samples were taken at a height of 60e70 cm above the floor, corresponding to the height at which the children were positioned when sleeping. All the samples collected in Shanghai were transported to Japan by air and analyzed by the same researchers (Kiyoshi Sakai and Md Khalequzzaman). The VOCs and NO2 were analyzed by Sakai et al. (2004, 2006). In brief, HCHO and NO2 were extracted with distilled water and analyzed on the spectrophotometer with the 4amino-3-hydrazino-5-mercapto-1,2,4-triazole and sulfonamide methods, respectively. The VOCs were extracted with carbon disulfide and analyzed with gas chromatography - mass spectrometry (GCeMS). The GCeMS (5980 Series II/5971A, Hewlett Packard, USA) was equipped with a 60 m 0.25 mm i.d. capillary column coated with a 1.5 mm film of NB-1 (GL Sciences, Japan). The GC oven temperature was first maintained at 45 C for 5 min then programmed to increase to 300 C at 10 C/ min intervals and held at 300 C for 7 min. The analysis was performed with a helium flow rate of 0.9 ml/min under a selected-ion monitoring mode targeting 16 VOCs. For some samples, the analysis was performed under a total-ion monitoring mode in order to examine all of the major peaks followed by selected-ion monitoring mode targeting the 16 VOCs. The detection limit for HCHO and NO2 was defined as the concentration equivalent to the absorbance of 0.03. The detection limit for the other VOCs was defined as 10 times the standard deviation of the response of the lowest standard. This limit was determined each day that analysis occurred, and the worst (highest) limit was considered to be the overall detection limit for this study. The field and media blanks were tested each year of the analysis. There was no contamination of the samples by VOCs. 2.4. Statistical analysis The SPSS software package (version 15.0, SPSS, USA) was used for all of the analyses. The descriptive statistics of the questionnaire data were calculated with a conditional logistic regression model. When the concentration of an indoor chemical substance was below the limit of detection (LOD), the concentration was set at half of the LOD. Because the data remained skewed after lognormal transformation, median and quartiles were used for the descriptive statistics. The Wilcoxon test was used to compare the concentrations of each chemical between the cases and controls, and the Spearman correlation test was employed to determine the correlation among the personal VOCs. The total VOCs were calculated by two methods: a) the sum of the concentrations of the sixteen individual VOCs; b) the sum of the concentrations of the sixteen individual VOCs and HCHO. In addition to analyzing air pollutants individually, we categorized the VOCs into families as alkanes (hexane), aromatics (benzene, toluene, xylene and styrene), chlorinated hydrocarbons (trichloroethylene, tetrachloroethylene, chloroform, 1,1,1trichloroethane, 1,2-dichloroethane, carbon tetrachloride, and p-dichlorobenzene), esters (butyl acetate), aldehydes (HCHO), ketones (methyl ethyl ketone and methyl isobutyl ketone) and alcohols (butyl alcohol) to avoid the problem of multiple comparisons. The distribution of concentrations within each VOC family in the controls was stratified into quartiles. The exposure to each pollutant was categorized as a binary variable by classifying the exposure level as low or high, using the 3rd quartile value of the controls’ exposure as a threshold value. A conditional multiple logistic regression model was used to determine the possible risk factors for childhood AL. We calculated adjusted odds ratios (ORs) and
Y. Gao et al. / Environmental Pollution 187 (2014) 81e89
83
3. Results
their 95% confidence intervals (CIs) in the model, and the potential effects of the covariates were assessed using a forward step-wise regression process. Each possible risk factor was entered separately into the model, and the covariates were selected for the analyses if they were related to childhood leukemia in the literature or were significantly associated (p < 0.10) with the risk of childhood AL in our study (e.g. parental education levels, parental occupations, parental smoking histories, annual household income, household use of pesticides, season of indoor air collection, and outdoor pollution). We repeated the above logistic regression analysis, this time including other potential covariates which were not significantly different between the cases and controls in our study (p > 0.10) (i.e. parental age at delivery and childhood/maternal passive smoking history). As these covariates did not markedly alter the observed results, we did not include them in the final model. The test for linear trends was evaluated by modeling the quartile values as continuous variables in the logistic regression model, again adjusted for the covariates mentioned above. The VOC concentrations were categorized by quartile, and the spearman correlation coefficients were used to test for associations between potential risk factors from questionnaire and indoor air pollutants. Two-tailed tests of significance were used, and p < 0.05 indicated statistical significance.
Of the 105 patients included in the study, 85 were diagnosed with ALL and 20 were diagnosed with AML. The general demographics for overall childhood AL and ALL yielded similar results. The analyses of the questionnaires of all 105 pairs of cases and controls also yielded similar results to those of the 64 pairs for whom the indoor air pollution was measured, so we reported the questionnaire results for the 105 pairs of cases and controls. Table 1 shows the demographic characteristics of the participants. Nearly forty-five percent of the cases and controls were males, and the average age was 5.8 years (standard deviation (SD) ¼ 4.0). Cases and controls had similar distributions of age, gender, birth weight, parental age at delivery, paternal education level, parental occupations, smoking (for parents), passive smoking (for both parents and children), alcohol history (for parents), family
Table 1 Description of characteristics for childhood AL cases and controls. Characteristic
Values
Cases (n ¼ 105) No. (%) or mean SD
Controls (n ¼ 105) No. (%) or mean SD
c2
p
Gender
Male Female
47 (45) 58 (55) 3349.3 487.3 9 (8) 88 (84) 8 (8) 5.8 4.1 3 (3) 55 (52) 47 (45) 27.7 4.0 28 (27) 44 (42) 33 (31) 31.6 6.0 7 (7) 38 (36) 60 (57)
e
e
0.000 4.547
0.989 0.103
e e
e e
0.353 1.519
0.552 0.468
20w 25w 30w
47 (45) 58 (55) 3348.6 411.9 3 (3) 98 (93) 4 (4) 5.8 4.0 5 (5) 52 (49) 48 (46) 27.4 3.6 25 (24) 52 (49) 28 (27) 30.5 5.4 12 (11) 40 (38) 53 (51)
1.972 1.613
0.160 0.446
>High school degree High school diploma High school degree High school diploma
