ENVIRONMENTAL RESEARCH 55, 31--39 (1991)
Methylene Chloride Exposure and Birthweight in Monroe County, New York 1 BETH P. BELL, *'2 PETER FRANKS,t NANCY HILDRETH,-~ AND JAMES MELIUS§ *Departments of Family Medicine and Preventive Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; ?Department of Family Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; SDepartment of Preventive Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; and §New York State Department of Health, Division of Occupational Health and Environmental Epidemiology, Albany, New York 12003 Received July 12, 1990 This study examined the relationship between birthweight and exposure to emissions of methylene chloride (DCM) from manufacturing processes of the Eastman Kodak Company at Kodak Park in Rochester, Monroe County, New York. County census tracts were categorized as exposed to high, moderate, low or no DCM based on the Kodak Air Monitoring Program (KAMP) model, a theoretical dispersion model of DCM developed by Eastman Kodak Company. Birthweight and information on variables known to influence birthweight were obtained from 91,302 birth certificates of white singleton births to Monroe County residents from 1976 to 1987. No significant adverse effects of exposure to DCM on birthweight were found. Adjusted birthweight in high exposure census tracts was 18.7 g less than in areas with no exposure (95% confidence interval for the difference between high and no exposure -51.6, 14.2 g). Problems inherent in the method of estimation of exposure, which may decrease power or bias the results, are discussed. Better methods to estimate exposure to emissions from multiple industrial point sources are needed. © 1991 AcademicPress,Inc.
INTRODUCTION Methylene chloride (dichloromethane, DCM) is a chlorinated hydrocarbon solvent, widely used in industry and in household products. In humans, exposure to DCM causes central nervous system depression and nervous system disorders. (Tabacova, 1986; Putz et al., 1976; Winneke, 1974). The chemical is ranked class B2 by the Environmental Protection Agency, indicating sufficient animal data to suggest carcinogenicity (Maltoni et al., 1986, 1988; Environmental Protection Agency, 1987). There are two studies of adverse reproductive effects of exposure to DCM, both in occupational settings. Both found increases in the spontaneous abortion rate of borderline statistical significance for women exposed to DCM during the first trimester of pregnancy (Taskinen et al., 1986; Axelsson et al., 1984). Only one of these studies examined birthweight; no measurable effect of exposure was found (Axelsson et al., 1984). ' Results presented in part at the Society for Epidemiologic Research annual meeting, Snowbird, Utah, June, 1990. 2 Currently Associate Medical Officer of Health, Waterloo, Ontario, Canada, N2G IE8. 31 0013-9351/91 $3.00 Copyright© 1991by AcademicPress, Inc. All fightsof reproductionin any formreserved.
32
BELL ET AL.
No studies have examined the possible adverse reproductive effects of residential exposure to DCM, although birthweight has been shown to be affected by exposure to other environmental contaminants. Decrements in birthweight have been demonstrated in women exposed to airborne heavy metals (McMichael et al., 1986; Nordstrom et al., 1978) and in populations residing near dumpsites (Vianna and Polan, 1984; New Jersey Department of Health, 1989), which often contain DCM (New Jersey Department of Health, 1989; Ozonoff et al., 1987). In 1987, the Kodak Park facility of the Eastman Kodak Company in Rochester, N.Y. reported environmental emissions of 62 chemicals under SARA Title III, Section 313 regulations. The plant has emitted DCM into the air for more than 40 years. In 1988, DCM emissions totalled nearly 9 million pounds (Times Union, 1988). Public awareness of the magnitude of air emissions, particularly of DCM, raised concern about possible health effects. As a result, this study was initiated to explore the possible association of the adverse pregnancy outcome of decreased birthweight and exposure to airborne DCM, using a theoretical computergenerated air dispersion model to define DCM exposure. MATERIALS AND METHODS
Birth certificates were obtained for all births in Monroe County in 1976--1987. Multiple births were excluded from the analysis. Infants weighing less than 750 grams were also excluded because of incomplete ascertainment and less accuracy of recorded birthweight of births at very low weights (Ferrara et al., 1988). Because of the known major differences in the distribution of birthweight and in the relationship of risk factors to birthweight between whites and nonwhites, the two groups were not considered together (Jason et al., 1986; Institute of Medicine, 1985; Hughes et al., 1989; Franks, 1987). Only 309 of the total 27,313 births to nonwhites in the county were in women living in the moderate and high exposure areas combined. Of these, 19 weighed less than 2500 grams at birth. Because of the few births among nonwhites in the areas of higher exposure, this study was restricted to white births. Thus the study population included the 91,302 white singleton births weighing 750 g or more, born to mothers residing in Monroe County in 1976-1987. Data available on birth certificates from all years included date of birth, census tract of residence, age, race, and educational level of mother and father, sex, gestational age, multiple births (plurality), month of pregnancy that prenatal care began, total previous births (called "gravidity" in this paper) and total previous live births (called "parity") of the mother, and conditions present during pregnancy. Other data available on the birth certificates only in some years was not included for study. A study of the concordance of Upstate New York (excluding New York City) birth certificate data and hospital records found agreement on infant's sex and race, and mother's and father's age, education, and number of previous live births to be 95-99.9% (New York State Department of Health, 1979). Month prenatal care began agreed 82% of the time. Hospital charts consistently indicated more conditions present during pregnancy than birth certificates. Birthweight was within 28 grams on 91.7% of the records.
METHYLENE CHLORIDE AND BIRTHWEIGHT
33
Dichotomized variables created from birth certificate data included race (white or not), complications of pregnancy ( present or not), late prenatal care (prenatal care begun in the third trimester or not), and term birth (gestationat age greater than 37 weeks or not). Other categorical variables included education (less than high school, high school, or greater than high school), parity (0, 1-3, >/4) and previous pregnancy loss, defined as gravidity minus parity (0, 1-2,/>3). Maternal age was considered both as a continuous variable and categorized (~34). Cutpoints for categorical variables were determined by examining the distribution of the variable within the sample, taking into consideration known risk factors for decreased birthweight. For example, both young (usually ~4 Previous L o s s e s 0 1-2 />3 Maternal Age 34 Exposure None Low Mod High
Statisticb
N
Mean
SE ~
(P)
46858 44444
3468.1 3339.7
2.6 2.4
t9220.3 = - 36.4 (0.0001)
87786 1437
3411.6 3287.7
1.8 14.5
t148o.1 = 8.5 (0.0001)
77961 7493
3428.2 3290.2
1.8 8.2
t82o8.6 = 16.4 (0.0001)
10109 35841 44166
3287.5 3388.7 3453.0
5.8 2.8 2.5
F2 = 444.58 (0.0001)
41521 47925 1799
3346.7 3457.7 3488.6
12.7 12.8 12.5
F 2 = 517.1 (0.0001)
70320 19451 1474
3404.9 3419.7 3361.1
14.0 14.4 13.9
F 2 = 11.49
1777 81871 7590
3275.0 3405.3 3460.9
12.6 1.9 6.1
F 2 = 93.6 (0.0001)
82076 6044 1795 1085
3407.7 3403.9 3400.8 3396.1
16.3 17.6 20.5 16.2
(0.0001)
F 3 = .35 (0.79)
a SE, standard error. b t , t test; F , F statistic.
during the period of greatest exposure weighed 66-74 g less than those in the nonexposed area (New Jersey Department of Health, 1989). It is important to emphasize that the negative findings of this study apply only to the levels of exposure in this particular setting and have little bearing on the possible effect of higher DCM exposures, such as in occupational settings. If DCM does adversely affect birthweight, it may occur at exposure levels higher than those estimated in this population. The highest predicted average concentration of DCM in this population, 50 p~g/m3, is below the proposed standard for community exposure to DCM, 60 p~g/m3, and an order of magnitude lower than that often found in occupational settings. In addition, the exposure levels, derived from the KAMP model, may in fact overestimate exposure. Ambient air sampling, conducted by Kodak between 1978
36
BELL
ET AL.
TABLE2 A D J U S T E D E F F E C T OF RISK F A C T O R S O N B I R T H W E I G H T ( G R A M S ) ( N Risk
factor
Parameter estimate a
= 82514)
95% Confidence interval b
Exposure (None)
Low Mod High Maternal Education Less than H S HS
(Greater than Parity
1.4
- 10.7, 13.5
- 1.9
-27.7,
23.9
- 18.7
-51.6,
14.2
- 160.8 -64.6
- 173.7, - 147.9 -72.4,
56.8
HS)
0 1-3
- 171.3
- 197.8, - 144.8
- 63.8
- 89.3, - 38.3
(/>4)
Previous losses 0
18.3
-9.9,
1-2
28.4
- 0.6, 5 7 . 4
46.5
(/>3) Maternal age (continuous) Late care Male sex Comp preg
-0.04 - 58.6
-0.92, 0.84 - 86.8, - 30.4
132.8
125.9, 139.7
- 114.2
- 126.6, - 1 0 1 . 9
a Parameter estimate is adjusted effect of the factor on birthweight compared to the group in parentheses (reference group). b Confidence interval of the effect of factor present compared to reference group.
and 1985, measured 1-hr composite samples of several chemicals six times per year. The KAMP estimate for DCM concentration is more than twice the levels found in these composite samples (personal communication, New York State Department of Health). Thus actual ambient air concentrations of DCM in this community may be half those predicted, a dose perhaps too low to produce a detectable clinical effect on birthweight. DCM is metabolized in vivo to carbon monoxide. In occupational studies, exposure to low levels of DCM (33-50 ppm) raised mean blood carboxyhemoglobin (CoHgb) concentrations to up to 4%, with higher concentrations seen with higher levels of exposure (Stewart et al., 1972; DiVincenzo and Kaplan, 1981a, 1981b; Ratney et al., 1974; Anders and Sunram, 1982). Both DCM and carbon monoxide cross the placenta and have been found in fetal tissue of pregnant women occupationally exposed to high levels of DCM. (Tabacova, 1986; Schwetz et al., 1975). The estimated exposure levels in our study are also below the level at which discernable increases in CoHgb concentration would be expected. To the extent that any effect of DCM exposure on birthweight is mediated through CoHgb formation, decreases in birthweight would not be anticipated at this exposure level. Misclassification of exposure may have arisen in a number of ways, contributing to the lack of a detectable effect of DCM exposure on birthweight. In contrast
37
METHYLENE CHLORIDE AND BIRTHWEIGHT TABLE 3 ADJUSTED EFFECT OF RISK FACTORS ON ODDS OF LOW BIRTHWEIGHT (N = 82514) Risk factor Exposure (None) Low Moderate High Maternal education Less than HS HS (greater than HS) Parity
Odds ratio n
95% Confidence interval
1.05 1.06 1.00
0.93, 1.17 0.93, 1.22 0.81, 1.24
1.42 0.99
1.34, 1.51 0.93, 1.05
0.82 1.19
0.75, 0.90 1.01, 1.23
1.16 1.06 1.14 0.91 2.05
1.06, 0.97, 1.00, 0.87, 1.97,
(0) l-3 I>4 Previous losses
(0) 1-2 >/3 Late care Male sex Comp preg
1.27 1.17 1.29 0.94 2.14
a Group in parentheses is the reference group, odds ratio = 1.
to the model's exposure isopleths, the actual level of exposure does not change abruptly. Thus, even if the model accurately predicts the concentration and dispersion pattern of DCM, these artificial cutpoints estimate exposure imprecisely, diluting any effect and reducing the power of the study. Similarly, the exposure of women living in census tracts crossed by an exposure isopleth was also poorly estimated because all births in a census tract crossed by an exposure isopleth were classified into the exposure category of the majority of the population of the census tract. Finally, as the distribution of births in a census tract is not fixed and may not conform to the distribution of the population, the actual proportion of births misclassified by this method is unknown and varied over time. Unmeasured confounding variables also may be playing a role. Although adjusting for maternal education may partially account for the confounding effect of smoking (Stein and Kline, 1983; Office on Smoking and Health, 1980), a small decrement in birthweight attributable to DCM exposure might be difficult to detect. Similarly, exposure to other environmental toxins could mask the detection of the effect of any one substance such as DCM. Employment is another unmeasured variable of potential importance. Assuming residence near employment, those occupationally exposed to DCM might also live in high exposure census tracts. Data on Kodak employees from a 1980 New York State Department of Health pilot study support this assumption in this study population (personal communication Joan P. Cooney, New York State Department of Health). This biases the result in favor of an effect and is of less importance in this negative study. As public awareness of potential effects of environmental pollutants grows and
38
BELL ET AL.
the government shows more interest in regulating such emissions, more studies of effects of airborne contaminants are likely. It is therefore important that modeling techniques such as the KAMP model used in this study be validated with extensive actual monitoring. The relationship between estimates of airborne concentrations of substances, generally calculated as annual averages, and individual dose should be clarified. Verification of the validity of these models will allow for more confidence in the findings of studies of the effects of airborne contaminants. However, this study's negative results suggest that the effect, if any, of this level of DCM exposure on birthweight is likely to be small. Further studies of effects of DCM exposure on reproductive outcomes probably should examine more highly exposed groups such as pregnant women workers, and other reproductive outcomes. ACKNOWLEDGMENT Research partially supported by a grant from the Centers of Disease Control and the Association of Teachers of Preventive Medicine.
REFERENCES Anders, M. W., and Sunram, J. M. (1982). Transplacental passage of dichloromethane and carbon monoxide. Toxicol. Lett. 12, 231-234. Axelsson, G., Lutz, C., and Rylander, R. (1984). Exposure to solvents and outcome of pregnancy in university laboratory employees. Brit. J. Ind. Med. 41, 305-312. DiVincenzo, G. D., and Kaplan, C. J. (1981a). Effect of exercise or smoking on the uptake, metabolism, and excretion of methylene chloride vapor. Toxicol. Appl. Pharmacol. 59, 141-148. DiVincenzo, G. D., and Kaplan, C. J. (1981b). Uptake, metabolism, and elimination of methylene chloride vapor by humans. Toxicol. Appl. Pharmacol. 59, 130-140. Eastman Kodak Co. (1988). Draft environmental impact statement for Eastman Kodak Company's proposed cellulose triacetate film base machine. Vol. II: Dispersion modelling analysis for air emissions from the manufacture of cellulose triacetate film base at Kodak Park. Environmental Protection Agency (1987). Health Advisory for Dichloromethane. Draft. Environmental Protection Agency, Office of Drinking Water. Washington, DC. Ferrara, A., Atakent, Y. S., and Levinson, B. (1988). Discrepancies in birth weights between hospital records and health department data for low birth weight infants in New York City. Public Health Rep. 103, 472-478. Franks, P. (1987). Social Inequities in Health: Rochester, New York. Faro. Med. 19, 438-443. Hughes, D., Johnson, K., Rosenbaum, S., and Liu, J. (1989). "The Health of America's Children." Children's Defense Fund. Institute of Medicine (1985). "Preventing Low Birthweight," p. 38, National Academy Press, Washington, DC. Jason, C. J., Samuhel, M. E., Glick, B. J., and Welsh, A. K. (1986). Geographic distribution of unexplained low birth weight. J. Occup. Med. 28:8, 728-740. Maltoni, C., Cotti, G., and Perino, G. (1988). Long-term carcinogenicity bioassays on methylene chlorine administered by ingestion to Sprague-Dawley rats and Swiss mice and by inhalation to Sprague-Dawley rats. Ann. N Y A c a d . Sci. 534, 352-366. Maltoni, C., Cotti, G., and Perino, G. (1986). Experimental research on methylene chloride carcinogenesis. "Archives of Research on Industrial Carcinogenesis. Vol IV." Princeton Scientific Publications, Princeton, NJ. McMichael, A. J., Vimpani, G. V., Robertson, E. F., Baghurst, P. A., and Clark, P. D. (1986). The Port Pirie cohort study: Maternal blood lead and pregnancy outcome. J. Epidemiol. Community Health 40, 18-25. New Jersey Department of Health, Division of Occupational and Environmental Health (1989). A Report on the health study of residents living near the Lipari landfill.
METHYLENE CHLORIDE AND BIRTHWEIGHT
39
New York State Department of Health (1979). "Reliability of Statistical and Medical Information Reported on Birth and Death Certificates." New York State Department of Health Monograph, No. 15. Nordstrom, S., Beckman, O., and Norderson, I. (1978). Occupational and environmental risks in and around a smelter in northern Sweden. Hereditas 88, 43--46. Office on Smoking and Health (1980). "The Health Consequences of Smoking for Women: A Report of the Surgeon General." Public Health Service. US Government Printing Office, Washington, DC. Ozonoff, D., Colten, M. E., Cupples, A., Heeren, T., Schatzkin, A., Mangione, T., Dresner, M,, and Colton, T. (1987). Health problems reported by residents of a neighborhood contaminated by a hazardous waste facility. Am. J. Ind. Med. 11, 581-597. Putz, V. R., Johnson, B. L., and Setzer, J. V. (1976). A comparative study of the effects of carbon monoxide and methylene chloride on human performance. J. Environ. Pathol. Toxicol. 2, 97-112. Ratney, R. S., Wegman, D. H., and Elkins, H. B. (1974). In vivo conversion of methylene chloride to carbon monoxide. Arch. Environ. Health 29, 223-226. SAS Institute Inc. (1988). "SAS Statistical Guide for Personal Computers, Release 6.03 Edition." SAS Institute Inc., Cary, NC. Schwetz, B. A., Leong, B. K. J., and Gehring, P. J. (1975). The effect of maternally inhaled trichloroethylene, perchloroethylene, methyl chloroform, and methylene chloride on embryonal and fetal development in mice and rats. Toxicol. Appl. Pharmacol. 32, 84-96. Stein, Z., and Kline, J. (1983). Smoking, alcohol and reproduction. Am. J. Public Health 73, 11541156. Stewart, R. D., Fisher, T. N., Hosko, M. J., Peterson, J. E., Baretta, E. D., and Dodd, H. C. (1972). Carboxyhemoglobin elevation after exposure to dichloromethane. Science 176, 295-296. Tabacova, S. (1986). Maternal exposure to environmental chemicals. Neurotoxicology 7, 421--440. Taskinen, H., Lindbohm, M. L., and Hemminki, K. (1986). Spontaneous abortions among women working in the pharmaceutical industry. Brit. J. Ind. Med. 43, 199-205. Times-Union (November 28, 1988). Expanding chemical use at Kodak. Times-Union, Rochester, NY. Vianna, N. J., and Polan, A. K. (1984). Incidence of low birth weight among Love Canal residents. Science 226, 1217-1219. Winneke, G. (1974). Behavioral effects of methylene chloride and carbon monoxide as assessed by sensory and psychomotor performance. In "Behavioral Toxicology." US Government Printing Office, Washington, DC.
Methylene Chloride Exposure and Birthweight in Monroe County, New York 1 BETH P. BELL, *'2 PETER FRANKS,t NANCY HILDRETH,-~ AND JAMES MELIUS§ *Departments of Family Medicine and Preventive Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; ?Department of Family Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; SDepartment of Preventive Medicine, University of Rochester School of Medicine and Dentistry, Rochester, New York 14642; and §New York State Department of Health, Division of Occupational Health and Environmental Epidemiology, Albany, New York 12003 Received July 12, 1990 This study examined the relationship between birthweight and exposure to emissions of methylene chloride (DCM) from manufacturing processes of the Eastman Kodak Company at Kodak Park in Rochester, Monroe County, New York. County census tracts were categorized as exposed to high, moderate, low or no DCM based on the Kodak Air Monitoring Program (KAMP) model, a theoretical dispersion model of DCM developed by Eastman Kodak Company. Birthweight and information on variables known to influence birthweight were obtained from 91,302 birth certificates of white singleton births to Monroe County residents from 1976 to 1987. No significant adverse effects of exposure to DCM on birthweight were found. Adjusted birthweight in high exposure census tracts was 18.7 g less than in areas with no exposure (95% confidence interval for the difference between high and no exposure -51.6, 14.2 g). Problems inherent in the method of estimation of exposure, which may decrease power or bias the results, are discussed. Better methods to estimate exposure to emissions from multiple industrial point sources are needed. © 1991 AcademicPress,Inc.
INTRODUCTION Methylene chloride (dichloromethane, DCM) is a chlorinated hydrocarbon solvent, widely used in industry and in household products. In humans, exposure to DCM causes central nervous system depression and nervous system disorders. (Tabacova, 1986; Putz et al., 1976; Winneke, 1974). The chemical is ranked class B2 by the Environmental Protection Agency, indicating sufficient animal data to suggest carcinogenicity (Maltoni et al., 1986, 1988; Environmental Protection Agency, 1987). There are two studies of adverse reproductive effects of exposure to DCM, both in occupational settings. Both found increases in the spontaneous abortion rate of borderline statistical significance for women exposed to DCM during the first trimester of pregnancy (Taskinen et al., 1986; Axelsson et al., 1984). Only one of these studies examined birthweight; no measurable effect of exposure was found (Axelsson et al., 1984). ' Results presented in part at the Society for Epidemiologic Research annual meeting, Snowbird, Utah, June, 1990. 2 Currently Associate Medical Officer of Health, Waterloo, Ontario, Canada, N2G IE8. 31 0013-9351/91 $3.00 Copyright© 1991by AcademicPress, Inc. All fightsof reproductionin any formreserved.
32
BELL ET AL.
No studies have examined the possible adverse reproductive effects of residential exposure to DCM, although birthweight has been shown to be affected by exposure to other environmental contaminants. Decrements in birthweight have been demonstrated in women exposed to airborne heavy metals (McMichael et al., 1986; Nordstrom et al., 1978) and in populations residing near dumpsites (Vianna and Polan, 1984; New Jersey Department of Health, 1989), which often contain DCM (New Jersey Department of Health, 1989; Ozonoff et al., 1987). In 1987, the Kodak Park facility of the Eastman Kodak Company in Rochester, N.Y. reported environmental emissions of 62 chemicals under SARA Title III, Section 313 regulations. The plant has emitted DCM into the air for more than 40 years. In 1988, DCM emissions totalled nearly 9 million pounds (Times Union, 1988). Public awareness of the magnitude of air emissions, particularly of DCM, raised concern about possible health effects. As a result, this study was initiated to explore the possible association of the adverse pregnancy outcome of decreased birthweight and exposure to airborne DCM, using a theoretical computergenerated air dispersion model to define DCM exposure. MATERIALS AND METHODS
Birth certificates were obtained for all births in Monroe County in 1976--1987. Multiple births were excluded from the analysis. Infants weighing less than 750 grams were also excluded because of incomplete ascertainment and less accuracy of recorded birthweight of births at very low weights (Ferrara et al., 1988). Because of the known major differences in the distribution of birthweight and in the relationship of risk factors to birthweight between whites and nonwhites, the two groups were not considered together (Jason et al., 1986; Institute of Medicine, 1985; Hughes et al., 1989; Franks, 1987). Only 309 of the total 27,313 births to nonwhites in the county were in women living in the moderate and high exposure areas combined. Of these, 19 weighed less than 2500 grams at birth. Because of the few births among nonwhites in the areas of higher exposure, this study was restricted to white births. Thus the study population included the 91,302 white singleton births weighing 750 g or more, born to mothers residing in Monroe County in 1976-1987. Data available on birth certificates from all years included date of birth, census tract of residence, age, race, and educational level of mother and father, sex, gestational age, multiple births (plurality), month of pregnancy that prenatal care began, total previous births (called "gravidity" in this paper) and total previous live births (called "parity") of the mother, and conditions present during pregnancy. Other data available on the birth certificates only in some years was not included for study. A study of the concordance of Upstate New York (excluding New York City) birth certificate data and hospital records found agreement on infant's sex and race, and mother's and father's age, education, and number of previous live births to be 95-99.9% (New York State Department of Health, 1979). Month prenatal care began agreed 82% of the time. Hospital charts consistently indicated more conditions present during pregnancy than birth certificates. Birthweight was within 28 grams on 91.7% of the records.
METHYLENE CHLORIDE AND BIRTHWEIGHT
33
Dichotomized variables created from birth certificate data included race (white or not), complications of pregnancy ( present or not), late prenatal care (prenatal care begun in the third trimester or not), and term birth (gestationat age greater than 37 weeks or not). Other categorical variables included education (less than high school, high school, or greater than high school), parity (0, 1-3, >/4) and previous pregnancy loss, defined as gravidity minus parity (0, 1-2,/>3). Maternal age was considered both as a continuous variable and categorized (~34). Cutpoints for categorical variables were determined by examining the distribution of the variable within the sample, taking into consideration known risk factors for decreased birthweight. For example, both young (usually ~4 Previous L o s s e s 0 1-2 />3 Maternal Age 34 Exposure None Low Mod High
Statisticb
N
Mean
SE ~
(P)
46858 44444
3468.1 3339.7
2.6 2.4
t9220.3 = - 36.4 (0.0001)
87786 1437
3411.6 3287.7
1.8 14.5
t148o.1 = 8.5 (0.0001)
77961 7493
3428.2 3290.2
1.8 8.2
t82o8.6 = 16.4 (0.0001)
10109 35841 44166
3287.5 3388.7 3453.0
5.8 2.8 2.5
F2 = 444.58 (0.0001)
41521 47925 1799
3346.7 3457.7 3488.6
12.7 12.8 12.5
F 2 = 517.1 (0.0001)
70320 19451 1474
3404.9 3419.7 3361.1
14.0 14.4 13.9
F 2 = 11.49
1777 81871 7590
3275.0 3405.3 3460.9
12.6 1.9 6.1
F 2 = 93.6 (0.0001)
82076 6044 1795 1085
3407.7 3403.9 3400.8 3396.1
16.3 17.6 20.5 16.2
(0.0001)
F 3 = .35 (0.79)
a SE, standard error. b t , t test; F , F statistic.
during the period of greatest exposure weighed 66-74 g less than those in the nonexposed area (New Jersey Department of Health, 1989). It is important to emphasize that the negative findings of this study apply only to the levels of exposure in this particular setting and have little bearing on the possible effect of higher DCM exposures, such as in occupational settings. If DCM does adversely affect birthweight, it may occur at exposure levels higher than those estimated in this population. The highest predicted average concentration of DCM in this population, 50 p~g/m3, is below the proposed standard for community exposure to DCM, 60 p~g/m3, and an order of magnitude lower than that often found in occupational settings. In addition, the exposure levels, derived from the KAMP model, may in fact overestimate exposure. Ambient air sampling, conducted by Kodak between 1978
36
BELL
ET AL.
TABLE2 A D J U S T E D E F F E C T OF RISK F A C T O R S O N B I R T H W E I G H T ( G R A M S ) ( N Risk
factor
Parameter estimate a
= 82514)
95% Confidence interval b
Exposure (None)
Low Mod High Maternal Education Less than H S HS
(Greater than Parity
1.4
- 10.7, 13.5
- 1.9
-27.7,
23.9
- 18.7
-51.6,
14.2
- 160.8 -64.6
- 173.7, - 147.9 -72.4,
56.8
HS)
0 1-3
- 171.3
- 197.8, - 144.8
- 63.8
- 89.3, - 38.3
(/>4)
Previous losses 0
18.3
-9.9,
1-2
28.4
- 0.6, 5 7 . 4
46.5
(/>3) Maternal age (continuous) Late care Male sex Comp preg
-0.04 - 58.6
-0.92, 0.84 - 86.8, - 30.4
132.8
125.9, 139.7
- 114.2
- 126.6, - 1 0 1 . 9
a Parameter estimate is adjusted effect of the factor on birthweight compared to the group in parentheses (reference group). b Confidence interval of the effect of factor present compared to reference group.
and 1985, measured 1-hr composite samples of several chemicals six times per year. The KAMP estimate for DCM concentration is more than twice the levels found in these composite samples (personal communication, New York State Department of Health). Thus actual ambient air concentrations of DCM in this community may be half those predicted, a dose perhaps too low to produce a detectable clinical effect on birthweight. DCM is metabolized in vivo to carbon monoxide. In occupational studies, exposure to low levels of DCM (33-50 ppm) raised mean blood carboxyhemoglobin (CoHgb) concentrations to up to 4%, with higher concentrations seen with higher levels of exposure (Stewart et al., 1972; DiVincenzo and Kaplan, 1981a, 1981b; Ratney et al., 1974; Anders and Sunram, 1982). Both DCM and carbon monoxide cross the placenta and have been found in fetal tissue of pregnant women occupationally exposed to high levels of DCM. (Tabacova, 1986; Schwetz et al., 1975). The estimated exposure levels in our study are also below the level at which discernable increases in CoHgb concentration would be expected. To the extent that any effect of DCM exposure on birthweight is mediated through CoHgb formation, decreases in birthweight would not be anticipated at this exposure level. Misclassification of exposure may have arisen in a number of ways, contributing to the lack of a detectable effect of DCM exposure on birthweight. In contrast
37
METHYLENE CHLORIDE AND BIRTHWEIGHT TABLE 3 ADJUSTED EFFECT OF RISK FACTORS ON ODDS OF LOW BIRTHWEIGHT (N = 82514) Risk factor Exposure (None) Low Moderate High Maternal education Less than HS HS (greater than HS) Parity
Odds ratio n
95% Confidence interval
1.05 1.06 1.00
0.93, 1.17 0.93, 1.22 0.81, 1.24
1.42 0.99
1.34, 1.51 0.93, 1.05
0.82 1.19
0.75, 0.90 1.01, 1.23
1.16 1.06 1.14 0.91 2.05
1.06, 0.97, 1.00, 0.87, 1.97,
(0) l-3 I>4 Previous losses
(0) 1-2 >/3 Late care Male sex Comp preg
1.27 1.17 1.29 0.94 2.14
a Group in parentheses is the reference group, odds ratio = 1.
to the model's exposure isopleths, the actual level of exposure does not change abruptly. Thus, even if the model accurately predicts the concentration and dispersion pattern of DCM, these artificial cutpoints estimate exposure imprecisely, diluting any effect and reducing the power of the study. Similarly, the exposure of women living in census tracts crossed by an exposure isopleth was also poorly estimated because all births in a census tract crossed by an exposure isopleth were classified into the exposure category of the majority of the population of the census tract. Finally, as the distribution of births in a census tract is not fixed and may not conform to the distribution of the population, the actual proportion of births misclassified by this method is unknown and varied over time. Unmeasured confounding variables also may be playing a role. Although adjusting for maternal education may partially account for the confounding effect of smoking (Stein and Kline, 1983; Office on Smoking and Health, 1980), a small decrement in birthweight attributable to DCM exposure might be difficult to detect. Similarly, exposure to other environmental toxins could mask the detection of the effect of any one substance such as DCM. Employment is another unmeasured variable of potential importance. Assuming residence near employment, those occupationally exposed to DCM might also live in high exposure census tracts. Data on Kodak employees from a 1980 New York State Department of Health pilot study support this assumption in this study population (personal communication Joan P. Cooney, New York State Department of Health). This biases the result in favor of an effect and is of less importance in this negative study. As public awareness of potential effects of environmental pollutants grows and
38
BELL ET AL.
the government shows more interest in regulating such emissions, more studies of effects of airborne contaminants are likely. It is therefore important that modeling techniques such as the KAMP model used in this study be validated with extensive actual monitoring. The relationship between estimates of airborne concentrations of substances, generally calculated as annual averages, and individual dose should be clarified. Verification of the validity of these models will allow for more confidence in the findings of studies of the effects of airborne contaminants. However, this study's negative results suggest that the effect, if any, of this level of DCM exposure on birthweight is likely to be small. Further studies of effects of DCM exposure on reproductive outcomes probably should examine more highly exposed groups such as pregnant women workers, and other reproductive outcomes. ACKNOWLEDGMENT Research partially supported by a grant from the Centers of Disease Control and the Association of Teachers of Preventive Medicine.
REFERENCES Anders, M. W., and Sunram, J. M. (1982). Transplacental passage of dichloromethane and carbon monoxide. Toxicol. Lett. 12, 231-234. Axelsson, G., Lutz, C., and Rylander, R. (1984). Exposure to solvents and outcome of pregnancy in university laboratory employees. Brit. J. Ind. Med. 41, 305-312. DiVincenzo, G. D., and Kaplan, C. J. (1981a). Effect of exercise or smoking on the uptake, metabolism, and excretion of methylene chloride vapor. Toxicol. Appl. Pharmacol. 59, 141-148. DiVincenzo, G. D., and Kaplan, C. J. (1981b). Uptake, metabolism, and elimination of methylene chloride vapor by humans. Toxicol. Appl. Pharmacol. 59, 130-140. Eastman Kodak Co. (1988). Draft environmental impact statement for Eastman Kodak Company's proposed cellulose triacetate film base machine. Vol. II: Dispersion modelling analysis for air emissions from the manufacture of cellulose triacetate film base at Kodak Park. Environmental Protection Agency (1987). Health Advisory for Dichloromethane. Draft. Environmental Protection Agency, Office of Drinking Water. Washington, DC. Ferrara, A., Atakent, Y. S., and Levinson, B. (1988). Discrepancies in birth weights between hospital records and health department data for low birth weight infants in New York City. Public Health Rep. 103, 472-478. Franks, P. (1987). Social Inequities in Health: Rochester, New York. Faro. Med. 19, 438-443. Hughes, D., Johnson, K., Rosenbaum, S., and Liu, J. (1989). "The Health of America's Children." Children's Defense Fund. Institute of Medicine (1985). "Preventing Low Birthweight," p. 38, National Academy Press, Washington, DC. Jason, C. J., Samuhel, M. E., Glick, B. J., and Welsh, A. K. (1986). Geographic distribution of unexplained low birth weight. J. Occup. Med. 28:8, 728-740. Maltoni, C., Cotti, G., and Perino, G. (1988). Long-term carcinogenicity bioassays on methylene chlorine administered by ingestion to Sprague-Dawley rats and Swiss mice and by inhalation to Sprague-Dawley rats. Ann. N Y A c a d . Sci. 534, 352-366. Maltoni, C., Cotti, G., and Perino, G. (1986). Experimental research on methylene chloride carcinogenesis. "Archives of Research on Industrial Carcinogenesis. Vol IV." Princeton Scientific Publications, Princeton, NJ. McMichael, A. J., Vimpani, G. V., Robertson, E. F., Baghurst, P. A., and Clark, P. D. (1986). The Port Pirie cohort study: Maternal blood lead and pregnancy outcome. J. Epidemiol. Community Health 40, 18-25. New Jersey Department of Health, Division of Occupational and Environmental Health (1989). A Report on the health study of residents living near the Lipari landfill.
METHYLENE CHLORIDE AND BIRTHWEIGHT
39
New York State Department of Health (1979). "Reliability of Statistical and Medical Information Reported on Birth and Death Certificates." New York State Department of Health Monograph, No. 15. Nordstrom, S., Beckman, O., and Norderson, I. (1978). Occupational and environmental risks in and around a smelter in northern Sweden. Hereditas 88, 43--46. Office on Smoking and Health (1980). "The Health Consequences of Smoking for Women: A Report of the Surgeon General." Public Health Service. US Government Printing Office, Washington, DC. Ozonoff, D., Colten, M. E., Cupples, A., Heeren, T., Schatzkin, A., Mangione, T., Dresner, M,, and Colton, T. (1987). Health problems reported by residents of a neighborhood contaminated by a hazardous waste facility. Am. J. Ind. Med. 11, 581-597. Putz, V. R., Johnson, B. L., and Setzer, J. V. (1976). A comparative study of the effects of carbon monoxide and methylene chloride on human performance. J. Environ. Pathol. Toxicol. 2, 97-112. Ratney, R. S., Wegman, D. H., and Elkins, H. B. (1974). In vivo conversion of methylene chloride to carbon monoxide. Arch. Environ. Health 29, 223-226. SAS Institute Inc. (1988). "SAS Statistical Guide for Personal Computers, Release 6.03 Edition." SAS Institute Inc., Cary, NC. Schwetz, B. A., Leong, B. K. J., and Gehring, P. J. (1975). The effect of maternally inhaled trichloroethylene, perchloroethylene, methyl chloroform, and methylene chloride on embryonal and fetal development in mice and rats. Toxicol. Appl. Pharmacol. 32, 84-96. Stein, Z., and Kline, J. (1983). Smoking, alcohol and reproduction. Am. J. Public Health 73, 11541156. Stewart, R. D., Fisher, T. N., Hosko, M. J., Peterson, J. E., Baretta, E. D., and Dodd, H. C. (1972). Carboxyhemoglobin elevation after exposure to dichloromethane. Science 176, 295-296. Tabacova, S. (1986). Maternal exposure to environmental chemicals. Neurotoxicology 7, 421--440. Taskinen, H., Lindbohm, M. L., and Hemminki, K. (1986). Spontaneous abortions among women working in the pharmaceutical industry. Brit. J. Ind. Med. 43, 199-205. Times-Union (November 28, 1988). Expanding chemical use at Kodak. Times-Union, Rochester, NY. Vianna, N. J., and Polan, A. K. (1984). Incidence of low birth weight among Love Canal residents. Science 226, 1217-1219. Winneke, G. (1974). Behavioral effects of methylene chloride and carbon monoxide as assessed by sensory and psychomotor performance. In "Behavioral Toxicology." US Government Printing Office, Washington, DC.
