- Email: [email protected]
Assessing combined chemical exposures as risk factors for neural tube defects1
Reproductive Toxicology 15 (2001) 631– 635
www.elsevier.com/locate/reprotox
Assessing combined chemical exposures as risk factors for neural tube defects Gary M. Shawa,*, Steve Selvinb, Suzan L. Carmichaela, Donna M. Schafferc, Verne Nelsona, Eric Neria a
March of Dimes Birth Defects Foundation, California Birth Defects Monitoring Program, Oakland, CA, USA b University of California, Berkeley, Berkeley, CA, USA c Kaiser Permanente, Oakland, CA, USA Received 30 May 2001; received in revised form 19 July 2001; accepted 22 July 2001
Abstract Many studies have investigated whether chemical exposures early in pregnancy increase risks to women of delivering offspring with congenital anomalies. We investigated whether periconceptional exposures to chemicals in combination increased risks to women of having neural tube defect (NTD)-affected pregnancies. Women were asked about occupational tasks performed during the periconceptional period. These tasks were assigned by an industrial hygienist to a priori defined exposure categories. The exposure categories included 74 chemical groups. Two population-based case control studies were analyzed. Information on tasks was obtained from mothers of 538 NTD cases and their 539 controls in one study, and mothers of 265 NTD cases and 481 controls from another study. We used data from the first study to identify clues. Specifically, we estimated NTD risks for maternal occupational exposures to all possible pairs, triplets, and quadruplets of 74 chemical groups. Chemical combinations revealing elevated NTD risks in these “clue generation” analyses were then investigated in the second population-based case-control study for their contribution to risk of NTDs. We computed odds ratios for each of the total 192,374 possible comparisons and identified all combinations that produced odds ratios of 5 or more. A 5-fold elevated risk criterion revealed 53 combinations. These 53 reflected various combinations of exposures exclusive to 12 of 74 chemical groups. Analyses of data from the second study did not identify odds ratios of 2.0 or greater for maternal exposures to the 12 chemical groups that resulted in 5-fold elevated risks in the first study. Despite the use of a labor-intensive method to categorize exposures, we were unable to substantiate clues associated with combined chemical exposures identified in one large case-control study as NTD risk factors in a second case-control study. © 2001 Elsevier Science Inc. All rights reserved. Keywords: Anencephaly; Congenital anomalies; Environment; Maternal occupation; Spina bifida; Teratogens
1. Introduction Many studies have investigated whether occupational exposures early in pregnancy increase risks to women to deliver offspring with congenital anomalies [1,2]. Most studies have relied on industry/job titles as surrogates for maternal exposures. We recently reported the results of an
* Corresponding author. Tel.: ⫹1-510-532-5337; fax: ⫹1-510-5321004. E-mail address: [email protected] (G.M. Shaw). The opinions expressed are the views of the authors. They do not necessarily reflect the official position of the California Department of Health Services or other organizations listed. This research was partially supported by funds from the Centers of Disease Control, Centers of Excellence Award No. U50/CCU913241.
investigation to assess risks to women for having neural tube defect (NTD)-affected pregnancies from both occupational and nonoccupational chemical exposures [3]. In that study, we utilized a classification approach that has been successfully used to investigate carcinogenic risks associated with occupational exposures [4,5]. Such an approach attempts to minimize classification errors that may be associated with proxy exposure measures like maternal occupational title, and is a combination of job-exposure linkage and individualized exposure assignment [6]. In an earlier investigation of periconceptional exposures to 74 groups of chemicals associated with maternal occupational and nonoccupational activities, we found no substantial contribution to NTD risk [3]. However, these analyses restricted effect estimations of NTDs among women exposed to each chemical group compared to women who
0890-6238/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved. PII: S 0 8 9 0 - 6 2 3 8 ( 0 1 ) 0 0 1 8 0 - 0
632
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
were not exposed to that chemical group. Thus, such comparisons may have resulted in underestimations of NTD risk if exposures to chemical combinations were predictive of risk. Because information on such exposures and their specific impact on NTD risks is lacking [1], we concurred with Sever [7] who has suggested that maternal occupational and environmental exposures deserve more attention as contributors to the etiology of NTDs. The approach we used in the earlier investigation to classify exposures (i.e. assessment by an industrial hygienist) has been argued to be the method of choice for assessing occupational risks in case-control studies [8]. Thus, any further analytic activity using data classified by this approach could potentially be informative about its methodologic underpinnings, as well as about its utility specific to epidemiologic research on congenital anomalies. Because only a few chemical combinations might be suggested a priori as meaningful to explore owing to the lack of understanding about the underlying etiology of NTDs, we chose to use data from our earlier investigation as a datasource to identify clues. Specifically, we estimated NTD risks for maternal occupational exposures for all possible paired, tripled, and quadrupled combinations of 74 chemical groups. Chemical combinations revealing elevated (5-fold or more) NTD risks in these “clue generation” analyses were then investigated in a different population-based case-control study for their contribution to risk of NTDs.
2. Materials and methods Details of the NTD case-control data used in these analyses have been described [3,9,10]. Data come from two case-control studies. “study 1” [3,9] included infants or fetuses with an NTD (anencephaly, spina bifida cystica, craniorachischisis, or iniencephaly) that were ascertained by reviewing medical records, including ultrasonography, at all hospitals and clinics for those infants/fetuses delivered in select California counties. Singleton liveborn infants and fetuses (included those prenatally diagnosed and electively terminated) among the cohort of 708,129 births and fetal deaths during 1989 to 1991 comprised the case study population. Controls were randomly selected from each area hospital in proportion to the hospital’s estimated contribution to the total population of singleton infants born alive in a given month from 1989 to 1991. Selected were 644 singleton infants without a reportable congenital anomaly [11] whose mother was a California resident. Women who only spoke languages other than English or Spanish, or who had a previous NTD-affected pregnancy were not eligible. Inperson interviews were completed with mothers of 538 (88% of eligible) cases and of 539 (88%) controls after an average of 4.9 months for cases and 4.6 months for controls from the date of delivery or estimated date of delivery. The second case-control study [10], “study 2,” included NTDs not included in study 1. Similar case ascertainment
procedures as used in study 1 were used for deliveries occurring from 1987 through 1988 to women residing in most California counties. A total of 652 control infants was randomly selected from all infants born alive (n ⫽ 341,839) in the same geographic area and time period as cases. Control infants had no major congenital anomalies prior to the first birthday [11]. Telephone interviews were completed with 265 (84% of eligible) case mothers and 481 (76%) control mothers. Interviews were completed an average of 3.7 years after the date of delivery for cases and 3.8 years for controls. In addition to information on maternal medical conditions, reproductive histories, and activities associated with various lifestyles, interviews in both studies elicited detailed work histories (paid and unpaid) from women for the periconceptional period. The requested information included employer name and address, type of industry, period of employment, weekly work hours, job title, and a detailed description of job tasks, which included inquiries about materials handled or machines used. The exposure assessment strategy, more fully described elsewhere [6], employed an industrial hygienist, unaware of whether a woman was a case or a control mother, who characterized occupational activities into tasks. A task corresponded to use of a certain machine or process, contact with a commercial chemical product or tradename product, contact with a type of product defined by its end use, behaviors associated with exposures, or working in certain work environments. Several information sources [12–15] were used to determine task-specific exposures, including inquiries to persons working in the industry and Material Safety Data Sheets, available from product manufacturers for hazardous commercial products. Information about tasks and task-specific exposures permitted the industrial hygienist to further classify women as “likely” exposed, “maybe” exposed, or “not” exposed. These assignments were made for each of 74 chemical agent groups. These groups were defined a priori based on potential reproductive toxicity, including adult systemic toxicity, fetotoxicity, and teratogenicity [6]. A case or control mother could be assigned to multiple chemical agent groups. Risks were estimated by the odds ratio, and the precision of the odds ratio was assessed by its 95% confidence interval. In previous analyses of study 1 data, risks of NTDaffected pregnancies were estimated for each of the 74 possible maternal exposures (displayed in Table 1) relative to those persons without that particular exposure [3]. That is, a risk was estimated for a “likely” exposure to 1 of the 74 chemical groups relative to those persons “not” exposed to that particular exposure (“maybe” exposed individuals were excluded). In the current analyses, we used a clue-generation approach to identify chemical risk factors for NTDaffected pregnancies. The approach was to first identify elevated risks based on combinations, i.e. pairs, triplets, and quadruplets, of exposures to these 74 chemicals based on data from study 1. Next, using the combinations of chemi-
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
633
Table 1 Frequencies of NTD cases and controls associated with periconceptional maternal occupational exposuresa to 74 chemical agent groups, Study 2 only (see Methods) Chemical Agent Groups
Cases
Controls
Chemical Agent Groups
Cases
Controls
Acetic acid and derivatives Alcohols, aliphatic Aldehydes Aluminum compounds Amines, aliphatic Amines, aromatic (including aminophenols) Ammonia and ammonium hydroxide Antibiotics Antimony compounds Antineoplastic drugs Arsenic compounds Bipyridyls Boron compounds Bromides (inorganic) Brominated compounds Cadmium compounds Carbamate insecticides Carbon monoxide Carbon dioxide Chlorinated hydrocarbons, aliphatic Chlorinated hydrocarbons, aromatic Chlorophenoxy herbicides Chromium compounds Copper compounds Dithiocarbarmate fungicides/thioureas Drugs, not otherwise classified Epoxides Esters, including formates Ethers Fluorides Fluorinated organics (excludes medical anesthetics) Formamides and other amides Glycol ethers and derivatives Glycols Halogenated organics, not otherwise classified Halophenols Hydrocarbons, aliphatic (C1–C4)
7 56 23 3 7 9 10 0 0 0 0 0 1 1 0 1 0 9 18 10 2 0 3 9 0 1 0 9 1 2 2 0 11 5 8 0 23
7 63 17 3 3 5 5 0 0 1 0 0 2 4 0 2 0 12 15 5 1 0 2 8 2 1 1 7 1 0 3 0 3 8 7 0 24
Hydrocarbons, aliphatic (C5–C12) Hydrocarbons, aromatic, mononuclear Iodine compounds (organic and inorganic) Isocyanates Ketones Lead compounds Lithium compounds Manganese compounds Mercury compounds Methacrylates and related compounds Nickel compounds Nitrates and nitrites Nitriles, cyanides, cyanogens Nitro compounds (aliphatic and aromatic) Nitrogen oxides (except nitrous) Nitrosamines and other N-nitroso compounds Nitrous oxide Organic dyes (excludes aromatic amines) Organic acids, derivatives, not otherwise classified Organophosphates Oxygen and ozone Peroxides Pesticides not otherwise classified Phenol compounds Phthalates and derivatives Phthalimide fungicides Polycylic aromatic hydrocarbons Pyrethrins, pyrethrum Ribavirin and other antiviral drugs Selenium and telluruim compounds Steroids Sulfides and disulfides Surfactants Terpenes and derivatives Tin compounds (organic only) Volatile anesthetics (except nitrous oxide) Zinc compounds
31 34 0 3 8 2 2 1 1 8 0 1 8 0 12 0 1 15 8 10 6 10 1 7 10 0 5 0 0 1 1 8 49 6 1 0 2
29 31 0 1 8 5 0 0 5 2 3 1 3 0 11 0 4 22 4 4 10 5 1 6 11 1 6 1 0 0 1 3 46 8 0 3 4
a
Those with “likely” exposure to that particular chemical, “maybe” exposures are excluded.
cals resulting in elevated risks in study 1, we investigated whether any of these chemical agent groups in isolation or in combination was associated with elevated risks in study 2. For all analyses, periconceptional periods were defined as the 6-month period 3 months before and after conception (study 1), and the 4-month period 1 month before through 3 months after conception (study 2). Because exposure assessments focused on maternal occupational exposures, analyses were limited to comparisons among women who reported working during the periconceptional period.
3. Results Among the 538 case and 539 control mothers from study 1, 353 case and 396 control mothers were employed during the periconceptional period. Among the 265 case and 481
control mothers from study 2, 143 case and 265 control mothers were employed. Further, among the employed women from study 1, 84 case and 81 control mothers were classified as not exposed to any of the 74 chemical groups, 137 case and 152 control mothers were classified as “likely” exposed and the remainder as “maybe” exposed and not further considered. Among the employed women from study 2, 56 case and 100 control mothers were classified as not exposed, 46 case and 78 control mothers were classified as “likely” exposed, and the remainder as “maybe” and were not further considered. Specific to study 1, “likely” exposed mothers were compared to “not” exposed mothers based on all possible pair, triple, and quadruple chemical combinations of the 74 chemical groups. We computed odds ratios for each of the total 192,374 possible comparisons. As a criterion for further study, we identified all combinations that produced
634
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
Table 2 Risks (odds ratios) of NTD-affected pregnancies associated with periconceptional maternal exposures to selecteda chemical groups, NTD Study 2 (see Methods) Maternal Exposure
Study 2 Cases
Study 2 Controls
Odds Ratio
95% Confidence Interval
Not exposedb Any Chemical 1 chemical 2 chemicals 3 chemicals ⱖ4 chemicals
56 43 16 15 6 6
100 60 19 16 14 11
Reference 1.3 1.5 1.7 0.8 1.0
0.7–2.2 0.7–3.4 0.7–3.9 0.2–2.3 0.3–3.1
a Exposures to the following chemical groups: aromatic hydrocarbons; sulfides and disulfides; halogenated organics; peroxides; aromatic amines; aliphatic hydrocarbons C1–C4; aliphatic hydrocarbons C5–C12; aliphatic alcohols; surfactants; ammonia and ammonia hydroxide; aldehydes; and carbon dioxide. b Unexposed to any of the 74 chemical groupings displayed in Table 1.
odds ratios of 5 or more. A 5-fold elevated risk criterion revealed 53 exposure combinations. Elevated risk estimates were the result of 6 or 7 “exposed” case mothers and 1 “exposed” control mother. These 53 reflected various combinations of the following 12 chemical groups: aromatic hydrocarbons; sulfides and disulfides; halogenated organics; peroxides; aromatic amines; aliphatic hydrocarbons C1-C4; aliphatic hydrocarbons C5-C12; aliphatic alcohols; surfactants; ammonia and ammonia hydroxide; aldehydes; and carbon dioxide. These 12 chemical groupings were investigated to determine whether they were associated with elevated NTD risks in data from study 2. Risks of having an NTD-affected pregnancy associated with periconceptional maternal exposures to any or some of these 12 chemical groupings are displayed in Table 2. These risks corresponded to “likely” exposures (to a particular agent group). Odds ratios of 2.0 or greater were not observed for any maternal exposures to the 12 chemical groups that had resulted in 5-fold elevated risks in study 1. Owing to sparse data, however, many of these effect estimates were imprecise. Additional analyses on data from study 2, using the same chemical combinations (i.e. single, pair, triplet, and quadruplet chemical(s)) that produced odds ratios of 5 or more in data from study 1 produced odds ratios ranging from 1.2 to 2.4 (data not shown).
4. Discussion This investigation used a clue-generation approach to identify chemical combinations that were potential risk factors for NTD-affected pregnancies. The approach identified elevated risks associated with chemical combinations in one case-control study (study 1) and investigated whether those chemicals were associated with elevated risks in a separate case-control study (study 2). Despite the fact that 12 of 74 a priori defined chemical groups were found to produce estimated effects of 5-fold or more in study 1, such highly elevated effects were not observed for these same 12 chemical groups when investigated in a separate case-control study (study 2).
Several tenable explanations exist for the discrepancy in findings between the two data sets. First, the elevated risks observed for the 12 chemicals in study 1 may have been a function of chance, particularly given the small numbers of “exposed” women producing the 5-fold elevated effects. We estimated effects based on all possible chemical agent group combinations (pairs, triples, and quadruples), therefore generating nearly 200,000 comparisons. Identifying 53 exposure scenarios that resulted in 5-fold elevated risks is not beyond what one would expect to observe by chance alone. Second, the methods of obtaining occupational activities between the two studies varied. In study 1, interviews were performed in-person on average 6 months postdelivery, whereas in study 2 the information was collected over the telephone on average 4 years post-delivery. It is possible that depth of probing for occupational activities and the quality of respondent recall was less in study 2 than in study 1. Some indirect evidence for this possibility was the observation that the number of case and control women in study 2 considered not exposed was substantially larger than the number of not exposed in study 1. This difference may be indicative of misclassifying women as not exposed when they should have been classified as exposed. Assuming these errors were unrelated to case or control status the resulting bias would attenuate effects estimated in study 2. Third, the discrepancy in findings may have occurred as a result of differential participation between studies. Participation in study 1 was 88% for cases and controls, and in study 2 was 84% for cases and 76% for controls. Thus, the lower percentage of study 2 controls who particpated could have resulted in different exposure profiles. This analysis was advantaged by using two populationbased, relatively large, case-control studies, along with exposures assessed by an industrial hygienist rather than relying on maternal reporting of specific exposures by product name or inferences from occupational title, to generate clues regarding combined chemical risk factors for NTDs. Few studies have investigated maternal chemical exposures as risk factors exclusively for NTDs and none to our knowledge has specifically investigated the risks potentially associated with exposures to combinations of chemicals. Some
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
studies of maternal occupational exposures have observed elevated risks associated with maternal exposures to chemicals such as organic solvents [16,17], glycol ethers [18], maternal occupations in agriculture [19,20], or maternal occupation as a nurse [21] or cleaner [19], whereas others have not observed elevated risks for some of the same exposures and occupations [3,22,23]. Because no estimate of exposure dose was available, the exposure assessment approach used assumed that women who were considered to have exposures to particular chemical agents all had the same level of exposure. This limitation, however, should not have significantly compromised our ability to substantiate the risks identified in study 1 in the data from study 2. Another potential limitation is that all exposure classifications were made by one industrial hygienist, i.e. exposure designations made for each reported task were not evaluated against the designations of another hygienist. However, errors associated with this single rater approach also should not have compromised our ability to substantiate the risks identified in study 1 in the data from study 2. An additional potential limitation is that the number of women considered “exposed” (likely exposed) was modest, particularly in study 2, even though the numbers of the overall study populations were relatively large. Despite the use of a labor-intensive method to categorize exposures, a method of choice for assessing chemical risks in case-control studies [8], and the use of two large casecontrol studies, we were unable to substantiate clues associated with combined chemical exposures as NTD risk factors.
References [1] Blatter BM, van der Star M, Roelveld N. Review of neural tube defects: risk factors in parental occupation and the environment. Environ Health Perspect 1994;102:140 –5. [2] Shaw GM, Gold EB. Methodological considerations in the study of parental occupational exposures and congenital malformations in offspring. Scand J Work Environ Health 1988;14:344 –55. [3] Shaw GM, Velie EM, Katz EA, Morland KB, Schaffer DM, Nelson V. Maternal occupational and hobby chemical exposures as risk factors for neural tube defects. Epidemiology 1999;10:124 –9. [4] Gerin M, Siemiatycki J, Kemper H, Begin D. Obtaining occupational exposure histories in epidemiologic case-control studies. J Occup Med 1985;27:420 – 6. [5] Siemiatycki J, Day NE, Fabry J, Cooper JA. Discovering carcinogens in the occupational environment: a novel epidemiologic approach. J Natl Cancer Inst 1981;66:217–25.
635
[6] Katz EA, Shaw GM, Schaffer DM. Exposure assessment in epidemiologic studies of birth defects by industrial hygiene review of maternal interviews. Am J Ind Med 1994;26:1–11. [7] Sever LE. Looking for causes of neural tube defects: where does the environment fit in. Environ Health Perspect 1995;103(suppl 6):165– 71. [8] Bouyer J, Hemon D. Retrospective evaluation of occupational exposures in population-based case-control studies: general overview with special attention to job exposure matrices. Int J Epidemiol 1993;22: (suppl.2):S57–S64. [9] Shaw GM, Schaffer D, Velie EM, Morland K, Harris JA. Periconceptional vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiol 1995;6:219 –26. [10] Wasserman, CR, Shaw GM, O’Malley CD, Tolarova MM, Lammer EJ. Parental cigarette smoking and risk for congenital anomalies of the heart, neural tube, or limb. Teratology 1996;53:261–7. [11] Croen LA, Shaw GM, Jensvold NJ, et al. Birth defects monitoring in California: a resource for epidemiological research. Ped Perinatol Epidemiol 1991;5:423–7. [12] Gosselin RE, Smith RP, Hodge HC, Braddock JE. Clinical toxicology of commercial products, 5th ed. Baltimore: Williams & Wilkins, 1984. [13] Hawley GG. The condensed chemical dictionary, 10th ed. New York: van Nostrand Reinhold Company, 1981. [14] MicromedexPOISINDEX (CD-Rom database), Denver. Micromedex, 1991. [15] Hazardous Materials Planning Program. AB2185, and AB 2187, California Health, and Safety Code, Division 20, Chapter 6.95, 1988. [16] Holmberg PC. Central-nervous-system defects in children born to mother exposed to organic solvents during pregnancy. Lancet 1979; 8135:177–9. [17] Kurppa K, Holmberg PC, Hernberg S, Rantala K, Riala R, Nurminen T. Screening for occupational exposures and congenital malformations. Scand J Work Environ Health 1983;9:89 –93. [18] Cordier S, Bergeret A, Goujard J, Ha M-C, Ayme S, Bianchi F, Calzolari E, De Walle HEK, Knill-Jones R, Candela S, Dale I, Dananche B, de Vigan C, Fevotte J, Kiel G, Mandereau L. Congenital malformations and maternal occupational exposure to glycol ethers. Epidemiol 1997:8355– 63. [19] Blatter BM, Roeleveld N, Zielhuis GA, Gabreels FJM, Verbeek ALM. Maternal occupational exposure during pregnancy and the risk of spina bifida. Occup Environ Med 1996;53:80 – 6. [20] Hammond FG, de Canache MF. Some epidemiological aspects of neural tube defects (NTD) in Barquisimeto, Venezuela. Am J Human Genet 1991;49:470. [21] Matte TD, Mulinare J, Erickson JD. Case-control study of congenital defects and parental employment in health care. Am J Ind Med 1993;24:11–23. [22] Cordier S, Ha M-C, Ayme S, Goujard J. Maternal occupational exposure and congenital malformations. Scand J Work Environ Health 1992;18:11–7. [23] Polednak AP, Janerich DT. Uses of available record systems in epidemiologic studies of reproductive toxicology. Am J Ind Med 1983;4:329 – 48.
www.elsevier.com/locate/reprotox
Assessing combined chemical exposures as risk factors for neural tube defects Gary M. Shawa,*, Steve Selvinb, Suzan L. Carmichaela, Donna M. Schafferc, Verne Nelsona, Eric Neria a
March of Dimes Birth Defects Foundation, California Birth Defects Monitoring Program, Oakland, CA, USA b University of California, Berkeley, Berkeley, CA, USA c Kaiser Permanente, Oakland, CA, USA Received 30 May 2001; received in revised form 19 July 2001; accepted 22 July 2001
Abstract Many studies have investigated whether chemical exposures early in pregnancy increase risks to women of delivering offspring with congenital anomalies. We investigated whether periconceptional exposures to chemicals in combination increased risks to women of having neural tube defect (NTD)-affected pregnancies. Women were asked about occupational tasks performed during the periconceptional period. These tasks were assigned by an industrial hygienist to a priori defined exposure categories. The exposure categories included 74 chemical groups. Two population-based case control studies were analyzed. Information on tasks was obtained from mothers of 538 NTD cases and their 539 controls in one study, and mothers of 265 NTD cases and 481 controls from another study. We used data from the first study to identify clues. Specifically, we estimated NTD risks for maternal occupational exposures to all possible pairs, triplets, and quadruplets of 74 chemical groups. Chemical combinations revealing elevated NTD risks in these “clue generation” analyses were then investigated in the second population-based case-control study for their contribution to risk of NTDs. We computed odds ratios for each of the total 192,374 possible comparisons and identified all combinations that produced odds ratios of 5 or more. A 5-fold elevated risk criterion revealed 53 combinations. These 53 reflected various combinations of exposures exclusive to 12 of 74 chemical groups. Analyses of data from the second study did not identify odds ratios of 2.0 or greater for maternal exposures to the 12 chemical groups that resulted in 5-fold elevated risks in the first study. Despite the use of a labor-intensive method to categorize exposures, we were unable to substantiate clues associated with combined chemical exposures identified in one large case-control study as NTD risk factors in a second case-control study. © 2001 Elsevier Science Inc. All rights reserved. Keywords: Anencephaly; Congenital anomalies; Environment; Maternal occupation; Spina bifida; Teratogens
1. Introduction Many studies have investigated whether occupational exposures early in pregnancy increase risks to women to deliver offspring with congenital anomalies [1,2]. Most studies have relied on industry/job titles as surrogates for maternal exposures. We recently reported the results of an
* Corresponding author. Tel.: ⫹1-510-532-5337; fax: ⫹1-510-5321004. E-mail address: [email protected] (G.M. Shaw). The opinions expressed are the views of the authors. They do not necessarily reflect the official position of the California Department of Health Services or other organizations listed. This research was partially supported by funds from the Centers of Disease Control, Centers of Excellence Award No. U50/CCU913241.
investigation to assess risks to women for having neural tube defect (NTD)-affected pregnancies from both occupational and nonoccupational chemical exposures [3]. In that study, we utilized a classification approach that has been successfully used to investigate carcinogenic risks associated with occupational exposures [4,5]. Such an approach attempts to minimize classification errors that may be associated with proxy exposure measures like maternal occupational title, and is a combination of job-exposure linkage and individualized exposure assignment [6]. In an earlier investigation of periconceptional exposures to 74 groups of chemicals associated with maternal occupational and nonoccupational activities, we found no substantial contribution to NTD risk [3]. However, these analyses restricted effect estimations of NTDs among women exposed to each chemical group compared to women who
0890-6238/01/$ – see front matter © 2001 Elsevier Science Inc. All rights reserved. PII: S 0 8 9 0 - 6 2 3 8 ( 0 1 ) 0 0 1 8 0 - 0
632
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
were not exposed to that chemical group. Thus, such comparisons may have resulted in underestimations of NTD risk if exposures to chemical combinations were predictive of risk. Because information on such exposures and their specific impact on NTD risks is lacking [1], we concurred with Sever [7] who has suggested that maternal occupational and environmental exposures deserve more attention as contributors to the etiology of NTDs. The approach we used in the earlier investigation to classify exposures (i.e. assessment by an industrial hygienist) has been argued to be the method of choice for assessing occupational risks in case-control studies [8]. Thus, any further analytic activity using data classified by this approach could potentially be informative about its methodologic underpinnings, as well as about its utility specific to epidemiologic research on congenital anomalies. Because only a few chemical combinations might be suggested a priori as meaningful to explore owing to the lack of understanding about the underlying etiology of NTDs, we chose to use data from our earlier investigation as a datasource to identify clues. Specifically, we estimated NTD risks for maternal occupational exposures for all possible paired, tripled, and quadrupled combinations of 74 chemical groups. Chemical combinations revealing elevated (5-fold or more) NTD risks in these “clue generation” analyses were then investigated in a different population-based case-control study for their contribution to risk of NTDs.
2. Materials and methods Details of the NTD case-control data used in these analyses have been described [3,9,10]. Data come from two case-control studies. “study 1” [3,9] included infants or fetuses with an NTD (anencephaly, spina bifida cystica, craniorachischisis, or iniencephaly) that were ascertained by reviewing medical records, including ultrasonography, at all hospitals and clinics for those infants/fetuses delivered in select California counties. Singleton liveborn infants and fetuses (included those prenatally diagnosed and electively terminated) among the cohort of 708,129 births and fetal deaths during 1989 to 1991 comprised the case study population. Controls were randomly selected from each area hospital in proportion to the hospital’s estimated contribution to the total population of singleton infants born alive in a given month from 1989 to 1991. Selected were 644 singleton infants without a reportable congenital anomaly [11] whose mother was a California resident. Women who only spoke languages other than English or Spanish, or who had a previous NTD-affected pregnancy were not eligible. Inperson interviews were completed with mothers of 538 (88% of eligible) cases and of 539 (88%) controls after an average of 4.9 months for cases and 4.6 months for controls from the date of delivery or estimated date of delivery. The second case-control study [10], “study 2,” included NTDs not included in study 1. Similar case ascertainment
procedures as used in study 1 were used for deliveries occurring from 1987 through 1988 to women residing in most California counties. A total of 652 control infants was randomly selected from all infants born alive (n ⫽ 341,839) in the same geographic area and time period as cases. Control infants had no major congenital anomalies prior to the first birthday [11]. Telephone interviews were completed with 265 (84% of eligible) case mothers and 481 (76%) control mothers. Interviews were completed an average of 3.7 years after the date of delivery for cases and 3.8 years for controls. In addition to information on maternal medical conditions, reproductive histories, and activities associated with various lifestyles, interviews in both studies elicited detailed work histories (paid and unpaid) from women for the periconceptional period. The requested information included employer name and address, type of industry, period of employment, weekly work hours, job title, and a detailed description of job tasks, which included inquiries about materials handled or machines used. The exposure assessment strategy, more fully described elsewhere [6], employed an industrial hygienist, unaware of whether a woman was a case or a control mother, who characterized occupational activities into tasks. A task corresponded to use of a certain machine or process, contact with a commercial chemical product or tradename product, contact with a type of product defined by its end use, behaviors associated with exposures, or working in certain work environments. Several information sources [12–15] were used to determine task-specific exposures, including inquiries to persons working in the industry and Material Safety Data Sheets, available from product manufacturers for hazardous commercial products. Information about tasks and task-specific exposures permitted the industrial hygienist to further classify women as “likely” exposed, “maybe” exposed, or “not” exposed. These assignments were made for each of 74 chemical agent groups. These groups were defined a priori based on potential reproductive toxicity, including adult systemic toxicity, fetotoxicity, and teratogenicity [6]. A case or control mother could be assigned to multiple chemical agent groups. Risks were estimated by the odds ratio, and the precision of the odds ratio was assessed by its 95% confidence interval. In previous analyses of study 1 data, risks of NTDaffected pregnancies were estimated for each of the 74 possible maternal exposures (displayed in Table 1) relative to those persons without that particular exposure [3]. That is, a risk was estimated for a “likely” exposure to 1 of the 74 chemical groups relative to those persons “not” exposed to that particular exposure (“maybe” exposed individuals were excluded). In the current analyses, we used a clue-generation approach to identify chemical risk factors for NTDaffected pregnancies. The approach was to first identify elevated risks based on combinations, i.e. pairs, triplets, and quadruplets, of exposures to these 74 chemicals based on data from study 1. Next, using the combinations of chemi-
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
633
Table 1 Frequencies of NTD cases and controls associated with periconceptional maternal occupational exposuresa to 74 chemical agent groups, Study 2 only (see Methods) Chemical Agent Groups
Cases
Controls
Chemical Agent Groups
Cases
Controls
Acetic acid and derivatives Alcohols, aliphatic Aldehydes Aluminum compounds Amines, aliphatic Amines, aromatic (including aminophenols) Ammonia and ammonium hydroxide Antibiotics Antimony compounds Antineoplastic drugs Arsenic compounds Bipyridyls Boron compounds Bromides (inorganic) Brominated compounds Cadmium compounds Carbamate insecticides Carbon monoxide Carbon dioxide Chlorinated hydrocarbons, aliphatic Chlorinated hydrocarbons, aromatic Chlorophenoxy herbicides Chromium compounds Copper compounds Dithiocarbarmate fungicides/thioureas Drugs, not otherwise classified Epoxides Esters, including formates Ethers Fluorides Fluorinated organics (excludes medical anesthetics) Formamides and other amides Glycol ethers and derivatives Glycols Halogenated organics, not otherwise classified Halophenols Hydrocarbons, aliphatic (C1–C4)
7 56 23 3 7 9 10 0 0 0 0 0 1 1 0 1 0 9 18 10 2 0 3 9 0 1 0 9 1 2 2 0 11 5 8 0 23
7 63 17 3 3 5 5 0 0 1 0 0 2 4 0 2 0 12 15 5 1 0 2 8 2 1 1 7 1 0 3 0 3 8 7 0 24
Hydrocarbons, aliphatic (C5–C12) Hydrocarbons, aromatic, mononuclear Iodine compounds (organic and inorganic) Isocyanates Ketones Lead compounds Lithium compounds Manganese compounds Mercury compounds Methacrylates and related compounds Nickel compounds Nitrates and nitrites Nitriles, cyanides, cyanogens Nitro compounds (aliphatic and aromatic) Nitrogen oxides (except nitrous) Nitrosamines and other N-nitroso compounds Nitrous oxide Organic dyes (excludes aromatic amines) Organic acids, derivatives, not otherwise classified Organophosphates Oxygen and ozone Peroxides Pesticides not otherwise classified Phenol compounds Phthalates and derivatives Phthalimide fungicides Polycylic aromatic hydrocarbons Pyrethrins, pyrethrum Ribavirin and other antiviral drugs Selenium and telluruim compounds Steroids Sulfides and disulfides Surfactants Terpenes and derivatives Tin compounds (organic only) Volatile anesthetics (except nitrous oxide) Zinc compounds
31 34 0 3 8 2 2 1 1 8 0 1 8 0 12 0 1 15 8 10 6 10 1 7 10 0 5 0 0 1 1 8 49 6 1 0 2
29 31 0 1 8 5 0 0 5 2 3 1 3 0 11 0 4 22 4 4 10 5 1 6 11 1 6 1 0 0 1 3 46 8 0 3 4
a
Those with “likely” exposure to that particular chemical, “maybe” exposures are excluded.
cals resulting in elevated risks in study 1, we investigated whether any of these chemical agent groups in isolation or in combination was associated with elevated risks in study 2. For all analyses, periconceptional periods were defined as the 6-month period 3 months before and after conception (study 1), and the 4-month period 1 month before through 3 months after conception (study 2). Because exposure assessments focused on maternal occupational exposures, analyses were limited to comparisons among women who reported working during the periconceptional period.
3. Results Among the 538 case and 539 control mothers from study 1, 353 case and 396 control mothers were employed during the periconceptional period. Among the 265 case and 481
control mothers from study 2, 143 case and 265 control mothers were employed. Further, among the employed women from study 1, 84 case and 81 control mothers were classified as not exposed to any of the 74 chemical groups, 137 case and 152 control mothers were classified as “likely” exposed and the remainder as “maybe” exposed and not further considered. Among the employed women from study 2, 56 case and 100 control mothers were classified as not exposed, 46 case and 78 control mothers were classified as “likely” exposed, and the remainder as “maybe” and were not further considered. Specific to study 1, “likely” exposed mothers were compared to “not” exposed mothers based on all possible pair, triple, and quadruple chemical combinations of the 74 chemical groups. We computed odds ratios for each of the total 192,374 possible comparisons. As a criterion for further study, we identified all combinations that produced
634
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
Table 2 Risks (odds ratios) of NTD-affected pregnancies associated with periconceptional maternal exposures to selecteda chemical groups, NTD Study 2 (see Methods) Maternal Exposure
Study 2 Cases
Study 2 Controls
Odds Ratio
95% Confidence Interval
Not exposedb Any Chemical 1 chemical 2 chemicals 3 chemicals ⱖ4 chemicals
56 43 16 15 6 6
100 60 19 16 14 11
Reference 1.3 1.5 1.7 0.8 1.0
0.7–2.2 0.7–3.4 0.7–3.9 0.2–2.3 0.3–3.1
a Exposures to the following chemical groups: aromatic hydrocarbons; sulfides and disulfides; halogenated organics; peroxides; aromatic amines; aliphatic hydrocarbons C1–C4; aliphatic hydrocarbons C5–C12; aliphatic alcohols; surfactants; ammonia and ammonia hydroxide; aldehydes; and carbon dioxide. b Unexposed to any of the 74 chemical groupings displayed in Table 1.
odds ratios of 5 or more. A 5-fold elevated risk criterion revealed 53 exposure combinations. Elevated risk estimates were the result of 6 or 7 “exposed” case mothers and 1 “exposed” control mother. These 53 reflected various combinations of the following 12 chemical groups: aromatic hydrocarbons; sulfides and disulfides; halogenated organics; peroxides; aromatic amines; aliphatic hydrocarbons C1-C4; aliphatic hydrocarbons C5-C12; aliphatic alcohols; surfactants; ammonia and ammonia hydroxide; aldehydes; and carbon dioxide. These 12 chemical groupings were investigated to determine whether they were associated with elevated NTD risks in data from study 2. Risks of having an NTD-affected pregnancy associated with periconceptional maternal exposures to any or some of these 12 chemical groupings are displayed in Table 2. These risks corresponded to “likely” exposures (to a particular agent group). Odds ratios of 2.0 or greater were not observed for any maternal exposures to the 12 chemical groups that had resulted in 5-fold elevated risks in study 1. Owing to sparse data, however, many of these effect estimates were imprecise. Additional analyses on data from study 2, using the same chemical combinations (i.e. single, pair, triplet, and quadruplet chemical(s)) that produced odds ratios of 5 or more in data from study 1 produced odds ratios ranging from 1.2 to 2.4 (data not shown).
4. Discussion This investigation used a clue-generation approach to identify chemical combinations that were potential risk factors for NTD-affected pregnancies. The approach identified elevated risks associated with chemical combinations in one case-control study (study 1) and investigated whether those chemicals were associated with elevated risks in a separate case-control study (study 2). Despite the fact that 12 of 74 a priori defined chemical groups were found to produce estimated effects of 5-fold or more in study 1, such highly elevated effects were not observed for these same 12 chemical groups when investigated in a separate case-control study (study 2).
Several tenable explanations exist for the discrepancy in findings between the two data sets. First, the elevated risks observed for the 12 chemicals in study 1 may have been a function of chance, particularly given the small numbers of “exposed” women producing the 5-fold elevated effects. We estimated effects based on all possible chemical agent group combinations (pairs, triples, and quadruples), therefore generating nearly 200,000 comparisons. Identifying 53 exposure scenarios that resulted in 5-fold elevated risks is not beyond what one would expect to observe by chance alone. Second, the methods of obtaining occupational activities between the two studies varied. In study 1, interviews were performed in-person on average 6 months postdelivery, whereas in study 2 the information was collected over the telephone on average 4 years post-delivery. It is possible that depth of probing for occupational activities and the quality of respondent recall was less in study 2 than in study 1. Some indirect evidence for this possibility was the observation that the number of case and control women in study 2 considered not exposed was substantially larger than the number of not exposed in study 1. This difference may be indicative of misclassifying women as not exposed when they should have been classified as exposed. Assuming these errors were unrelated to case or control status the resulting bias would attenuate effects estimated in study 2. Third, the discrepancy in findings may have occurred as a result of differential participation between studies. Participation in study 1 was 88% for cases and controls, and in study 2 was 84% for cases and 76% for controls. Thus, the lower percentage of study 2 controls who particpated could have resulted in different exposure profiles. This analysis was advantaged by using two populationbased, relatively large, case-control studies, along with exposures assessed by an industrial hygienist rather than relying on maternal reporting of specific exposures by product name or inferences from occupational title, to generate clues regarding combined chemical risk factors for NTDs. Few studies have investigated maternal chemical exposures as risk factors exclusively for NTDs and none to our knowledge has specifically investigated the risks potentially associated with exposures to combinations of chemicals. Some
G.M. Shaw / Reproductive Toxicology 15 (2001) 631– 635
studies of maternal occupational exposures have observed elevated risks associated with maternal exposures to chemicals such as organic solvents [16,17], glycol ethers [18], maternal occupations in agriculture [19,20], or maternal occupation as a nurse [21] or cleaner [19], whereas others have not observed elevated risks for some of the same exposures and occupations [3,22,23]. Because no estimate of exposure dose was available, the exposure assessment approach used assumed that women who were considered to have exposures to particular chemical agents all had the same level of exposure. This limitation, however, should not have significantly compromised our ability to substantiate the risks identified in study 1 in the data from study 2. Another potential limitation is that all exposure classifications were made by one industrial hygienist, i.e. exposure designations made for each reported task were not evaluated against the designations of another hygienist. However, errors associated with this single rater approach also should not have compromised our ability to substantiate the risks identified in study 1 in the data from study 2. An additional potential limitation is that the number of women considered “exposed” (likely exposed) was modest, particularly in study 2, even though the numbers of the overall study populations were relatively large. Despite the use of a labor-intensive method to categorize exposures, a method of choice for assessing chemical risks in case-control studies [8], and the use of two large casecontrol studies, we were unable to substantiate clues associated with combined chemical exposures as NTD risk factors.
References [1] Blatter BM, van der Star M, Roelveld N. Review of neural tube defects: risk factors in parental occupation and the environment. Environ Health Perspect 1994;102:140 –5. [2] Shaw GM, Gold EB. Methodological considerations in the study of parental occupational exposures and congenital malformations in offspring. Scand J Work Environ Health 1988;14:344 –55. [3] Shaw GM, Velie EM, Katz EA, Morland KB, Schaffer DM, Nelson V. Maternal occupational and hobby chemical exposures as risk factors for neural tube defects. Epidemiology 1999;10:124 –9. [4] Gerin M, Siemiatycki J, Kemper H, Begin D. Obtaining occupational exposure histories in epidemiologic case-control studies. J Occup Med 1985;27:420 – 6. [5] Siemiatycki J, Day NE, Fabry J, Cooper JA. Discovering carcinogens in the occupational environment: a novel epidemiologic approach. J Natl Cancer Inst 1981;66:217–25.
635
[6] Katz EA, Shaw GM, Schaffer DM. Exposure assessment in epidemiologic studies of birth defects by industrial hygiene review of maternal interviews. Am J Ind Med 1994;26:1–11. [7] Sever LE. Looking for causes of neural tube defects: where does the environment fit in. Environ Health Perspect 1995;103(suppl 6):165– 71. [8] Bouyer J, Hemon D. Retrospective evaluation of occupational exposures in population-based case-control studies: general overview with special attention to job exposure matrices. Int J Epidemiol 1993;22: (suppl.2):S57–S64. [9] Shaw GM, Schaffer D, Velie EM, Morland K, Harris JA. Periconceptional vitamin use, dietary folate, and the occurrence of neural tube defects. Epidemiol 1995;6:219 –26. [10] Wasserman, CR, Shaw GM, O’Malley CD, Tolarova MM, Lammer EJ. Parental cigarette smoking and risk for congenital anomalies of the heart, neural tube, or limb. Teratology 1996;53:261–7. [11] Croen LA, Shaw GM, Jensvold NJ, et al. Birth defects monitoring in California: a resource for epidemiological research. Ped Perinatol Epidemiol 1991;5:423–7. [12] Gosselin RE, Smith RP, Hodge HC, Braddock JE. Clinical toxicology of commercial products, 5th ed. Baltimore: Williams & Wilkins, 1984. [13] Hawley GG. The condensed chemical dictionary, 10th ed. New York: van Nostrand Reinhold Company, 1981. [14] MicromedexPOISINDEX (CD-Rom database), Denver. Micromedex, 1991. [15] Hazardous Materials Planning Program. AB2185, and AB 2187, California Health, and Safety Code, Division 20, Chapter 6.95, 1988. [16] Holmberg PC. Central-nervous-system defects in children born to mother exposed to organic solvents during pregnancy. Lancet 1979; 8135:177–9. [17] Kurppa K, Holmberg PC, Hernberg S, Rantala K, Riala R, Nurminen T. Screening for occupational exposures and congenital malformations. Scand J Work Environ Health 1983;9:89 –93. [18] Cordier S, Bergeret A, Goujard J, Ha M-C, Ayme S, Bianchi F, Calzolari E, De Walle HEK, Knill-Jones R, Candela S, Dale I, Dananche B, de Vigan C, Fevotte J, Kiel G, Mandereau L. Congenital malformations and maternal occupational exposure to glycol ethers. Epidemiol 1997:8355– 63. [19] Blatter BM, Roeleveld N, Zielhuis GA, Gabreels FJM, Verbeek ALM. Maternal occupational exposure during pregnancy and the risk of spina bifida. Occup Environ Med 1996;53:80 – 6. [20] Hammond FG, de Canache MF. Some epidemiological aspects of neural tube defects (NTD) in Barquisimeto, Venezuela. Am J Human Genet 1991;49:470. [21] Matte TD, Mulinare J, Erickson JD. Case-control study of congenital defects and parental employment in health care. Am J Ind Med 1993;24:11–23. [22] Cordier S, Ha M-C, Ayme S, Goujard J. Maternal occupational exposure and congenital malformations. Scand J Work Environ Health 1992;18:11–7. [23] Polednak AP, Janerich DT. Uses of available record systems in epidemiologic studies of reproductive toxicology. Am J Ind Med 1983;4:329 – 48.