Parental occupation and risk of small-for-gestational-age births: a nationwide epidemiological study in Sweden

Abstract

BACKGROUND

Although evidence suggests that some occupations may be a risk factor for small-for-gestational age (SGA) birth, associations between a wide range of maternal and paternal occupations and risk of SGA births remain unclear. Our objective was to analyze the risk of SGA births by parental occupation, including the entire Swedish population of mothers (≥20 years) and fathers.

METHODS

We linked nationwide data (1990–2004) on singletons born to employed mothers to nationwide data on maternal and paternal occupation and other individual-level variables. Information on parental occupations was obtained from the 1990 census. Approximately 95% of SGA births (calculated using normative data) were defined on the basis of ultrasound. Odds ratios of SGA birth were calculated with 95% confidence intervals. Women and men were analyzed separately.

RESULTS

There were 816 310 first singleton live births during the study period, of which 29 603 were SGA events. Families with low incomes had an increased risk of SGA births. After accounting for maternal age at the infant's birth, period of birth, family income, region of residence, marital status and smoking habits, several maternal occupational groups (including ‘mechanics and iron and metalware workers’ and ‘packers, loaders and warehouse workers’) had a significantly higher risk of SGA birth than the reference group (all women in the study population). Among paternal occupational groups, only waiters had an increased risk of SGA birth.

CONCLUSIONS

This large-scale follow-up study shows that maternal occupation affects risk of SGA birth, whereas paternal occupation does not seem to have an impact on SGA birth. Further studies are required to examine the specific agents in those maternal occupations that are associated with an increased risk of SGA birth.

Introduction

Small-for-gestational age (SGA) is a term used to describe an infant who is smaller than expected on the basis of the number of weeks of completed pregnancy and typically refers to an infant who is below the tenth percentile of length and weight for gestational age (World Health Organization, 2007). The causes of SGA are still partly unknown. Some of the known risk factors include a maternal age of less than 20 or more than 35 years (Zeitlin et al., 2001; Canadian Institute for Health Information, 2009), shorter maternal stature (Clausson et al., 1998; Thompson et al., 2001), primiparity (Thompson et al., 2001; Canadian Institute for Health Information, 2009), smoking (Thompson et al., 2001; Cnattingius, 2004), maternal undernutrition (Thompson et al., 2001; Hobel and Culhane, 2003), low maternal BMI (Clausson et al., 1998; Zeitlin et al., 2001), high maternal BMI (Zeitlin et al., 2001), preexisting hypertension (Thompson et al., 2001; Canadian Institute for Health Information, 2009), gestational hypertension (Canadian Institute for Health Information, 2009), pre-eclampsia (Clausson et al., 1998; Ananth et al., 1999; Thompson et al., 2001) and vascular lesions (Zeitlin et al., 2001).

Research has also shown that ethnicity is associated with SGA births (Thompson et al., 2001), as are neighborhood socioeconomic characteristics, and living in an urban as opposed to a rural area (Canadian Institute for Health Information, 2009). Another factor that may be associated with SGA births is maternal stress. Maternal stress is known to be associated with low birthweight (Hobel and Culhane, 2003). Low birthweight can result from intrauterine growth retardation, which causes SGA, or from premature birth.

There is also a growing body of evidence that suggests that individual-level socioeconomic status (SES) is a risk factor for SGA birth (Parker et al., 1994; Koupilova et al., 1998, Fairley and Leyland, 2006; Luo et al., 2006). Low SES may influence the risk of SGA birth in multiple ways. For example, exposure to harmful agents may result from residential, lifestyle or occupational factors, all of which may be related to SES.

Although a person's job is often associated with his or her SES, occupation can also be a proxy for occupational exposure. At least one occupational study has found increased risks of SGA birth in mothers whose occupations involved exposure to organic solvents (Ahmed and Jaakkola, 2007a). A few studies have reported associations between broader occupational categories and risk of SGA birth (Savitz et al., 1996b; Ahmed and Jaakkola, 2007b) and a few have reported associations between specific parental occupations and incidence of SGA birth (Savitz et al., 1996a; Rylander and Kallen, 2005; Ahmed and Jaakkola, 2007b; Meyer et al., 2008, Simcox and Jaakkola, 2008). However, the association between a wider range of specific occupations and risk of SGA birth remains unclear, and to the best of our knowledge, no previous nationwide study has investigated the association between a wide range of maternal and paternal occupational groups and risk of SGA births, which was the purpose of the present study.

Data on SGA births were obtained from the nationwide Medical Birth Register, which allowed us to analyze large sample sizes in each occupational category. The use of nationwide birth data and data from other nationwide registers permitted almost complete follow-up of all women and men during the study period. The aim of this study was to analyze risks of first singleton SGA birth in the employed Swedish population by maternal and paternal occupation, controlling for potential confounding variables.

Materials and Methods

Data sources

We linked nationwide data (1990–2004) on singletons born to employed mothers to nationwide data on maternal and paternal occupation and other individual-level variables. One important reason for choosing the year 1990 as the start of the study period was that since that year, ultrasound examination has been offered to all pregnant women in Sweden.

Data used in this study were retrieved from the nationwide WomMed II Database, located at the Center for Primary Health Care Research at Lund University. This database contains information from the Swedish Medical Birth Register, which covers 99% of all births in Sweden beginning in 1973, and includes both birth records and prenatal care data such as prospectively collected information about complications during pregnancy and delivery (The Swedish Centre for Epidemiology, 2003). The WomMed II database also contains nationwide, individual-level hospital diagnoses and death register data from the National Board of Health and Welfare, as well as the population register (census) data from Statistics Sweden, the Swedish Government-owned statistics bureau.

Outcome variable: small-for-gestational-age births

Babies with a birthweight of more than 2 standard deviations below the mean for gestational age (i.e. the 2.5th percentile) according to the Swedish references curve for estimated fetal weight were categorized as SGA. This corresponds to more than 24% (approximately 850 g) lower birthweight than expected for a full-term baby. Information on birthweight and gestational age was gathered from the Medical Birth Register. When available, ultrasound performed during the second trimester was used to estimate the gestational age; otherwise gestational age was estimated from the date of the last menstrual period. All pregnant women in Sweden are offered free antenatal care. At the first visit, usually in gestational weeks 10–12, information on date of the last menstrual period is obtained. Since 1990, between gestational weeks 16 and 18, women have been offered an ultrasound examination to date their pregnancy and calculate the expected date of delivery. A total of 95% accepted the offer since 1990 (Swedish Council on Technology Assessment in Health Care, 1999). Once an expected date of delivery is obtained from ultrasound, this date is used in all medical files and birth records regardless of the date of the last menstrual period.

The ultrasound examinations were performed by trained midwives. In ultrasound examinations conducted prior to gestational week 14, gestational age was assessed using biparietal diameter (Robinson and Fleming, 1975; Saltvedt et al., 2004; Sladkevicius et al., 2005). In examinations conducted during weeks 16 through 18, the midwives relied on femur length and biparietal diameter to estimate gestational age (Persson and Weldner, 1986).

Predictor variable: parental occupation

We used parental occupation as the predictor variable in this study because it was the best proxy for parental occupational exposure that was available to us. Moreover, occupation has been used as a proxy for occupational exposures in previous studies of SGA birth and low birthweight (Savitz et al., 1996a; Rylander and Kallen, 2005; Ahmed and Jaakkola, 2007b, Meyer et al., 2008; Simcox and Jaakkola, 2008). Information on parental occupations was obtained from the 1990 census, from which fairly complete (92%) individual-level employment data were available for the entire population. Census data include a numerical occupational code for each employed individual. These occupational codes were created by Statistics Sweden on the basis of a national adaptation of the Nordic Occupational Classification (NYK) (Swedish National Central Bureau of Statistics, 1982). Because many of the occupational groups defined by the census codes contain too few individuals for a meaningful statistical analysis, we used 53 NYK-based occupational groups that have been used in several large-scale epidemiological studies from the Nordic countries (Andersen et al., 1999; Mutanen and Hemminki, 2001; Pukkala et al., 2009; Li et al., in press). These occupational groups were combined on the basis of occupational similarities. A detailed list of the Swedish Census codes included in each of the 53 occupational groups can be found in Appendix A of a recent article by Pukkala et al. (2009).

Occupation was assessed separately for mothers and fathers. An infant was excluded from the analysis of both parental categories if its mother was categorized as unemployed or as a student in the 1990 census, or if maternal employment information was missing from the census. If an infant's father was categorized as unemployed or a student in the 1990 census, or if paternal employment information was missing, that infant was included only in the analysis of maternal occupation.

Individual-level sociodemographic variables

Maternal age at infant's birth was divided into 5-year age groups as follows: 20–24, 25–29, 30–34 and ≥35 years. Only mothers aged ≥20 years at the birth of their child were included in the analysis. Period of birth was divided into 5-year groups from 1990 to 2004. The variable family income was defined as the mother's family income from the year of childbirth divided by the number of people in the family; that is, individual family income per capita. This variable was provided by Statistics Sweden. The income parameter also took into consideration the ages of people in the family and used a weighted system whereby small children were given lower weights than adolescents and adults. The calculation procedure was performed as follows: the sum of all family members’ incomes was multiplied by the individual's consumption weight divided by the family members’ total consumption weight. The final variable was calculated as empirical quartiles from the distribution. Region of residence was categorized on the basis of Statistics Sweden's ‘H-regions,’ which are regions with homogenous populations (Swedish National Central Bureau of Statistics, 2003). We used three categories, all of which were based on Statistics Sweden's 1990 H-regions. The first category was ‘large cities.’ This category included Stockholm/Södertälje, Gothenburg, Malmö/Lund/Trelleborg, plus counties in which more than 90 000 inhabitants lived within a 30-km radius of the center of the county. The second category was ‘middle-size towns.’ This category included counties in which more than 27 000 and less than 90 000 inhabitants lived within a 30-km radius of the center of the county and in which more than 300 000 inhabitants lived within a 100-km radius of the same point. The third category was ‘small towns/rural areas.’ This category included counties with fewer than 27 000 inhabitants within a 30-km radius of the center of the county. Mother's region of residence was used to define the region of residence. Marital status was defined as mother's marital status during the year of childbirth and was divided into two groups: (i) married or cohabiting with a partner or (ii) unmarried, divorced or widowed. We chose to include the variable marital status, although many women in Sweden cohabit with a partner without being married. For example, data from the time period 1991–1998 showed that only 35–40% of all women were registered as being married. Non-married, cohabiting women without children are not classified as cohabiting in Swedish national census data. This means that potentially increased risks in the unmarried group would be diluted because this group also includes women who are cohabiting with a partner. Mothers were divided into two groups by smoking habits: non-smokers and smokers. Smoking data were obtained from the Medical Birth Register and represented the mother's smoking habits at the first visit to the maternal clinic.

Statistical analysis

Using logistic regression analysis, we estimated risk of giving birth to a singleton who was SGA by family income, region of residence, marital status, smoking habits and parental occupation (the reference group was composed of all women or all men in the study population). All risk estimates were adjusted for maternal age at infant's birth and period of birth (in 5 year periods). The risk estimates of SGA birth by occupational status were also adjusted for family income, region of residence, marital status and smoking habits. We also conducted an ancillary analysis of the risk of term SGA birth and preterm SGA birth by family income, region of residence and marital status. The sex of the infant showed no specific effects; data are therefore given for female and male infants together. We used SAS version 9.1 for the statistical analyses (SAS, 2003).

Ethical considerations

This study was approved by the Ethics Committee of Lund University.

Results

During the study period, there were 816 310 first singleton live births in Sweden in which mothers were employed (according to the 1990 Census) and linkable to their infants in the Medical Birth Register. Of these, 29 603 (3.6%) were SGA births. Table I shows the risk of SGA birth for first singleton live births by family income, region of residence, mother's marital status and mother's smoking habits. All risks are also adjusted for maternal age at infant's birth and period of birth (in 5-year groups). Statistically significant differences between the subcategories and the reference group were found for most variables. The strongest associations with SGA were found for the variables smoking habits and family income.

Table I.

Population sizes, number of mothers with SGA events and risk of SGA birth by family income, region of residence, marital status and smoking habits.

	Population	Mothers with SGA events	OR	95% CI
Family income
Low income	226 762	9895	1.22	1.19	1.24
Middle-low income	217 996	7874	1.02	1.00	1.04
Middle-high income	186 703	6187	0.95	0.93	0.97
High income	184 849	5647	0.85	0.81	0.88
Region of residence
Large cities	255 385	9400	1.02	1.00	1.04
Middle-sized towns	275 351	9621	0.97	0.95	0.98
Small towns/rural areas	285 574	10 582	1.01	0.99	1.04
Marital status
Married/cohabiting	416 092	14 526	1.00	0.99	1.02
Unmarried, divorced, or widowed	400 218	15 077	1.00	0.98	1.01
Smoking habits
Non-smoker	684 884	21 601	0.69	0.69	0.70
Smoker	131 426	8002	1.44	1.42	1.46
All	816 310	29 603	Reference

Reference group is all women in the study population.

SGA, small-for-gestational age; OR, odds ratio; CI, confidence interval. All analyses are adjusted maternal age at infant's birth and period of birth.

Bold type: 95% CI does not include 1.00.

A total of 79% of all SGA events were term SGA events. The results of an ancillary analysis showed no major difference between risk of term SGA birth and risk of preterm SGA birth. For example, low income affected risk of term and preterm SGA birth in a similar way (data not shown).

Table II shows the odds ratio (ORs) for singleton SGA births by mother's and father's occupation after adjustment for maternal age at infant's birth, period of birth (in 5-year groups), family income, region of residence, marital status and smoking habits. Significantly increased ORs of singleton SGA births were found in mothers who worked as ‘textile workers,’ ‘mechanics and iron and metalware workers,’ ‘electrical workers,’ ‘wood workers,’ ‘beverage manufacture workers,’ ‘glass, ceramic and tile workers,’ or ‘packers, loaders and warehouse workers.’ Only fathers who worked as ‘waiters’ had significantly increased ORs of singleton SGA births.

Table II.

Number of SGA events and risk of SGA birth by parental occupation.

Occupational categories (based on the Nordic Occupational Classification)	Mother's occupation				Father's occupation
	SGA events	OR	95% CI		SGA events	OR	95% CI
Technical, science research related workers and physicians	536	1.15	0.95	1.40	1850	0.94	0.88	1.00
Dentists	27	1.12	0.73	1.70	29	1.00	0.70	1.44
Nurses	605	1.06	0.87	1.28	112	0.97	0.81	1.18
Assistant nurses	1958	1.09	0.91	1.30	227	1.00	0.87	1.14
Other health and medical workers	341	1.10	0.90	1.36	66	1.16	0.91	1.48
Teachers	983	1.08	0.90	1.31	412	0.87	0.79	0.97
Religious, juridical and other social-science-related workers	1001	1.14	0.95	1.38	924	0.99	0.92	1.06
Artistic workers	150	1.10	0.87	1.40	151	0.83	0.71	0.98
Journalists	58	0.96	0.70	1.32	66	0.89	0.70	1.14
Administrators and managers	242	1.20	0.97	1.49	274	0.86	0.76	0.97
Clerical workers	2675	1.13	0.95	1.36	596	0.96	0.88	1.05
Sales agents	488	1.07	0.88	1.31	1101	0.95	0.88	1.02
Shop managers and assistants	948	1.04	0.86	1.25	508	0.95	0.87	1.05
Farmers	60	0.74	0.54	1.00	284	0.82	0.72	0.92
Gardeners and related workers	77	0.98	0.73	1.30	147	0.95	0.81	1.12
Fishermen, whalers and sealers	-*	–			31	1.05	0.74	1.49
Forestry workers	-*	–			141	0.87	0.73	1.03
Miners and quarry workers	–*	–			64	1.07	0.84	1.37
Seamen	-*	–			29	1.08	0.75	1.55
Transport workers	65	1.22	0.90	1.65	173	1.09	0.94	1.27
Drivers	126	1.05	0.82	1.35	1005	0.95	0.89	1.02
Postal and communication workers	394	1.19	0.97	1.46	296	0.99	0.88	1.12
Textile workers	150	1.31	1.03	1.66	92	1.07	0.87	1.32
Shoe and leather workers	-*	–			17	1.25	0.78	2.00
Smelters and metal foundry workers	23	1.36	0.86	2.14	200	1.02	0.88	1.18
Mechanics and iron and metalware workers	337	1.29	1.05	1.58	2295	1.05	0.99	1.11
Plumbers	-*	–			223	0.95	0.83	1.09
Welders	21	0.97	0.61	1.54	357	1.05	0.94	1.17
Electrical workers	193	1.28	1.02	1.60	928	0.99	0.92	1.06
Wood workers	93	1.51	1.15	1.99	1018	0.95	0.89	1.02
Painters and wall paperhangers	23	0.93	0.59	1.46	345	0.95	0.85	1.06
Other construction workers	-*	–			586	0.91	0.83	0.99
Bricklayers	-*	–			106	1.09	0.90	1.32
Printers and related workers	122	1.27	0.99	1.64	202	1.02	0.88	1.18
Chemical process workers	49	1.16	0.83	1.62	205	0.99	0.86	1.14
Food manufacture workers	157	1.20	0.95	1.53	302	1.06	0.94	1.20
Beverage manufacture workers	13	2.02	1.12	3.64	13	0.97	0.57	1.66
Glass, ceramic and tile workers	180	1.32	1.04	1.66	366	1.11	0.99	1.24
Packers, loaders and warehouse workers	342	1.24	1.01	1.52	730	1.02	0.94	1.10
Engine and motor operator workers	36	1.02	0.70	1.48	471	1.02	0.92	1.12
Public safety and protection workers	124	0.98	0.76	1.25	308	0.85	0.75	0.95
Cooks and stewards	647	1.13	0.93	1.37	272	1.05	0.92	1.19
Home helpers	2012	1.08	0.90	1.30	125	1.02	0.86	1.22
Waiters	225	0.95	0.76	1.18	106	1.24	1.02	1.51
Building caretakers and cleaners	787	1.15	0.95	1.39	481	1.06	0.96	1.17
Chimney sweeps	-*	–			14	0.61	0.36	1.01
Hairdressers	210	1.21	0.97	1.51	19	1.27	0.81	1.99
Launderers and dry cleaners	256	0.99	0.80	1.22	131	0.92	0.78	1.10

Reference group is all women or men in the study population.

SGA, small-for-gestational age; OR, odds ratio; CI, confidence interval. All analyses are adjusted for family income, region of residence, marital status, smoking habits, maternal age at infant's birth and period of birth. Bold type: 95% CI does not include 1.00. -* Cases <10 are not shown.

Discussion

To the best of our knowledge, this is the first large-scale study to investigate the association between a wide range of maternal and paternal occupation groups and SGA birth; in total over 800 000 live births were included in the study. The main finding of this study was that specific maternal occupations carried a significantly increased risk of SGA birth, whereas among paternal occupational groups, only waiters had an increased risk of SGA birth.

We also found that singletons whose mothers had a low level of family income, which can be seen as a proxy for low SES, had an increased risk of SGA. Our results are in agreement with the results of earlier studies from Sweden and the USA (Parker et al., 1994; World Health Organization, 1995; Kramer et al., 2000), which found a positive association between lower SES and an increased risk of SGA birth. Low SES may be a risk factor for SGA birth because social and economic deprivation is associated with occupational exposure (Savitz et al., 1996a; Ahmed and Jaakkola, 2007b; Meyer et al., 2008; Simcox and Jaakkola, 2008), low social participation (Dejin-Karlsson and Ostergren, 2003), smoking (Horta et al., 1997; Cnattingius, 2004; Bell et al., 2008) and poor nutrition (Hobel and Culhane, 2003, Mitchell et al., 2004).

Previous studies have found that some maternal occupations are associated with an increased risk of SGA birth. A recent study from Finland, which included 2568 singleton newborns, found that mothers employed in the fields of farming and forestry, factory work and mining and construction had a higher risk of giving birth to an SGA infant than housewives (Ahmed and Jaakkola, 2007b). Other previous studies have also reported associations between specific maternal occupations and incidence or risk of SGA birth (Savitz et al., 1996a; Simcox and Jaakkola, 2008; Jakobsson and Mikoczy, 2009); for example, in newborns of nurses (Simcox and Jaakkola, 2008), food service workers (Savitz et al., 1996a; Meyer et al., 2008), electrical equipment operators (Savitz et al., 1996a), hairdressers (Rylander and Kallen, 2005) and rubber workers (Jakobsson and Mikoczy, 2009).

There are several mechanisms through which parental occupation might affect intrauterine growth during pregnancy (Ahmed and Jaakkola, 2007a), including direct exposure of harmful agents through the placenta, exposure to parents’ clothing (which might contain various agents from work) and possible additional exposure from living in close proximity to an industrial workplace. The data in the present study did not include information about specific occupational exposures. It is, however, likely that many of the maternal occupations associated with an increased risk of SGA birth in the present study included occupational exposures associated with increased risk of SGA birth in previous research. For example, previous research indicates that exposure to specific chemicals at work is associated with SGA births (Seidler et al., 1999) and other adverse birth outcomes (Zhang et al., 1992). Moreover, maternal exposure to welding fumes and metal dusts or fumes during pregnancy is associated with reduced fetal growth (Quansah and Jaakkola, 2009). Additionally, an occupational cohort study from Taiwan showed that increased levels of lead in the blood were associated with SGA birth (Chen et al., 2006). The association between environmental lead exposure and SGA birth has also been found in other parts of the world (Pietrzyk et al., 1996; Jelliffe-Pawlowski et al., 2006).

It is also likely that many of the maternal occupations that were associated with an increased risk of SGA in the present study included strenuous physical activity such as heavy lifting (e.g. among packers, loaders and warehouse workers). Previous research has shown that strenuous physical activity is associated with SGA birth (Artal and O'Toole, 2003).

An important limitation of our study is that in our population-based database, information was not available on detailed job tasks or on potential exposure to chemicals on the job. It was therefore necessary to use occupation as a proxy for occupational exposure, and occupational group is an imperfect measure of occupational exposure. Furthermore, we had no information on exposure to harmful agents outside the workplace. Collapsing occupations into broader occupational categories, moreover, creates the potential for bias caused by misclassification. The 53 occupational groups used in our study have, however, been used in several large-scale epidemiological studies from the Nordic countries. A further considerable limitation of this study is that a proportion of parents are likely to have changed occupational category during the study period, and a number of parents of SGA infants will have been classified as students in the 1990 census, and therefore excluded from this study. It is likely that the errors introduced into the data set by this problem are greater for births later in the study period than for births earlier in the study period because the time elapsed between exposure and birth is larger toward the end of the study period. Furthermore, as the study population was of childbearing age, they would most likely obtain ‘better’ jobs with time, i.e. jobs with less exposure to potentially harmful agents. Misclassification of younger subjects to the potentially more demanding or highly exposed jobs held at younger ages would likely result in an underestimation of risk for SGA in some occupations, although the magnitude of this misclassification is uncertain.

This study also has a number of strengths. Approximately 95% of SGA births were classified as such on the basis of ultrasound examination (Swedish Council on Technology Assessment in Health Care, 1999). A further strength was the use of the civic registration number (changed to a serial number to ensure anonymity) assigned to each individual in Sweden, which made it possible to track the records of every individual during the whole follow-up period. This ensured that there was no loss to follow-up. Furthermore, the 1990 data on occupational status used in this study were 92% complete. The quality of data on occupational titles in the Swedish census has been assessed and found to be reasonable (Warnryd et al., 1989). The proportion of concordant occupational titles was 72%. In terms of reliability, the coding showed that about 10% of occupations were misclassified. A further strength of this study was the availability of family income for the year of birth. Finally, we had individual-level data on smoking, which is a major risk factor for SGA birth (Horta et al., 1997; Cnattingius, 2004; Bell et al., 2008). It was also possible to compare parental occupations that were associated with high risk of SGA in singleton births with parental occupations that carry a high risk of lung cancer for the parents (Hemminki and Li, 2003; Ji and Hemminki, 2005).

Conclusion

The present study showed that several maternal occupations carried a significantly higher risk of SGA than the risk in the reference group (all women in the study population). Father's occupation had a minor effect on risks. Further studies are needed to examine the causative agents in those maternal occupations that are associated with an increased risk of having an SGA infant.

Funding

The project described was supported by grant number R01 HD052848-01 A1 from the National Institute of Child Health and Human Development, and its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Child Health and Human Development. This work was also supported by grants to Drs Kristina and Jan Sundquist from the Swedish Research Council, the Swedish Council for Working Life and Social Research, the Swedish Research Council Formas and the Region of Skåne in Sweden. The grantors played no role in the formulation of the research questions, choice of research design, data collection and analyses, preparation of the manuscript or the decision to publish.