Preterm Birth and Prenatal Maternal Occupation: The Role of Hispanic Ethnicity and Nativity in a Population-Based Sample in Los Angeles, California

Abstract

Objectives. We investigated preterm birth (PTB) in relation to maternal occupational exposure and whether effect measures were modified by Hispanic ethnicity and nativity in a population-based sample with high proportion of Hispanics.

Methods. We used a case-control study (n = 2543) nested within a cohort of 58 316 births in Los Angeles County, California, in 2003. We categorized prenatal occupations using the US Census Occupation Codes and Classification System and developed a job exposure matrix. Odds ratios for PTB were estimated using logistic regression.

Results. Odds ratios for PTB were increased for all women in health care practitioner and technical occupations, but the 95% confidence intervals included the null value; effects were more pronounced among Hispanics. We estimated elevated odds ratios for foreign-born Hispanic women in building and grounds cleaning and maintenance occupations. Shift work and physically demanding work affected births among US-born but not foreign-born Hispanics.

Conclusions. Hispanic women are at particular risk for PTB related to adverse prenatal occupational exposure. Nativity may moderate these effects on PTB. Maternal occupational exposures likely contribute to ethnic disparities in PTB.

The prevalence of preterm birth (PTB) in the United States increased over the past decades with persistent racial/ethnic disparities,^1–3 and it remains the main cause of infant mortality.^4,5 Hispanic women have on average a higher prevalence of PTB than non-Hispanic White women,^1,6 and risks tend to vary by nativity (US-born vs foreign-born).^6,7 These disparities remain poorly understood and have not to date been explained by social or demographic factors.^1,6–9 The US Hispanic population is highly diverse with regard to cultural, behavioral, and socioeconomic factors. Research considering underlying differences in occupational or environmental influences to explain ethnic disparities in risk for PTB is sparse and has largely not considered nativity. Differences related to ethnicity and nativity have been reported for behavioral reproductive risk factors including maternal smoking, alcohol consumption, or diet,^10,11 but research has not addressed whether PTB risks related to maternal occupation differ by ethnicity and nativity. Occupational exposure may increase risks for PTB by interrupting the prenatal neuroendocrine balance, thereby promoting premature parturition,^12,13 and these adverse occupational influences may possibly affect Hispanic populations in the United States disproportionally and may possibly also be modified by nativity.

Several studies involving mainly European populations and a handful of US studies including mainly non-Hispanic populations have attempted to elucidate associations between PTB and maternal work, but findings have been inconclusive.^14–24 These earlier studies^14–24 have generally not considered potential influences of ethnicity, race, or nativity on associations between maternal work and PTB. In fact, non-US studies have not considered ethnicity, race, or nativity at all.^{18–20,25,26} US studies have generally adjusted for race/ethnicity but have not accounted for nativity.^21,22,27,28 A few of the earlier studies were registry based, involving a range of occupations, but lacked information regarding whether women worked in the registered job during pregnancy.^22,25,27 Registry-based findings from Sweden did not consider racial/ethnic population composition or nativity but suggested that working as building cleaner, mechanic, or food manufacturer was associated with slightly higher odds ratios for PTB.²⁵ One earlier survey in North Carolina that reported increased odds ratios for PTB among electrical equipment workers and janitors adjusted for ethnicity and race but did not consider nativity.^21,29

Several previous studies from different countries focused on selected types of exposures, such as physical or psychological demand^{14,18–20,27,28} or shift work.¹⁷ Findings were inconclusive, and possible modifying influences of ethnicity or nativity were not examined. Reports from Canada and Europe suggested moderately increased risks for PTB related to physical and psychological job strain.^19,20 By contrast, a Dutch study involving a population that was part northern European, part northern African, part Surinamese, and part other, unspecified origins showed that neither physical demand nor exposure to toxic substances influenced PTB; however, results by race/ethnicity or nativity were not reported.¹⁸ The Danish National Birth Cohort, a mainly northern European population, found no association between PTB and occupation-related infectious exposures.³⁰ Finally, effects for single occupational groups, especially nurses, have been reported, ^26,31 with increased risks seen in North Carolina²² but not in Finland.³⁰ Findings from the US Nurses’ Health Study suggested increased PTB risks in relation to some work exposures.³¹ Potential differences in effects by race/ethnicity and nativity among nurses^26,31 or health care workers²² were not examined in any of these studies. A sole study to date has been conducted among Hispanic women in a rural area of California, investigating farm work (vs no work) and low birth weight and PTB, but no increase in risk was reported.³² Overall, studies from Europe, where much of the previous research was done,^{18,20,25,30,33–35} are not directly applicable to the United States because of differences in racial/ethnic population composition, health care and maternity leave policies, and working conditions.²⁴ In summary, although a few studies investigated effects of maternal work on PTB, to our knowledge, no previous US population-based study has considered effects particularly for Hispanic women differentiated by nativity status and a wide range of maternal occupations.

Thus, we hypothesized that occupational exposure is related to preterm delivery and that the odds ratios may be higher among Hispanic women with effects potentially further modified according to nativity. We used a case-control study with a high proportion of Hispanics nested within a population-based cohort of 58 316 eligible births in Los Angeles County, California, and categorized maternal occupations according to the 2000 US Census Occupational Classification System.³⁶ We also used a job exposure matrix (JEM).

METHODS

The University of California, Los Angeles, Environment and Pregnancy Outcomes Study was originally designed to assess effects of air pollution on birth outcomes.³⁷ Briefly, we selected all 66 795 records for children born in 2003 to mothers who resided in 111 Los Angeles County zip codes (41% of all Los Angeles County births). We excluded births with recorded defects (n = 202), with extreme or missing values for gestational age (< 140 days or > 320 days; n = 5948) and weight (< 500 g or > 5000 g; n = 130), multiple gestations (n = 1574), births that were not eventually reported to the state (n = 110), and births outside Los Angeles County (n = 515), resulting in a final cohort of 58 316 eligible births (87% of the original total). From this cohort, we selected all cases of low birth weight (< 2500 g) or PTB (< 37 weeks) and an equal number of randomly selected controls (≥ 2500 g, full term) from a set of 24 zip codes located in close proximity to air monitoring stations and randomly selected 30% of cases and an equal number of controls from a set of 87 zip codes located near or intersected by major roadways. Cases and controls were thus matched on zip code set (i.e., 24 or 87 zip codes) and birth month by design. We conducted interviews (in English or Spanish) 3 to 6 months postpartum with 2543 of the 6374 women originally selected from the cohort.

Occupational Exposure Assessment

Mothers provided work-related information: primary job including job title, type of business or industry, number of months worked, and weekly and typical daily working hours. To categorize occupations, we used the coding scheme of the 2000 US Census Occupational Classification System, based on occupation codes classifying jobs according to similar duties, demands, and education and training.³⁶ According to each woman’s reported job title and business the corresponding occupation code was assigned, resulting in 20 categories; additionally, we assigned high school and college students to 2 categories (Table A, available as a supplement to the online version of this article at http://www.ajph.org). Women with assigned categories were classified as working during pregnancy or not working during pregnancy; only 26 participants had missing information for occupation during pregnancy and were excluded.

We developed a JEM, assessing strenuous physical labor (physical demand; standing long periods, bending, lifting), psychologically demanding work (psychological demand; e.g., dealing with sick patients), shift work, embryotoxicants (e.g., metals, pesticides), infectious diseases, and indoor air pollution (e.g., combustion). Each job title–business combination was rated as not, maybe, or likely exposed to each of these factors independently by an industrial hygienist and a second reviewer (A. W.); rater discrepancies were resolved by 2 additional reviewers (B. R. and M. W., an exposure specialist). All were blinded to any study-related information.

Pregnancy and Demographic Information

Information retrieved from birth certificates included maternal race/ethnicity, date and place of birth, pregnancy complications, and education. Interview-assessed data included marital status, living with a smoker, maternal smoking, alcohol consumption, and income.

Statistical Methods

To examine potential effects related to occupational categories, we used office and administrative support occupations as the reference on the basis of the assumption that they are less likely than other occupations to produce harmful work-related exposure; previously similar reference categories were used.^21,26 We estimated crude and adjusted odds ratios of PTB (< 37 weeks; with the reference > 37 weeks, not low birth weight) using single and multiple variable logistic regressions, computed separately for occupational categories and for each JEM exposure (reference = not exposed). We conducted separate analyses among all White Hispanic women (hereinafter referred to as Hispanic) and in subgroups by nativity (US-born or foreign-born). Potential confounders were selected a priori; most potential confounders had few missing data (< 1%–2%) except income. We performed multiple imputations with 5 imputation data sets to replace missing values for all covariates in models adjusting for income. We adjusted final models for nonimputed variables: race/ethnicity (among all), maternal education (reference = 12 years), and age (reference = 20–24 years); we selected these variables because they are considered independent risk factors for the outcome and in combination changed 1 or more of the estimates of interest by at least 5% to 10%. Adding parity, marital status, living with a smoker, maternal smoking, alcohol consumption, pregnancy complications, or income to the models did not further change the estimates of interest more than 10%. Covariates were categorized as shown in Table 1. We also examined potential effects of average and typical daily work hours and restricted the sample to those working 7 to 9.5 months during pregnancy. All analyses were done with SAS version 9.2 (SAS Institute, Cary, NC).

TABLE 1—

Demographic and Pregnancy Characteristics Among Women by Work During Pregnancy Status and Among Working Women by Hispanic Ethnicity and Nativity Status: Environment and Pregnancy Outcomes Study; Los Angeles County, CA; 2003

		Working During Pregnancy, No. (%)
Characteristic	Not Working During Pregnancy (n = 1176), No. (%)	Total (n = 1341)	White Non-Hispanic (n = 277)a	Hispanic Foreign-Born (n = 476)a	Hispanic US-Born (n = 253)a
Race/ethnicity
Hispanic White	875 (74.4)	800 (59.7)
Non-Hispanic White	132 (11.2)	299 (22.3)
African American or Black	81 (6.9)	100 (7.5)
Asian	42 (3.6)	64 (4.8)
Otherb	40 (3.4)	72 (5.4)
Missing	6 (0.5)	6 (0.5)
Born in the United States	385 (32.7)	653 (48.7)
Missing	2 (0.2)	1 (0.1)
Maternal age, y
< 20	178 (15.1)	88 (6.6)	3 (1.1)	29 (6.1)	36 (14.2)
20–24	279 (23.7)	259 (19.3)	19 (6.9)	101 (21.2)	76 (30.0)
25–29	302 (25.7)	357 (26.6)	62 (22.4)	134 (28.2)	82 (32.4)
30–34	264 (22.5)	384 (28.6)	104 (37.6)	128 (26.9)	47 (18.6)
≥ 35	153 (13.0)	253 (18.9)	89 (32.1)	84 (17.7)	12 (4.7)
Parity ≥ 1	793 (67.4)	718 (53.5)	96 (34.7)	320 (67.2)	139 (54.9)
Pregnancy complicationsc	89 (7.6)	129 (9.6)	28 (10.1)	42 (8.8)	28 (11.1)
Smoking
Former smokers	234 (19.9)	409 (30.5)	128 (46.2)	108 (22.7)	93 (36.8)
Pregnancy smokers	56 (4.8)	71 (5.3)	22 (7.9)	8 (1.7)	15 (5.9)
Nonsmokers	883 (75.1)	861 (64.2)	127 (45.9)	360 (75.6)	145 (57.3)
Missing	3 (0.3)	0	0	0	0
Living with smoker in pregnancy
Yes	214 (18.2)	245 (18.3)	41 (14.8)	70 (14.7)	52 (20.6)
Missing	18 (1.53)	5 (0.37)	2 (0.7)	0	2 (0.8)
Alcohol use in pregnancy	116 (9.9)	170 (12.7)	50 (18.1)	40 (8.4)	38 (15.0)
Maternal education
≤ 8, y	225 (19.1)	112 (8.4)	1 (0.4)	97 (20.4)	1 (0.4)
9–11, y	343 (29.2)	189 (14.1)	6 (2.2)	114 (24.0)	38 (15.0)
12, y	308 (26.2)	363 (27.1)	28 (10.1)	161 (33.8)	107 (42.3)
13–15, y	139 (11.8)	261 (19.5)	47 (17.0)	55 (11.6)	71 (28.1)
≥ 16, y	140 (11.9)	395 (29.5)	194 (70.0)	40 (8.4)	34 (13.4)
Missing	21 (1.8)	21 (1.6)	1 (0.4)	9 (1.9)	2 (0.8)
Income
< $10 000	312 (26.5)	198 (14.8)	11 (4.0)	102 (21.4)	41 (16.2)
$10 000–29 999	399 (33.9)	427 (31.8)	34 (12.3)	216 (45.4)	93 (36.8)
$30 000–49 999	86 (7.3)	194 (14.5)	34 (12.3)	59 (12.4)	51 (20.2)
≥ $50 000	113 (9.6)	383 (28.6)	185 (66.8)	29 (6.1)	50 (19.8)
Missing	266 (22.6)	139 (10.4)	13 (4.7)	70 (14.7)	18 (7.1)
Married or living together
Yes	929 (79.0)	1053 (78.5)	247 (89.2)	384 (80.7)	175 (69.2)
Missing	9 (0.77)	5 (0.37)	1 (0.4)	3 (0.6)	0
Time worked in pregnancya
≤ 3 mo		144 (11.9)	28 (10.1)	68 (14.3)	29 (11.5)
4–6 mo		271 (22.4)	47 (17.0)	114 (24.0)	61 (24.1)
7–9.5 mo		789 (65.2)	200 (72.2)	291 (61.1)	162 (64.0)
Missing		7 (0.6)	2 (0.7)	3 (0.6)	1 (0.4)

Note. Percentages may not add to 100% because of rounding.

^aExcludes low birth weight term births.

^bIncludes Native American/American Indian, Indian, Filipino, Hawaiian, Guamanian, Samoan, Eskimo, Aleut, Pacific Islander, other (specified).

^cPregnancy complications include any complication recorded on birth certificate.

RESULTS

Basic population characteristics are displayed in Table 1 by pregnancy work status and among working women by ethnicity/nativity (for Hispanics combined see Table B, available as a supplement to the online version of this article at http://www.ajph.org). Hispanic women were the largest ethnic/racial group at 66.6% (n = 1675), of whom 71.1% were foreign-born (n = 1191), with 78.6% born in Mexico (n = 936), 18.4% in Central America (n = 219), and 2.4% in South America (n = 29). The mean gestational ages for PTB among working and nonworking women were 34.5 weeks (SD = 2.8) and 34.6 weeks (SD = 2.7), respectively.

Compared with not working in pregnancy, working was not associated with PTB among all women (adjusted odds ratio [AOR] = 0.93; 95% confidence interval [CI]  = 0.78, 1.11); similarly, odds ratios were not increased among all Hispanics (AOR = 0.93; 95% CI = 0.75, 1.15), with similar estimates among US-born and foreign-born women (data not shown). Among the women who worked in pregnancy, odds ratios for PTB and occupation categories were increased for health care practitioners and technical occupations, building and grounds cleaning and maintenance occupations, and food preparation and serving occupations (Table 2); however, the CIs based on adjusted models included the null value. We also found increased ORs with wide confidence intervals for community and social service occupations and transport and material moving operations (Table 2).

TABLE 2—

Odds Ratios for Preterm Birth in Relation to Maternal Occupation Among All Working Women: Environment and Pregnancy Outcomes Study; Los Angeles County, CA; 2003

Occupation	No.	OR (95% CI)	AORa (95% CI)
Office and administrative support occupations (Ref)	229	1.00	1.00
Management occupations	79	0.95 (0.56, 1.60)	1.02 (0.58, 1.79)
Business operations specialists	19	0.41 (0.13, 1.28)	0.43 (0.13, 1.35)
Financial specialists	26	1.13 (0.50, 2.58)	1.35 (0.58, 3.16)
Computer and mathematical occupations	10	1.03 (0.28, 3.75)	1.21 (0.32, 4.58)
Life, physical, and social science occupations	15	0.56 (0.17, 1.82)	0.70 (0.21, 2.37)
Community and social services occupations	22	2.23 (0.92, 5.43)	2.28 (0.90, 5.80)
Legal occupations	23	1.42 (0.60, 3.35)	1.67 (0.68, 4.10)
Education, training, and library occupations	107	0.96 (0.60, 1.54)	1.09 (0.65, 1.82)
Arts, design, entertainment, sports, and media occupations	34	0.84 (0.40, 1.79)	0.92 (0.40, 2.10)
Health care practitioners and technical occupations	63	1.40 (0.80, 2.46)	1.79 (0.96, 3.33)
Health care support occupations	52	0.97 (0.52, 1.79)	0.93 (0.49, 1.78)
Food preparation and serving occupations	74	1.54 (0.91, 2.62)	1.64 (0.94, 2.85)
Building and grounds cleaning and maintenance occupations	54	2.25 (1.23, 4.11)	1.86 (0.95, 3.63)
Personal care and service occupations	47	0.96 (0.50, 1.83)	0.84 (0.43, 1.63)
Sales occupations	162	1.33 (0.89, 2.00)	1.33 (0.87, 2.03)
Production occupations	91	1.45 (0.89, 2.36)	1.43 (0.83, 2.45)
Transportation and material moving operations	41	1.79 (0.92, 3.49)	1.79 (0.89, 3.61)
Studentsb
High school	17	1.37 (0.51, 3.69)	1.25 (0.43, 3.68)
College or university	27	1.06 (0.47, 2.39)	1.20 (0.51, 2.82)

Note. AOR = adjusted odds ratio; CI = confidence interval; OR = odds ratio. Results are shown for occupation categories (n ≥ 10). Preterm birth < 37 weeks vs term birth ≥ 37 weeks, non-low birth weight (≥ 2500 g). The sample size was n = 1211 (excludes term low birth weight, n = 130). Occupations were grouped according to the 2000 US Census Occupational Classification System.³⁶

^aAdjusted for maternal age, race/ethnicity, and maternal education.

^bHigh school and college or university students are additional categories not based on occupation codes.

Using the JEM approach among all women, we found increased ORs for PTB related to likely physically demanding work and to likely shift work. For other exposures assessed with the JEM, no association with PTB was indicated (Table C, available as a supplement to the online version of this article at http://www.ajph.org).

Restricting the sample to Hispanic women who worked during pregnancy, the OR for health care practitioners and technical occupations increased (AOR = 4.46; 95% CI = 1.24, 16.06), whereas most other estimates related to the occupational categories and the JEM-based exposure remained similar to estimates based on the full sample but with wider CIs (Table 3). In several occupational categories, numbers were too small (n < 10) to analyze (Table A). Stratifying by US-born (n = 253) and foreign-born (n = 476) among Hispanics and using the JEM approach, odds ratios for likely shift work and physical demand, respectively, were increased only among US-born but not among foreign-born Hispanic women (Table 4). However, the OR for the category building and grounds cleaning and maintenance occupations was increased only among foreign-born Hispanic women (AOR = 2.46; 95% CI = 1.10, 5.50), who represented 96.6% of all women holding such a job (Table A).

TABLE 3—

Odds Ratios for Preterm Birth Among White Hispanic Women Working During Pregnancy Related to Maternal Occupational Exposures Based on a Job Exposure Matrix: Environment and Pregnancy Outcomes Study; Los Angeles County, CA; 2003

Job Exposure Matrix Variablesa	No.	OR (95% CI)	AORb (95% CI)
Shift work
Maybe	99	0.79 (0.51, 1.24)	0.70 (0.44, 1.12)
Likely	173	1.35 (0.95, 1.92)	1.30 (0.89, 1.90)
Physical demand
Maybe	165	0.97 (0.66, 1.42)	0.90 (0.60, 1.34)
Likely	248	1.56 (1.11, 2.18)	1.40 (0.95, 2.06)
Psychological demand
Maybe	73	0.94 (0.57, 1.53)	1.19 (0.71, 2.00)
Likely	36	0.76 (0.38, 1.52)	0.95 (0.45, 1.98)
Indoor air pollution
Maybe	133	1.41 (0.96, 2.06)	1.29 (0.86, 1.95)
Likely	54	1.05 (0.60, 1.84)	0.99 (0.55, 1.79)
Embryotoxicants
Maybe	120	0.96 (0.64, 1.43)	0.93 (0.61, 1.42)
Likely	81	1.35 (0.85, 2.16)	1.05 (0.63, 1.75)
Infectious disease
Maybe	156	1.06 (0.74, 1.53)	1.11 (0.76, 1.62)
Likely	94	0.82 (0.52, 1.28)	0.97 (0.60, 1.56)

Note. AOR = adjusted odds ratio; CI = confidence interval; OR = odds ratio. Preterm birth < 37 weeks vs term birth ≥ 37 weeks, non-low birth weight. The sample size was n = 729 (excludes term low birth weight, n = 71).

^aReference: “not exposed” category for each job exposure variable.

^bAdjusted for maternal age and maternal education.

TABLE 4—

Odds Ratios for Preterm Birth in Relation to Maternal Occupational Exposures Assessed With a Job Exposure Matrix Among US-Born and Foreign-Born White Hispanic Working Women: Environment and Pregnancy Outcomes Study; Los Angeles County, CA; 2003

Job Exposure Matrix Variablesa	No.	OR (95% CI)	AORb (95% CI)
Shift work
US-born
Maybe	31	0.77 (0.35, 1.68)	0.67 (0.28, 1.56)
Likely	26	3.13 (1.30, 7.54)	3.52 (1.36, 9.14)
Foreign-born
Maybe	68	0.79 (0.46, 1.35)	0.69 (0.39, 1.22)
Likely	147	1.12 (0.74, 1.67)	1.06 (0.68, 1.64)
Physical demand
US-born
Maybe	46	1.94 (1.01, 3.73)	2.06 (1.04, 4.07)
Likely	31	2.58 (1.18, 5.64)	2.48 (1.05, 5.83)
Foreign-born
Maybe	119	0.65 (0.39, 1.07)	0.60 (0.35, 1.02)
Likely	217	1.24 (0.81, 1.90)	1.07 (0.66, 1.74)
Psychological demand
US-born
Maybe	43	0.88 (0.45, 1.72)	1.13 (0.56, 2.30)
Likely	21	0.75 (0.30, 1.90)	0.75 (0.27, 2.11)
Foreign-born
Maybe	30	1.04 (0.50, 2.18)	1.32 (0.59, 2.93)
Likely	15	0.79 (0.28, 2.26)	1.22 (0.40, 3.75)
Infectious disease
US-born
Maybe	54	1.62 (0.87, 3.03)	1.57 (0.80, 3.08)
Likely	48	0.92 (0.47, 1.78)	1.01 (0.49, 2.08)
Foreign-born
Maybe	102	0.86 (0.55, 1.34)	0.95 (0.59, 1.51)
Likely	46	0.80 (0.43, 1.49)	0.97 (0.50, 1.88)

Note. AOR = adjusted odds ratio; CI = confidence interval; OR = odds ratio. Preterm birth < 37 weeks vs term birth ≥ 37 weeks, non-low birth weight. Numbers are shown for cells of job exposure matrix categories including n ≥ 10 in US-born and in foreign-born groups. The sample size was n = 253 US-born White Hispanic women and n = 476 foreign-born White Hispanic women (excludes 29 term low birth weight among US-born and 42 among foreign-born).

^aReference: “not exposed” category for each job exposure variable.

^bAdjusted for maternal age and maternal education.

When restricting the sample to all women who worked for 7 to 9.5 months during pregnancy, the association with shift work strengthened (likely: AOR = 1.74, 95% CI = 1.12, 2.69). For jobs grouped according to occupational category, the CIs for all estimates widened; only for production occupations did associations strengthen (AOR = 2.11; 95% CI = 1.03, 4.30). Among all working women or among Hispanics, we found no association with PTB for the number of average hours worked per week (> 40 hours and 20–39 hours vs < 20 hours). Reported typical daily hours worked, that is, alternating day and night time of work, working nights, or working more than 8 hours between 6:00 am and 10:00 pm (reference: ≤8 hours 6:00 am–10:00 pm) were also not associated with PTB in either group (results not shown).

DISCUSSION

We found evidence for increased ORs for PTB and maternal work in health care practitioners and technical occupations, with pronounced effects among Hispanics, and in building and grounds cleaning and maintenance among foreign-born Hispanics, that is, the group almost exclusively found in these jobs. Shift work and physically demanding work increased the ORs for PTB, with pronounced effects among US-born but not among foreign-born Hispanic women. In other words, elevated ORs for PTB related to certain demanding types of maternal work, and effects were more pronounced for Hispanic women, with modification of effects by nativity. These findings suggest that prenatal occupational exposures are contributing to the ethnic disparities in PTB in the United States.

Although health care workers have been suspected to be at elevated risks for adverse pregnancy outcomes,^31,38,39 to our knowledge no earlier study has reported risks separately for Hispanics. The pronounced effects we found for this group might indicate that Hispanic women working in health care are doing especially heavy physical work, although we currently have no evidence for this. In a North Carolina registry investigation, nurses’ and nurse aides’ risks for PTB increased²²; however, results were not presented by ethnicity. The US Nurses’ Health Study showed increased risks for early PTB (< 32 weeks) related to working at night but found little evidence for effects of any other work-related exposures.³¹ Because the latter study was conducted solely among nurses, a possible overall excess risk for PTB related to maternal work as a nurse could not be assessed, and differences by ethnicity were not examined. A Finnish study using a Northern European sample showed no increase for PTB risk for nurses compared with office workers.²⁶ Country-specific working conditions and differences in ethnic composition and in health care may contribute to the contrasting findings between the United States and particularly the European Nordic countries.²⁴

Part of the increased risks seen particularly among Hispanic health care workers may be explained by demanding work schedules involving shift work, which is in line with the increased ORs for PTB for likely shift work we found among US-born Hispanics and among those working 7 to 9.5 months during pregnancy. It may be that women who worked longer and harder were generally healthier with the effect that our estimates may even have been attenuated by a potential healthy worker selection bias. Previous findings for shift work and PTB have been inconclusive, and no previous study has specifically addressed effects for Hispanic women.^17,35 Our findings for US-born Hispanic women are slightly stronger than those of a recent systematic review that reported summary relative risk estimates between 1.03 (95% CI = 0.93, 1.14) and 1.16 (95% CI = 1.00, 1.33) for PTB related to different types of shift work not examined by ethnicity and nativity.¹⁷ We also asked about typical hours, including night or alternating day and night work and found no relation with PTB. This result may be explained by the fact that our question did not capture early morning and late evening shifts with possibly similar impacts on the neuroendocrine balance in pregnancy as night work or day–night changes.⁴⁰ Sustained changes in the balance of neuroendocrine mediators can result from irregular sleep related to shift work.^40,41 Because neuropeptide hormones may play a role in parturition control processes,¹² the disturbance of circadian rhythms might interrupt the neuroendocrine balance of pregnancy.^42,43

Physically demanding work moderately increased ORs among all women, and effects were pronounced among US-born but not foreign-born Hispanics. No previous study on occupational physical demands has considered effects according to Hispanic ethnicity and nativity. Two studies reported moderately increased risks for PTB related to physical job strain¹⁹ and standing more than 6 hours per day.²⁰ Standing long hours slightly increased pooled risk estimates for PTB in a systematic review.¹⁵ In Italy, increased risks were reported for being employed in heavy work,⁴⁴ whereas in the Netherlands, no association with PTB was found.¹⁸ There is no comparable measure of objective levels of physical demand across studies and countries. Job-related demands likely differ between countries and populations and may be lower overall in northern European countries, which are known to provide high standards of worker protection.²⁴

The higher ORs related to physical demands and shift work that we found among US-born but not among foreign-born Hispanics may have several explanations. Differences in contextual factors related to acculturation may possibly have contributed to differential effects.⁴⁵ Lower acculturation has been reported for foreign-born compared with US-born Hispanics, and it has been associated with better support networks, healthier nutrition, and lower rates of smoking and alcohol consumption,^9–11,46,47 reflecting healthier practices among women in the country of origin.^9,47 Among foreign-born Hispanics in our sample, fewer reported smoking or alcohol consumption, whereas US-born Hispanics had better education and fewer were in the low-income categories (< $30 000; Table 1). The behavioral factors point toward higher PTB risks among US-born Hispanics, and we would expect better income and education to contribute to lower risk. We adjusted all models for maternal education; we initially included income, but because it did not change the estimates by more than 10%, we did not retain it in the final models. Thus, income or education differences are unlikely to explain the modification of effects by nativity reported here. Adding other factors likely related to nativity^9,47 (Table 1)—that is, parity, pregnancy complications, living with a smoker, maternal smoking or alcohol consumption, and marital status—also did not appreciably change associations. It is possible that other factors, for example those related to diet and physical activity or larger support networks not captured in this study, may differ by nativity and potentially buffer against adverse effects.⁴⁶ Another explanation may be a higher rate of migration of healthier women, likely resulting in a healthy immigrant effect benefiting foreign-born Hispanics. Along those lines, the most powerful influences on birth outcomes may be related to a woman’s health long before pregnancy starts, most likely involving influences operating during her own early childhood and in utero life period.^9,48 Nativity may possibly be an indicator of a combination of these factors, which may in turn moderate the risks that occupational exposures exhibit on PTB.

We estimated strongly increased odds ratios for PTB related to building and grounds cleaning and maintenance occupations among foreign-born Hispanics. Women doing these jobs were almost exclusively foreign-born Hispanics, which supports the notion that jobs and possibly tasks within certain jobs may differ between and among ethnic groups by nativity status.

Among those working 7 to 9.5 months of pregnancy, odds ratios were increased for production occupations; more than 90% of workers in these jobs were foreign-born Hispanics. We also found some suggestion of associations for jobs involving food preparation and serving, in line with 2 earlier studies,^21,25 whereas in the Danish National Birth Cohort no association of food handling and PTB was observed.³⁰ The earlier studies did not report on ethnicity or nativity, whereas our data show that the majority of women working in these jobs were again foreign-born Hispanics.

Limitations and Strengths

Limitations of our study relate to the lack of actual measurement of occupational exposure. However, using women’s reported job titles and tasks and business or industry during pregnancy allowed us to apply the standardized US Census Occupational Classification System.³⁶ Although the determination of specific exposures has limitations, our approach captures complex real-world exposure situations. Additionally, by applying a JEM we could assess job-related impacts without depending on women’s recall or report of exposures, which has advantages and also limitations⁴⁹ because it may still result in nondifferential misclassification of exposure and likely underestimation of risks. Another limitation is the relatively low response to our survey, mainly because of the difficulty in locating women randomly selected from birth records.³⁷ Responders differed somewhat from nonresponders and the original birth cohort with regard to some PTB risk factors. Responders had more education, and a higher percentage were US-born.³⁷ Thus, the inclusion of better educated women may have led to a higher proportion of women with less strenuous jobs. Another limitation is the small number in some occupational categories, which constrained our ability to further investigate effects in subgroups by ethnicity and nativity. Confounding is very unlikely to explain our findings because we evaluated many potential confounders for which we had detailed information. We conducted several sensitivity analyses, including multiple imputations, and added variables to the models without the findings changing appreciably. We also conducted analyses not excluding LBW in the reference group, and estimates remained similar.

Our study had several important strengths. Particularly, this study is, to our knowledge, the first population-based study on a range of maternal occupational exposures and PTB with a large group of Hispanic women, including a high proportion of first-generation immigrants. Other strengths include the detailed maternal interview information about prenatal occupation and detailed information about potential confounders. Our findings may be generalizable to urban US populations with similar racial/ethnic and nativity composition.

Conclusions

We provide evidence that Hispanic women are at particular risk for PTB attributable to adverse prenatal occupational exposures.^12,13,41,43 Nativity may moderate effects on PTB for certain occupational exposures. Thus, prenatal occupational exposures likely contribute to the persistent and yet insufficiently understood ethnic disparities in premature birth occurrence. Preventive measures aiming at reducing risks of PTB related to maternal occupation need to target Hispanic women and consider the implications of nativity status.