Occupational Exposures and Adverse Pregnancy Outcomes Among Nurses: A Systematic Review and Meta-Analysis

Abstract

Background:

The nursing profession has been associated with several adverse pregnancy outcomes. However, the associations between occupational exposures and adverse pregnancy outcomes among this group have not been systematically examined. This review collates all epidemiological evidence to examine the strength of associations and consistency among eligible studies.

Methods:

A computer search of EMBASE and PubMed from 1966 through August 2009 was performed, followed by a search of reference lists of relevant studies and narrative reviews

Results:

Fourteen studies explored the relation between anesthetic gases and spontaneous abortion, 8 the relation between anesthetic gases and congenital malformations, 7 the relation between chemotherapy agents and congenital malformations, and 4 the relation between shift work and spontaneous abortion. In the random-effects models the summary odds ratio (OR) was moderately elevated for all the relations: OR = 1.27, 95% confidence interval (CI) 0.99-1.63 for anesthetic gases and spontaneous abortion. The summary OR was between 1.05 and 1.09 in high-quality studies, registry-based studies, and cohort studies: OR = 1.33, 95% CI 1.09-1.68 for anesthetic gases and congenital malformation. The summary OR was between 0.97 and 1.22 for high-quality studies, registry-based studies, and cohort studies: OR = 1.35; 95% CI 0.91-2.01 for chemotherapy agent and spontaneous abortion. The summary OR was between 1.34 and 1.69 for high-quality studies, registry-based studies, and cohort studies: OR = 1.44, 95% CI 1.06-1.95 for shift work and spontaneous abortion.

Conclusions:

Nurses were found to be at increased risk of adverse pregnancy outcomes, but the strength of association was weaker in the well-designed studies. The significance of the findings is limited by the number and heterogeneity of the studies.

Introduction

According to the World Health Organization (WHO), the nursing profession is increasingly becoming one of the fastest growing workforces in the healthcare sector in many countries, with women of reproductive age constituting about 70%–80% of this group.¹ Diverse occupational hazards, including chemical, physical, and biological hazards, have been reported in the healthcare sector.^2

–8 The work is more often than not stressful and ergonomically challenging, and the occupational hazards have been associated with several adverse pregnancy outcomes, including spontaneous abortion, congenital malformation, preterm delivery, and low birth weight.^2

–5,9,10 Studies conducted in the 1970s systematically reported an association between anesthetic gases and spontaneous abortion and congenital malformation and noted odds ratios (OR) around 3.0.^11,12 These findings have been confirmed in recent studies.^12

–15 Limited studies have also reported on the relation between shift work, chemotherapy agents, strenuous physical activity, and ethylene oxide and adverse pregnancy outcomes, including low birth weight, preterm delivery, small for gestational age (SGA), time to pregnancy, infertility, and menstrual problems.^4,13
–15

The validity of the findings for the relations between occupational exposures and adverse pregnancy outcomes have been questioned because of inherent methodological problems.^6
–8 More importantly, questions relating to the strength and consistency of the associations, as well as confounding, also have not been systematically addressed. Earlier reviews of the association between occupational hazards and adverse pregnancy outcomes among nurses have been qualitative^16
–18 rather than quantitative or have included a limited number of published studies or more diverse occupational groups,^19
–21 and they have not taken into account validity issues relating to heterogeneity and publication bias.

The overall aim of this review was to synthesize the epidemiological evidence on the associations between occupational exposures and adverse pregnancy outcomes among nurses, to establish the overall strengths and consistency of these associations, and to explore the sources of heterogeneity.

Materials and Methods

The meta-analysis reported here adopted the recent guidelines for conducting and reporting meta-analyses proposed by the Strengthening the Reporting of Observational Studies in Epidemiology group (STROBE).²²

Search strategy and study selection

We conducted a computerized search of PubMed and EMBASE for all epidemiological studies reporting on the relations between occupational exposures and adverse pregnancy outcomes using the following keywords: Nurses [Mesh] or Nurse midwives [Mesh] or Nurse anesthetists [Mesh] or Dental assistants [Mesh] or Nurses aides [Mesh] or Operating room nursing [Mesh] or Nitrous oxide [Mesh] or Anesthetics, inhalation [Mesh] and Abortion, spontaneous [Mesh] or Congenital abnormalities [Mesh] or Abnormalities [subheading] Infertility [Mesh] or Premature birth [Mesh] or Infant, low birth weight [Mesh]. The search period was January 1966 through August 2009. We also searched the reference lists of relevant articles included in this review and of narrative reviews.

A study was considered eligible for inclusion if (1) it is in the English language in a peer-reviewed journal, (2) it is a case-control or cohort or cross-sectional design, (3) it provides sufficient information (e.g., OR, incidence rate ratio, cumulative incidence ratio, hazard ratio [HR], or prevalence ratio) that could be used to estimate the relation between occupational exposure(s) and adverse pregnancy outcomes, (4) it is an original study, (5) the exposed population included female nurses, and (6) it assessed occupational exposure(s). When studies overlapped, the most recent one or the one with sufficient adjustment for confounding or the larger study was included in the analysis. Again, when a study reported data on healthcare workers, we only reviewed the data on female nurses. All letters were excluded. A study was also excluded if the authors reported data on nurses and other occupational groups and opted not to examine the data on nurses separately. Studies using the job title as a measure of occupational exposures were excluded.

Several studies reported on the relation between occupational exposure(s) and adverse pregnancy outcomes, but the relations considered in this systematic review were those that at least four studies have reported on. The method adopted in retrieving relevant articles is depicted in Fig. 1.

FIG. 1.

Selection process of studies for systematic review and meta-analysis. Only relations for which sufficient studies have reported on and shown.

Study appraisal

The Newcastle-Ottawa Scale (NOS)²³ recommended by the Cochrane Non-Randomised Studies Methods Working Group was used to assess the quality of the eligible studies. The NOS uses a star rating system to judge quality based on three aspects of the study: (1) selection of cohorts or cases/controls (4 items), (2) comparability of cohorts or cases/controls (2 items), and (3) ascertainment of exposure and outcome (3 items). For cross-sectional studies, we used the first 6 items of the NOS (i.e., selection of cohorts or cases/controls, 4 items; comparability cohorts or cases/controls, 2 items).

High-quality characteristics within each of the items of the NOS were awarded a star up to a maximum of 4 stars for selection, 2 stars for comparability, and 3 stars for ascertainment. The methodological quality of the eligible studies based on the NOS is shown in Tables 2 and 3. The items of the scale (NOS) are shown as footnotes in Tables 2 and 3. We further stratified the methodological quality of the studies according to an a priori chosen cutoff point (≥5) applied in a previous study to represent high quality.²⁴ The studies were critically appraised by the authors, and any discrepancies were resolved between them.

Data extraction

Characteristics of eligible articles were recorded in a data extraction form similar to one used previously.²⁵ The data extraction was performed by the authors, and any discrepancies were resolved in consensus.

Statistical methods

The ORs and 95% confidence intervals (CIs) for the outcome were calculated if the original study provided prevalence ratios. The fixed-effects and the random-effects models^26,27 were used to calculate the overall summary OR, but it was decided a priori to apply the random-effects model to allow for the possibility of variation in the study-specific estimates between occupational exposures and adverse pregnancy outcomes. To explore the sources of heterogeneity, subgroup and meta-regression analyses were undertaken. Statistical test of heterogeneity was performed using the Q-statistic with a conservative p value of 0.10.²⁸ We also calculated I² statistic, which measures the percentage of total variations across studies due to heterogeneity rather than chance: I² = 25% represents trivial heterogeneity, I² = 25%–50% suggests moderate heterogeneity, and I² >50% suggests substantial and important heterogeneity.²⁹ Forest plot was used to illustrate the study-specific relations between occupational exposures and adverse pregnancy outcomes. The tests of Egger et al.³⁰ and of Begg and Mazumdar³¹ were applied to assess small study effects and publication bias, respectively.

The meta-analysis was run on STATA 10.³² EpiCal 2000 was used to calculate the ORs for studies presenting effect measures as prevalence ratios. For studies providing more than one effect estimates (OR) for a relation, the general variance-based method was applied to compute a single effect estimate.²⁸

Results

General overview of excluded and included studies

Twenty-four eligible studies (Table 1) were identified, of which 13 were cohort studies,^{2
–4,10,12,33,35

–40} 9 were cross-sectional,^{9,11,13,14,41

–45} and 2 were case-control studies.^46,47 A total of 22 studies^{1,8,10,48

–66} were excluded because either the definition of outcome or exposed population was incompatible with that of the present study or the studies provided insufficient data to calculate an effect estimate. Selevan et al.⁶⁷ (Fig. 1) and Hemminki et al.⁴⁶ overlapped in both exposure and outcome. Nevertheless, the Hemminki et al. study⁴⁶ was included in the analysis because it included the larger study population. Effect estimates (OR) were calculated for 10 studies,^{9,12,14,33,35,37,39,40,44,45} because the authors provided prevalence ratios as measures of effect, but ORs were computed for 3 studies^3,41,46 that provided more than one effect estimate (ORs) for the relation between the occupational exposures and adverse pregnancy outcomes.

Table 1.

Selected Characteristics and Quality Scores of 26 Observational Studies Concerning Work as a Nurse and Adverse Pregnancy Outcomes

Study design, reference/location	Date of data collection	Exposed	Unexposed	Response rate, %	Exposure of interest	Outcome of interest	Effect estimate used in meta-analysis OR (95 %CI)
Cohort studies Rosenberg and Kirves¹²/ Finland	1965–1972	94 nurse anesthetists^a 177 scrub nurses^b	206 other nurses	Overall 70–75	Anesthetic gases	SA	1.38 (0.62-3.06)^a,c 2.14 (1.15-3.97)^b,c
Axelsson and Rylander³³/Sweden	1970–1979	166 nurses	470 other nurses	Overall 84	Anesthetic gases	SA	1.86 (1.06-3.27)
				Exposed 85Unexposed 84
Axelsson et al.³/Sweden	1980–1988	1,243 midwives	1,262 midwives	Overall 84.3	Anesthetic gases	SA	0.95 (0.62-1.47)
Axelsson et al.²/Sweden	1980–1984	450 nurses	185 nurses	Overall 78.1	Shift work	SA	1.05 (0.62-1.77)
Ericson and Källen³⁴/Sweden	1980–1981	143 dental workers	10,355 women in gainful employment	>70	Anesthetic gases	SA	0.85 (0.72-1.02)
Guirguis et al.³⁵/Canada	1981–1985	3038 nurses	911 nurses	Exposed 97.4	Anesthetic gases	SA	1.29 (1.06-1.58)
				Unexposed 90
Ericson and Källen³⁹/Sweden	1973–1978	1323 infants of operating room nurses	1382 infants of internal ward nurses	Overall >70	Anesthetic gases	SA	1.07 (0.79-1.46)
Heidam³⁷/Denmark	Entire reproductive life before May 1980	179 dental assistants	80 dental assistants	Overall 92.1	Anesthetic gases	SA	0.90 (0.40-2.10)^c
				Exposed 94Unexposed 91
Skov et al.¹⁰/Denmark	1973–1988	1,282 oncology nurses	2572 other nurses	Not specified	Antineoplastic drugs	SA	0.80 (0.38-1.69)
Lauwerys et al.³⁸/Belgium	Unknown	176 operating room nurses	104 other nurses	Exposed 65	Anesthetic gases	SA	0.26 (0.18-0.39)^c
				Unexposed 38.5		CGM	1.00 (0.65-1.55)^c
Saurel-Cubzolles et al.⁴/France	1987–1989	284 operating room nurses	480 other nurses	Overall 77	Anesthetic gases	SA	1.90 (1.10-3.50)
						CGM	1.40 (0.60-3.60)
Ericson³⁶/Sweden	1973–1975	494 operating room nurses	19,127 women in medical profession	Overall 93.7	Anesthetic gases	CGM	0.86 (0.56-1.32)
Whelan et al.⁴⁰/USA	1998–2001	2,446 nurses	5,242 nurses	Overall 76.0	Shift work	SA	1.25 (1.14-1.37)
Cross-sectional studies
Rowland et al.⁴¹/USA	1987	113 dental assistants	559 dental assistants	Overall 69	Nitrous oxide	SA	2.60 (1.30-5.00)
Cohen et al.¹¹/USA	1965–1970	67 operating room nurses	92 other nurses	Overall 77	Anesthetic gases	SA	3.15 (0.95-10.5)^c
Stucker et al.⁴²/France	1985–1986	170 oncology nurses	195 other nurses	Overall 87	Cytostatic drugs	SA	2.20 (1.20-4.10)
Fransman et al.¹³/Netherlands	1996–2003	532 oncology nurses	663 other nurses	Overall 79	Antineoplastic drugs	SA	1.20 (0.60-2.70)
McAbee et al.⁴³/USA	Not specified	318 nurses	815 nurses	Overall 30.4	Antineoplastic drugs	SA	0.62 (0.08-4.99)
Medkova¹⁴/Czech Republic	1985	10 infants	50 infants	Not specified	Cytostatic agents	SA	0.28 (0.04-1.84)^c
Corbett et al.⁹/USA	Not specified	434 operating room nurses	261 operating room nurses	Overall 84.5	Anesthetic gases	CGM	3.21 (1.80-5.73)
Cohen et al.⁴⁴/USA	1968–1978	400 dental assistants	3,184 dental assistants	Overall 71.8	Anesthetic gases	CGM	1.18 (0.59-2.34)
				Exposed 70Unexposed 73.6		SA	2.26 (1.83-2.83)
Report of an Ad Hoc Committee⁴⁵/USA	1978	1,826 nurse anesthetists	2,256 other healthcare workers	Overall 52.8	Anesthetic gases	CGM	Nurse aneathetists 1.71 (1.16-2.54)
		2,781 operating room technicians and operating room nurses				SA	Operating room nurses and technicians 1.11 (0.85-1.45)Nurse anesthetists 1.22 (0.95-1.56)Operating room nurses and technicians 1.36 (1.15-1.61)
Case-control studies
Hemminki et al.⁴⁶/Finland	1973–1979	217 nurses	571 nurses	Not specified	Cytostatic drugs	SA¹	1.31 (0.75-2.28)
					Anesthetic gases	CGM	1.20 (0.70, 2.40)
							1.20 (0.30, 4.60)
Matte et al.⁴⁷/USA	1968–1980	161 infants of nurses	70 infants of other nurses	Not specified	Anesthetic gases	CGM	1.42 (1.06-1.88)

a,b

in column 3 or 6 corresponds to ^a,bin right column.

Crude OR.

CGM, congenital malformation; CI, confidence interval; OR, odds ratio, SA, spontaneous abortion.

The eligible studies were conducted in the United States,^{9,12,40,43
–45,47} Finland,^33,46 Sweden,^2,3,34,37,40 Canada,³⁶ Denmark,^10,38 France,^5,42 Belgium,³⁹ the Netherlands,¹³ and the Czech Republic¹⁴ (Table 1). The most relevant characteristics of the eligible studies are shown in Table 1. The quality scores assessed according to the NOS are also shown in Tables 2 and 3. The methodological quality was high in 12 studies. The mean NOS was 4.7, standard deviation (SD) 2.41.

Anesthetic gases and spontaneous abortion

Fifteen studies provided 16 effect estimates (Rosenberg and Kirves¹² and Report of Ad Hoc Committee⁴⁵ reported 2 effect estimates each) for the relation between anesthetic gases and spontaneous abortion. The forest plot, the effect estimates, and their 95% CIs for the relation, grouped by year of publication, are shown in Fig. 2. There was strong evidence of heterogeneity among the study-specific effect estimates (I² = 88.1%, p = 0.000, Q-statistic = 125.75). The summary OR for the 16 effect estimates of the 15 studies was moderately elevated (fixed effects model: OR 1.21, 95% CI 1.12-1.31; random effects model: OR 1.27, 95% CI 0.99-1.63). In the subgroup analysis, there was evidence of excess risk of spontaneous abortion across the study subgroups, but this excess was not corroborated in the cohort studies,^{3,4,12,33
–35,37
–39} in the high-quality studies,^{4,33

–37,46} (Table 2), in the north European studies,^{3,12,33,34,37
–39,45} and in the registry-based studies^{3,33,34,39,45} (Table 3). Except for studies published before 1980, there was substantial heterogeneity across all the study subgroups. Meta-regression analyses indicated no statistically significant (p = 0.295) effect of the year of publication on the effect estimates: exp (b) = 1.20, 95% CI 0.96-1.50). Similarly, meta-regression showed no effect of the studied determinants on the effect estimate (not shown). The Begg's test for publication bias (p = 0.418) and Egger's test (p = 0.914) for small study bias was statistically significant. The rest of the results are shown in Table 4.

FIG. 2.

Odds ratio (OR) and 95% confidence interval (CI) for the relation between anaesthetic gases and spontaneous abortion among nurses, shown chronologically. The summary odds ratios (diamond) were calculated from the random-effects models. srn, scrub nurses; ann, for anaesthetists; aort, operating room technicians and operating room nurses.

Table 2.

Quality Assessment Using Newcastle-Ottawa Scale for Cohort and Cross-Sectional Studies

	Selection ^a				Comparability		Ascertainment of exposure/outcome
Eligible studies	Item 1	Item 2	Item 3	Item 4	Item 5	Item 6	Item 7	Item 8	Item 9	Total score
Cohort studies
Rosenberg and Kirves¹²	Yes	Yes	No	No	No	No	No	No	No	2
Axelsson and Rylander³³	Yes	Yes	Yes	Yes	Yes	No	Yes	Yes	No	7
Axelsson et al.²	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	8
Axelsson and Rylander³³	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	No	8
Ericson and Kallen³⁶	Yes	No	Yes	Yes	No	No	Yes	Yes	Yes	6
Guirgius et al.³⁵	Yes	Yes	Yes	No	Yes	No	No	No	Yes	5
Ericson and Källen³⁴	Yes	No	Yes	Yes	Yes	No	Yes	Yes	Yes	7
Heidam³⁷	Yes	Yes	Yes	No	Yes	No	No	No	Yes	5
Skov et al.¹⁰	Yes	Yes	Yes	Yes	No	No	Yes	Yes	No	6
Lauwerys et al.³⁸	Yes	Yes	No	No	No	No	No	No	No	2
Saurel-Cubzolles et al.⁴	Yes	Yes	No	No	Yes	No	No	No	No	3
Ericson and Källen³⁹	Yes	Yes	Yes	Yes	No	No	Yes	Yes	Yes	7
Whelan et al.⁴⁰	Yes	Yes	No	No	Yes	No	No	No	No	3
Cross-sectional studies
Rowland et al.⁴¹	Yes	Yes	No	No	Yes	Yes	NA	NA	NA	4
Cohen et al.⁴⁴	Yes	Yes	No	No	Yes	Yes	NA	NA	NA	4
Stucker et al.⁴²	Yes	Yes	No	No	Yes	No	NA	NA	NA	3
Fransman et al.¹³	Yes	Yes	Yes	Yes	Yes	Yes	NA	NA	NA	6
McAbee et al.⁴³	Yes	No	No	No	Yes	Yes	NA	NA	NA	3
Medkova¹⁴	Yes	No	No	No	No	No	NA	NA	NA	1
Corbett et al.⁹	Yes	No	No	No	No	No	NA	NA	NA	2
Cohen et al.¹¹	Yes	Yes	No	No	No	No	NA	NA	NA	2
Report of Ad Hoc Committee⁴⁵	Yes	No	Yes	No	No	No	NA	NA	NA	2

Item 1, representativeness of cohort?; Item 2, appropriateness of non-exposed?; Item 3, appropriateness of assessment of exposure?; Item 4, outcome present at the start of study?; Item 5, primary confounders?; Item 6, additional confounders?; Item 7, appropriateness of assessment of outcome?; Item 8, was follow-up complete?; Item 9, was response rate between compared groups >85% or nonresponse similar in exposed and nonexposed?

NA, not applicable.

Table 3.

Quality Assessment Using Newcastle-Ottawa Scale for Case-Control Studies

	Selection ^a				Comparability		Assessment of Outcome
Eligible studies	Item 1	Item 2	Item 3	Item 4	Item 5	Item 6	Item 7	Item 8	Item 9	Total score
Hemminki et al.⁴⁶	Yes	Yes	Yes	Yes	Yes	No	Yes	No	No	6
Matte et al.⁴⁷	Yes	Yes	Yes	Yes	Yes	No	Yes	No	No	6

Item 1: is the case definition adequate? Item 2: representativeness of cases? Item 3: appropriate selection of controls? Item 4: adequate definition of controls? Item 5: primary confounders? Item 6: additional confounders? Item 7: appropriateness of assessment of exposure? Item 8: method of assessment similar in cases and controls? Item 9: was response rate between compared groups >85% or nonresponse similar in exposed and non-exposed?

Table 4.

Summary Odds Ratios of Random Effects Model for Relation Between Anesthetic Gases and Spontaneous Abortion

	Summary OR for random effects model
Study characteristic	n	OR (95% CI)	Q-statistic	I²%	p value
Effect estimates of all studies	16	1.27 (0.99-1.62)	125.75	88.1	0.00
Study design
Cohort	10	1.07 (0.77-1.50)	72.28	87.5	0.000
Cross-sectional	5	1.73 (1.26-2.39)	20.17	80.2	0.000
Regional location
Northern Europe	9	1.00 (0.68-1.46)	56.90	85.9	0.000
North America	6	1.60 (1.25-2.06)	22.77	78.0	0.000
Year of publication
1980 or later	11	1.17 (0.82-1.67)	116.53	91.4	0.002
Before 1980	5	1.37 (0.99-1.63)	4.67	14.4	0.322
Assessment of spontaneous abortion
Register/medical records	5	1.05 (0.83-1.33)	8.15	50.9	0.086
Self-report	11	1.36 (0.97-1.93)	100.45	90.0	0.000
Adjustment for ≥3 potential confounders
Yes	6	1.51 (1.00-2.09)	26.96	81.5	0.000
No	10	1.12 (0.79-1.59)	79.4	88.6	0.000
Quality
High	7	1.09 (0.89-1.34)	14.28	57.8	0.027
Low	9	1.36 (0.97-1.92)	99.16	91.9	0.000

One case-control study and one study published in Western Europe were not reported.

n = number of effect estimate; I² = heterogeneity statistics; high-quality refers to total score ≥5 on the Newcastle-Attawa Scale.

Anesthetic gases and congenital malformations

Eight studies (3 cross-sectional studies^9,44,45 3 cohort studies,^4,36,38 and 2 case-control studies^46,47) provided 9 effect estimates (Report of Ad Hoc Committee⁴⁶ reported 2 effect estimates) for the relation between anesthetic gases and congenital malformations. The forest plot for the study-specific relations, grouped by year of publication, is shown in Fig. 3. There was substantial heterogeneity among the study-specific effect estimates (I² = 55.6%, p = 0.021, Q-statistic = 18.022). The overall summary OR for all 9 effect estimates was moderately elevated (fixed effects model: OR 1.29, 95% CI 1.12-1.48; random effects models: OR 1.33, 95% CI 1.06-1.68). In a sensitivity analysis, the between-study heterogeneity was trivial in the cohort studies^4,36,38 (I² = 0.0%, p = 0.617, Q-statistic = 0.97), in the studies published after 1980^{4,38,44,46,47} (I² = 0 %, p = 0.765, Q-statistic = 1.84), and in the studies adjusting for at least three potential confounders (I² = 19.6%, p = 0.292, Q-statistics = 3.73). Moderate heterogeneity was observed, however, in the high-quality studies and in the studies retrieving information on congenital malformations from national registers or medical records^36,46,47 (I² = 45%, p = 0.163, Q-statistic = 3.63), and substantial heterogeneity was noted in the north American studies^9,44,45,47 (I² = 66%, p = 0.017, Q-statistic = 12.03). The risk of congenital malformation was increased in the North American studies (random effects model: OR 1.41, 95% CI 1.20-1.66) and moderately increased in the remaining study subgroups, including the studies relying on national registers or medical records for data on the outcome (random effects model: OR 1.22, 95% CI 0.96-1.54), in the studies published after 1980 (random effects model: OR 1.27, 95% CI 1.02-1.58), in the high-quality studies (Table 2) (random effects model: OR 1.15, 95% CI 0.78-1.70), and in the studies adjusting for at least three potential confounders (random-effects model: OR 1.20, 95% CI 0.92-1.56), but not in the cohort studies (random-effects model: OR 0.97, 95% CI 0.72-1.29). The Begg's test (p = 0.436) for publication bias and the Egger's test (p = 0.718) for small study bias were not statistically significant. There were not enough studies to undertake meta-regression analysis.

FIG. 3.

Odds ratio (OR) and 95% confidence interval (CI) for the relation between anaesthetic gases and congenital malformation among nurses, shown chronologically. The summary odds ratios (diamond) were calculated from the random-effects model.

Chemotherapy agents and spontaneous abortion

Four cross-sectional studies,^13,14,42,48 a case-control study,⁴⁶ and a cohort study¹⁰ examined the relation between shift work and spontaneous abortion. There was evidence of moderate heterogeneity among the study-specific effect estimates (I² = 45.0%, p = 0.190, Q-statistic = 7.44). The summary OR for all the 6 studies was moderately elevated (fixed effects models: OR 1.28, 95% CI 0.93-1.75; random effects model: OR 1.21, 95% CI 0.80-1.83). In a sensitivity analysis, trivial heterogeneity was noted in the high-quality studies^10,13,46 (I² = 0.0%, p = 0.572, Q-statistic = 1.02), in the studies adjusting for at least three potential confounders^13,42,43 (I² = 16.4%, p = 0.302, Q-statistic = 2.39), and in the studies^3,31,41 relying on national registers or medical records for data on spontaneous abortion (I² = 7.3%, p = 0.299, Q-statistic = 1.08).

Only one cohort study¹⁰ reported on the above relation. The summary ORs in the high-quality studies (random effects model: OR 1.34, 95% CI 0.89-2.02), in the studies controlling for at least three potential confounders (random effects model: OR 1.59, 95% CI 0.93-2.72), and in the studies relying on national registers or medical records for data on the outcome (random effects model: OR 1.37, 95% CI 0.78-2.39) were elevated. The summary OR from the North American studies was reduced (random effects model: OR 1.15, 95% CI 0.78-1.70). All the studies were published after 1980. The Begg's test (p = 0.176) for publication bias and Egger's test (p = 0.129) for small study bias both yielded nonsignificant results. The were not enough data to undertake meta-regression.

Shift work and spontaneous abortion

The risk of spontaneous abortion was related to shift work in 3 cohort studies^3,33,40 and in a case-control study.⁴⁶ The forest plot for the study-specific effect estimates is shown in Figure 4. There was substantial heterogeneity (I² = 61.9%, p = 0.049, Q-statistic = 7.88) and moderate risk of spontaneous abortion (fixed effects model: OR 1.28, 95% CI 1.17-1.39; random effects model: OR 1.44, 95% CI 1.06-1.95). There were very limited data to continue with further analysis. However, the Begg's test (p = 1.00) for publication bias and the Egger's test (p = 0.427) for small study bias were both statistically nonsignificant.

FIG. 4.

Odds ratios (OR) and 95% confidence intervals (CI) for the relation between chemotherapy agent and spontaneous abortion among nurses, shown chronologically. The summary odds ratio (diamond) was calculated from the random-effects model.

Discussion

The present systematic review and meta-analysis provided evidence of a moderate association between exposure to anesthetic gases and risk of spontaneous abortion, but the findings were not consistent across the study subgroups. The associations between anesthetic gases and congenital malformation and that between chemotherapy agents and spontaneous abortion were moderate and consistent. The association between shift work and spontaneous abortion was elevated but suggestive, based on limited data.

Although there have been several narrative reviews^16

–21 exploring this topic, only 2 were systematic reviews,^19,20 with no elements of meta-analysis. These studies^19,20 focused on all healthcare workers. Our study covers 12 years after the most recent review¹⁶ and focused on nurses. Dranitsaris et al.²⁰ did not define any eligibility criteria and included 2 studies rejected in the present meta-analysis. The strength of association for the relation between anesthetic gases and spontaneous abortion and that between chemotherapy agents and congenital malformation in the previous studies^19,20 were stronger than that observed in the present meta-analysis. Unlike one previous study,¹⁹ we explored the sources of heterogeneity for the relation between anesthetic gases and spontaneous abortion, but none of the studied factors explained the inconsistencies. The number of studies reporting on the remaining relations was insufficient for a detailed investigation of the sources of heterogeneity.

In practice, nurses and healthcare workers in general are simultaneously exposed to several occupational hazards in medical facilities; the lack of data on other occupational hazards in the original studies makes it difficult to attribute the risk of spontaneous abortion and congenital malformation to anesthetic gases, chemotherapy agents, and shift work alone.

Strengths and limitations

The present meta-analysis had a number of strengths. The guidelines for reporting and conducting systematic reviews and meta-analysis proposed by the STROBE²² group were applied. All published studies on the subject were retrieved by applying a logical strategy to search several sources. The eligibility criteria for inclusion of relevant published studies were prespecified. The studies included in the meta-analyses were restricted to English language articles published in peer-reviewed journals, which could introduced selection bias. Nevertheless, very few studies in the search published in other languages were identified; therefore, we believed the magnitude of selection bias is minimal. The small number of studies limited a thorough investigation of heterogeneity. The cutoff point applied in the present systematic review and meta-analysis was arbitrarily, and a study using a different cutoff point may arrive at different conclusions.

The main study characteristics to be considered in the context of validity, including selection of study population, assessment of exposure and outcome, sample size, response rate, control of confounders and statistical limitations, are discussed below.

Selection of study populations

With regard to anesthetic gases, different populations at risk were studied, including operating room workers (excluding physicians),^{4,9,11,12,35,36,38,39,45
–47} midwives,³ and dental assistants.^36,37,41,44 These population groups are not expected to have similar levels of exposure to anesthetic gases, and the risk of adverse pregnancy outcomes may differ. Chemotherapy agents were also studied among oncology nurses,^{10,13,14,43,46} and shift work was investigated among nurses (unspecified)^2,40 and operating room workers.^3,46 Most studies selected the comparison group from the healthcare sector, who may differ from the exposed group on other aspects of the work, such as prolonged standing, stress, and radiation exposure.^11,35 In other studies, the comparison groups were from the same occupation as the study population but did not work during pregnancy.^{4,14,36,38,44,45} The problem with these studies was that timing of exposure may differ between the comparison groups. Thus, adjusting for differences in occupational exposure (e.g., shift work) may present a unique challenge. Another problem is the healthy worker effect, which could attenuate the strength of the effect estimate. Few studies, however, selected a valid comparison group.^{3,12,13,37,39,40,41,42,46} The comparison group in some studies was from the general population.^2,34 Healthcare workers are generally known to differ markedly from the general population in their knowledge of the impact of occupational hazards on the fetus. They have better knowledge of pregnancy diagnosis compared to the general population.^4,5 They may also become pregnant at more advanced maternal age compared to the women in the general population.⁶⁸ Furthermore, the general population includes unemployed and unhealthy women, women who are less educated, heavy smokers, and those who are more likely to indulge in illicit drugs and to start prenatal care late. They are likely to have a higher number of previous births and unfavorable birth outcomes.⁶⁹ These factors are potential confounders of the relations between occupational exposure and the risk of adverse pregnancy outcomes.^10,34,36,39

Assessment of occupational exposures

There are important methodological issues related to the assessment and source of exposure and timing and dose of exposure. Most studies used a dichotomous exposure classification. Two studies^40,41 measured the frequency of exposure, however, and the dose of exposure was measured in three studies.^3,10,13 The generic terms anesthetic gases, chemotherapy agents, and antineoplastic drugs were mostly applied in the studies, making it difficult to relate any association to any particular occupational agents (e.g., nitrous oxide, cyclophosphamide). The definition of shift work also varied, with a vague definition provided in one study.⁴⁶

Most studies included in the present systematic review primarily focused on assessing exposure occurring any time during pregnancy^{3,12,13,35

–39,42,44,47} or exposure in the first trimester.^{2,3,5,41,43,46} Exposure during the last menstrual period⁴¹ or around the period of conception¹¹ was also reported, but some studies failed to identify a specific time window of susceptibility.^{9,10,33,34,36
–38,43} Trimesters and any time during pregnancy exposures do not completely represent critical windows of embryonic or fetal development.^70
–72 Thus, using trimesters or a broad exposure window (e.g., any time during pregnancy) to define the critical window of exposure may inaccurately define the susceptibility period and, hence, bias the effect estimate.^70,71

Potential errors in assessing exposure in the original studies were also associated with the sources of exposure information (Table 1). Exposure information was derived mostly from postal questionnaires to the mothers^{3,9,11,12,14,33,34,39,41

–45,47} or by questionnaire administration to head nurses.^10,46 Querying mothers allows collection of detailed data on characteristics of the exposures and personal data of the study population, but when exposures are assessed from memory after delivery, there is a potential for information bias, with exposures perceived as hazards overreported by cases and underreported by controls, particularly if the adverse pregnancy outcomes are obvious to participants and the exposure of interest is difficult to remember. Information bias because of maternal recall is also likely to increase when public concern for the exposure in question increases, but concentration measurement as applied in two studies^10,13 may be free of these problems.

Assessment of adverse pregnancy outcomes

Studies reporting on spontaneous abortion have investigated late spontaneous abortion, but Axelsson et al.³ and Whelan et al.⁴⁰ reported on both early (<12 weeks) and late spontaneous abortions in women working irregular shift. Even for studies reporting on late spontaneous abortion, the end point definition was heterogeneous, including loss of product of conception up to 28 weeks³ or 29 weeks⁴² of gestation or before 21 weeks⁴¹ or 20 weeks of gestation.^12,35,40,44 Hemminki et al.,⁴⁶ however, applied the hospital discharge diagnoses corresponding to codes 643 and 645 of the eighth revision of the International Classification of Diseases (ICD). Some studies, failed to provide any definition.^12,14,37

Congenital malformations were grouped into a single outcome usually undertaken to increase statistical power, the result of which could decrease the biological validity.⁷² Some studies, however, focused on specific malformations, but the studied malformations varied.^44

–47,49

Another issue of concern is the source of information for adverse pregnancy outcomes. The eligible studies used a variety of sources, including maternal interview,^4,35,42 postal questionnaires,^{9,11

–14,37,38,40,41,43
–45} and medical records or national registers.^{10,34,36,39,46,47} The use of different sources may result in different prevalence estimates for a given adverse pregnancy outcome. Maternal reports of adverse pregnancy outcomes may be sensitive to such factors as culture, educational level, ethnicity, medical experience, and concern and knowledge of environmental and occupational risk factors influencing the outcome.^6,7,72 Variations in these factors between the study population and comparison groups may influence the effect estimates.^7,72 Data from national registers or medical records may provide reliable data but underreporting of minor malformations, and failure to capture early pregnancy losses (including loss from severe malformation) may undermine the integrity of the registers or medical records and, thus, lead to underestimation of the effect. Geographical and socioeconomic differences in hospitalization may also influence identification of adverse pregnancy outcomes in registers and medical records.⁷²

Sample size, response rate control of confounders, and statistical limitations

Sample sizes in the original studies were woefully inadequate (Table 1), causing increased variability and decreased stability for some results. The response rates were often fairly low, ranging from 30.3 to 97.4% (Table 1), with response rate comparison between the study group and the comparison group mostly not reported. Twelve^{3,4,13,35,36,39,41

–44,47} of the 24 studies had controlled for at least three potential confounders of the relation between occupational exposures and congenital malformation and spontaneous abortion, but 5 studies^{11,12,14,34,38} provided only crude effect estimates, and the rest accounted for one or two potential confounders of adverse pregnancy outcomes. Except for two studies,^3,40 the dependency of an adverse pregnancy outcome from the same woman was not taken into account in the analysis.

Conclusions

The present meta-analysis strengthened previous evidence that occupational exposure to anesthetic gases may increase the risk of spontaneous abortion. Chemotherapy agents may increase the risk of congenital malformations. The association between shift work and spontaneous abortion was suggestive. A new summary effect estimate was provided for the relation between anesthetic gases and congenital malformation. The significance of these associations is limited by the small number and heterogeneity of the studies.

Footnotes

Acknowledgments

R.Q. was funded by a Ph.D. Scholarship from the Medical School of the University of Birmingham, United Kingdom.

Disclosure Statement

The authors have no conflicts of interest to report.

References

World Health organization. Statistics on health workforce demographics. apps.who.int/globalatlas/default.asp

Axelsson

, Rylander

, Molin

. Outcome of pregnancy in relation to irregular and inconvenient work schedules. Br J Ind Med, 1989; 46:393–398.

Axelsson

, Ahlborg

Jr , Bodin

. Shift work, nitrous oxide exposure, and spontaneous abortion among Swedish midwives. Occup Environ Med, 1996; 53:374–378.

Saurel-Cubizolles

, Hays

, Estryn-Behar

. Work in operating rooms and pregnancy outcome among nurses. Int Arch Occup Environ Health, 1994; 66:235–241.

Saurel-Cubizolles

, Kaminski

, Llado-Arkhipoff

et al. Pregnancy and its outcome among hospital personnel according to occupation and working conditions. J Epidemiol Community Health, 1985; 39:129–134.

Lindbohm

. Effects of parental exposure to solvents on pregnancy outcome. J Occup Environ Med, 1995; 37:908–914.

Ahlborg

Jr , Hemminki

. Reproductive effects of chemical exposures in health professions. J Occup Environ Med, 1995; 37:957–961.

Ekblad

. Biological agents and pregnancy. J Occup Environ Med, 1995; 37:962–965.

Corbett

, Cornell

, Endres

, Lieding

. Birth defects among children of nurse-anesthetists. Anesthesiology, 1974; 41:341–344.

10.

Skov

, Maarup

, Olsen

, Rørth

, Winthereik

, Lynge

. Leukaemia and reproductive outcome among nurses handling antineoplastic drugs. Br J Ind Med, 1992; 49:855–861.

11.

Cohen

, Bellville

, Brown

Jr . Anesthesia, pregnancy, and miscarriage: A study of operating room nurses and anesthetists. Anesthesiology, 1971; 35:343–347.

12.

Rosenberg

, Kirves

. Miscarriages among operating theatre staff. Acta Anaesthesiol Scand, 1973; 53:37–42.

13.

Fransman

, Roeleveld

, Peelen

, de Kort

, Kromhout

, Heederik

. Nurses with dermal exposure to antineoplastic drugs: Reproductive outcomes. Epidemiology, 2007; 18:112–119.

14.

Medkova

. On the problems of agents with mutagenic and carcinogenic effects in health care. Cytostatic drugs. Acta Univ Palacki Olomuc Fac Med, 1990; 126:107–115.

15.

Bodin

, Axelsson

, Ahlborg

Jr . The association of shift work and nitrous oxide exposure in pregnancy with birth weight and gestational age. Epidemiology, 1999; 10:429–436.

16.

Figa-Talamanca

. Reproductive problems among women healthcare workers: Epidemiologic evidence and preventive strategies. Epidemiol Rev, 2000; 22:249–260.

17.

Ahlborg

Jr , Hemminki

. Reproductive effects of chemical exposures in health professions. J Occup Environ Med, 1995; 37:957–961.

18.

Olfert

. Reproductive outcomes among dental personnel: A review of selected exposures. J Can Dent Assoc, 2006; 72:821–825.

19.

Boivin

. Risk of spontaneous abortion in women occupationally exposed to anaesthetic gases: A meta-analysis. Occup Environ Med, 1997; 54:541–548.

20.

Dranitsaris

, Johnston

, Poirier

et al. Are health care providers who work with cancer drugs at an increased risk for toxic events? A systematic review and meta-analysis of the literature. J Oncol Pharm Pract, 2005; 11:69–78.

21.

Buring

, Hennekens

, Mayrent

, Rosner

, Greenberg

, Colton

. Health experiences of operating room personnel. Anesthesiology, 1985; 62:325–330.

22.

von Elm

, Altman

, Egger

et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement: Guidelines for reporting observational studies. Lancet, 2007; 370:1453–1457.

23.

Wells

, Shea

, O'Connell

et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. www.ohri.ca/programs/clinical_epidemiology/oxford.htm

24.

Roffey

, Wai

, Bishop

et al. Causal assessment of awkward occupational postures and low back pain Results of a systematic review. Spine J, 2010; 10:89–99.

25.

Jaakkola

, Ahmed

, Ieromnimon

et al. Preterm delivery and asthma: A systematic review and meta-analysis. J Allergy Clin Immunol, 2006; 118:823–830.

26.

Mantel

, Haenszel

. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst, 1959; 22:719–748.

27.

DerSimonian

, Laird

. Meta-analysis in clinical trials. Control Clin Trials, 1986; 7:177–188.

28.

Petitte

. Approaches to heterogeneity in meta-analysis. Stat Med, 2001; 20:3625–3633.

29.

Higgins

, Thompson

. Quantifying heterogeneity in a meta-analysis. Stat Med, 2002; 21:1539–1558.

30.

Egger

, Davey Smith

, Schneider

, Minder

. Bias in meta-analysis detected by a simple, graphical test. BMJ, 1997; 315:625–634.

31.

Begg

, Mazumdar

. Operating characteristics of a rank correlation test for publication bias. Biometries, 1994; 50:1088–1101.

32.

Stata Corporation. Stata statistical software, release 10. College Station, TX: Stata Corporation, 2005.

33.

Axelsson

, Rylander

. Exposure to anaesthetic gases and spontaneous abortion: Response bias in a postal questionnaire study. Int J Epidemiol, 1982; 11:250–256.

34.

Ericson

, Källen

. Pregnancy outcome in women working as dentists, dental assistants or dental technicians. Int Arch Occup Environ Health, 1989; 61:329–333.

35.

Guirguis

, Pelmear

, Roy

, Wong

. Health effects associated with exposure to anaesthetic gases in Ontario hospital personnel. Br Ind Med, 1990; 47:490–497.

36.

Ericson

, Kallen . Survey of infants born in 1973 or 1975 to Swedish women working in operating rooms during pregnancies. Anesth Analg, 1979; 58:302–305.

37.

Heidam

. Spontaneous abortions among dental assistants, factory workers, painters, and gardening workers: A follow-up study. J Epidemiol Community Health, 1984; 38:149–155.

38.

Lauwerys

, Siddons

, Misson

et al. Anaesthetic health hazards among Belgian nurses and physicians. Int Arch Occup Environ Health, 1981; 48:195–203.

39.

Ericson

, Källen

. Hospitalization for miscarriage and delivery outcome among Swedish nurses working in operating rooms 1973–1978. Anesth Analg, 1985; 64:981–988.

40.

Whelan

, Lawson

, Grajewski

. Work schedule during pregnancy and spontaneous abortion. Epidemiology, 2007; 18:350–355.

41.

Rowland

, Baird

, Shore

, Weinberg

, Savitz

, Wilcox

. Nitrous oxide and spontaneous abortion in female dental assistants. Am J Epidemiol, 1995; 141:531–538.

42.

Stucker

, Caillard

, Collin

, Gout

, Poyen

, Hemon

. Risk of spontaneous abortion among nurses handling antineoplastic drugs. Scand J Work Environ Health, 1990; 16:102–107.

43.

McAbee

, Gallucci

, Checkoway

. Adverse reproductive outcomes and occupational exposures among nurses: An investigation of multiple hazardous exposures. AAOHN J, 1993; 41:110–119.

44.

Cohen

, Gift

, Brown

et al. Occupational disease in dentistry and chronic exposure to trace anesthetic gases. J Am Dent Assoc, 1980; 101:21–31.

45.

American Society of Anesthesiologists. Report of Ad Hoc Committe. Effect of trace anesthetics on the health of operating room personnel. Anesthesiology, 1974; 41:321–340.

46.

Hemminki

, Kyyrönen

, Lindbohm

. Spontaneous abortions and malformations in the offspring of nurses exposed to anaesthetic gases, cytostatic drugs, and other potential hazards in hospitals, based on registered information of outcome. J Epidemiol Community Health, 1985; 39:141–147.

47.

Matte

, Mulinare

, Erickson

. Case-control study of congenital defects and parental employment in health care. Am J Ind Med, 1993; 24:11–23.

48.

Saurel-Cubizolles

, Job-Spira

, Estryn-Behar

. Ectopic pregnancy and occupational exposure to antineoplastic drugs. Lancet, 1993; 341:1169–1171.

49.

Luke

, Mamelle

, Keith

. The association between occupational factors and preterm birth: A United States nurses' study. Research Committe of the Association of the Women's Health, Obstetric, and Neonatal Nurses. Am J Obstet Gynecol, 1995; 173:849–862.

50.

Lawson

, Whelan

, Hibert

, Grajewski

, Spiegelman

, Rich-Edwards

. Occupational factors and the risk of preterm birth in nurses. Am J Obstet Gynecol, 2009; 200:e1–8.

51.

Simcox

, Jaakkola

. Does work as a nurse increase the risk of adverse pregnancy outcomes? J Occup Environ Med, 2008; 50:590–592.

52.

Katz

, Miller

, Bowes

Jr . Pregnancy complications of physicians. West J Med, 1988; 149:704–707.

53.

Grunebaum

, Minkoff

, Blake

. Pregnancy among obstetricians: A comparison of births before, during, and after residency. Am J Obstet Gynecol, 1987; 157:79–83.

54.

Pinhas-Hamiel

, Rotstein

, Achiron

et al. Pregnancy during residency—An Israeli survey of women physicians. Health Care Women Int, 1999; 20:63–70.

55.

Osborn

, Harris

, Reading

, Prather

. Outcome of pregnancies experienced during residency. J Fam Pract, 1990; 31:618–622.

56.

Miller

, Katz

, Cefalo

. Pregnancies among physicians. A historical cohort study. J Reprod Med, 1989; 34:790–796.

57.

Schwartz

. Pregnancy in physicians: Characteristics and complications. Obstet Gynecol, 1985; 66:672–676.

58.

Phelan

. Pregnancy during residency: I. The decision “to be or not to be.” Obstet Gynecol, 1988; 72:425–431.

59.

Gabbe

, Morgan

, Power

, Schulkin

, Williams

. Duty hours and pregnancy outcome among residents in obstetrics and gynecology. Obstet Gynecol, 2003; 102:948–951.

60.

Klebanoff

, Shiono

, Rhoads

. Outcomes of pregnancy in a national sample of resident physicians. N Engl J Med, 1990; 323:1040–1045.

61.

Heinonen

, Saarikoski

. Reproductive risk factors, pregnancy characteristics and obstetric outcome in female doctors. Br J Obstet Gynaecol, 2002; 109:261–264.

62.

Seneviratne

, Fernando

. Influence of work on pregnancy outcome. Int J Gynaecol Obstet, 1994; 45:35–40.

63.

Ortayli

, Ozugurlu

, Gokcay

. Female health workers: An obstetric risk group. Int J Gynaecol Obstet, 1996; 54:263–270.

64.

Whelan

, Lawson

, Grajewski

, Hibert

, Spiegelman

, Rich-Edwards

. Work schedule during pregnancy and spontaneous abortion. Epidemiology, 2007; 18:350–355.

65.

Koemeester

, Broersen

, Treffers

. Physical work load and gestational age at delivery. Occup Environ Med, 1995; 52:313–315.

66.

Valanis

, Vollmer

, Steele

. Occupational exposure to antineoplastic agents: Self-reported miscarriages and stillbirths among nurses and pharmacists. J Occup Environ Med, 1999; 41:632–638.

67.

Selevan

, Lemasters

. The dose-response fallacy in human reproductive studies of toxic exposures. J Occup Med, 1987; 29:451–454.

68.

Klebanoff

, Shiono

, Rhoads

. Spontaneous and induced abortion among resident physicians. JAMA, 1991; 265:2821–2825.

69.

Savitz

, Hertz-Picciotto

, Poole

, Olshan

. Epidemiologic measures of the course and outcome of pregnancy. Epidemiol Rev, 2002; 24:91–101.

70.

Hertz-Picciotto

, Pastore

, Beaumont

. Timing and patterns of exposures during pregnancy and their implications for study methods. Epidemiology, 1996; 143:597–607.

71.

Hinckley

, Bachand

, Reif

. Late pregnancy exposures to disinfectant by-products and growth-related birth outcomes. Environ Health Perspect, 2005; 113:1808–1813.

72.

Shaw

, Gold

. Methodological considerations in the study of parental occupational exposures and congenital malformations in the offspring. Scand J Work Environ Health, 1988; 14:344–355.