Indoor and outdoor air pollution and couple fecundability: a systematic review

Abstract

BACKGROUND

Air pollution is both a sensory blight and a threat to human health. Inhaled environmental pollutants can be naturally occurring or human-made, and include traffic-related air pollution (TRAP), ozone, particulate matter (PM) and volatile organic compounds, among other substances, including those from secondhand smoking. Studies of air pollution on reproductive and endocrine systems have reported associations of TRAP, secondhand smoke (SHS), organic solvents and biomass fueled-cooking with adverse birth outcomes. While some evidence suggests that air pollution contributes to infertility, the extant literature is mixed, and varying effects of pollutants have been reported.

OBJECTIVE AND RATIONALE

Although some reviews have studied the association between common outdoor air pollutants and time to pregnancy (TTP), there are no comprehensive reviews that also include exposure to indoor inhaled pollutants, such as airborne occupational toxicants and SHS. The current systematic review summarizes the strength of evidence for associations of outdoor air pollution, SHS and indoor inhaled air pollution with couple fecundability and identifies gaps and limitations in the literature to inform policy decisions and future research.

SEARCH METHODS

We performed an electronic search of six databases for original research articles in English published since 1990 on TTP or fecundability and a number of chemicals in the context of air pollution, inhalation and aerosolization. Standardized forms for screening, data extraction and study quality were developed using DistillerSR software and completed in duplicate. We used the Newcastle-Ottawa Scale to assess risk of bias and devised additional quality metrics based on specific methodological features of both air pollution and fecundability studies.

OUTCOMES

The search returned 5200 articles, 4994 of which were excluded at the level of title and abstract screening. After full-text screening, 35 papers remained for data extraction and synthesis. An additional 3 papers were identified independently that fit criteria, and 5 papers involving multiple routes of exposure were removed, yielding 33 articles from 28 studies for analysis. There were 8 papers that examined outdoor air quality, while 6 papers examined SHS exposure and 19 papers examined indoor air quality. The results indicated an association between outdoor air pollution and reduced fecundability, including TRAP and specifically nitrogen oxides and PM with a diameter of ≤2.5 µm, as well as exposure to SHS and formaldehyde. However, exposure windows differed greatly between studies as did the method of exposure assessment. There was little evidence that exposure to volatile solvents is associated with reduced fecundability.

WIDER IMPLICATIONS

The evidence suggests that exposure to outdoor air pollutants, SHS and some occupational inhaled pollutants may reduce fecundability. Future studies of SHS should use indoor air monitors and biomarkers to improve exposure assessment. Air monitors that capture real-time exposure can provide valuable insight about the role of indoor air pollution and are helpful in assessing the short-term acute effects of pollutants on TTP.

Graphical Abstract

Graphical Abstract.

Exposure to outdoor and indoor air pollutants may increase time to pregnancy.

Introduction

Air pollution is both a sensory blight and a threat to human health. Inhaled environmental pollutants can be naturally occurring or human-made (such as secondhand smoke (SHS)), and include gaseous pollutants, such as ozone (O₃), sulfur dioxide (SO₂) and nitrogen oxides (NO_x), and non-gaseous components, such as particulate matter (PM), volatile organic compounds (VOCs) and polycyclic aromatic hydrocarbons (PAHs). Acute and chronic health effects have been documented, impacting multiple organs and systems, including the respiratory and cardiovascular systems (Stieb et al., 2009; Hamanaka and Mutlu, 2018), as well as reducing overall life expectancy (Dockery et al., 1993; Wu et al., 2020). The impact of air pollution on reproductive and endocrine systems has been studied to a lesser degree, although studies have reported associations of traffic-related air pollution (TRAP), e.g. exposure to PM and NO_x, with birth outcomes, including pregnancy loss (Kioumourtzoglou et al., 2019), preterm birth (Chang et al., 2012), low birth weight (Pedersen et al., 2013) and small- and large for gestational age infants, as well as with incident diabetes, obesity and preeclampsia (Krämer et al., 2010; Jerrett et al., 2014; Chen et al., 2021; Daniel et al., 2021; Li et al., 2021a). Indoor domestic and occupational air pollutants including SHS, biomass fuel smoke and organic solvents are also linked to adverse birth outcomes (Ha et al., 2002; Wahabi et al., 2013). Potential mechanisms for these adverse effects include direct disruption of the endocrine system (Darbre, 2018), the induction of oxidative stress and inflammation (Zhang et al., 2016), the formation of DNA adducts (Binková et al., 1999) and altered DNA methylation (Yauk et al., 2008). While some evidence suggests that air pollution contributes to infertility, the extant literature is inconsistent (Frutos et al., 2015; Carré et al., 2017). As a result, it is difficult for health care providers, researchers and policymakers to distill the information for effective decision-making.

Impaired fecundity, the biological ability to reproduce, commonly manifests as delayed conception or as diagnosed infertility, defined as the inability to conceive a child within 12 months of unprotected intercourse (Zegers-Hochschild et al., 2009), which affects an estimated 45 to 53 million couples worldwide (Mascarenhas et al., 2012). In addition to being a marker of reproductive fitness, fecundity has been linked to overall health and future survival in both men and women (Ahrenfeldt et al., 2021). A couple’s ability to reproduce is operationalized as time to pregnancy (TTP), also known as fecundability, and is historically quantified as the per-cycle probability of conception (Baird et al., 1986). Although some reviews have studied the association of exposure to common outdoor air pollutants such as PM with a diameter of ≤2.5 µm (PM_2.5), nitrogen dioxide (NO₂) and SO₂ with TTP (Checa Vizcaíno et al., 2016; Conforti et al., 2018), there is currently no comprehensive systematic review that also examines exposure to indoor inhaled pollutants, including airborne occupational toxicants and SHS. To fill this gap, we undertook a systematic review of the following air pollutants in relation to TTP.

Outdoor air pollutants

Outdoor air pollution, originating from natural sources or anthropogenic activities such as fossil fuel combustion or traffic, is among the major causes of morbidity and mortality in adults (Klepac et al., 2018; Rajagopalan et al., 2018; Fong et al., 2019). Health effects of outdoor air pollutants, including cardiovascular disease, cancer, asthma and poor reproductive outcomes, contribute significantly to the global burden of disease worldwide (Cohen et al., 2017). Vehicle emissions are the major source of pollutants such as NO₂ and PM_2.5 in urban areas. TRAP, a mixture of vehicle exhaust, evaporative emissions and road dust, also increases surface O₃ concentrations, particularly in summertime. Evidence in humans is accumulating to suggest that TRAP is associated with reduced fecundability and longer TTP by influencing both male and female reproductive health. In men, TRAP has been associated with impaired spermatogenesis and poor semen quality, and in women, TRAP has been associated with increased risk of endometriosis and polycystic ovarian syndrome (PCOS) (Mahalingaiah et al., 2014; Nobles et al., 2018a; Lin et al., 2019).

Emissions from a combustion of oil, diesel fuel and/or wood in industrial settings and wildfires produce PM_2.5 and PM with a diameter of ≤10 µm (PM₁₀). PM₁₀ also includes dust and particles released from landfills, agricultural processes, wildfires and waste burning (Harrison et al., 2004; Nguyen et al., 2017; Ali et al., 2019; Harrison, 2020). PM is composed of a mixture of many chemicals that reflects the emission source, e.g. PAHs and heavy metals. PM and other outdoor air pollutants including PAHs, NO₂, O₃ and SO₂ induce systematic inflammation and oxidative stress. Some of these pollutants can also disrupt the endocrine system. (Rudel and Perovich, 2009; Lanki et al., 2015; Viehmann et al., 2015; Li et al., 2016). In women, an inverse relationship has been observed between exposure to PM_2.5 and ovarian reserve measured using transvaginal ultrasonography (Gaskins et al., 2019), as well as associations of exposure to PM₁₀, PM_2.5, carbon monoxide (CO), NO₂ and SO₂ with spontaneous abortion (Moridi et al., 2014; Gaskins et al., 2019). In men, PM exposure has been inversely associated with sperm concentration and count (Wu et al., 2017), and exposure to PM₁₀, PM_2.5, CO and NO_x has been associated with lower testosterone levels (Radwan et al., 2016).

The endocrine-disrupting effects of PAHs have also been well-documented (Zhang et al., 2016): PAHs have been shown to interact with estrogen and androgen receptors, and to activate aromatic hydrocarbon receptor, resulting in altered steroid function and anti-estrogenic activity (Vondráček et al., 2018). In male rats, inutero exposure to PAHs results in a lifelong decrease in sperm production (Takeda et al., 2004), and inhalation exposure has been associated with both decreased sperm production and lowered plasma testosterone (Inyang et al., 2003). In men, serum concentrations of PAHs have been inversely linked to both sperm motility (Song et al., 2013) and overall semen quality (Chen et al., 2021), and serum PAHs have been found to be positively correlated with idiopathic male infertility (Xia et al., 2009). In women, PAHs have been detected in placental tissue (Madhavan and Naidu, 1995) and are associated with first-trimester fetal loss (Wu et al., 2010) and preterm birth (Padula et al., 2014).

Indoor air pollutants

Indoor air pollutants originate from cigarette smoke, the burning of biomass fuel for cooking and heat, consumer products (e.g. phthalates contained in air fresheners) and occupational use of chemicals in laboratories, factories and other workplaces. While non-occupational inhaled pollutants like biomass fuel smoke have been linked to adverse birth outcomes (Boy et al., 2002; Mishra et al., 2004), we focus here on SHS and occupational inhaled pollutants as there is no peer-reviewed literature on fecundability and biomass fuel smoke, despite an association with adverse birth outcomes (Boy et al., 2002; Mishra et al., 2004). We also did not include chemical pollutants from consumer products as these were covered in our prior review (Hipwell et al., 2019) and often have multiple exposure routes.

Secondhand smoke

While the adverse effect of cigarette smoking on fecundity is commonly accepted, the effect of SHS is less well-documented. Both the exhaled portion of mainstream smoke and sidestream smoke (from the burning of tobacco products) make up environmental tobacco smoke (ETS), the source of SHS exposure (National Research Council (US) Committee on Passive Smoking, 1986). Approximately 300–400 of the 4000-plus compounds identified in tobacco smoke have been measured in ETS, including nicotine, CO, PAHs such as benzo[a]pyrene (BaP), nitric oxide (which oxidizes to NO₂), and various heavy metals, including lead and cadmium. Some of these chemicals, including BaP (Neal et al., 2008) and cotinine (Sterzik et al., 1996), the major metabolite of nicotine, have been detected in follicular fluid and, in the case of cotinine, in granulosa cells of women undergoing IVF who were exposed to SHS (Zenzes et al., 1997), indicating that they may directly impact oocyte development. In an experimental study, ovaries retrieved from mice exposed to ETS had increased levels of primordial follicle depletion, antral follicle oocyte apoptosis and oxidative stress, resulting in fewer follicles available for ovulation; among oocytes that survived, elevated levels of lipid peroxidation and mitochondrial reactive oxygen species were observed, signaling reduced fertilization potential (Sobinoff et al., 2013). Several other rodent studies have reported reduced number of pregnancies and offspring per pregnancy in ETS-exposed animals (Florek and Marszalek, 1999; Florek et al., 2002; Khan et al., 2008). In addition to increased oxidative stress, endocrine disruption may play a role. Healthy women exposed to SHS have been shown to have lower levels of estrogens, progesterone and aldosterone (Soldin et al., 2011), and SHS-exposed women with PCOS have been found to have a higher total testosterone and free androgen index, and lower sex hormone-binding globulin compared with unexposed women (Li et al., 2018). In the context of IVF, SHS exposure has been associated with implantation failure and spontaneous abortion (Neal et al., 2005; Meeker et al., 2007; Benedict et al., 2011), although the biological mechanism underlying these associations may in part be explained by male partners’ smoking, which is independently associated with poor semen quality (Sharma et al., 2016) and IVF outcomes (Joesbury et al., 1998; Zitzmann et al., 2003).

Indoor inhaled occupational pollutants

Common occupational inhaled exposures include formaldehyde, solvents and other VOCs. These chemicals are employed in a variety of workplaces including laboratories, manufacturing and processing plants and healthcare settings.

VOCs are solid or liquid compounds containing carbon that emit vapors at room temperature including gasoline, benzene, formaldehyde, toluene, xylene, styrene and perchloroethylene, as well as others (Anand et al., 2005). Overall exposure to solvents has been associated with reduced fertility or infertility in women, while more specific exposures have varying outcomes (Lindbohm, 1999). Occupational exposure to benzene, toluene and xylene has been associated with decreased sperm vitality and activity (Xiao et al., 2001; Webb et al., 2014). In addition, there is evidence that benzene exposure can lead to chromosomal abnormalities in sperm cells and that toluene metabolites may directly target reproductive organs, resulting in DNA damage in the testes (Xiao et al., 2001; Webb et al., 2014). In women, toluene has been linked to menstrual disorders and reduced fertility (Correa et al., 1996; Lindbohm, 1999).

Formaldehyde is a flammable, colorless vapor used in the biomedical field and in the manufacturing of plastics and other materials. Humans produce very small amounts of the chemical during metabolic processes, but it can be toxic upon inhalation (Suh et al., 2000). In studies of male rats, inhalation of formaldehyde led to damage of testicular tissue and seminiferous tubules, as well as decreased levels of testosterone (Duong et al., 2011). Formaldehyde exposure in women has also been associated with higher rates of endometriosis, irregular menstruation and miscarriage (Taskinen et al., 1999; Duong et al., 2011). Similarly, higher rates of miscarriage have been reported in female partners of men exposed to formaldehyde in the workplace (Wang et al., 2012).

Biological mechanisms and windows of exposure

The biological mechanisms through which air pollution may impact fecundability point to relevant windows of exposure for each pollutant. Pollutants such as PM and PAHs induce inflammatory processes and hormonal disruption, leading to low sperm count or poor semen quality that may explain the acute influence of air pollution exposure on couples’ fecundability in a single cycle. In contrast, the impact of pollutants on ovarian reserve may explain the link between chronic exposure to air pollutants and reduced fecundability. Long-term exposure to air pollution (exposure beyond one reproductive cycle) also increases the risk of chronic conditions such as endometriosis and PCOS, which are in turn risk factors for reduced fecundability. That said, it can be difficult to distinguish between acute and chronic exposures as acute exposure to air pollution is often highly correlated with chronic exposure. Keeping these limitations in mind, the current systematic review summarizes the strength of evidence for associations of outdoor air pollution, SHS and indoor occupational inhaled pollutants with couple fecundability, and identifies gaps and limitations in the literature to inform policy decisions and future research.

Methods

This review was conducted according to the Preferred Reporting Items for Systematic review and Meta-Analysis Protocol: (PRISMA) (Moher et al., 2009, 2015; Shamseer et al., 2015). The protocol is registered (CRD42020208340) on PROSPERO (www.crd.york.ac.uk/prospero).

Search strategy

We performed an electronic search of literature on TTP or fecundability and a number of chemicals and chemical classes (see Table I) in the context of air pollution, inhalation and aerosolization. The following six databases were searched: MEDLINE via PubMed, EMBASE.com, Global Health on the Ovid platform, Scopus at Elsevier, Cochrane Central (Wiley) for trials and GreenFile via EBSCOhost. A professional health sciences librarian (M.K.-F.) developed and executed the bibliographic searches using controlled vocabulary terms, Boolean operators and text words to define the search strategy (see Supplementary File S1).

Table I.

Summary of chemicals and abbreviations.

Carbon monoxide (CO)

Formaldehyde

Nitrogen dioxide (NO2)

Nitrogen oxides (NOx)

Ozone (O3)

Particulate matter (PM)

Polycyclic aromatic hydrocarbons (PAHs)

Secondhand smoke (SHS)

Solvents

Styrene

Sulfur dioxide (SO2)

Traffic-related air pollution (TRAP)

Volatile organic compounds (VOCs)

Searches were limited to articles published in the English language through to 11 February 2021 and beginning in 1990, as that year marked the passage of the Clean Air Act Amendment. The Amendment was ‘designed to curb three major threats to the nation’s environment and to the health of millions of Americans: acid rain, urban air pollution, and toxic air emissions’ (US Environmental Protection Agency, 1990). To ensure that all relevant papers were captured by the search, the systematic reviews that were screened and deemed relevant were reviewed to verify that no papers within the scope had escaped.

Screening and eligibility

Article screening was conducted with DistillerSR software (Evidence Partners, Ontario, Canada) using standardized forms for title and abstract screening and for full-text review. Each level of review (see Fig. 1) was completed in duplicate by the authors, and any conflicts were resolved through discussion.

Figure 1.

Flow chart of systematic review inclusion and exclusion process.

Inclusion and exclusion criteria

At the title- and abstract screening level, we included all human studies that related to air pollution or chemical exposures and to TTP or fecundability. We screened out editorials, opinion pieces or introductions to special sections. We also screened out articles describing animal, plant or invitro studies, and those that described behavioral exposures (e.g. personal care product use, smoking). SHS (a.k.a., ETS) was included as an exposure. We obtained full-text reports for all titles that did not meet these exclusion criteria or about which there was uncertainty. At the next level of screening, we included only original empirical research papers that considered TTP or fecundability as an outcome and examined exposure to outdoor and indoor air pollutants. Prospective and retrospective cohort, cross-sectional and case–control studies were included. Systematic reviews were quarantined and then reviewed to guarantee that no relevant articles had been missed by the search. Included studies measured exposures directly from biospecimens and air monitors and/or indirectly via self-report, occupation or residential/workplace location. We excluded studies of populations that were solely composed of participants undergoing IVF procedures. We also excluded studies that examined chemicals with multiples routes of exposure where the primary source may not have been inhalation (e.g. exposure to agrochemicals may be dermal or via ingestion or via inhalation), as well as studies where exposure was in utero (e.g. inutero exposure to SHS), as neither type of study would aid in answering our main question: Do inhaled air pollutants affect fecundability? Further, we excluded disasters, both natural and human-made (e.g. World Trade Center attacks, wildfires), because they cannot be regulated, and a primary goal of this review was to inform future policy. Of the studies that examined SHS, several examined the effects of SHS during multiple exposure windows over the life course. We included papers reporting effects of SHS exposure during childhood and adulthood, but excluded inutero and multi-generational exposures, as exposure was not via the inhalation route.

Outcome criteria

We included studies that operationalized fecundability as TTP measured continuously in months or cycles, or dichotomously as >6 or >12 months of unprotected intercourse without a conception. Studies with an unspecified outcome of infertility, either self-reported or medically diagnosed, were excluded.

Data extraction

We created standardized forms for data extraction in DistillerSR, which were completed in duplicate. Where available, prior articles that described study methods in greater detail were reviewed. Discrepancies were recorded and resolved through discussion. The review authors were not blind to the journal titles, study authors or institutions.

Risk of bias assessment

We used a modified version of the Newcastle-Ottawa Scale (NOS) risk-of-bias tool for observational studies (Stang, 2010; Wells et al., 2011; Zeng et al., 2015) to evaluate domains of participant selection, comparability of groups/confounder control and assessment of exposure (Table II). Study characteristics considered to reduce the risk of bias were awarded a point. We followed recommendations to convert the NOS score to Agency for Healthcare Research and Quality (AHRQ) standards of good, fair and poor (Singh et al., 2015). Good-quality studies were identified as those awarded 3–4 points in the selection/exposure domain AND 1–2 points in the comparability domain AND 2–3 points in the outcome domain. Fair studies were indicated by 2 points for selection/exposure AND 1–2 points for comparability AND 1–2 points for outcome, whereas poor-quality studies scored 0–1 points for selection/exposure OR 0 points for comparability OR 0 points for outcome. In addition, we devised five additional quality metrics not captured by the NOS, specifically for this review to identify methodological features of inhaled air pollutant and fecundability studies that distinguish those of highest quality: (i) inhaled air pollution exposure was measured using either personal air monitors or a combination of environmental air monitors and geolocation; (ii) multiple measures were taken either over time or across space; (iii) inhaled air pollution exposure was measured within the TTP window (i.e. during the time period in which participants were trying to conceive); (iv) participants were actively trying to conceive; and (v) daily pregnancy tests were administered to accurately assess timing of conception, or daily urine were collected and retrospectively tested, or urine was tested within a window of biological relevance (Table II).

Table II.

Assessment of risk of bias and study quality.

Newcastle-Ottawa Scale domains	Criteria for higher quality
Representativeness of the exposed group	Representative/population-based sample; selected group, addressed in analyses by inverse probability weighting (IPW) or other appropriate method*
Selection of the non-exposed cohort	Drawn from the same group as the exposed cohort over the same time frame*
Adequacy of exposure assessment	Individual level, personal air monitor; individual level, geospatial/residential; environment: dust/air sampling, city monitors
Exposure measured in both men and women	Women and men*
Comparability/control for confounders	Controls for age*
	Controls for maternal smoking*
	Controls for at least one additional factor*
Assessment of outcome	Independent, biological measure of pregnancy; medical record confirmation; recruited when already pregnant (for retrospective design)*
Study design	Prospective, longitudinal study*
Completeness of outcome data	No missing outcome data or complete follow-up, all subjects accounted for; missing outcome data unlikely to be related to TTP or subjects lost to follow up unlikely to introduce bias—number lost ≤20% or description of those lost suggested no different from those followed*
Additional quality metrics
Was inhaled air pollution exposure measured using either personal air monitors or a combination of environmental/city air monitors and geocoding?	Yes
Were multiple measures taken either over time or over space?	Yes
Was inhaled air pollution exposure measured within the TTP window?	Yes
Were participants actively trying to conceive?	Yes
Were daily pregnancy tests administered during TTP, or were daily urine samples collected and tested retrospectively, or was urine tested within a window of biological relevance?	Yes

^*Study characteristics used to convert Newcastle-Ottawa Scale (NOS) scores to Agency for Healthcare Research and Quality (AHRQ) standards of good, fair and poor.

TTP, time to pregnancy.

Air quality metrics

Studies of air quality and fecundability measured exposure at specified times relative to conception (e.g. days or weeks prior) or during hypothesized critical periods in reproductive development (e.g. childhood). They also measured exposure across different durations of time and via different measures, e.g. self-report, biomarkers or personal or environmental air monitors. We awarded a point to studies that measured exposure via air monitoring because self-reported exposure tends to be imprecise and susceptible to recall bias, and biomarkers do not distinguish between inhalation and other routes of exposure, instead (imperfectly) capturing overall body burden regardless of exposure route. This point was awarded to studies that measured exposure to ambient air pollution using personal and/or environmental air monitors, environmental dust/air sampling and geographical information system (GIS)-based methods that relied on highly resolved temporal and spatial air-pollutant modeling. We also awarded an extra point if multiple measures were taken either over time or over space, whether via monitors or via biospecimens. Note that if exposure to air pollution was measured by self-report or occupation, no point was awarded on this quality metric.

Fecundability metrics

We chose to award the NOS study design quality point to prospective studies or those that measured retrospective TTP during pregnancy, when recall may still be considered reliable, and an additional quality point to studies of couples actively trying to conceive. We expand on the intrinsic tradeoffs between prospective and retrospective studies of TTP in the discussion section below. We elected not to penalize retrospective studies, but we acknowledge there may be issues of selection bias, which could be relevant for quality. Given the nascent field of TTP research, for now we chose to cast a broader net. We awarded an additional quality point to studies that measured exposure during the time when couples were trying to conceive, as contrasted to exposure recalled post-conception or measured via assays of biosamples collected during the prenatal period. Finally, as daily pregnancy tests are considered the gold standard for studies of fecundability and have been included in quality metrics utilized in past reviews of environmental exposures and TTP (Hipwell et al., 2019; Kahn et al., 2021), we awarded an additional point for studies that employed this method or for those that employed daily urine collection with the potential for retrospective testing or, at a minimum, conducted urine testing within a window of biological relevance.

Results

Our initial search yielded 6122 results, 922 of which were excluded as duplicates (see Fig. 1). Of the 5200 articles that remained, 4994 articles were excluded at the level of title and abstract screening. After a full-text screening and confirmation of relevant chemicals and exposure routes, 35 papers remained for data extraction and synthesis. An additional three papers were identified independently to fit the criteria, and five papers were removed, as the chemicals analyzed involved multiple routes of exposure and had been included in previous reviews of TTP we conducted on persistent and non-persistent chemicals (Hipwell et al., 2019; Kahn et al., 2021). Of the resulting 33 papers from 28 studies, 8 papers examined outdoor air quality and TTP, 6 papers examined SHS and TTP and 19 papers examined indoor air quality related to occupational exposure and TTP (Tables III–V).

Table III.

Characteristics, studies of outdoor air quality and time to pregnancy (TTP).

First author, year	Study acronym	Country	Period of exposure	Study design	No. couples	No. women	No. men	Mode of exposure measurement
								Air monitor (environmental)	Proxy of exposure	Self-report
Dejmek, 2000		Czech Republic	1994–1998	Retrospective cohort, pregnancy-based	0	857	0	SO2 at central monitor
Hariparsad, 2019		South Africa	2015	Retrospective	0	305	0			Occupation: market/street trader (cooking and dust exposure vs unexposed)
Joffe, 2008		Wales	1998–2000	Retrospective	0	499	0		Residential proximity to landfill site
Li, 2021a,b	CFSMW	China	2010–2011	Retrospective cohort		10 211		PM2.5 at residential address
Mendola, 2017	LIFE	USA	2005–2009	Prospective cohort	500	500	500		Distance of residential address to nearest major roads
Nobles, et al., 2018b	LIFE	USA	2005–2009	Prospective cohort	500	500	500	PM2.5, PM10, SO2, CO, NO2, NOx, O3 at residential address
Slama, 2013		Czech Republic	1993–1997	Retrospective cohort, pregnancy-based	0	1916	0	PM2.5, SO2, NO2, O3, c-PAH at central monitor
Wesselink, 2020	PRESTO	USA and Canada	2013–2018	Prospective cohort		8790			Distance of residential address to nearest major roads

SO₂, sulfur dioxide; PM_2.5, particulate matter with a diameter of ≤2.5 µm; PM₁₀, particulate matter with a diameter of ≤10 µm; CO, carbon monoxide; NO₂, nitrous dioxide; NO_x, nitrous oxides; O₃, ozone; c-PAH, carcinogenic polycyclic aromatic hydrocarbons.

Table V.

Characteristics, studies of inhaled occupational pollutants and time to pregnancy (TTP).

First author, year	Study acronym	Country	Period of exposure	Study design	No. couples	No. women	No. men	Mode of exposure measurement
								Air monitor	Biomonitoring in urine/serum	Self-report
Ahlborg, 1996		Sweden	1940–1983	Retrospective	0	972	0			Occupation: midwives, number of deliveries with NOx per month
Dahl, 1999		Norway	1991	Retrospective	0	1008 women (1408 pregnancies)	0			Occupation: dental surgeons (VOC exposed) vs high school teachers
Figà-Talamanca, 2000		Italy	Air and blood samples from 1993 to 1998	Retrospective	0	0	167	Environmental air concentrations of nickel, chromium and lead	Mean blood concentration of nickel, chromium, and lead of machine operators	Occupation: mint workers Administrative vs technical vs stamper (solvents exposed) vs founder (metals exposed), self-reported exposure to metals and solvents
Hooiveld, 2006		The Netherlands	2001	Cross-sectional	0	0	700			Painters (exposed to organic solvents) vs carpenters (unexposed to solvents)
Kolstad, 1999		Denmark, Italy and The Netherlands	1960–1995	Retrospective	0	0	541	Measurements of workroom air	Urine samples of styrene	Occupation: reinforced plastics workers
Kolstad, 2000		Denmark, Italy and The Netherlands	1970–1997	Retrospective	0	0	602	Dust and air sampling for styrene	Serum samples of styrene	Occupation: Reinforced plastics workers
Luderer, 2004		USA	Not provided	Retrospective	0	0	97			Occupation: millwrights and painters (as exposed to solvents) to carpenters (unexposed)
Plenge-Bönig, 1999		Germany		Retrospective	0	150	90			Printing industry workers (toluene exposure)
Rowland, 1992		USA		Retrospective	0	418	0			Nitrous oxide exposure among dental assistants, hours/week reported working in a room with unscavenged NOx
Sallmén, 1995		Finland		Retrospective	0	197	0		Solvents in urine and blood	Frequency of use (days/week and type of work reported)
Sallmén, 1998		Finland	1965–1983		316	0	0		Solvents in urine	Solvents
Sallmén, 2006	AHS	USA (Iowa and North Carolina)	1993–1997	Cross-sectional	2112	2112	2112			Solvent use at home or work at least once/month
Sallmén, 2008		Portugal	1986–1997	Retrospective	0	406	0	Personal and static for solvents		Occupation: shoe manufacturing workers vs workers in stores of food units and storehouses
Spinelli, 1997		Italy	1993		622	622	622			Solvents and industrial occupation
Taskinen, 1999		Finland	1985–1995	Retrospective	0	602	0			Occupation: wood processing industry (solvents, formaldehyde and wood dust)
Wang, 2012		China	2007–2009	Retrospective	0	0	302	Formaldehyde		Occupational exposure via wood processing industry
Wennborg, 2001		Sweden	1990–1994	Retrospective	0	560 women (735 pregnancies)	0			Occupation: laboratory personnel vs personnel in non-laboratory departments
Wulff, 1999		Sweden	1972–1981 and 1982–1992	Retrospective	803	703				Distance from smelter, occupation (smelter workers, men and women), occupation within smelter
Zhu, 2005	DNBC	Denmark	1997–2003	Retrospective, pregnancy-based	0	829	0			Occupation: laboratory technicians vs teachers

VOC, volatile organic compound; NO_x, nitrous oxides.

Description of studies

Studies with both preconception (prospective) and post-conception (retrospective) recruitment were included in our review. Articles from three preconception cohorts met all inclusion criteria. The Pregnancy Study Online (PRESTO), which yielded two articles, collected data on both outdoor air quality and self-reported SHS via electronic survey (Wesselink et al., 2019, 2020). The Longitudinal Investigation of Fertility and the Environment (LIFE) study, which also yielded two articles, followed couples actively trying to conceive for 1 year in Michigan and Texas and utilized residential addresses and Environmental Protection Agency (EPA) air monitoring to track air pollution exposure during the preconception period (Mendola et al., 2017; Nobles et al., 2018b). The ‘Snart Gravid’ (Pregnancy Planning) study was conducted among pregnancy planners in Denmark; women were asked about exposure to SHS and then followed for up to 12 cycles using bimonthly online questionnaires (Radin et al., 2014).

Within the retrospective studies, we included both pregnancy-based cohorts as well as studies that recruited participants after live births had occurred. The Avon Longitudinal Study of Pregnancy and Childhood (ALSPAC) enrolled women during pregnancy, at which time data on TTP were collected (Hull et al., 2000). Dejmek et al. (2000) and Slama et al. (2013) reported on a study that obtained TTP information at delivery from all full-term singleton births born between April 1994 and March 1998 in one district in the Czech Republic. Li et al. (2021b) presented data from a study of 18,571 couples registered with the local National Health and Family Planning Commission (NHFPC) of the People’s Republic of China; couples were asked about TTP during the previous year. Joffe et al. (2008) reported on a study that identified pregnant women through local midwifery records, then enlisted health visitors and midwives to obtain information on TTP from those whose pregnancies reached at least 24 weeks of gestation. Zhu et al. (2005) used TTP recalled mid-pregnancy by participants in the Danish National Birth Cohort (DNBC).

All of the studies of occupational inhaled pollutants obtained data on TTP retrospectively. Many used a combination of self-report, occupation, air monitoring and/or biological samples to measure exposure.

A note on effect sizes

Studies reported either adjusted fecundability odds ratios (aFORs) or adjusted odds ratios (aORs) with 95% CIs. In general, fecundability odds ratios (FORs) model TTP as a discrete, time-to-event outcome using months or menstrual cycles. Analogous to hazard ratios, FORs represent the probability of conceiving in a specified time period (month or cycle), conditional on not having conceived in the prior time period, per unit of exposure. Because different units of change in exposure were often used, it was difficult to compare FORs across studies. While some studies reported FORs per unit of exposure (modeled continuously, sometimes log-transformed), others reported FORs per quantiles of exposure or per binary exposure (exposed versus unexposed). Quantiles and binary (high/low or yes/no) exposure categories are sample-determined, thus they are not comparable across studies with different exposure ranges.

In general, though not always, diminished fecundability, or longer TTP, is indicated by FOR <1 (reduced probability of conception in a given month/cycle) or by OR for infertility >1. Where odds ratios are employed, the outcome has either been dichotomized and compares TTP > 12 versus ≤12 months or TTP > 6 versus ≤6 months.

Associations between outdoor air pollution and fecundability

Eight studies from six cohorts examined outdoor air quality and fecundability (Table III). Research was conducted on populations in China (Li et al., 2021b), the Czech Republic (Dejmek et al., 2000; Slama et al., 2013), South Africa (Hariparsad and Naidoo, 2019), the USA (Mendola et al., 2017; Nobles et al., 2018b), the USA and Canada (Wesselink et al., 2020) and Wales (Joffe et al., 2008) (Table III). The two studies from the Czech Republic examined data from the same cohort, as did the two studies from the USA.

GIS-based methods using data from environmental air monitors or operationalized as proximity to major roads (Mendola et al., 2017; Wesselink et al., 2020) or to a landfill site (Joffe et al., 2008) were used to estimate exposure to ambient air pollution. Some analyses such as those conducted by Nobles et al. (2018b) and Li et al. (2021b) relied on GIS-based methods that generated high-resolution spatial and temporal air pollution data. Dejmek et al. (2000) used central monitors (low spatial resolution), and Joffe et al. (2008), Mendola et al. (2017) and Wesselink et al. (2020) examined distance from major roads as a proxy for exposure. Hariparsad and Naidoo (2019) compared exposure in participants whose stalls in a South African market were located close to the road or who sold food cooked over a fire (using biomass fuel) with exposure in participants who worked elsewhere in the market. Some studies examined multiple pollutants (Nobles et al., 2018b; Slama et al., 2013) and others focused on one, for example, only PM_2.5 (Li et al., 2021b) or SO₂ (Dejmek, 2000). While most of the studies were conducted retrospectively (Dejmek et al., 2000; Joffe et al., 2008; Slama et al., 2013; Hariparsad and Naidoo, 2019; Li et al., 2021b), much of the exposure data were collected prospectively (Dejmek et al., 2000; Slama et al., 2013; Li et al., 2021b), reducing concerns about recall bias due to self-reported residential address. As a negative control and to enhance the robustness of findings, Slama et al. (2013) compared associations of exposure with TTP in the month after pregnancy to associations of exposure with TTP in the months during the preconception period.

Some analyses examined specific windows of exposure. For example, Nobles et al. (2018b) considered a time-varying exposure averaged over the past month prior to conception, as well as acute exposure 5–10 days following ovulation. Li et al. (2021b) examined exposure 1, 3 and 5 years prior to conception, arguing that the etiologically relevant period of exposure for TTP is unknown. Dejmek et al. (2000) studied monthly average exposure over the 4 months prior to conception. Other studies that examined distance from major roads (e.g. Wesselink et al., 2020) did not consider timing of exposure.

Evidence for longer TTP

Li et al. (2021b) found reduced FORs for every 10 μg/m³ increase of PM_2.5 averaging exposures over 1, 3 and 5 years during the preconception period. Slama et al. (2013) reported reduced FORs for exposure to PM_2.5 using a 2-month lag model and for exposure to NO₂ using a one-month lag model during the preconception period. Nobles et al. (2018b) reported reduced fecundability for exposure to NO_x 8 days after ovulation (aFOR: 0.84, 95% CI 0.71–0.99), and for exposure to O₃ 5 days (aFOR: 0.87, 95% CI 0.76–0.99) and 1 day (aFOR: 0.83, 95% CI 0.72–0.96) before ovulation. Dejmek et al. (2000) found reduced fecundability with increased exposure to SO₂ 2 months before conception when comparing medium and low exposure groups (aOR: 0.57, 95% CI 0.37–0.88) and high and low exposure groups (aOR = 0.49, 0.29–0.81).

Wesselink et al. (2020) found that women from the USA who lived <50 m from the closest major road had lower fecundability compared with those who lived ≥400 m from the closest major road (0.88, 95% CI = 0.80, 0.98) where close proximity to the roadway served as a surrogate for increased air pollution exposure. The FOR for Canadian women was similar in magnitude, but less precise (0.93, 95% CI 0.74, 1.16). Likewise, the length of major roadways within buffers of 50 and 100 m from women’s residential address was associated with lower fecundability in both countries; associations were attenuated within larger buffers. Finally, Hariparsad and Naidoo (2019) reported an aOR for TTP >12 months of 2.6 (95% CI = 1.6–4.3) comparing female traders classified as exposed to biomass fuel use and traffic emissions to traders classified as unexposed.

Evidence for no association with TTP

Nobles et al. (2018b) found no association between PM_2.5, CO or SO₂ exposure and fecundability regardless of the timing of exposure within the menstrual cycle. Slama et al. (2013) reported no association with exposure to carcinogenic PAHs (c-PAHs, such as BaP) or SO₂. Joffe et al. (2008) found no difference in the odds of fecundability between women living <3 km from a landfill site compared with those living >3 km from the landfill, although it should be noted that the theoretical route of exposure was not clearly specified.

Evidence for shorter TTP

Nobles et al. (2018b) reported PM₁₀ exposure 6 days post-ovulation to be associated with increased fecundability (aFOR: 1.25, 95% CI 1.01–1.54).

Summary

There is some evidence of a relationship between exposure to outdoor air pollution and longer TTP (Table VI). Exposures to SO₂, NO₂, NO_x, O₃ and PM_2.5 were associated with decreased fecundability, as was proximity to major roads and traffic emissions. However, it should be noted that SO₂ and PM_2.5 exposure was not consistently related to fecundability, and the studies where associations were found had been given a lower quality rating. CO and c-PAH were not associated with TTP, though they were only examined in one study each. Additionally, the timing and magnitude of exposure varied across studies, and few of the included studies applied methods to identify windows of susceptibility for exposure to ambient air pollutants, e.g. a distributed lagged model.

Table VI.

Effect sizes, outdoor air pollution and time to pregnancy (TTP).

First author, year	Study acronym	Analysis	Effect size, fully adjusted model (95% CI)								Covariates◊	AHRQ/additional quality metrics
			CO	NOx	O3	PM	PAH	SO2	Geo-spatial proximity	Biomass fuel and dust exposure
Dejmek, 2000		Tertiles						aOR: 0.49 for conception (0.29–0.81) (high vs low) for exposure 2 months before conception			Age, parity, season, marital status, respiratory epidemic in previous month, year of pregnancy, temp average, temp maxima, high pollution signal to stay indoors, conception seasonality	Fair/2
Hariparsad, 2019		Occupation: Market trader vs unexposed vendor								aOR: 2.6 (1.6–4.3)	Smoking, chronic comorbidity, years working, use of biomass fuel in home	Poor/0
Joffe, 2008									aFOR: 1.17 (0.81–1.68)		Age, income, education level, paid employment, multivitamin use, birth control use, planned pregnancy, medication use, alcohol, smoking, partner smoking, holiday	Good/2
Li, 2021a,b	CFSMW	For 10 μg/m3 increase in 1, 3 and 5-year average exposures				PM<2.5: aFOR: 0.89 (0.86–0.92) for 1 year aFOR: 0.89 (0.86–0.92) for 3 year aFOR: 0.89 (0.86–0.93) for 5 year					Household income, education level, race/ethnicity, BMI, partner’s age, alcohol consumption, smoking occupation, physical activity, geographic region, ambient temperature	Fair/4
Mendola, 2017	LIFE	Continuous and categorical (<200 m, 200 to <500 m, 500 to <1000 m, 1000 m or more							aFOR: 1.03 (1.01–1.06) for every 200 m further away the couples residence was from a major roadway		Age, parity, study site, and salivary alpha-amylase, a stress marker	Good/4
Nobles, et al., 2018b	LIFE	Change in IQR for each pollutant	Null results for all time windows	8 days post-ovulation: aFOR: 0.84 (0.71–0.99)	aFOR: 0.87 (95% CI: 0.76–0.99) 5 days before ovulation; aFOR: 0.83, (0.72–0.96) 1 day before ovulation	PM2.5: Null results PM10: aFOR: 1.25 (1.01–1.54) 6 days post-ovulation		Null results for all time windows			Age, household income, education level, race/ethnicity, BMI, parity, gravidity, research site, serum cotinine	Good/5
Slama, 2013		Lag variables for months pre and post outcome for an increase of 10 units (ug/m3)		Lag 1: 0.71 (0.57–0.87); Months 1–2: 0.72 (0.53–0.97)	Null results for all pre-outcome time lags	Lag 2: 0.86 (0.77–0.97); Months 1–2: 0.78 (0.65–0.94)	c-PAH: Null results for all time lags	Null results for all time lags			Age, education level, BMI, smoking, parity, time off contraception, marital status, respiratory epidemic in previous month	Poor/3
Wesselink, 2020	PRESTO	Living <50 m vs ≥400 m from closest major road							US: aFOR: 0.88 (0.80, 0.98)		Age, parity income, race, education, lifestyle, BMI, frequency of intercourse, sugar-sweetened beverage intake, doing something to improve chances of conception, census tract median household income, census tract education level, census tract racial ethnic composition	Fair/3

aOR, adjusted odds ratio; aFOR, adjusted fecundability odds ratio; SO₂, sulfur dioxide; PM_2.5, particulate matter with a diameter of ≤2.5 µm; PM₁₀, particulate matter with a diameter of ≤10 µm; CO, carbon monoxide; NO₂, nitrous dioxide; NO_x, nitrous oxides; O₃, ozone; c-PAH, carcinogenic polycyclic aromatic hydrocarbons; AHRQ, Agency for Healthcare Research and Quality.

^◊Potential mediators.

Associations between secondhand smoke and fecundability

Six studies examined SHS and fecundability (Table IV). Data were from a limited range of countries that included Canada (Wesselink et al., 2019), Denmark (Jensen et al., 2006; Radin et al., 2014), the UK (Hull et al., 2000) and the USA (Peppone et al., 2009; Hyland et al., 2016; Wesselink et al., 2019) (Table IV). All exposure measures were self-reported.

Table IV.

Characteristics, studies of secondhand smoke (SHS) and time to pregnancy (TTP).

First author, year	Study acronym	Country	Period of exposure	Study design	No. couples	No. women	No. men	Self-reported measure of SHS
Hull, 2000	ALSPAC	UK	1991–1992	Retrospective, pregnancy-based	12 106	12 106	12 106	SHS exposure among smokers and non-smokers
Hyland, 2016	WHI OS	USA	1993–1998	Retrospective				Lifetime exposure
Jensen, 2006		Denmark	1931–1970	Retrospective twin study	0	1653	1598	One or both parents smoking during childhood
Peppone, 2009		USA	1982–1998	Retrospective		4794		Childhood, adult and joint childhood and adult SHS exposure
Radin, 2014		Denmark	2007–2011	Prospective cohort		2346		Childhood and adult SHS exposure among never smokers
Wesselink, 2019	PRESTO	USA and Canada	2013–2018	Prospective cohort		5473	1411	Adult SHS exposure among men and women

Analyses by Jensen et al. (2006), Hyland et al. (2016) and Peppone et al. (2009), all used retrospectively reported SHS exposure and TTP recalled after live births had occurred. In the article by Hyland et al. (2016), both exposure and outcome were ascertained many years later in a cohort of post-menopausal women.

Evidence for longer TTP

Hull et al. (2000) examined the effect of SHS on the odds of conception among both smoking and non-smoking participants in ALSPAC. Among non-smoking women, those exposed to SHS had an aOR of 1.17 (95% CI: 1.02–1.37) for taking >6 months to conceive and an aOR of 1.14 (95% CI: 0.92–1.42) for taking >12 months to conceive compared with women unexposed to SHS. Among women who smoked, the researchers reported aORs of 1.51 (95% CI: 1.27, 1.78) and 1.57 (95% CI: 1.26–1.96) for taking >6 and >12 months to conceive, respectively, comparing those exposed to SHS with those unexposed. Hyland et al. (2016) examined data from never-smoker post-menopausal women on lifetime exposure to SHS and reported an aOR comparing the highest versus lowest levels of lifetime SHS exposure of 1.18 (95% CI: 1.02–1.35) for infertility (defined as the inability to conceive after 12 months of actively trying). Similarly, Peppone et al. (2009) investigated the effect of SHS exposure during childhood, adulthood and both childhood and adulthood on TTP. The aOR for difficulty becoming pregnant lasting >12 months was 1.57 (95% CI: 1.15–2.15) for those exposed to SHS in childhood but not in adulthood, 1.41 (95% CI: 1.06–1.87) for those exposed to SHS in adulthood only and 1.68 (95% CI: 1.28–2.20) for those exposed to SHS in both childhood and adulthood. The greater odds ratio for those exposed to both adult and childhood SHS may suggest evidence of a dose–response relationship, a fundamental toxicological concept.

Evidence for no association with TTP

Wesselink et al. (2019) examined both men’s and women’s SHS exposure in relation to TTP using data from PRESTO but found no associations. Jensen et al. (2006) employed a twin study designed to examine the effect of childhood SHS exposure on male and female fecundability and reported null results for both male and female twins regardless of whether one or both parents smoked during childhood. Radin et al. (2014) examined SHS exposure during childhood and adulthood among women who never smoked and also reported no association with TTP.

Evidence for shorter TTP

There was no evidence for shorter TTP in any of the studies reviewed.

Summary

There is some evidence for a relationship between SHS exposure and delayed TTP in both smoking and non-smoking populations (Table VII). Lifetime exposure to SHS during childhood and adulthood was found to be associated with delayed TTP, and a more pronounced effect was observed for cumulative exposure occurring in both childhood and adulthood. These results are not consistent across all reviewed sample populations, however, as null results were reported in three (Jensen et al., 2006; Radin et al., 2014; Wesselink et al., 2019) out of the six included studies. Notably, two of the studies with null findings (Radin et al., 2014; Wesselink et al., 2019) were given higher quality ratings than the rest of the studies (although no SHS study earned a ‘Good’ quality rating); these two studies were also the only ones to examine both chronic and acute exposure to SHS.

Table VII.

Effect sizes, secondhand smoke (SHS) and time to pregnancy (TTP).

First author, year	Study acronym	Analysis	Self-reported measure of SHS	Effect size, fully adjusted model (95% CI)	Covariates◊	AHRQ/additional quality metrics
Hull, 2000	ALSPAC	Active, passive, both and number of cigarettes smoked	Adult exposure among smokers and non-smokers	Couples: Among non-smokers: aOR: 1.17 (1.02–1.37) for conception after 6 months; 1.14 (0.92–1.42) for conception after 12 months; Among smokers: 1.51 (1.27–1.78) for conception after 6 months, 1.57 (1.26–1.96) for conception after 12 months	Age, education level, BMI, use of birth control, alcohol, smoking, partner smoking, ethnicity, length of OC use, type of housing, gravidity, parity, years of cohabitation	Fair/0
Hyland, 2016	WHI OS	Highest vs lowest levels	Lifetime exposure	Among ever-smokers: aOR: 1.14 (1.03–1.26) for infertility; Among never-smokers: aOR: 1.18 (1.02–1.35) for infertility	Age decades, race, education, alcohol use, insecticide exposure, OC use, BMI at 18 years and hard exercise at 18 years	Poor/0
Jensen, 2006		Neither parents smoked used as reference group	Childhood exposure to smoking parents	Among female twins: one smoking parent: 1.07 (0.91, 127); two smoking parents: 1.05 (0.83, 1.33) Among male twins: one smoking parent: 0.95 (0.79–1.15); two smoking parents: 1.27 (0.91–1.76)	Female age at start of TTP, use of OCs, in utero smoking exposure, current smoking	Poor/0
Peppone, 2009		Sample-specific tertiles, no exposure used as reference group	Childhood, adult, and joint childhood and adult exposure	aOR for difficulty becoming pregnant per childhood secondhand exposure: 1.27 (1.03–1.56); aOR for difficulty becoming pregnant per adult secondhand exposure: 1.24 (1.03–1.51) Childhood SHS but no adult SHS: 1.57 (1.15–2.15) Adult SHS but no childhood SHS: 1.41 (1.06–1.87) Both adult and childhood SHS: 1.68 (1.28–2.20)	Age, parity, income, race, education level, BMI, marital status, BC use, alcohol, smoking, period irregularity, problem getting pregnant	Poor/0
Radin, 2014		Passive smoking in adulthood, hours per day	Childhood and adult exposure	Among never-smokers, childhood exposure: aFOR: 1.04 (0.89–1.21) Among never-smokers, adult exposure: aFOR: 0.92 (0.82–1.03)	Age, education, partner smoking (where applicable)	Fair/4
Wesselink, 2019	PRESTO	No current SHS exposure and partner not regular smoker used as reference group	Male partners reported smoking behaviors	Women: Any current SHS exposure, partner is regular smoker: 0.66 (0.42, 1.04) Null results for lifetime SHS exposure Men: null results for current and lifetime SHS exposure	Women: age, parity, income, race, education level, BMI, frequency of intercourse, stress, diet, sleep, depression, doing something to improve chances of conception, supplement use Men: age, income, race, education level, BMI, maternal age, female BMI, female education, doing something to improve chances of conception, prior successful impregnation of partner	Fair/3

aOR, adjusted odds ratio; aFOR, adjusted fecundability odds ratio; BC, birth control; OC, oral contraceptive.

^◊Potential mediators.

Associations between inhaled occupational pollutants and fecundability

Nineteen studies examined inhaled occupational pollutants and fecundability (Table V). While the majority of the studies were conducted in Europe (Sallmén et al., 1995, 1998, 2006, 2008; Ahlborg et al., 1996; Spinelli et al., 1997; Dahl et al., 1999; Kolstad et al., 1999; Plenge-Bönig and Karmaus, 1999; Taskinen et al., 1999; Wulff et al., 1999; Figà-Talamanca et al., 2000; Kolstad et al., 2000; Wennborg et al., 2001; Zhu et al., 2005; Hooiveld et al., 2006), several were conducted in the USA (Rowland et al., 1992; Luderer et al., 2004) and one was conducted in China (Wang et al., 2012) (Table V).

All studies were retrospective. However, a range of techniques were used for exposure assessment including occupation, environmental air monitors and biomarkers. Some studies examined relative exposure dose within an occupational cohort (Ahlborg et al., 1996), whereas others used occupation as a proxy for exposure. Hooiveld et al. (2006) compared painters (proxy for ‘exposed to organic solvents’) to carpenters (proxy for ‘unexposed to organic solvents’); similarly, Luderer et al. (2004) used millwrights and painters as their exposed group and carpenters as their unexposed group, Sallmén et al. (2008) compared shoe manufacturing workers to store and storehouse workers, and Wennborg et al. (2001) compared laboratory personnel with personnel in non-laboratory departments.

Several studies combined these two strategies and compared occupations while also reporting dose within the ‘exposed’ occupation. Dahl et al. (1999) compared dental surgeons (‘exposed’) to high-school teachers (‘unexposed’), and then compared frequency of self-reported exposure to a variety of VOCs (chloroform-based root fillings, methylated ethanol with and without benzene and isopropyl alcohol, chlorhexidine with ethanol, aldehydes, phenols and chlorides) within the group of dental surgeons. Similarly, Zhu et al. (2005) compared laboratory technicians (‘exposed’) to teachers (‘unexposed’), and then created an exposure index by multiplying exposure level by exposure frequency among laboratory technicians.

Studies also combined data from environmental air monitors with self-reported occupational exposure. Figà-Talamanca et al. (2000) assessed male mint (metal coin production) workers’ exposure to solvents and metals based on self-reported occupation and exposure and environmental air concentrations of nickel, chromium and lead; they also measured blood levels of the metals in machine operators. Wang et al. (2012) measured formaldehyde exposure among workers in the wood-processing industry using a portable formaldehyde detector. Sallmén et al. (2008) used both personal and environmental air sampling to measure solvent exposure.

Biomarkers were used in conjunction with other measurements in several studies. Sallmén et al. (1995) measured solvent biomarkers in urine and serum as well as self-reported occupation and self-reported exposure, while Sallmén et al. (1998) relied on a combination of self-reported occupation, self-reported exposure and urinary solvent concentrations. Kolstad et al. (1999) used measurements of styrene in workroom air and urine samples to quantify exposure; in a later paper, Kolstad et al. (2000) used urine and serum samples as well as air sampling to measure styrene exposure.

Finally, Wulff et al. (1999) created categories of exposure among men and women working at a smelter (extracting metal from ore) for at least 3 months, while they were trying to become pregnant based on job type and compared them to women living within 20 km of the smelter and a random sample of women living in two more distant towns.

Evidence for longer TTP

Ahlborg et al. (1996) found reduced fecundability (aFOR: 0.63, 95% CI = 0.43–0.94) among midwives assisting more than 30 deliveries/month where NO_x was used compared to midwives who did not assist deliveries where NO_x was used. Rowland et al. (1992) found reduced fecundability among dental assistants exposed to unscavenged NO_x (scavenging equipment captures unused and exhaled gas and vents it away from workers), with an aFOR of 0.94 (0.90–0.98) for every hour of exposure per week. Taskinen et al. (1999) reported reduced fecundability (adjusted fecundability density ratio: 0.64, 95% CI = 0.43–0.93) for exposure to formaldehyde among female woodworkers when comparing those with the highest level of exposure to those who were unexposed. Similarly, Wang et al. (2012) reported an aOR of 2.83 (95% CI = 1.08–7.41) for TTP >12 months, comparing men exposed to formaldehyde to men classified as unexposed. Sallmén et al. (2006) found that while male or female exposure to solvents was associated with subfertility when comparing exposure >1/month to exposure <1/month, the effect was more pronounced when both partners reported exposure, and most pronounced if the male was exposed on a weekly or daily basis. Finally, Plenge-Bönig and Karmaus (1999) reported an aFOR of 0.47 (95% CI: 0.29, 0.77) for toluene exposure after comparing time of unprotected intercourse while working in the printing industry with time of unprotected intercourse while working in other industries.

Evidence for no association with TTP

No association was found between exposure to styrene and TTP by Kolstad et al. (1999) or Kolstad et al. (2000) when comparing sample-specific quartiles of serum levels. Neither Luderer et al. (2004) nor Hooiveld et al. (2006) reported any association between exposure solvents and TTP in men, and Dahl et al. (1999), Taskinen et al. (1999) and Zhu et al. (2005) found no association between exposure to solvents and TTP in women. Spinelli et al. (1997) found no association between solvents and fecundability among men or women, and Plenge-Bönig and Karmaus (1999) reported no association between toluene exposure and TTP in men. Figà-Talamanca et al. (2000) found no association between exposure to solvents or metal fumes and TTP in a sample of men, comparing different occupations within the mint with unexposed (administrative) staff. Wulff et al. (1999) reported no association between proximity or occupational exposure to heavy metals emitted by a smelter and TTP in women.

Evidence for shorter TTP

Zhu et al. (2005) reported an aFOR of 1.45 (95% CI = 1.15–1.81) using TTP among those with the highest level of exposure to epoxy substances among laboratory technicians compared to those with the lowest level of exposure.

Summary

There is mixed evidence for a relationship between inhaled occupational pollutants and TTP (Table VIII). In general, studies were low quality and focused on chronic, long-term exposure, making the isolation of critical windows of exposure difficult. However, associations of exposure to NO_x and formaldehyde with TTP were consistently positive. In contrast, the majority of included studies found no association between exposure to solvents and reduced fecundability.

Table VIII.

Effect sizes, inhaled occupational pollutants and time to pregnancy (TTP).

First author, year	Study acronym	Analysis	Effect size, fully adjusted model (95% CI)				Covariates*	AHRQ/additional quality metrics
			Formaldehyde	Solvents	Heavy metals	NOx
Ahlborg, 1996		5 categories (number of deliveries with NOx/month): 0, 1–10, 11–20, 21–30, ≥31				aFOR: 0.63 (0.43–0.94) for ≥31 deliveries/month vs none	Cycle order, age, pregnancy order, previous fertility problem, OC use, tea consumption	Poor/2
Dahl, 1999		Daily exposure to individual VOCs, placing chloroform-based root fillings		No associations			Age, smoking, medical history indicating reduced fertility in respondent or spouse	Poor/1
Figà-Talamanca, 2000				No association	No association		Maternal and paternal age, smoking, alcohol consumption, education level, contraceptive use	Poor/1
Hooiveld, 2006		Tertiles for estimated exposure; dichotomized for exposed vs not exposed via questionnaire measures		No association			Maternal age, maternal smoking, maternal and paternal alcohol use, year of pregnancy, chemical occupational exposures, physical occupational exposures, medication use during pregnancy	Poor/0
Kolstad, 1999				No association			TTP starting date, study center, maternal age, maternal and paternal smoking, OC use, parity	Poor/3
Kolstad, 2000		Serum levels categorized into four groups		No unbiased associations (aFOR of 0.68 (0.48–0.97) in one country for lowest vs unexposed; authors believe due to residual confounding)			Maternal age, research site, female smoking, use of OCs, TTP starting year, length of employment	Poor/3
Luderer, 2004				No associations			Age, female smoking, attempting to conceive	Fair/2
Plenge-Bönig, 1999		Low, medium and high exposure		Women: aFOR: 0.47 (0.29–0.77) No association in men			Age, parity, smoking, ethnicity	Poor/0
Rowland, 1992						aFOR, unscavenged gas: 0.94 (0.90–0.98) for every h/week	OC use, smoking, age, history of PID, number of previous sexual partners, frequency of intercourse, race	Fair/0
Sallmén, 1995		Not exposed, low and high exposure categories		aIRR: 0.41 (0.27–0.62) for daily or high solvent exposure compare to not exposed aIRR: 0.28 (0.11–0.71) for shoe factory workers; 0.44 (0.22–0.86) for dry cleaning shop workers; and 0.58 (0.34–0.98) for metal industry workers vs unexposed workers			Age, previous induced abortion or extrauterine pregnancy, recent use of IUD, alcohol use, older age at menarche, unplanned pregnancy, frequency of intercourse	Poor/1
Sallmén, 1998		Unexposed, low-intermediate, high-frequent		Men: aFOR: 0.68 (0.48–0.98) for aromatic hydrocarbons No association for other solvents measured			Maternal age, partner exposure, frequency of intercourse, TTP starting year, older maternal age at menarche, female cycle duration	Poor/0
Sallmén, 2006	AHS			Exposure to solvents >1/month vs <1/month aOR 1.42 (1.15–1.76) Either parent exposed >1/month vs <1/month aOR: 1.62 (1.20–2.17) Only ‘wife’ exposed >1/month vs <1/month aOR: 1.56 (1.17–2.09); both parents exposed, man monthly exposed aOR: 1.65 (1.17–2.33); man daily or weekly exposed aOR: 2.10 (1.22–3.60)			Age, BMI, research site, smoking, partner BMI, partner alcohol use	Poor/2
Sallmén, 2008				aFDR for low solvent exposure: 0.55 (0.40–0.74); aFDR for high solvent exposure: 0.70 (0.52–0.94)			Age, parity, race, age at menarche, partner’s environmental contaminants, duration of employment, planned pregnancy, alcohol, partner smoking, partner alcohol use, period irregularity, last method of contraception	Poor/3
Spinelli, 1997				Women: no association Men: aFOR: 0.74 (0.62–0.89 for industrial occupation			Age, parity, frequency of intercourse, partner’s environmental contaminants, job shift work, hours worked/week, stress, alcohol, smoking, coffee and tea consumption, video display terminal use	Poor/1
Taskinen, 1999		Four categories: not exposed, low, medium and high	aFDR: 0.64 (0.43–0.93) comparing highest level of exposure to unexposed	No association			Parity, menstrual cycle length, employment, smoking, alcohol consumption	Poor/0
Wang, 2012	Binary classification of exposure (high/low)		aOR: 2.83 (1.08–7.41)				BMI, alcohol use	Fair/2
Wennborg, 2001				aFOR: 0.79 (0.68–0.93)			Age, partner’s age, duration of employment, cycle order, known fertility problems, partner’s laboratory work	Fair/2
Wulff, 1999					No association		Age, parity, income, education level, partner’s age, menstrual cycle length, smoking, caffeine consumption, alcohol consumption, job shift work, hours worked/week, stress, video display terminal use, time trend	Poor/1
Zhu, 2005	DNBC	Categories created from exposure index		EI ≥6: aFOR: 1.28 (1.02–1.61)			Age, gravidity, BMI, smoking, partner’s occupation	Fair/2

aOR, adjusted odds ratio; aFOR, adjusted fecundability odds ratio; aFDR, adjusted fecundability density ratio; aIRR, adjusted Incidence Rate Ratio; BC, birth control; OC, oral contraceptive; PID, pelvic inflammatory disease; IUD, intrauterine device.

^*Potential mediators.

Discussion

The effect of poor air quality on health has been an area of in-depth investigation in environmental epidemiology for quite some time, although reproductive outcomes have not been a central focus. To our knowledge, this is the first systematic review to summarize the effects of both outdoor and indoor inhaled environmental pollutants on fecundability. The primary findings from the review revealed evidence for associations between outdoor air pollution on reduced fecundability, particularly NO_x and PM_2.5, as well as exposure to SHS and to formaldehyde in the context of occupational exposure.

Prospective versus retrospective study design in TTP studies

Our review identified studies that employed both prospective and retrospective designs, each of which has limitations that potentially impact the validity and generalizability of results. On the one hand, while prospective studies have the advantage of measuring TTP in real time, they must enroll couples at or near the time they begin trying to conceive, and thus are limited to ‘pregnancy planners’. Given that ∼50% of pregnancies are unplanned in the USA (Wellings et al., 2013; Finer and Zolna, 2016), these samples are unlikely to be representative of the general population. Internet-based preconception studies, such as PRESTO, are further restricted to participants who have internet access and are comfortable with sharing personal health data online. On the other hand, retrospective studies that recruit participants during pregnancy or postpartum and ask participants about TTP at enrollment are susceptible to the effects of recall bias, although short-term retrospectively recalled TTP has been shown to have reasonable validity (Zielhuis et al., 1992). These studies are also limited to couples who successfully conceived and are thus more fertile than the general population. In addition, retrospective studies may enroll couples who experienced an unplanned pregnancy, and self-reported preconception data may be of uneven quality, as couples with unintended pregnancies may less accurately recall exposures during the preconception period compared with pregnancy planners, and time ‘at risk’ may be less precise. Nevertheless, simulations have found that while some bias may be created in a retrospective design, it is likely minor (Eijkemans et al., 2019). Within the studies we reviewed, there was no identifiable trend in results with respect to prospective versus retrospective studies, i.e. both kinds of studies presented evidence for longer TTP as well as evidence for no association with TTP.

Both types of TTP study also miss pregnancies that result in fetal loss before they are clinically recognized. In theory, this can be avoided in prospective studies by collecting daily urine samples for analysis of hCG (Wilcox et al., 1999) to identify even occult pregnancies, although this is expensive, time-consuming and rarely done in practice. None of the included studies in this review employed this method.

Despite these limitations, TTP studies are a useful epidemiologic approach that have identified delays in conception due to a wide variety of factors.

Live birth bias in retrospective studies that enroll participants postnatally

Live birth bias is a form of left truncation where selective survival between conception and birth results in biased estimates of exposure. Because exposure is only assessed among the live births, and not in the entire population of conceptuses (Leung et al., 2021), a difference in exposure distribution between the live births and all conceptuses will introduce bias. While previous work has simulated live birth bias when considering the associations between prenatal environmental exposures and child outcomes (Liew et al., 2015; Raz et al., 2018; Leung et al., 2021), no study to date has considered the possibility of live birth bias in TTP studies when participants are recruited postnatally. Live birth bias is not an issue if TTP studies are conducted in preconception cohorts, with frequent and repeated ascertainment of pregnancy status. However, it is possible that for retrospective studies of TTP among women recruited upon a live birth, there are some variables, such as air pollution, that may affect both TTP and delivery of a live birth. In this case, delivery of a live birth would act as a collider in the relationship between air pollution and TTP. Simulation studies suggest that the bias would be toward the null and small in magnitude (Liew et al., 2015; Raz et al., 2018; Leung et al., 2021); however, without further knowledge regarding such a variable, we cannot evaluate this possibility.

Appropriate confounder control and comparability between exposed and unexposed groups

Some included studies controlled for variables that, depending on the temporal relationship of the variables to the exposure, may be downstream factors of exposure and/or on the pathway between the exposure and outcome (e.g. exposure to SHS during childhood). Conditioning on mediators or colliders that are downstream of exposure can introduce bias (Schisterman et al., 2009). Examples of these adjustments are controlling for ‘difficulty getting pregnant’ (Peppone et al., 2009) and ‘known fertility problems’ (Ahlborg et al., 1996). This is of particular concern when the exposure of interest is long-term and cumulative, or when the exposure occurs long prior to the reproductive period, as is the case for studies of childhood/lifetime SHS exposure. A careful consideration of potential confounders, based on the causal diagram of the study question while accounting for the temporal relation between the exposure and third variable, is warranted to obtain an unbiased estimate of the association between the exposure and TTP.

The challenge of constructing a cohort where exposed and unexposed groups are comparable is especially difficult when exposure status is related to occupation, or when the proxy for exposure is occupation. In either situation, given that different professions attract different kinds of people, the prevalence of covariates also associated with the outcome may vary with occupation and result in a biased effect estimate. For example, Luderer et al. (2004) compared millwrights and painters (exposed group) with carpenters (unexposed group), and while the researchers controlled for age, female smoking and attempting to conceive, there may be multiple additional covariates that are associated with the outcome that vary between the different professions.

Windows of exposure

Identifying sensitive windows of exposure and consideration of the lag time between exposure and outcome in environmental epidemiology are critical both for intervention and from a policy development perspective. In studies of air pollution, they are also closely tied to understanding biological mechanisms. Studies included in this review used a variety of exposure windows, as no consensus exists whether acute (recent with respective to TTP in that particular cycle), distal or chronic exposure contributes to longer TTP. Some studies were able to examine windows of susceptibility because advances in exposure assessment of ambient air pollution have provided us with refined measures of exposure with high temporal resolution. Interestingly, outdoor air pollution studies that examined both short-term and acute exposure around the cycle that led to conception and long-term cumulative exposure (e.g. in the 3 years prior to the cycle that led to the conception) showed lower fecundability. These observations suggest that the effect of air pollution exposure on TTP might result from various underlying mechanisms, such as acute effects on semen quality or chronic influences on inflammatory conditions such as PCOS. Each of these mechanisms could affect fecundability during a different window of susceptibility. It is important to note that many acute exposures are highly correlated with longer-term exposures (e.g. higher seasonal pollution with living in a low socioeconomic status area), so isolating critical exposure windows can be difficult. With time-varying exposures such as ambient air pollution, specifying the hypothesized mechanism and sensitive window of the exposure-outcome relationship is particularly important. In addition, the type of air pollution might play a role in observed short- vs long-term associations, e.g. short-term exposure to NO_x and TRAP and long-term, cumulative exposure to PM was associated with reduced fecundability. Experimental data confirm that each of these pollutants might induce specific biological processes, i.e. inflammation or hormone disruption; nevertheless, many of the exposures are collinear (and for example, PM is a component of TRAP), so it is difficult to draw firm conclusions about timing for specific compounds from the available data.

Limitations of air quality metrics chosen

A large body of work has been dedicated to investigating the best way to capture exposure to ambient air pollution (Watson et al., 1988; Sellier et al., 2014). While many researchers assert that personal air monitors offer the most accurate picture, others argue that more objective measurements of exposure such as environmental air monitors may both introduce less bias into studies and be less susceptible to reverse causation (Weisskopf and Webster, 2017). In addition, environmental air monitors capture exposures that are likely most susceptible to policy intervention. For these reasons, in the review, we did not make a distinction in quality between studies that measure exposure to ambient air pollution using personal and/or residential air monitors, environmental dust/air sampling or GIS-based methods. Nevertheless, the determination of ‘high quality’ for studies of air quality and fecundability is likely dependent on the ultimate goals of the research and the corresponding proposed mitigation strategies.

Lack of air monitoring and biomarker data in SHS studies

All of the studies of SHS and fecundability relied on self-reported exposure, which may contribute to the inconsistency of their findings, as these measures may be affected by recall and may not provide comparable assessments of exposure intensity within or between studies. Biomarkers of SHS would be far preferable and are readily available: cotinine is most commonly measured in urine, but assays are also available for blood, saliva, hair and nails. Nicotine is measurable in all of these biospecimens as well as skin, and concentrations of other combustion byproducts including PAHs, tobacco-specific nitrosamines and heavy metals may be quantified in various matrices (Torres et al., 2018). Nicotine and cotinine are rapidly metabolized (Benowitz et al., 2009), so repeated prospective biosample collection would be most appropriate in the context of TTP studies; however, hair and nail samples collected during pregnancy could be used to assess exposure during the prepregnancy period and may provide a more reliable measure of cumulative or chronic exposure than a single spot urine sample, which is more appropriate for acute exposure. To assess potential critical windows of exposure during earlier life stages, longitudinal cohorts that collect biospecimens during childhood and adolescence could be leveraged.

Comparability across studies

When comparing the findings of various studies of air pollution and TTP, several factors should be considered. First, inhaled pollutants that were examined varied from TRAP to general ambient pollutants such as PM or SHS exposure, to pollutants that are only relevant in the occupational context. These exposures might influence fecundability through different biological pathways and thus windows of susceptibility might differ. Moreover, these studies used various units of exposure and varying reference groups, some of which were sample-specific (e.g. comparing relative high and low groups within the exposure range of the given sample).

In this review, all studies of indoor air quality and TTP examined SHS or exposure in occupational settings, despite the large body of literature showing associations between indoor cooking-related biomass fuel use and birth outcomes (Boy et al., 2002; Mishra et al., 2004). With the advancement in exposure assessment technology, including low-cost and Internet of Things-connected monitors (i.e. physical devices such as smart appliances and wearables with sensors that connect to the internet and transmit data), and the availability of sensors capable of monitoring air quality indoors, such studies are feasible. These technologies would provide valuable insights into the role of indoor air pollution on reproductive outcomes and TTP, particularly in the residential setting (Boyle et al., 2015; Kelly, 2017; Gaskins and Hart, 2020). Air monitors that capture real-time exposure would be particularly valuable in assessing the short-term acute effect of pollutants on TTP.

The need to better assess exposure was also notable in the studies of indoor occupational settings and TTP, as occupation was often used as a proxy for exposure, resulting in less reliable results and a lower quality rating. Especially given the recent emphasis on Total Worker Health (a United States Centers for Disease Control initiative) and Total Exposure Health (supported by the United States Department of Defense), further collaborations between occupational exposure scientists and reproductive epidemiologist can aid in the advancement of our understanding of the relationship between indoor air quality and fecundability (Lundrigan et al., 2018; Phillips et al., 2020).

Conclusions

In summary, the evidence suggests that exposure to outdoor air pollutants such as NO_x and PM_2.5, SHS and some occupational inhaled pollutants may reduce fecundability. Future studies of SHS and occupational inhaled pollutants should use indoor air monitors and biomarkers to improve exposure assessment, and studies of biomass fuel smoke are needed.

Supplementary data

Supplementary data are available at Human Reproduction Update online.

Data availability

No new data were generated or analyzed in support of this research.