Introduction

The production of many organochlorine compounds including polychlorinated biphenyls (PCBs) started in the 1930s. PCBs were used as heat transfer fluids, hydraulic fluids, solvent extenders, plasticizers, flame retardants, and dielectric fluids until the use was restricted in 1972 and prohibited since the late 1970s in most industrialized countries including Denmark (Medehouenou et al. 2011; Noren and Meironyte 2000; Takeuchi et al. 2011). As alternatives to PCBs, other compounds such as polybrominated biphenyls (PBBs) and polybrominated diphenyl ethers (PBDEs) were then introduced as flame retardants. PBBs and two types of PBDE mixtures (i.e., pentaBDEs and octaBDEs) have been phased out globally, and the use of a third type (i.e., decaBDEs) is being regulated (Sjodin et al. 2004; Wu et al. 2011). Several organochlorine pesticides (OCPs) including dichlorodiphenyltrichloroethane (DDT) and hexachlorobenzene (HCB) have also been banned from use (Andersen et al. 2002; Noren and Meironyte 2000). Many of these polychlorinated and polybrominated compounds are persistent and bioaccumulate in the food chain, causing humans to be exposed through the diet such as fish, meat, and dairy products (Brauner et al. 2011; Brauner et al. 2012; Diamanti-Kandarakis et al. 2009; Halldorsson et al. 2008b; Thomsen et al. 2007). Several OCPs, PBDEs, and PCBs are registered as persistent organic pollutants (POPs) under the Stockholm convention (Stockholm Convention 2009).

Human elimination of the POPs is slow. The median half-life of the DDT metabolite p,p’-dichlorodiphenyldichloroethylene (p,p’-DDE) among Brazilian malaria control workers was 28 months, although with large inter-individual variations (Ferreira et al. 2011). Using serum from former capacitor workers, Seegal et al. found that the geometrical mean half-lives over a study period from 1976 to 2004 were 17.8 years for the heavy PCB congeners (i.e., PCBs eluting before DDE using gas chromatography) and 9.6 years for the light congeners (i.e., PCBs eluting after DDE using gas chromatography) (Seegal et al. 2011). In female capacitor workers, the half-lives were between 1.5 and 10 times longer than in males, possibly because the initial PCB concentrations were higher in the males and thus were influenced less by background exposure (Seegal et al. 2011). PBDE levels in serum of Swedish electronic workers decreased during vacations with estimated half-lives of 14 days for BDE-209, 37 days for BDE-203, 110 days for BDE-183, 270 days for BDE-154, and 680 days for BDE-153 (Jakobsson et al. 2003).

Although, the lipophilic POPs have been banned for many years, they may still be of concern to public health, especially for the fetus, which is at a vulnerable stage of life (Bonefeld-Jorgensen 2010). The temporal changes of lipophilic POP concentrations have not been studied much in Europe in recent years, and to our knowledge, no time trend studies have been conducted in Denmark. The lipophilic POP concentrations in human serum may be dependent on factors such as age, body mass index (BMI), parity, and gender (Caspersen et al. 2016; Fernandez-Rodriguez et al. 2015; Long et al. 2015; Medehouenou et al. 2011; Porta et al. 2012; Rylander et al. 2012; Sandau et al. 2000; Wolff et al. 2005). Hence, it is important to control for these covariates when conducting a time trend study by e.g., selecting a homogenous population such as nulliparous pregnant women. To our knowledge, no previous studies of lipophilic POP trends have included only nulliparous pregnant women.

The objectives of the present study was (1) to determine the temporal trends of the most abundant lipophilic POPs in serum of Danish pregnant nulliparous women between 2011 and 2013 and (2) to investigate associations between the lipophilic POP concentrations and the pregnant women’s age at delivery, pre-pregnancy BMI, number of miscarriages, and educational level.

Materials and methods

Samples

Blood samples were collected from 197 pregnant women, randomly sampled among Aarhus Birth Cohort participants, who fulfilled the selection criteria of being nulliparous and giving birth to a live born singleton neonate between 2011 and 2013 at the Aarhus University Hospital, Denmark (Mortensen et al. 2013). The blood samples were taken in the first trimester before the end of gestational week 13 and were processed within 2 h after blood drawing and stored as serum at −80° C (Mortensen et al. 2013). The gestational age at the time of blood draw was estimated by ultrasound.

The women filled out a questionnaire including questions about height, pre-pregnancy weight, miscarriages, country of birth, smoking status, alcohol consumption, and educational level (Bjerregaard-Olesen et al. 2016a). The Aarhus Birth Cohort also holds information about the delivery and the newborn reported by the attending midwife.

All participants provided written consent to the storing of their blood samples in the biobank and that the samples and information could be used for research. The study was approved by the Danish National Committee on Health Research Ethics (j.nr: M-20110054) and the Danish Data Protection Agency (j.nr: 2011-41-6014).

Chemical analyses

The chemical analysis of PCBs, PBDEs, and OCPs was conducted at Centre de Toxicologie, Institute National de Santé Publique du Québec, a contract laboratory certified by the Canadian Association for Environmental Analytical Laboratories. The serum samples were analyzed for lipophilic POPs including OCPs (aldrin, alphachlordane, cisnonachlor, gammachlordane, HCB, mirex, oxychlordane, p,p’-DDE, p,p’-DDT, β-hexachlorocyclohexan (β-HCH), and trans-nonachlor), PCBs (28, 52, 99, 101, 105, 118, 128, 138, 153, 156, 170, 180, 183, and 187), BB-153, and PBDEs (15, 17, 25, 28, 33, 47, 99, 100, and 153). Briefly, 2 mL serum was spiked with internal standards (13C12-PCB 141, 13C12-PCB 153, 13C12-PCB 180, 13C12-PCB 194, 13C6-HCB, 13C6-α-HCH, 13C10-trans-nonachlor, 13C10-oxychlordane, 13C12-p,p’-DDE, 13C10-Parlar 26, 13C10-Parlar 50, 13C12-PBDE 77, 3,6-F2-PBDE 99, and PBDE 101). The serum proteins were denatured using reagent grade alcohol followed by addition of a saturated ammonium sulfate solution and liquid-liquid extraction using n-hexane. The extracts were evaporated, reconditioned in 0.5 mL hexane, and purified on activated florisil columns (60–100 mesh; Fisher Scientific, Nepean, Ontario, Canada), which were eluted using 9 mL dichloromethane:hexane (1:3). The purified extracts were concentrated, and 3 μL was analyzed using gas chromatography (Agilent 7890A) equipped with an Agilent 60 m DB-XLB column (0.25 mm i.d., 0.25 μm film thickness) and coupled to a 5975C mass spectrometer (MS; Agilent technologies, Mississauga, ON, Canada). Helium was used as the carrier gas, and the injections were in splitless mode. The MS used selected ion monitoring (SIM) with negative chemical ionization (NCI) and methane as the reagent gas. The limit of quantification (LOQ) was determined as a signal-to-noise ratio of 3:1, corresponding to 0.02 μg/L for HCB, p,p’-DDE and p,p’-DDT, 0.01 μg/L for the other OCPs and PCBs, and 0.05 μg/L for the PBDEs. The recoveries varied between 68 and 90% for all analytes, the intra-day precision was 1.2–8.3%, and the inter-day precision was 2.8–13%. Further details have been described elsewhere (Chevrier et al. 2013; Fisher et al. 2016).

Serum lipids were analyzed using a standard enzymatic procedure, and the total lipid concentration was obtained by summation of the concentrations of cholesterol esters, free cholesterol, triglycerides, and phospholipids. Further details have been described elsewhere (Phillips et al. 1989).

The serum concentrations of perfluorinated alkyl acids (PFAAs) were analyzed at the Department of Environmental Science, Aarhus University, as described in a recent study (Bjerregaard-Olesen et al. 2016b). Briefly, 100 μL serum samples were spiked with 13C–labeled internal PFAA standards, diluted with 3 mL 0.1 M formic acid, and extracted by solid phase extraction using OASIS HLB cartridges (3 mL, 60 mg; Waters, Milford, MA, USA). The cartridges were washed with 1 mL Millipore water and conditioned with 2 mL methanol and 2 mL 0.1 M formic acid before addition of the diluted serum samples. The PFAAs were eluted using 2 mL methanol. The extracts were dried and reconstituted in 100 μL 2 mM ammonium acetate/methanol (50/50, v/v). Sixteen PFAAs (C4, C6-C8, and C10 sulfonates; C5-C14 carboxylates; and C8 sulfonamide) were analyzed in the reconstituted extracts using LC-MS/MS on an Agilent 1100 series HPLC (Agilent technologies, Palo Alto, CA, USA) equipped with a C18 Kinetex column (2.1 × 150 mm, Phenomenex, Torrance, CA, USA) and coupled to a triple quadrupole QTrap 5500 (Sciex, Concorde, ON, Canada). The relative inter-day standard deviation was between 1.2 and 16.7% for all the PFAAs.

Statistics

For the samples that contained POPs at concentrations below the LOQ, we assigned the samples with values of LOQ/2 for the statistical analysis. The lipophilic POP concentrations were corrected for total serum lipids (μg/kg lipid). Only POPs that could be detected in at least 50% of the samples were included in the statistical analysis. In addition to the single compounds, we analyzed the data for three groups of lipophilic POPs:

$$ \begin{array}{c}\hfill \sum \mathrm{PCB}=\mathrm{PCB}\ 28+\mathrm{PCB}\ 52+\mathrm{PCB}\ 99+\mathrm{PCB}\ 101+\mathrm{PCB}\ 105+\mathrm{PCB}\ 118+\mathrm{PCB}\ 128+\mathrm{PCB}\ 138\hfill \\ {}\hfill +\mathrm{PCB}\ 153+\mathrm{PCB}\ 156+\mathrm{PCB}\ 170+\mathrm{PCB}\ 180+\mathrm{PCB}\ 183+\mathrm{PCB}\ 187\hfill \end{array} $$
$$ \begin{array}{c}\hfill \sum \mathrm{OCP}=\mathrm{aldrin}+\mathrm{alphachlordane}+\mathrm{cisnonachlor}+\mathrm{gammachlordane}+\mathrm{HCB}+\mathrm{mirex}+ p,{p}^{'}\hbox{-} \mathrm{DDT}\hfill \\ {}\hfill + p,{p}^{'}\hbox{-} \mathrm{DDE}+\mathrm{oxychlordane}+\upbeta \hbox{-} \mathrm{HCH}+\mathrm{trans}\hbox{-} \mathrm{nonachlor}\hfill \end{array} $$
$$ \begin{array}{c}\hfill \sum \mathrm{lipPOP}=\sum \mathrm{PCB}+\sum \mathrm{OCP}+\mathrm{BB}\hbox{-} 153+\mathrm{BDE}\hbox{-} 100+\mathrm{BDE}\hbox{-} 15+\mathrm{BDE}\hbox{-} 153+\mathrm{BDE}\hbox{-} 17\hfill \\ {}\hfill +\mathrm{BDE}\hbox{-} 25+\mathrm{BDE}\hbox{-} 28+\mathrm{BDE}\hbox{-} 33+\mathrm{BDE}\hbox{-} 47+\mathrm{BDE}\hbox{-} 99\hfill \end{array} $$

The women were divided into four categories according to their educational level (low: municipal primary or lower secondary school; lower middle: upper secondary school or 1–2 years of vocational training; higher middle: additional 3–4 years of education; high: >4 years of additional education). The linear associations between the lipophilic POP concentrations and the age at delivery, pre-pregnancy BMI, number of miscarriages, and educational level were determined for the crude data and after adjustment for the following potential explanatory factors: age (continuous), BMI (continuous), gestational age at blood draw (continuous), and smoking status (three categories: non-smoker, smoked until pregnancy, and smoked during pregnancy) using linear regression analysis.

Spearman correlation analysis was performed to assess the bivariate correlation between the POPs. Only samples with POP concentrations above the LOQ were included in the bivariate correlation.

Results

Study participants

The characteristics of the 197 study participants are presented by year in Table 1. The median age at delivery, pre-pregnancy BMI, and gestational age at blood draw were similar throughout the study years. The distribution of women across the categories of education, smoking status, and alcohol intake were also similar over the study years. Most of the pregnant women were born in Denmark (91%), and all of the women were living in Denmark at the time of delivery. The women with a non-Danish origin had statistically significantly higher ∑OCP concentrations than the women with a Danish origin, but only slightly higher ∑PCB concentrations (Table S1 in the supplementary material).

Table 1 Characteristics of the 197 randomly selected nulliparous women from the Aarhus Birth Cohort and median concentrations of the lipophilic POPs (μg/kg serum lipids), 2011–2013
Full size table

Detection frequency

Twelve lipophilic POPs were detected in more than 50% of the samples: PCB 118 (93%), PCB 138 (100%), PCB 153 (100%), PCB 156 (58%), PCB 170 (97%), PCB 180 (100%), PCB 187 (91%), HCB (99%), oxychlordane (55%), p,p’-DDE (95%), β-HCH (68%), and trans-nonachlor (51%) (Table S2 in the supplementary material). The brominated flame retardants including PBDEs and PBB were detected in less than 5% of the samples (Table S3 in the supplementary material).

Temporal trends

The median concentrations of 11 of the 12 included lipophilic POPs (i.e., PCB118, PCB138, PCB153, PCB156 PCB170, PCB180, PCB187, HCB, p,p’-DDE, β-HCH, and trans-nonachlor) were lower in 2013 compared to 2011 (Table 1 and Fig. 1). However, the median concentration of oxychlordane was higher in 2013 and 2012 compared to 2011 (Table 1 and Fig. 1). Moreover, the percentage of samples with oxychlordane concentrations above the LOQ increased from 36% in 2011 to 67% in 2012 and 62% in 2013 (Table S2 in the supplementary material).

Fig. 1
figure 1

Temporal trends of the lipophilic POP levels (ng/g serum lipids) in serum from 197 Danish pregnant women in the period from 2011 to 2013. The boxes display the 25th and 75th centiles, and the line inside the boxes represents the median value. The whiskers display the upper and lower values within 1.5 times the interquartile range beyond the box. For the sum of the PCBs, the sum of OCPs, and the sum of lipophilic POPs, see the Methods section

Full size image

The temporal trends were statistically significantly different from 0 for HCB (β = −5.4; 95% confidence interval [95% CI]: −10.6; −0.3), oxychlordane (β = 16.6; 95% CI: 5.7; 27.4), and β-HCH (β = −14.8; 95% CI: −25.1; −4.6) when all 197 women were included in the adjusted analysis. In addition to these three compounds, the temporal trends were also statistically significant for PCB138, PCB153, PCB170, PCB180, trans-nonachlor, ∑PCB, and ∑lipPOP when the analysis was restricted to women with a Danish origin (Tables 2 and 3 and supplementary Fig. S1). The temporal trends did not change much, when we used data that were not corrected for serum lipids (data not shown).

Table 2 Temporal trends of polychlorinated biphenyl levels (μg/kg serum lipids) in serum from 197 Danish pregnant women in the period 2011–2013
Full size table
Table 3 Temporal trends of organochlorine pesticide levels (μg/kg serum lipids) in serum from 197 Danish pregnant women in the period 2011–2013
Full size table

Associations between the lipophilic POP concentrations and the women’s characteristics

The associations between the lipophilic POP concentrations and the pregnant women’s age, BMI, number of miscarriages, and educational level are presented in Table 4. Higher age at delivery was associated with higher concentrations of all the studied lipophilic POPs, being statistically significant for all PCBs, oxychlordane, β-HCH, and trans-nonachlor (Table 4).

Table 4 Continuous linear regression between the serum concentrations (μg/kg serum lipids) of persistent organic pollutants and study variables in 197 Danish pregnant women, 2011–2013
Full size table

The lipophilic POP concentrations were mostly inversely associated with the pre-pregnancy BMI, although we found a positive association for β-HCH (β = 0.02, 95%CI: 0.00; 0.04). The associations were statistically significant for PCB153, PCB156, PCB170, PCB180, PCB187, HCB, trans-nonachlor, ∑PCB, and ∑POP (Table 4).

For nine of the 12 analyzed lipophilic POPs, we found that the concentrations were inversely associated with the number of miscarriages, although none of the associations were statistically significant (Table 4).

A higher educational level was associated with higher concentrations of PCB187, p,p’-DDE, and ∑OCP but lower concentrations of PCB118, PCB138, PCB153, PCB156, PCB180, HCB, oxychlordane, β-HCH, ∑PCB, and ∑lipPOP, although only statistically significantly for β-HCH (Table 4).

Correlations

The bivariate Spearman correlations were positive and statistically significant between ∑OCP, ∑PCB, and ∑lipPOP (Table 5 and Table S4), but the correlations between the lipophilic POPs and the PFAAs were weak (Table 5 and Table S5).

Table 5 Spearman correlation coefficients between the summed concentrations of polychlorinated biphenyls (∑PCB), organochlorine pesticides (∑OCP), lipophilic POPs (∑lipPOP), and perfluoroalkyl acids (∑PFAA) in serum of 197 Danish pregnant women, 2011–2013
Full size table

Discussion

Temporal trends of lipophilic POPs

As shown in Table 1 and Fig. 1, the median serum concentrations of 11 of the 12 analyzed lipophilic POPs were lower in 2013 compared to 2011. The temporal trend analyses were statistically significant only for three compounds, HCB, oxychlordane, and β-HCH, in the analysis where all 197 women were included (Tables 2 and 3). Three women with a non-Danish origin had up to 10 times higher p,p’-DDE levels and 3–5 times higher ∑lipPOP levels than the mean of the other 194 women. We conducted a secondary analysis with restriction to women with a Danish origin (n = 180), and in this analysis, the temporal trends were now statistically significant for eight compounds including HCB, oxychlordane, β-HCH, PCB138, PCB153, PCB170, PCB180, trans-nonachlor, ∑PCB, and ∑lipPOP (Tables 2 and 3). As the statistical significance was so highly affected by the few outliers, we will in the following focus on the direction of the temporal trend slopes rather than on the statistical significance. The downward trends were expected, since the production and use are regulated for all the lipophilic POPs that were included in the trend analysis (Stockholm Convention 2009).

To our knowledge, there are no studies on recent lipophilic POP time trends in serum from pregnant women in the non-Arctic parts of Europe. However, the Arctic Monitoring and Assessment Programme (AMAP) published a report in 2015 including time trends for pregnant women in the circumpolar region (AMAP 2015). Our results support the data from the AMAP report that showed recent downward trends of PCB and OCP concentrations in pregnant women’s whole blood, serum, and plasma in Alaska, Nunavik in Canada, Greenland, Iceland, northern Norway, and coastal Chukotka in Russia (AMAP 2015).

As shown in Table 6, the Danish women in the present study had much lower concentrations of most OCPs compared to the pregnant women from Alaska, Nunavik, Greenland, Murmansk, and Iceland, but the p,p’-DDE concentration was similar to northern Norway and Reykjavik (AMAP 2015). For PCBs, the concentrations in the Danish women were lower than for pregnant women from Nunavik, Greenland, and Iceland, but similar to northern Norway, and higher than in Alaska (AMAP 2015) (Table 6).

Table 6 Geometric mean plasma and serum concentrations (ng/g lipid) of lipophilic POPs in pregnant women in recent years
Full size table

A study from the US with slightly lower median serum PCB concentrations compared to the present study (Table 6) found downward trends of PCB74, PCB118, PCB138, PCB153, and PCB180 in US pregnant women between 2008–2009 and 2011–2012 (Zota et al. 2013).

Overall, our results agree with the time trends in the other studies, and only differ with regard to the trend for oxychlordane. Oxyclordane was detected in 55% of the samples in the present study, and the concentrations were close to the LOQ for most samples. Hence, we expect that the upward trend for oxychlordane in the present study was a chance finding, since the oxychlordane precursor chlordane is under regulation and is included in the Stockholm Conventions annex A (Stockholm_Convention 2009).

Associations between the lipophilic POP concentrations and the women’s characteristics

In agreement with most other studies (Caspersen et al. 2016; Fernandez-Rodriguez et al. 2015; Long et al. 2015; Medehouenou et al. 2011; Porta et al. 2012; Rylander et al. 2012; Sandau et al. 2000; Wolff et al. 2005), we found that the PCB and OCP concentrations were higher in older women (Table 4). This association is presumed to be a reflection of the persistency and long human serum half-life of the lipophilic POPs.

There are conflicting results in the literature about the association between lipophilic POP levels in serum and BMI. Our results support studies from Denmark, Norway, and the USA, which reported lower PCB levels for pregnant women with a higher BMI (Caspersen et al. 2016; Halldorsson et al. 2008b; Wolff et al. 2005). We suggest that the lipophilic POPs might be stored in the adipose tissue of women with a high BMI rather than being in circulation. Contrarily, two studies of the adult population in Spain as well as one study of older Canadians found higher serum or plasma levels of most OCPs and PCB congeners with higher BMI (Fernandez-Rodriguez et al. 2015; Medehouenou et al. 2011; Porta et al. 2012). Long et al. (2015) found no statistically significant association between BMI and OCP or PCB levels in 194 Greenlandic pregnant women (Long et al. 2015). The median lipophilic POP concentrations were much lower for the women in the present study compared to the study populations we have used for comparison (Fernandez-Rodriguez et al. 2015; Long et al. 2015; Medehouenou et al. 2011; Porta et al. 2012; Wolff et al. 2005). The association with BMI might be dependent on the lifestyle and diet of the women including their intake of fatty fish (Caspersen et al. 2013), which we do not have sufficient information to adjust for.

The plasma concentrations of PCBs and OCPs have been reported to be lower for pregnant women with a higher parity (Caspersen et al. 2016). This might be due to a placental transfer from the woman to the fetus during pregnancy or a subsequent transfer during the breastfeeding. In the present study, we studied the association between the number of previous miscarriages and the lipophilic POP concentrations in nulliparous women, thus, neglecting the influence from breast feeding. We found a tendency of lower lipophilic POP concentrations with a higher number of miscarriages (Table 4), which might support the hypothesis that some POPs were partly transferred from the pregnant woman to the fetus. This was emphasized by a study in Peruvian pregnant women, where the lipid-corrected lipophilic POP concentrations in maternal serum were found to decrease from the first to the third trimester, and several PCBs and p,p’-DDE were detected in cord blood (Adetona et al. 2013). However, lipophilic POPs have also been detected in breast milk (AMAP 2015). Our findings were not statistically significant, which might be because the serum samples were taken in early pregnancy (i.e., gestational week 11–13).

Contrary to our results, Leoni et al. (1989) found statistically significantly higher concentrations of the PCB mixture “Fenclor 54” in pooled serum from 120 Italian women who had miscarried prior to the blood sampling compared to 120 full-term pregnancy controls, whereas the levels of HCB, DDE, and DDT were insignificantly higher for women who experienced miscarriage compared to the controls (Leoni et al. 1989). We suggest that the higher concentration of Fenchlor 54 in the women who miscarried might be explained by a shorter duration of the pregnancy compared to the full-term controls. A case-control study from 2003 reported that 45 women with a history of three or more consecutive miscarriages had similar PCB, HCB, and DDE levels as 30 healthy non-pregnant women (Sugiura-Ogasawara et al. 2003). However, neither of these studies adjusted for other explanatory factors such as age, BMI, and smoking (Leoni et al. 1989; Sugiura-Ogasawara et al. 2003). Overall, the association between the number of miscarriages and the lipophilic POP concentrations may be weak if at all present.

Considering educational level as a proxy for socioeconomic status, our results agreed with those of a previous study, which found no linear associations between PCB levels and socioeconomic status in Danish pregnant women (Halldorsson et al. 2008b). The overall lack of associations with socioeconomic status might be an indication that the contact with sources of lipophilic POP exposure does not differ much across the social classes of pregnant women in Denmark. Contrarily, studies from the USA, Canada, and Australia found higher lipophilic POP levels with a higher educational level (Medehouenou et al. 2011; Reid et al. 2013; Wolff et al. 2005). Porta et al. (2012) found that Spanish women in the least affluent social classes from the Barcelona Health Survey had the highest levels of p,p’-DDT, p,p’-DDE, β-HCH, and PCB180, whereas women in the most affluent social classes had the highest levels of HCB, PCB118, PCB138, and PCB153 (Porta et al. 2012).

Bivariate correlations of POP concentrations

As expected, the bivariate Spearman analyses showed positive correlations between the lipid-corrected concentrations of the lipophilic POP groups (Table 5), thus, indicating that the sources of exposure were similar for all the lipophilic POPs. Contrarily, the lipophilic POPs were poorly correlated with the PFAAs (Table 5), which might indicate that the Danish pregnant women’s exposure to these compounds comes from different sources. For example, other studies have found that a higher dietary intake of fatty fish was significantly associated with higher PCB and OCP concentrations but not PFAA concentrations in the Danish population (Brauner et al. 2011; Brauner et al. 2012; Eriksen et al. 2011; Halldorsson et al. 2008a, b). In contrast to our results, Long et al. (2015) observed that higher concentrations of the ∑PCB and ∑OCP were associated with higher ∑PFAA concentrations in pregnant Inuit women from Greenland (Long et al. 2015). Due to cultural, dietary, and regional differences, the exposure sources might be very different for the Greenlandic women and the Danish women.

Strengths and limitations of the study

Ideally, all studies on trends should follow the same selection procedure—random sampling from all potential candidates or sampling according to fixed criteria. In the present study, we randomly selected the pregnant women from the Aarhus Birth Cohort. Moreover, the women were not aware of their exposure levels, so the potential selection bias is expected to be non-differential. One of the major strengths of the present study is the design of including only nulliparous women. Lipophilic POPs can be transferred from mother to fetus via the placenta and to the child via breastmilk (Adetona et al. 2013; Covaci et al. 2002; Longnecker et al. 1999), and the serum POP concentrations might therefore be lower for women of higher parity. To avoid confounding from the reproductive history, we included only nulliparous women in early pregnancy.

A limitation to this study is that we included participants over three study years only (i.e., 2011–2013). Thus, it is possible that the observed downward trends were only momentary observations, although we expect them to be long-lasting, as the production and use of PCBs and OCPs have been regulated for more than 30 years in Denmark. We included less than 200 women, which limits the precision of our findings. Our results might not be transferable to other study populations. We found that the lipophilic POP concentrations differed for the women born in Denmark compared to women born elsewhere, and the downward trends for PCB138, PCB153, PCB170, PCB180, trans-nonachlor, ∑PCB, and ∑lipPOP were statistically significant only when restricting the analysis to women born in Denmark. Moreover, the associations between lipophilic POP concentrations and BMI and educational level differ much across study populations.

In summary, there are few studies on temporal trends of lipophilic POPs from non-arctic regions, and to our knowledge, this is the first temporal trend study of lipophilic POPs in the Danish population. Except for oxychlordane, we found that the lipophilic POP concentrations followed downward trends in the Danish pregnant women, as also observed for pregnant women in the Arctic regions. The Danish women had much lower OCP concentrations compared to what have been reported for Arctic women in the most recent AMAP report (AMAP 2015). We found associations of higher lipophilic POP concentrations with higher age and lower BMI, whereas the associations with the educational level and number of miscarriages were less clear.