Phthalate levels in prenatal and postnatal bedroom dust in the SELMA study

Phthalates are common in polyvinyl chloride (PVC) plastics and numerous consumer goods in our homes from which they can migrate and adhere to indoor dust particles. It is known that indoor dust exposure contribute to human phthalate intake; however, there is a lack of large studies with a repeated-measure design investigating how phthalate levels in indoor dust may vary over time in people ’ s homes. This study investigated levels of seven phthalates and one alternative plasticiser di-iso-nonyl-cyclohexane-di-carboxylate (DiNCH) in bedroom dust collected prenatally around week 25 during pregnancy and postnatally at six months after birth, from 496 Swedish homes. Prenatal and postnatal phthalate levels were compared using correlation and season-adjusted general linear regression models. Over the nine-month period, levels of six out of seven phthalates were associated as indicated by a positive Pearson correlation (0.18 < r < 0.50, P < .001) and Lin ’ s concordance correlation between matched prenatal and postnatal dust samples. Compared to prenatal levels, the season-adjusted postnatal levels decreased for five phthalates, whilst di-ethyl-hexyl phthalate (DEHP), di-2-propylheptyl phthalate (DPHP) and DiNCH increased. The results suggest that families with higher phthalate levels in bedroom dust during pregnancy are likely to remain among those with higher levels in the infancy period. However, all average phthalate levels changed over this specific nine-month period suggesting that available phthalate sources or their use were altered between the dust collections. Changes in home characteristics, family lifestyle, and phthalate replacement trends may contribute to explain the differences.


Introduction
As many people spend a large proportion of their time indoors, indoor environments are important for human health and well-being (Salthammer et al., 2018).Interiors and consumer goods can release a wide range of chemicals that may accumulate in the indoor environment and become available for human intake (Salthammer et al., 2018).Phthalates are contaminants that are frequently detected in non-industrial indoor dust and air (Lucattini et al., 2018;Salthammer et al., 2018).They are high-volume compounds mainly used as plasticisers in polyvinyl chloride (PVC) interiors, synthetic leather, packaging, toys, etc.The more volatile phthalates are also used as solvents in fragrance, cleaning products, personal care products and paint (Swedish Chemical Agency [SCA], 2015; Wormuth et al., 2006).
Increasing evidence from observational epidemiology and experimental toxicology have linked phthalate exposure to adverse human health effects (Dorman et al., 2018;Engel et al., 2021;Radke et al., 2018).Di-n-butyl phthalate (DnBP), di-iso-butyl phthalate (DiBP), butyl-benzyl phthalate (BBzP), and di-2-ethylhexyl phthalate (DEHP), have been added to the Candidate List as Substances of Very High Concern in the European Union (SCA, 2015).Similar bans for use in children's toys and care articles have been introduced in e.g., the US and China (Christia et al., 2019).This has increased the use of substitute phthalates and alternative plasticisers (Bui et al., 2016;Sackmann et al., 2018).Still, phthalates continue to be global high-volume production chemicals.Since regulated phthalates remain in use in products with long life cycles (e.g., PVC floorings), many phthalates will remain ubiquitous indoor pollutants for a considerable time (Abb et al., 2009;Bu et al., 2020).
Pregnant women and children may be particularly vulnerable to exposure to phthalates such as BBzP and DEHP, due to their ability to disrupt early-life hormone-dependent development (Bølling et al., 2020;Engel et al., 2021).Higher levels of BBzP and DEHP in dust have been associated with asthma, allergy and croup in children (Bornehag et al., 2004;Bølling et al., 2020;Preece et al., 2021aPreece et al., , 2021b)).Phthalate intake from dust is of particular concern for young children.They spend more time on the floor and display frequent mouthing behaviour resulting in a higher dust ingestion rate relative to body weight than adults (Bekö et al., 2013;Lioy et al., 2015).
Previously, numerous studies have assessed phthalate levels in indoor dust.However, sample sizes have often been small and data on replacement phthalates and alternative plasticisers is limited.There is a lack of large, repeated-measure design studies to assess variations in phthalate levels in dust over time.The objectives of this study were to: 1) examine phthalate levels in prenatal and postnatal bedroom dust samples, 2) examine the impact of some home and family characteristics on phthalate levels in dust, 3) investigate the stability of phthalate levels in prenatal compared to postnatal dust (collected nine months apart) among families in the Swedish Environmental Longitudinal Mother and child Asthma and allergy (SELMA) study.

Study population
SELMA is an ongoing prospective pregnancy cohort with around 2000 participating mother-child pairs enrolled between late 2007 and early 2010.The study aims to investigate early life exposure to environmental factors during pregnancy and infancy and its importance for health and development in children.In brief, pregnant women were invited to participate at their first visit to their antenatal clinic (median week 10 of pregnancy) and 39% accepted.Self-administered questionnaires were used to collected information on lifestyle and living environment.The study was approved by the Regional Ethical Board in Uppsala, Sweden.Individual informed consent was obtained before each subject began participation in the study.The enrolment process and data collection is more precisely described by Bornehag et al. (2012).Prenatal and postnatal dust samples were available for 912 SELMA families from which 501 were randomly selected for phthalate analysis.

Collection of dust
The dust collection was performed by the participants following detailed procedure manuals which were distributed together with the sampling equipment.Sampling sites were the parent's bedroom during pregnancy and the room where the child slept during infancy.Dust was collected from surfaces above floor level (e.g., shelves, doorframes and skirting boards) into a filter sock attached to the nozzle of the family vacuum cleaner.The filter sock was made of Heavy Duty polyamide with an aperture of 25 μm (Allied Filter Fabrics, Australia).Following collection, the filter sock with its dust content was placed in airtight polypropylene (PP) tubes, submitted by post, and stored frozen (− 20 • C) until analysis.This sampling was conducted prenatally in gestational week 25 and repeated postnatally when the child was six months old.

Chemical analysis of dust
The dust was analysed for seven phthalates: di-ethyl phthalate (DEP), DnBP, DiBP, BBzP, DEHP, di-iso-nonyl phthalate (DiNP), di-2propylheptyl phthalate (DPHP), and one phthalate replacement plasticiser di-iso-nonyl-cyclohexane-di-carboxylate (DiNCH).Analysis limits of detection (LOD) are reported in Supplementary Table S1 together with information on transitions and internal standards, Table S2 reports the quality control, and Table S3 shows the CAS-numbers of included compounds.All phthalate standards and their corresponding D 4 -labelled internal standards were purchased from TRC Chemicals (Toronto, Canada).Hexamoll® DINCH, were kindly provided by BASF, Ludwigshafen, Germany.
Dust samples were analysed using gas chromatography tandem mass spectrometry (GC-MS/MS) as previously described by Weiss et al. (2018) and Preece et al. (2021aPreece et al. ( , 2021b)).More information on the chemical analysis is available in the Supplementary Information.In brief, to eliminate phthalate contamination, all glass vials, pipette tips and tweezers were washed twice with 70% ethanol before use.Any large objects were carefully removed from the dust.No further fractioning to smaller particle sizes was performed and sample preparation steps were kept to a minimum to avoid phthalate contamination.
Ten mg of dust was weighed in duplicates into glass vials and extracted with 1 mL toluene containing 50 μg of internal standards.
Blank samples and quality controls were added in each analysis, to control for phthalate contamination.Samples were sonicated for 30 min, centrifuged, and 1 μL supernatant injected onto to a Shimadzu GC 2010-Plus chromatograph (Shimadzu Corporation) coupled to a GC TQ8040 triple quadrupole mass spectrometer (Shimadzu Corporation).The phthalate levels were reported as the duplicate mean values (mass fractions, μg/g dust).

Determinants
Information on determinants was retrieved from two questionnaires distributed at enrolment and at 12 months after birth.At enrolment, information on residential area (urban or rural), type of housing (house or flat), bedroom flooring material (PVC or other material), size of home (m 2 ), and mother's age and education was retrieved.Vacuum cleaning frequency and cat or dog at home were retrieved at enrolment and 12 Maternal smoke exposure was assessed by cotinine concentration in maternal blood collected at enrolment.Blood plasma was stored frozen (− 80 • C) until analysis by LC-MS/MS as described by Axelsson et al. (2015).Mothers were categorised as smokers if plasma cotinine concentrations were greater than 15 ng/mL, as non-smokers if levels were below 0.2 ng/mL, and as passive smokers if the concentration was between (Jefferis et al., 2010).Mothers with plasma cotinine concentrations ≥0.2 ng/mL were categorised as exposed to tobacco smoke.

Statistical analysis
For data analysis, phthalate levels below the LOD were replaced with a value corresponding to LOD/ ̅̅̅ 2 √ .Due to skewed distributions, dust levels were log 10 -transformed to allow for parametric testing.The impact of the selected determinants on phthalate levels was assessed with independent t-tests (2-sided) and included dust collection season, home characteristics (residential area, type of housing, bedroom flooring, and size of home), family characteristics (vacuum cleaning frequency and cat or dog at home) and maternal characteristics (age, education, and smoking).The two continuous determinants, size of home and mother's age at birth, were dichotomised using the median value as the cut-off point.The impact of changing the vacuum cleaning frequency on prenatal and postnatal phthalate levels were tested with 2sided independent t-tests.Data on matched prenatal and postnatal phthalate levels was available for 496 families, whilst home and participant characteristics data was available for between 429 and 496 families.
The precision of association between the individual family's prenatal and postnatal phthalate levels was assessed by calculating Pearson correlation coefficients.To further test the accuracy of the associations, Lin's concordance correlation coefficients (CCCs) with 95% CIs and Bland-Altman plots were analysed.Differences between the study population average prenatal and postnatal phthalate levels were explored with general linear regression models and the results were reported as least square geometric means (LSGMs) with 95% CIs.The phthalate levels were adjusted for dust collecting season by adding the season variable (categorised as summer or winter) as a covariate in the regression models.Dust collection season was chosen as an adjustment factor a priori based on previous literature (Bi et al., 2018;Liu et al., 2020;Wei et al., 2018).Within-home differences were further tested with paired t-tests (2-sided).
Differences between prenatal and postnatal phthalate levels based on the time point of the dust collection were compared between two evenly sized groups.The groups were created by dichotomising the time point of child birth using the median as the cut-off point ( May

The study population and characteristics
Descriptive information of the study population are summarised in Table 1.The full study population was 496 families and data on participant and home characteristics was available for between 496 and 429 families.Most characteristics were available for more than 98% of families, except for size of home (available for 95.2%) and mother's smoke exposure (available for 86.5%).All subgroup size percentages in Table 1 are comparisons to the full study population (N = 496).
Vacuum cleaning frequency and having a cat or a dog in the home were included in the analysis as they may affect the amount or composition of the dust.The majority of families vacuum cleaned once a week or more (61% during pregnancy and 71% during infancy).More than half (57%) of families reported the same prenatal and postnatal

Table 1
Characteristics available for the full study population (N = 496 families) for whom phthalate levels were determined in bedroom dust.Characteristics were collected at enrolment (prenatal period) or at 12 months after birth of child (postnatal period).vacuum cleaning frequency (Table S4).Urban homes were more common (65%) than rural, and a majority of families lived in houses (69%) compared to flats.PVC flooring was reported by 24% of participants and the mean size of the home was 118.3 m 2 (SD 49.1).Regarding the maternal characteristics, the mean age was 31.4 years (SD 4.7) and median 31.3 years.Thirty percent of mothers were categorised in the lower education category, and 10% were exposed to tobacco smoke.Twenty-eight families (5.6%) reported a change of bedroom flooring.As this is a low number, the prenatal information on home characteristics was considered to be representative for the postnatal period (Table 1).

Phthalate levels in prenatal and postnatal dust
The detection frequency for phthalates were ≥90% whilst DiNCH was detected in considerably fewer samples (26% and 62% for prenatal and postnatal samples, respectively), as shown in Table 2. From the seven analysed phthalates, DEHP was found in highest levels with geometric mean (GM) levels of 150 μg/g (prenatal) and 130 μg/g dust (postnatal), followed by DiNP and BBzP (excluding samples below LOD).The least abundant phthalates were DEP and DPHP with GMs ranging between 1.3 and 1.9 μg/g.Prenatal and postnatal GM levels of DiNCH were 26 and 36 μg/g, respectively.The GMs decreased when levels below LOD were assigned their replacement values, particularly for prenatal DEHP for which the detection rate was 89.9% (Table S5).Statistical analysis of DiNCH was performed on participants with detectable levels only due to low detection rates, leaving 107 matched pairs (22% of the study population).

Determinants for phthalate dust levels
Dust collection during heating season was significantly associated with lower prenatal DnBP and DiNCH levels (P < .01)with similar nonsignificant trends in postnatal dust (Table 3).For DEHP, there was a nonsignificant trend of lower prenatal levels whilst postnatal levels were significantly higher during the heating season.
Median levels of DEHP were almost threefold higher and DiNP 70% lower in prenatal dust collected in homes during the later period (between May 2009 and July 2010) compared to the earlier (December 2007 to January 2009) (P < .001for both) (Table 3).The opposite was observed in postnatal dust; the median level for DEHP was 25% lower (P < .05),and DiNP 67% higher (P < .001) in dust collected later (November 2009 to January 2011) compared to earlier (September 2008 to October 2009).For DPHP, there were significantly higher levels in the later collection periods for both prenatal and postnatal dust.We found neither vacuum cleaning frequency nor having a cat or a dog in the home had a significant impact on phthalate dust levels, although prenatal DiNCH levels were higher among families who vacuum cleaned more frequently (n = 130) (Table 3).We found no major impact of changes in vacuum cleaning frequency on phthalate level differences (Table S6).
No phthalate levels were significantly different between the residential areas (urban or rural), whilst DiNCH levels were higher in prenatal dust from urban homes (P < .01)(n = 130).Dust from flats had higher levels of BBzP (P < .01).DiNCH levels were higher in prenatal dust from families living in flats (P < .01)(n = 130).PVC flooring was associated with higher levels of DnBP (postnatal only), BBzP and DEHP (prenatal and postnatal) compared to dust from rooms with other flooring materials.Smaller homes were associated with higher levels of BBzP (P < .01),DEHP (P < .05),and DiNP (postnatal P < .01),whilst prenatal levels of DiNCH were lower (n = 130, P < .001).
Younger mothers had significantly higher levels of DEP (at both time points) and postnatal BBzP (P < .05).Mothers in the lower education category had higher levels of DEP (at both time points) and postnatal levels of BBzP (P < .05) in their dust.Mothers who were exposed to tobacco smoke (10%) had higher prenatal dust levels of BBzP and lower postnatal levels of DPHP (P < .05)(Table 3).

Comparisons between prenatal and postnatal levels
Table 4 shows the correlation analysis results.A modest precision in associations over the nine months were found for levels of six phthalates (DEP, DnBP, DiBP, BBzP, DEHP, and DPHP), as indicated by a positive Pearson correlation between prenatal and postnatal phthalate dust levels.The phthalate with the strongest Pearson correlation was BBzP (r = .50,P < .001),followed by DEP (r = 0.48, P < .001),while it was weakest for DEHP (r = 0.18, P < .001).DiNP levels did not correlate between the two collection time points, with contradictive results between Pearson correlation and CCC.For DiNCH the Pearson correlation and the CCC were negative (r = − .29,P < .01,n = 107).
The accuracy of these associations was confirmed by the CCCs that were somewhat lower but within the same range of magnitude as the Pearson correlation coefficients (Table 4).Supplementary Figure S1 shows the scatter plots with best fit linear regression lines and Bland-Altman plots for prenatal and postnatal phthalate levels.The lines of best fit were close to the line of perfect concordance for DEP, DiBP, BBzP and DPHP, whilst for DnBP and DEHP, the difference was somewhat  S5.
larger.For DiNP and DiNCH, the lines of best fit have negative slopes and the scatter plots show no apparent trends.As indicated by a more narrow limit-of-agreement span in the Bland-Altman plots, the best degree of agreement was found for DnBP, DiBP, BBzP and DPHP.Considerably wider spans were found for DiNP and particularly DiNCH, indicating a lower degree of agreement between measurements of prenatal and postnatal phthalate levels (Fig. S1).Phthalate level differences were examined with general linear regression models adjusted for dust collection season.Levels of all compounds were significantly different (P < .01) between the two collection time points with LSGMs of DEHP, DPHP and DiNCH being higher whilst the other five phthalates were lower in postnatal compared to prenatal dust (Fig. 1).
Corresponding numeric data for crude and adjusted GM levels are available in Table S5.Season adjustment reduced the prenatal and increased the postnatal levels of DnBP and DEHP slightly, whilst the opposite was observed for DiNCH.Only marginal effects were observed on the levels of the other five phthalates.Equivalent within-home  A.-S. Preece et al.
differences were found for all compounds, which were significant for all phthalates but not for DiNCH (Fig. S2).

Discussion
When prenatal and postnatal dust levels were compared, we found that all phthalates except for DiNP were relatively stable over the nine month period, as indicated by the correlation results.However, we also found decreasing GM levels of DEP, DnBP, DiBP, BBzP and DiNP, whilst DEHP, DPHP and the alternative plasticiser DiNCH increased.For DEHP and DiNP, the difference was inconsistent over the 3.5 year duration of the dust collection.Low detection rates limits the statistical analysis of DiNCH and results must therefore be interpreted cautiously.

Phthalate dust levels
The average phthalate levels detected in this study are comparable or lower than previous results from other studies.Phthalate levels in settled dust from Swedish homes have been reported in three previous studies (Bergh et al., 2011;Bornehag et al., 2005;Luongo and Ostman, 2016) with relatively comparable overall results, DEHP was the most abundant phthalate with higher median levels (499-770 μg/g) compared to 130 μg/g in our study.DEP was the phthalate detected in the lowest dust levels (median 0-14 μg/g), probably due to its high volatility compared to the other phthalates.DnBP was more abundant in the three studies with median 103-150 μg/g, compared to 17-22 in our study.BBzP and DiBP were found in low levels similar to our results.DiNP was only reported by Luongo and Ostman (2016) with median levels of 106 μg/g, which is considerably higher than ours.Neither study analysed DPHP or DiNCH.
Two larger studies (n ≥ 100) from Denmark and Japan measured phthalate levels in residential settled dust collected around the same time as ours, reported lower BBzP and higher DEHP levels than ours (Ait Bamai et al., 2014;Langer et al., 2010).The Japanese study reported a higher median DiNP level (95 μg/g).A meta-analysis of 59 studies of phthalate levels in indoor dust from North America, Europe and Asia published 2000-2019 concluded that DEHP is dominant with a median level of 616 μg/g (Bu et al., 2020).The mean total concentrations of six phthalates were highest in Asia (945 μg/g) and lowest in Europe (580 μg/g).As differences between studies may also relate to method choices, comparisons between studies must be done with caution.Method differences may relate to the choice of dust type, the collection site, the collection method and the sample preparation method (Lucattini et al., 2018).Another problem is that some plasticisers (such as DiNP and DiNCH) exists in different isomers, making chemical analysis difficult.
Comparisons between different studies and laboratories might therefore not be conclusive.
In this study, dust levels of DEP and DPHP were lower than levels of the other phthalates.Phthalate levels in indoor dust are strongly associated with sources present in the indoor environment (Lucattini et al., 2018), but are also determined by the ability of each phthalate to migrate and adhere to dust (Salthammer et al., 2018).As DEP has relatively high volatility and low molar mass (Table S3), its ability to adhere to dust is lower than for the other included phthalates in this study.Low dust levels of DEP are consistently reported (Bu et al., 2020).For the less volatile and increasingly used phthalate DPHP, the low levels may reflect fewer available sources or a lower migration rate.

Determinants
We found that home characteristics mainly influenced dust levels of BBzP and DEHP.Levels were higher in flats compared to houses and in smaller homes compared to larger.We have previously reported higher dust levels of BBzP and DEHP in prenatal dust from SELMA families with PVC bedroom flooring (Preece et al., 2021a(Preece et al., , 2021b) ) and similar results were reported in other studies (Bi et al., 2018;Bornehag et al., 2004;Christia et al., 2019).The common use of PVC flooring in Sweden (Larsson et al., 2010) might also contribute to explain the lower Danish BBzP levels compared to this study.Since PVC flooring were more common in flats than in houses, and flats were smaller than houses, these three characteristics may be related (data not included).
BBzP levels were also positively associated with the three maternal characteristics measured in this study: lower age, lower education, and smoking.Education and smoking can be regarded as indicators of socioeconomic status (Haustein, 2006;Mackenbach et al., 2016;Vlismas et al., 2009).However, younger people in Sweden are more likely to live in flats rather than houses (Statistics Sweden, 2020) and have had less time to complete higher education.Hence, the characteristics lower age, lower education, living in flats, and PVC flooring may be related.
For mothers in the lower education category, DEP levels were higher in both prenatal and postnatal dust.DEP sources are unlikely to be related to floorings or other interiors, as it is mainly used in fragrance, personal care products, cosmetics, and cleaning agents (SCA, 2015).It has been suggested that socioeconomic status is related to an unequal risk of environmental exposure to harmful chemicals, including phthalates (Montazeri et al., 2019;Zota et al., 2014).However, higher phthalate metabolite concentrations have been reported among lower (Bloom et al., 2019;Montazeri et al., 2019;Tyrrell et al., 2013) and higher (Tyrrell et al., 2013) socioeconomic statuses.Such differences are less well studied in dust.Zhang et al. (2020) reported higher phthalate levels in dust from urban homes than rural where socioeconomic status is lower, a relationship strongly influenced by building materials.A possible association between DEP levels in indoor dust and education level must be investigated further, and should also consider indoor air concentrations due to the higher volatility of DEP.

Phthalate level correlation over nine months
The positive correlation between phthalate levels in prenatal and postnatal dust suggests that families with higher phthalate levels during pregnancy are likely to remain among those with high levels nine months later.This relationship was strongest for DEP and BBzP.Regular use of similar personal care products, cosmetic articles and cleaning agents could maintain DEP levels in a room over time.BBzP is more likely a plasticiser in products with longer use, such as synthetic leather, electric appliances, and interiors including PVC floorings (SCA, 2015) which may contribute to a stable migration and dust adsorption.However, DEHP which may be used in similar products as BBzP (SCA, 2015), showed the weakest significant correlation.Therefore, sources of DEHP may have been differently available in the bedrooms during the prenatal compared to postnatal dust collection.Prenatal and postnatal differences of indoor sources may also explain the lack of correlation and lower degree of agreement between levels of DiNP.The low detection rates for the alternative plasticiser DiNCH results in a less robust analysis.For DiNCH, although we found a significant negative correlation, the concordance analysis suggests that the data has the lowest degree of agreement of all eight analysed compounds in this study.The low detection rates and subsequently reduced amount of DiNCH data available for statistical analysis have likely contributed to the weaker results.

Differences between prenatal and postnatal phthalate levels
This study found differences between average prenatal and postnatal levels for all phthalates, with some increasing and others decreasing in the study population.The dust was collected over a time period during which the families adapted their homes and lifestyle to include a newborn.This may affect the indoor phthalate sources, e.g., through redecoration or replacement of furniture or other belongings (Lang et al., 2016).Lang et al. (2016) report that Canadian women (n = 80) reduced their use of cosmetic products during pregnancy.Results from Barrett et al. (2014) suggest that among pregnant women in the US (n = 894), those aware of risks from environmental chemicals also made active choices to reduce exposure.Such decisions could favour purchases of organic foods or "eco-friendly" personal care products (Barrett et al., 2014).It is known that expecting women can pursue a healthier lifestyle e.g., by reducing smoking (Graham et al., 2014) or improving their diet (Hillier and Olander, 2017).Our observed changes in phthalate levels may be influenced by family efforts to reduce chemical exposure.This could involve reducing phthalate sources in their homes, or choosing more "eco-friendly" products where regulated phthalates have been replaced.Such replacement alternatives could be DPHP or DiNCH which would help explain our observed increasing levels of these compounds.
Although nine months is a short time period, a further contribution to the changes in phthalate levels may be due to manufacturer's choices.DEHP was the dominant plasticiser in the EU, but the use decreased sharply in Sweden after 2000 and in the EU after 2009, whilst alternatives increased (Bui et al., 2016;Sackmann et al., 2018;SCA, 2015).The use of DiNP increased followed by replacement for DPHP and non-phthalate DiNCH (Bui et al., 2016;Sackmann et al., 2018;SCA, 2015).This might affect the indoor phthalate sources even if the families did not change their lifestyle or consumption behaviours.It may also explain why differences in prenatal and postnatal DEHP and DiNP levels were inconsistent over the dust collection period.However, our dust collection time span is short and not sufficient for a time-trend analysis.We have previously reported phthalate replacement trends in early pregnancy urine collected from the SELMA mothers from 2007 to 2010 (Shu et al., 2018).Similar exposure trends for urine biomarkers has also been observed in other countries (Koch et al., 2017;Wang et al., 2019;Zota et al., 2014), although trends for phthalate levels in dust are not well-studied.Larsson et al. (2017) concludes that levels of DEHP in preschool dust collected in the EU has decreased in a way that reflects the phase-out.Levels of DiNCH have increased in dust since its introduction on the market in 2002 (Salthammer et al., 2018).Today, the use of newer substitute plasticisers that were not investigated in this study are also increasing (Bui et al., 2016;Christia et al., 2019).
Phthalates are biodegradable and can be metabolized by microorganisms or degrade abiotically in the indoor environment.The DEHP monoester mono-ethyl-hexyl phthalate (MEHP) has been detected in indoor dust in levels up to 3-fold lower than the parent diester (Weiss et al., 2018).Increasing heat and humidity can speed up both biotic and abiotic degradation, which affects the stability of phthalate diester levels in the dust (Bope et al., 2019).Environmental factors also influence phthalate air-dust partitioning behaviour.A specific change in indoor temperature or relative humidity affects the degradation and partition behaviour of each phthalate differently (Salthammer et al., 2018).Indoor temperature and humidity are unlikely to be identical in a home nine months apart.Therefore, such unmeasured factors may have influenced the repeated measurement results and contributed to the observed phthalate level variance.

Strengths and limitations
The strengths of this study include a relatively large study population (N = 496) and a repeated prenatal and postnatal collection of settled dust from above floor level.The dust samples were collected and analysed using identical methods.
However, the collection was performed by participants which means the non-floor surface area and the duration of the dust collection varied.We expect any differences to be evenly distributed in the relatively large study population.Further, the collection process cannot be considered fully standardised and we were unable to assess the age and relative amount of dust collected from each home.More robust results could have been obtained if we had monitored indoor temperature and relative humidity during the dust collection.Also, dust samples from each family may not have been collected in the same bedroom, either because the mother and child slept in different bedrooms or because the family moved homes between the dust collections.However, it is common practice for Swedish infants to sleep in their parents' bedroom (>85%) (Wennergren et al., 2021), and we believe the number of families that moved was low (as indicated by the low frequency of flooring material change).Therefore, we expect these scenarios to have a limited influence on our main results.
Prior to chemical analysis, dust samples were not fractioned by particle size, resulting in more heterogeneous dust samples than sieved dust.This may affect how mass fraction levels are comparable between families.However, by collecting dust from elevated surfaces and excluding floor dust, we aimed to acquire expedient samples which require minimal preparation before phthalate analysis.Avoiding a sieving process reduces the risk of contamination which is substantial for ubiquitous contaminants like phthalates.Settled dust from above floor level has been suggested to be more representative of indoor phthalate emissions than floor dust.Phthalate levels are more directly related to the flooring materials (Ait Bamai et al., 2014).Floor dust or dust from vacuum cleaner bags is more conveniently available, but have a greater proportion of large size fractions that requires more processing before analysing (Weiss et al., 2018).Dust from elevated surfaces is more likely to consist of previously airborne particles, but it is a limitation that we do not know details of what surfaces were used for the collection.
Participants in the SELMA study have higher education and smoke less than the average Swedish population (Bornehag et al., 2012).Although this selection bias is unlikely to affect the main conclusions of this study, it limits the generalisability.We are also unable to assess how generalisable our results are among non-expecting families.The data collection time differences must also be considered when interpreting the results.The dust was collected during mid-pregnancy and when the child was 6 months old, whereas several determinants were ascertained during the first trimester of pregnancy (enrolment) or when the child was 12 months old.As many determinants were included in the statistical analysis of differences, there may be false positive results.Finally, the reported results relate to bedroom dust which may not represent dust from other parts of the home.

Conclusion
Our results suggest that families with high phthalate levels in their bedroom dust during pregnancy remain among those with higher levels during infancy.The average levels of several regulated phthalates decreased for the study population.Although we cannot explain the observed differences, a reduction in indoor phthalate sources and family lifestyle changes may contribute.The levels of replacements DPHP and DiNCH increased over the studied time period.Current knowledge on dust exposure contributions to intake of these compounds is insufficient.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
2009).Families with births before the cut-off (n = 247) collected prenatal dust from December 2007 until January 2009 and postnatal dust from September 2008 until October 2009.Families with births from May 2009 onwards (n = 249) collected prenatal dust from February 2009 until May 2010 and postnatal dust from November 2009 until January 2011.The statistical analysis was conducted using IBM® SPSS® Statistics for Windows, version 25.0 and 27.0 (Armonk, NY: IBM Corp).Lin's concordance correlation coefficients and best fit regression lines were calculated with MedCalc® version 20.027 (MedCalc Software,Ltd, Ostend, Belgium).General linear models and LSGMs were calculated using PROC GLM in SAS® version 9.3 of the SAS System for Windows.Copyright © 2012, SAS Institute Inc. Cary, NC, USA.

Fig. 1 .
Fig. 1.Prenatal and postnatal phthalate levels in bedroom dust from 496 families shown as LSGMs with 95% CIs adjusted for dust collection season (heating or non-heating).

Table 2
Detection frequency and phthalate levels (mass fractions, μg/g) in prenatal and postnatal bedroom dust collected approximately nine months apart (N = 496).The detection limits (LOD) refer to the extracted phthalates in 1 mL of toluene and are reported in Supplementary Information TableS1.Samples < LOD excluded, GMs for data where < LOD values are replaced with LOD/ ̅̅̅ 2 √ are available in Table a b

Table 3
Median dust phthalate levels (μg/g) with respect to investigated home and participant characteristics, based on available data for 496 families.Asterisks indicate statistically significant a phthalate level differences between participants with different characteristics in either prenatal or postnatal dust.

Table 4
Pearson correlation coefficients and Lin's concordance correlation coefficients calculated from log 10 -transformed prenatal and postnatal phthalate levels in bedroom dust from 496 families.
a Samples with DiNCH levels < LOD excluded due to low detection rate (n = 107).