Spatial and temporal trends of the Stockholm Convention POPs in mothers’ milk — a global review

Persistent organic pollutants (POPs) have been of environmental and health concern for more than half a century and have their own intergovernmental regulation through the Stockholm Convention, from 2001. One major concern is the nursing child’s exposure to POPs, a concern that has led to a very large number of scientific studies on POPs in mothers’ milk. The present review is a report on the assessment on worldwide spatial distributions of POPs and of their temporal trends. The data presented herein is a compilation based on scientific publications between 1995 and 2011. It is evident that the concentrations in mothers’ milk depend on the use of pesticides and industrial chemicals defined as POPs. Polychlorinated biphenyls (PCBs) and “dioxins” are higher in the more industrialized areas, Europe and Northern America, whereas pesticides are higher in Africa and Asia and polybrominated diphenyl ethers (PBDEs) are reported in higher concentrations in the USA. POPs are consequently distributed to women in all parts of the world and are thus delivered to the nursing child. The review points out several major problems in the reporting of data, which are crucial to enable high quality comparisons. Even though the data set is large, the comparability is hampered by differences in reporting. In conclusion, much more detailed instructions are needed for reporting POPs in mothers’ milk. Temporal trend data for POPs in mothers’ milk is scarce and is of interest when studying longer time series. The only two countries with long temporal trend studies are Japan and Sweden. In most cases, the trends show decreasing concentrations of POPs in mothers’ milk. However, hexabromocyclododecane is showing increasing temporal concentration trends in both Japan and Sweden.


Introduction
Mothers' milk is a source of nutrients, energy, and protection for the newborn child, and it carries essential elements from the mother to the child (Kramer and Kakuma 2012). Due to the lipophilic properties of a range of anthropogenic organic pollutants, ubiquitously distributed in human food and our environment, many of these chemicals are accumulated in mothers' milk. Accordingly, the nursing child is targeted by a vast number of undesirable pollutants (IPCS 2007;UNEP and WHO 2013). These pollutants are similar to those entering the fetus via the cord blood after transfer across the placental barrier (CDC 2013;Frederiksen et al. 2010), although there are differences in the presence of pollutants in the blood and in the mothers' milk. Due to the high chemical and metabolic stability and toxicity of some anthropogenic chemicals as well as their ability to spread globally, and bioaccumulate, 25 chemicals have been adopted under the Stockholm Convention (SCa), known as persistent organic pollutants (POPs). Among these listed POPs, polybrominated diphenyl ethers (PBDEs) are separated Table 1 Names and abbreviations of all POPs and two suggested POPs are presented, CAS numbers are given, as well as chemical structures or general structures. Some major review documents regarding the POPs are presented under BReview articles^(the column to the far right)
Accordingly, it is of interest to review the present exposure situation to POPs for nursing children worldwide, i.e., spatial exposure data as well as levels of POPs in mothers' milk over time. The objective of this review is to summarize the concentrations of POPs in mothers' milk during a delimited time period, 1995-2011. In addition, some recent data have been generated within the Swedish monitoring program and are included herein.
Analysis of some of the first identified POPs in mothers' milk was published in the 1960s (Norén and Westöö 1968) and then novel POPs were added to the list (Westöö and Norén 1978). Still, as shown herein, there is limited mothers' milk data for several of the POPs.

Materials and methods
Names, abbreviations, and (general) structures of the POPs applied in the present review are presented in Table 1, together with reviews discussing their toxicities/ecotoxicological effects.
Methods for data retrieval on POPs in mothers' milk A literature search was performed using the database Web of Science and the following search terms were used: BHuman Milk,^BBreast Milk,^BMother's Milk,^and BMothers Milk,ĉ ombined with the name of the substance of interest. All references found were compiled in one database and duplicates were removed.

Methods for inclusion of data
To reduce errors due to comparison of data from different sources and to avoid presenting a historical overview, the following limits were set for inclusion in the study: & The study must be a scientific peer-reviewed paper published 1995-2011. & Studies must report and quantify any of the POPs listed in Table 1, in at least six subjects (donors). In the case of pooled samples, a pool should contain a minimum of six donors. & Information about place and year of sampling are required. & Inclusion of a time series requires a minimum of five reported data points and is only included if the report/paper includes the original values of the time series.
Other sources of data Data from the Swedish Environmental Monitoring Program are included herein (http://www.imm.ki.se/Datavard/ aBiologiska_mätdata_-_organiska_ämnen), as well as time series data from Fång et al. (2013).

Substance summary tables
Concentration data are presented in the substance summary tables (Tables 2, 3 , 4, 5, 6, 7, 8, 9, 10, 11, 12, and 13) in the following manner: & One reported concentration from one sample location is represented by a single value, e.g., B5.& Two reported concentrations from one location are represented by two values separated by a semicolon, e.g., B3; 5.& Three or more reported concentrations from one sample location are represented by giving the range, e.g., B3-5.& If a study reports data from more than one sampling location, all are included, e.g., by presenting, BSweden 5^and BNorway 4.& In the case of a time series, i.e., more than one sample from one location, only the most recent value is included in the summary

Spatial distribution diagrams
If mean or median values are given for different sampling locations within the same country or subgroups (e.g., age, primiparae versus multiparae), a weighted mean or median value is calculated based on the number of individuals in each group: for example, reported mean concentrations of 2, 3, and 5 ng/g fat of BDE-47, based on 10, 10, and 20 samples (total of 40 samples) from cities X, Y and Z in Sweden, respectively. The weighted mean value for study A will thus be a bar at 3.75 ng/g fat example given below.

Methods applied for statistical reports
To test for significant log-linear trends, log-linear regression analyses were performed for the entire investigated time period and for the most recent 10 years using the annual arithmetic mean values. In cases where the regression line had a poor fit, a 3-point running mean smoother was checked for statistical significance in comparison with the regression through an ANOVA (Nicholson et al. 1998). Potential outliers in the temporal trends were detected using a method described by Hoaglin and Welsch (1978). The suspected outliers are merely indicated in the figures and were included in the statistical calculations. Values below level of quantification (LOQ) were replaced by LOQ/2 prior to the statistical analyses. Power analysis was also carried out. The power was fixed to 80 % and the minimum possible trend to be detected during a monitoring period of 10 years at a significant level of 5 % was estimated. A significance level of 5 % was used for all tests.

Results and discussion
A total of 253 scientific articles on POPs in mothers' milk were identified on the basis of the applied methodology (cf. above). Several of the articles included data on more than one of the POPs. The diagram (Fig. 1) visualizes the number of reported concentrations of the corresponding POP that were available for this review. The results are presented in this review in descending order, starting from the POPs that are most well researched in relation to occurrence and concentrations in mothers' milk, worldwide, i.e., going from DDT to SCCP and chlordecone (Fig. 1).
Looking into the distribution of the scientific articles published on POPs in mothers' milk in the chosen time period, it is clear that most of the studies originate from China, Japan, North America, and Western Europe (Fig. 2). However, publications are scattered throughout the globe making a spatial trend review possible. The results of the spatial distribution and concentrations of POPs are presented under the sections BDDT and DDTrelated compounds^to BSCCPs,^including tables and figures when applicable. Temporal trend data on POPs in mothers' milk are scarce but available data are reported herein under the sections BDDT and DDT related compounds^to BPFOS.^Some novel data are included for recent exposure assessments performed on mothers' milk from Sweden.
It is of importance to consider that a reported concentration might not be generally applicable to a country as a whole, for instance, samples might originate from a farming area where pesticides have been in use. This could be more important to consider for countries with large diversity, either geographically and/or cultural, i.e., rural versus urban life styles. In smaller, more homogenous countries such as Sweden, POP concentrations have been found to be quite uniform, independent of geographical distribution ).

DDT and DDT-related compounds
A very large total number of reports are dealing with DDT and related compounds in mothers' milk. Related compounds are 2,4′-DDT and the transformation products 4,4′-DDE, 4,4′-DDD, 2,4′-DDE, and 2,4′-DDD. The data shown in Table 2 refer only to the three individual compounds 4,4′-DDT, 4,4′-DDE, and 4,4′-DDD as well as ∑DDT. However, the ∑DDT may consist of some very different sums, sometimes including only the three main 4,4′-substituted DDTs mentioned but occasionally also including 2,4′-substituted DDTs. This makes the sum data less reliable for comparisons. However, we have still chosen to include sum data to visualize the larger data set, but avoiding the confusion with further differentiated data. Concentration data on 4,4′-DDT, 4,4′-DDE, and 4,4′-DDD are presented in detail in Table 2, subsectioned into four geographically large areas, i.e., Africa; Asia, Australia, and the Pacific region; Europe; and The Americas. Some of the results on DDT and related compounds in mothers' milk are highlighted below. Table 2 includes calculated ratio values of 4,4′-DDT/4,4′-DDE based on reported mean or median concentrations of the two compounds and gives an indication for recent discharges of DDT (with ratio values of 0.5 and above) or more historical use (ratios below 0.2) of this pesticide. Figure 3 displays the 4, 4′-DDT and 4,4′-DDE concentrations as reported throughout the world.

Africa
Concentrations of 4,4′-DDE are the highest among the three individual DDT compounds reported in Table 2, although the 4,4′-DDT/4,4′-DDE ratio indicates similar levels between the two major constituents in mothers' milk. The highest concentrations of 4,4′-DDT and 4,4′-DDE are reported from Zimbabwe (Chikuni et al. 1997) and South Africa (Okonkwo et al. 2008;Sereda et al. 2009). Still, a few studies indicate low concentrations of DDTs in mothers' milk. It is particularly clear that African mothers have high concentrations of 4,4′-DDT compared to most other samples from other regions. This is of course implying present or recent use of DDT for spraying, potentially indoors (Channa et al. 2012).

Asia, Australia, and the Pacific region
Reports from Asia, Australia, and the Pacific Region indicate that certain mothers have been highly exposed to both 4,4′-DDT and 4,4′-DDE (Nair et al. 1996;Nasir et al. 1998;Stuetz et al. 2001;Wong et al. 2002), while overall levels are above the common European concentrations, but below the concentrations reported for milk from African mothers (Table 2 and Fig. 3). The data indicate primarily old releases of DDT based on the low 4,4′-DDT/4,4′-DDE ratio, although there are exceptions (e.g., Nair et al. 1996;Nasir et al. 1998;Stuetz et al. Fig. 2 Global distribution of exposure assessment studies of POPs in mothers' milk, up to year 2011. The circles are placed at the site of the capital city for each country, and the sizes of the circles visualize the abundance of studies from the countries on which this review is built. Black circle = 1 study, green circle = 2-5 studies, blue circle = 6-10 studies, red circle=11-15 studies, and pink circle >15 studies      Venezuela 280-1000 a Brunetto et al. (1996) When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a  Wong et al. 2002). Accordingly, direct exposure to DDT cannot be excluded.

Europe
The majority of studies on DDT in mothers' milk are originating from Europe. The levels are in the lowest end (e.g., 20-250 ng/g fat of 4,4′-DDE) of all studies reviewed except for studies on DDTs in milk from the Eastern part of Europe, 250-2800 ng/g fat, as shown in Table 2. The higher concentrations in Eastern Europe is also followed by higher 4,4′-DDT/4,4′-DDE ratios indicating more recent use or unintentional release of DDT. However, the ratio is generally low indicating successful elimination of this POP from use in the society. The DDT and related compounds still present in mothers' milk are a mirror of intake via food. Some high exposure levels to DDT among Eastern European citizens, as determined by analysis of blood, are supporting the higher levels in mothers' milk from countries in this part of Europe (Hovander et al. 2006).

The Americas
Low concentrations of DDTs are reported from Canada and the USA, while Mexico in Central America ( Table 2) reported levels that are similarly high as in Africa and some Asian countries. It is notable that in the countries from which the mothers' milk contain the highest concentrations of DDTs, there is a more recent input of DDT (Fig. 3), which is confirmed by higher 4,4′-DDT/4,4′-DDE ratios, 0.12-0.4 (Table 2). A Brazilian study is reporting the highest ratio among all studies reviewed, i.e., 12 (Azeredo et al. 2008), indicating the present use of DDT. However, the actual concentration of ∑DDTs is lower than many other studies.

PCBs
Polychlorinated biphenyls (PCBs) reported as CB-153, sum of the six indicator CBs (CB-28, CB-52, CB-101, CB-138, CB-153, and CB-180, only if the concentrations of all six were reported), or the estimated total sum of PCB (∑PCB, the method of estimating the sum may vary between studies) in all 116 studies were tabulated (Table 3). Dioxin-like PCBs are not reported here but instead discussed together with the dioxins and furans (BPCDDs, PCDFs, and DL-PCBs^). Since PCBs are showing decreasing trends after the bans came into effect, studies from different time periods (1995-2011) may not be altogether comparable (Fig. 4).
Although the ban of PCB that was introduced stepwise during the 1970s and 1980s has led to significantly lowered concentrations in the environment, leakage due to inappropriate handling of waste material or from, e.g., building material, large capacitors, and hydraulic systems, still in use or stored at dumping sites, can still be expected and can thus cause elevated concentrations in mothers' milk from highly industrialized countries.

The Americas
The concentrations of CB-153 reported from most of the 14 studies from the Americas were fairly low to moderate, around or below 50 ng/g fat: Brazil (Paumgartten et al. 2000), Canada (Dewailly et al. 1996;Newsome et al. 1995), and the USA (Fitzgerald et al. 1998;Pan et al. 2010

HCB and HCHs
Mothers' milk concentrations of hexachlorobenzene (HCB) and the three more common hexachlorocyclohexane (HCH) isomers, α-HCH, β-HCH, and γ-HCH, are presented in Table 4, as well as the less commonly reported levels of δ-HCH. α-HCH, β-HCH, and γ-HCH represent the HCHs present in the Bold^technical-grade HCH pesticide, while the          commonly used pesticide, lindane, corresponds to γ-HCH.
All the HCH isomers are related to pesticide use, while HCB has both a pesticide history and is also an abundant by-product from industrial activities and poorly controlled incineration/ backyard burning. The pattern of HCH in the world is highly influenced by recent use of HCH as a pesticide. It is notable that the β-HCH isomer is the most abundant of the HCH isomers in mothers' milk even though this compound is related to the historical HCH pesticide use and not to lindane (γ-HCH). However, the half-life of γ-HCH is much shorter in humans and wildlife than the half-life of β-HCH, and the observations confirm the higher persistency and lower reactivity of the β-HCH isomer compared to the others. The data for HCB and the HCHs are dominated by studies of mothers' milk from Asia and Europe (Table 4) and are reported in 94 and 113 scientific reports worldwide, respectively (Figs. 5 and 6).

Africa
Most reported HCB concentrations from African countries range from about 2 to 5 ng/g fat, on either mean or median basis. Somewhat higher levels are reported in Ghananese mothers' milk, 2.5-40 ng/g fat (Darko and Acquaah 2008;Ntow 2001;UNEP 2009). The highest concentrations are reported from Tunisia, mean concentration of 0.4-290 ng/g fat (Ennaceur et al. 2007(Ennaceur et al. , 2008.
In general, one could consider the ∑HCH levels in African mothers' milk to be on the lower end on a global scale and comparable to concentrations reported from the Americas.
There are quite few studies from Africa reporting levels of ∑HCH in mothers' milk. Most of the studies report mean concentrations of approximately 10-100 ng/g fat (Table 4), the exceptions being one study from Ghana (Saleh et al. 1996) and one study from Libya (Elafi et al. 2001), reporting concentrations of 210 and 500 ng/g fat, respectively.

Asia, Australia, and the Pacific region
A few countries have reported HCB mean concentrations below 10 ng/g fat: Cambodia (Kunisue et al. 2004a), India (Devanathan et al. 2009), Korea (UNEP 2009), and the Philippines (Malarvannan et al. 2009). However, the majority of countries have reported mean values in the range 10-100 ng/g fat (Table 4). In the higher end of reported HCB, concentrations from the region include Australian levels between 370 and 460 ng/g fat (Behrooz et al. 2009;Quinsey et al. 1995), with levels up to 1500 ng/g fat, as well as Kazakhstan (Hooper et al. 1997) and Thailand (Stuetz et al. 2001), reporting concentrations above 100 ng/g fat (Table 4).
The reported levels of HCHs, both individual isomers as well as ∑HCH, are the highest in Asia, Australia, and the Pacific region, compared to the rest of the world, although there are a few studies reporting comparatively low mean concentrations, i.e., below 10 ng/g fat: Cambodia (Kunisue et al. 2004a), Kuwait (Saeed et al. 2000), the Philippines (Malarvannan et al. 2009), and Taiwan (Chao et al. 2006) but also studies from Japan (Nagayama et al. 2007a) and Australia (Khanjani and Sim 2006;Kunisue et al. 2004a). On the contrary, other studies from Japan and Australia report higher concentrations (Table 4). The majority of studies from Asia, Australia, and the Pacific region report ∑HCH concentrations in the range of hundreds of nanogram per gram fat, but a large number report concentrations in the range of thousands of nanogram per gram fat (Table 4). In India, there is a high HCH contamination according to the mothers' milk concentrations, which range from 120 to 22,000 ng/g fat ∑HCH, with several studies reporting values of thousands of nanogram per gram fat (Table 4). The highest concentrations of the HCH isomers in mothers' milk have been reported from India with means of 1800, 16,000, 1300, and 2300 ng/g fat for α-HCH, β-HCH, γ-HCH, and δ-HCH, respectively (Siddiqui et al. 2002).

Europe
In general, the HCB concentration in European mothers' milk is higher than the rest of the world, although the most extreme values of HCB in mothers' milk are not from Europe. Only two European countries, Croatia (Frkovic et al. 1996) and Finland (Shen et al. 2007), report mean HCB concentrations below 10 ng/g fat, while several Croatian studies report median concentrations above 10 ng/g fat (Krauthacker et al. 1998(Krauthacker et al. , 2009Romanic and Krauthacker 2006). The majority of the HCB concentrations reported from European countries are in the range of 10-100 ng/g fat, reported on either mean and median basis (Table 4). In Europe, it is primarily the Eastern countries that report highly elevated HCB levels in the analyzed mothers' milk, with the highest levels from the Czech Republic, median values up to 370 ng/g fat (Cajka and Hajslova 2003;Cerna et al. 2010). However, high levels are also reported in Spanish mothers' milk with medians of 630 and 910 ng/g fat (Ribas-Fito et al. 2005).
The HCH concentrations in mothers' milk from European mothers in general show significantly lower concentrations than milk from Asia, Australia, and the Pacific regions, but higher than concentrations reported from Africa and the Americas. The reported European levels of ∑HCH are for the most part homogenous, and the majority of mean and/or median concentrations are in the range 10-100 ng/g fat. However, there are a number of studies from Denmark (Shen et al. 2007), Finland (Shen et al. 2007), Romania (Covaci et al. 2001), and Russia (Polder et al. 1998;Tsydenova et al. 2007) that report ∑HCH concentrations of several hundreds of nanogram per gram fat (Table 4). The highest reported mean concentration is 1000 ng/g fat, in a study from Russia (Tsydenova et al. 2007).

The Americas
Overall, the HCB contamination seems to be lower in the Americas than any of the other regions with a higher portion of studies below 10 ng/g fat and no study reporting concentrations above 100 ng/g fat (Table 4). Antigua and Barbuda (UNEP 2009) along with a number of studies from the USA report levels below 10 ng/g fat. However, two studies report a mean concentration of 15 ng/g fat ) and a median concentration of 14 ng/g fat (Fitzgerald et al. 2001). The two highest mean concentrations of HCB were reported in studies of Mexican mothers' milk, reaching 92 and 53 ng/g fat (Rodas-Ortiz et al. 2008;Waliszewski et al. 1998), and the third highest was reported from Canada, 43 ng/ g fat (Newsome and Ryan 1999).
In the Americas, the ∑HCH concentrations are similar to the concentrations in Europe, albeit there are fewer reported observations. The two lowest concentrations are from Antigua and Barbuda (UNEP 2009) and Nicaragua (Romero et al. 2000), 5 and 7 ng/g fat, respectively. The majority of studies report values in the lower end of the range 10-100 ng/g (Table 4), although exceptions to this are reported concentrations in the range of hundreds of nanogram per gram from Brazil (Paumgartten et al. 2000) and Mexico (Elvia et al. 2000;Rodas-Ortiz et al. 2008;Waliszewski et al. 1998), with the highest mean concentration reported in mothers' milk from Mexico, at 750 ng/g fat.

Africa
Only ∑chlordanes from four countries on the African continent were reported. The highest concentrations were from Senegal, with a mean concentration of 11.7 ng/g fat (UNEP 2009).

Asia, Australia, and the Pacific region
Concentrations of oxychlordane vary greatly between countries, mostly between 0.5 and 10 ng/g fat. Extreme concentrations (140 and 150 ng/g fat) are reported from one Australian study (Quinsey et al. 1995), whereas the other studies from Australia report concentrations below 20 ng/g fat. Banned in most countries in 1997, chlordane was still allowed to be used as a termiticide in the Northern Territory (Australia) (UNEP Chemicals). Concentrations of αand γchlordane are only reported from a few countries, whereof extreme concentrations are reported from Jordan 460 and 590 ng/g fat, respectively (Nasir et al. 1998). The highest concentrations of ∑chlordanes were reported from Japan (Konishi et al. 2001;Kunisue et al. 2006), while studies from the rest of the countries in the region report concentrations generally below 10 ng/g fat.

Europe
Fourteen studies from Europe report mean concentrations of oxychlordane, most of them close to or below 5 ng/g fat, and the reported median concentrations are in general of the same magnitude. The highest values of oxychlordane in Europe are reported from Ukraine (16-22 ng/g fat) (Gladen et al. 1999(Gladen et al. , 2003. Only very few countries reported ∑chlordanes, whereof the highest values, 10-60 ng/g fat, are reported from Russia   (Polder et al. 1998(Polder et al. , 2008aTsydenova et al. 2007). One study reports concentrations below 4 ng/g fat from Western, Central, and Eastern Europe (UNEP 2011).

The Americas
A few studies from the American continent in general show higher concentration of oxychlordane (Johnson-Restrepo et al. 2007;Newsome et al. 1995;Newsome and Ryan 1999) compared to Europe. The highest concentrations are between 40 and 60 ng/g fat from Canada (Newsome and Ryan 1999) and Mexico (Elvia et al. 2000). A few studies reporting concentrations of αand γ-chlordane give values below 3 ng/g fat, except extreme concentrations reported from Mexico, 260 and 930 ng/g fat, respectively (Rodas-Ortiz et al. 2008).

Africa
Only two studies of PBDEs in mothers' milk from Africa were identified, and both studies confirm the occurrence of BDE-47 in mothers' milk at mean or median concentrations below 2 ng/g fat (Asante et al. 2011;Darnerud et al. 2011). BDE-209 was only reported in one study from Ghana (Asante et al. 2011).

Asia and the Pacific region
The concentrations of BDE-47 are rather uniform and low (below 2 ng/g fat) in mothers' milk in Asia and the Pacific region, even though a very large geographical area is covered. One study from Japan reports the highest level of BDE-47 outside the USA with a mean concentration of 19 ng/g fat (Akutsu and Hori 2004). From the Philippines, a study reports somewhat elevated levels, mean concentrations of 1.2 and 4.9 ng/g fat (13). Both low and high BDE-47 concentrations have been determined in samples from Turkey, with mean concentrations reaching 6.0 ng/g fat (Erdogrul et al. 2004;Ozcan et al. 2011). BDE-209 was reported in 6 out of 16 studies with similar levels (<1 ng/g fat) independent of the study (cf. Table 7). The highest BDE-209 concentrations are reported in mothers' milk from the Philippines and Vietnam with levels of around 2 ng/g fat (Malarvannan et al. 2009;Tue et al. 2010).
The reported ∑PBDEs in Asian mothers' milk confirm the observations of BDE-47 concentrations, with the highest concentrations from the Philippines, mean concentrations of 5-10 ng/g fat (Malarvannan et al. 2009;Sudaryanto et al. 2008b), as well as from Japan and Turkey with mean concentrations of 31 and 67 ng/g fat, respectively (Akutsu and Hori 2004;Ozcan et al. 2011). These concentrations are comparable with the levels reported in mothers' milk from the USA.

Europe
In general, the concentrations of BDE-47 in mothers' milk in Europe (approximately 1-2 ng/g fat) are higher compared to the levels in Asia and the Pacific region and Africa but lower than in the Americas. The lowest levels of BDE-47 in Europe are reported in samples from the Czech Republic, Germany, Italy, and Russia, with mean concentrations below 1 ng/g fat (Alivernini et al. 2011;Kazda et al. 2004;Polder et al. 2008a;Raab et al. 2008). The highest levels are reported in mothers' milk from the UK with median concentration of 2.7 ng/g fat (Kalantzi et al. 2004), and the remaining results from Europe are in between the mentioned BDE-47 concentrations (Table 7). Hence, the differences in the levels are rather small.
As few as 7 out of 25 of the European mothers' milk samples report BDE-209, with the highest concentrations in samples from France, 1.5 ng/g fat (Antignac et al. 2008;Athanasiadou and Bergman 2008). Since BDE-209 has a short half-life in humans, 14 days (Thuresson et al. 2006), the differences in concentrations of this PBDE congener vary greatly. Consequently, exposure levels of BDE-209 and nona-BDEs become uncertain when seen over time.
In Europe, the ∑PBDE concentrations are in general 3-4 ng/g fat, with a few exceptions ( Table 7). The highest ∑PBDE concentrations are reported in mothers' milk from the UK, with median levels of 6.3 ng/g fat, which is still three to five times lower than levels reported from the USA.

The Americas
Only studies from the USA could be found that report levels of PBDEs from the Americas and that met the criteria set for this review. In general, the reported levels of BDE-47 are much higher in the mothers' milk samples from the USA compared to the rest of the world. The concentrations are rather uniform, with mean values at 35-40 ng/g fat (Johnson-Restrepo et al. 2007;Schecter et al. 2003Schecter et al. , 2005Schecter et al. , 2010, but with levels reaching as high as 73 ng/g fat (Park et al. 2011). Also, the levels of BDE-209 are higher in the USA than the rest of the world (Table 7), which is indicating a higher prevalence of deca-BDE exposure.
Also, the concentrations of ∑PBDEs in the samples from the USA are overall similar, with means of 66-76 ng/g fat, and one median value of 51, reported in four different studies (Johnson-Restrepo et al. 2007;Schecter et al. 2003Schecter et al. , 2005Schecter et al. , 2010. However, one study reports a mean concentration as high as 130 ng/g fat (Park et al. 2011).
The results clearly show that US mothers' milk contains the highest concentrations of PBDEs. This is in line with any other exposure study from the USA, showing mothers and other individuals being subjected to environmental exposures of PBDEs that are the highest in the world.

Heptachlor
Heptachlor concentrations reported as cis-HCL-epoxide, HCL-epoxide, and heptachlor from 49 studies were selected and tabulated (Table 8). Concentrations of cis-HCL-epoxide are only reported from one or a few countries from each continent/region, and they range between 0.7 ng/g fat in China (Hedley et al. 2010) and 8 ng/g fat in Brazil (Paumgartten et al. 2000).

Africa
Heptachlor was only reported from four regions of the African continent with concentrations ranging between 0.9 and 2.25 ng/g fat (UNEP 2009(UNEP , 2011.

Asia, Australia, and the Pacific region
Fifteen studies report concentrations of HCL-epoxide. Most of these report concentrations below 10 ng/g fat, but higher concentrations are reported from Australia, 53 and 78 ng/g fat (Quinsey et al. 1995); Jordan, 190 ng/g fat (Nasir et al. 1998); Thailand, 60 ng/g fat (Stuetz et al. 2001); and Turkey, 61 ng/g fat (Cok et al. 2005). The highest reported concentration of heptachlor is from Jordan at 500 ng/g fat (Nasir et al. 1998).    BDE-47, BDE-209, and∑PBDE, in mothers' milk, as reported in studies from around the world, 1995-2011, are presented

Europe
Concentrations of HCL-epoxide reported from Europe are in most cases below or close to 3 ng/g fat, although higher concentrations are reported from Ukraine, 16-22 ng/g fat (Gladen et al. 1999(Gladen et al. , 2003, and the Netherlands, 30 ng/g fat (Albers et al. 1996).

The Americas
Concentrations of HCL-epoxide and heptachlor from the Americas are reported as less than 4 ng/g fat except for Mexico, 160 and 580 ng/g fat, respectively (Rodas-Ortiz et al. 2008).

Dieldrin, endrin, and aldrin
The OCPs discussed herein were regulated at an early stage in many countries. Despite of this, rather high concentrations of dieldrin are reported (Table 9).

Africa
No studies were retrieved that report aldrin and/or endrin in mothers' milk from any African nation. Dieldrin was reported in seven studies, three of which were part of the UNEP screening program with the lowest levels in mothers' milk from Ghana, Nigeria, and Senegal, with mean concentrations of 1.3-4.1 ng/g fat (UNEP 2009). When concentrations from more than one sampling site in the same country and study are reported, the concentrations are given as two values separated by a semicolon (two concentrations) or a dash (for more than two levels) a Recalculated from fresh weight, assuming 4 % fat content However, some reports indicate levels of up to 25-120 ng/g fat (Ennaceur et al. 2007(Ennaceur et al. , 2008Ntow et al. 2008).

Asia and the Pacific region
Although only few studies report aldrin concentrations in mothers' milk, the variation is great. The most comprehensive study originates from Australia showing a large national spatial distribution in a range 0.01-0.68 ng/g fat (Harden et al. 2007). Somewhat higher levels of aldrin are found in Kuwait and the highest concentration are from India and Jordan with mean concentrations up to 860 ng/g fat (Nasir et al. 1998;Siddiqui et al. 2002).
Dieldrin is more frequently reported in mothers' milk than aldrin and endrin and at rather high concentrations (Table 9). The levels of dieldrin are rarely above 100 ng/g fat, with the exceptions of one study from Jordan, which reports a mean concentration of 1400 ng/g fat (Nasir et al. 1998), and one study from China, 9300-10,000 ng/g fat (Wang et al. 2008). Only two studies have been retrieved reporting on endrin in mothers' milk, one from Jordan and one from Kuwait (Nasir et al. 1998;Saeed et al. 2000).

Europe
A study from Croatia reports aldrin and endrin in their samples, with mean concentrations of 1.3 and 2.0 ng/g fat, respectively (Frkovic et al. 1996). Again, the occurrence of dieldrin in mothers' milk is frequently reported from European countries in low concentrations, for example in samples from Germany and Croatia, with mean concentrations below 3.8 ng/g fat (Frkovic et al. 1996;Zietz et al. 2008). However, other studies from Germany show higher levels, with mean concentrations of 4 and 14 ng/g fat (Raab et al. 2008;Schlaud et al. 1995). Two reports investigating dieldrin levels in mothers' milk, both in Denmark and Finland, show similar results (Shen et al. 2007(Shen et al. , 2008 and likewise from the WHO-UNEP monitoring program (UNEP 2011). The highest levels are reported from the UK with a mean concentration of 48 ng/g fat (Harris et al. 1999) and the Netherlands with a median concentration of 50 ng/g fat (Albers et al. 1996).

The Americas
In the Americas, one study has reported levels of aldrin from a known pest-controlled area on the Yucatán peninsula in Mexico with a mean concentration of 280 ng/g fat (Rodas-Ortiz et al. 2008). In the lower end of reported levels of dieldrin in mothers' milk are samples from the Americas, Antigua and Barbuda, Chile, and Uruguay, with mean concentrations of 2.6, 5.0, and 4.9 ng/g fat (UNEP 2009), respectively, as well as from the Group of Latin American and Caribbean countries, with a median concentration of 4.9 ng/g fat (UNEP 2011). The highest levels of dieldrin in mothers' milk in the Americas are found in samples from a know pest-controlled area in on the Yucatán peninsula in Mexico with a mean concentration of 300 ng/g fat (Rodas-Ortiz et al. 2008). Endrin could only be found in a sample from Nicaragua (Romero et al. 2000) and from the abovementioned pestcontrolled area on the Yucatán peninsula in Mexico (Rodas-Ortiz et al. 2008).
Pentachlorobenzene, toxaphene, and mirex Pentachlorobenzene (PCBz), mirex, and toxaphene concentrations in mothers' milk are only reported in a few studies (Table 10). Mirex levels are reported as 16, 20, and 68 ng/g fat in mothers' milk from Germany, Russia, and Canada, respectively (Newsome and Ryan 1999;Polder et al. 2008a;Skopp et al. 2002). Both PCBz and mirex are reported in around 1 ng/ g fat in most mothers' milk samples, with a few exceptions (Table 10). The most profound exception is mirex found in a concentration of 200 ng/g fat in mothers' milk from Mexico (Rodas-Ortiz et al. 2008).
It is notable that so few reports have been published on these POPs. Therefore, any assessments of spatial differences are impossible.

HBCDD and PBB
Concentrations of hexabromocyclododecane (HBCDD) and polybrominated biphenyl (PBB) reported in mothers' milk are presented in Table 11 on a fat weight basis. From isomer-specific information, it is clear that α-HBCDD is the most abundant isomer of environmental HBCDDs (Eljarrat et al. 2009;Lankova et al. 2013). The spatial distribution of HBCDD is illustrated in Fig. 9.

Africa
Only two studies of HBCDDs in mothers' milk from all of Africa were retrieved, both showing subnanogram per gram fat concentrations in the mothers' milk analyzed (Asante et al. 2011;Darnerud et al. 2011). No study was found reporting on PBBs in mothers' milk from any African nation.

Asia and the Pacific region
The lowest concentrations of ∑HBCDDs reported in mothers' milk from Asia come from the Philippines and Vietnam with mean and median concentrations below 1 ng/g fat (Malarvannan et al. 2009;Tue et al. 2010). A study from China shows a greater range in ∑HBCDD concentrations, with means of 0.33-2.8 ng/g fat (Shi et al. 2009). In Japanese mothers' milk, ∑HBCDDs levels of 1.4 ng/g fat (Kakimoto et al. 2008) are reported. No study was found reporting on PBBs in mothers' milk from any nation in Asia or the Pacific region.

Europe
The levels of ∑HBCDDs in Europe are quite uniform and low (Table 11), i.e., <1 ng/g fat. However, the highest concentrations on a global scale are those reported from a Spanish study, with a mean concentration of 47 ng/g fat, in mothers' milk from a population living close to a textile processing plant (Eljarrat et al. 2009). The ∑HBCDD concentrations in Swedish samples reported median concentrations within the range of 0.3-0.4 ng/g fat . Somewhat higher concentrations, and comparable with the Japanese levels, are reported from Norway, with mean concentrations of 1.7 ng/g fat . The ∑HBCDDs in mothers' milk from the UK report a mean concentration of 6.0 ng/g fat (Abdallah and Harrad 2011), which is the highest background level worldwide, apart from the Spanish Bhot spot^samples. One study reports PBB concentrations from two countries, Denmark and Finland, both indicating mean concentrations below 0.3 ng/g fat (Shen et al. 2008).

The Americas
We have not found any study with our search criteria which has reported the presence of HBCDDs or PBBs in mothers' milk from any nation in the Americas.

PFOS
Thirteen studies report PFOS concentrations in mothers' milk from a total of 13 countries in Asia and Europe and one study from the USA (Table 12). The concentration of PFOS ranges between 39 and 200 pg/mL, with Hungary as the only exception, 330 pg/mL (Volkel et al. 2008). Due to the limited data, it is not possible to draw any conclusions regarding spatial exposure differences. The spatial distribution of PFOS is illustrated in the diagram in Fig. 10.

Endosulfan
Only some scattered endosulfan mothers' milk data are available (Table 13). The sum concentrations from India and Turkey are exceptionally high, i.e., concentrations above 1000 ng/g fat (Cok et al. 2011;Sanghi et al. 2003). These high levels are comparable to some data on DDT in mothers' milk (cf. Table 2) and are likely due to the very recent use of endosulfan in these countries.

Chlordecone
No studies were found with our search criteria reporting chlordecone concentrations in mothers' milk.

SCCPs
No studies were found with our search criteria reporting SCCP concentrations in mothers' milk, even though there is an When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a dash B-^(for more than two concentrations) a Recalculated from fresh weight, assuming 4 % fat content b Group of Latin America and Caribbean countries   Kostyniak et al. (1999) When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) a Recalculated from fresh weight, assuming 4 % fat content When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a dash B-^(for more than two concentrations) a Recalculated from fresh weight, assuming 4 % fat content Fig. 9 Graphical presentation of HBCDD concentrations in mothers' milk from countries worldwide When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a dash B-^(for more than two concentrations) a Quantified using BDE-139 as surrogate standard agency report on SCCPs in Swedish mothers' milk (Darnerud et al. 2012) indicating their presence in this matrix.

Global distribution trends
The data collected and compiled within the study indicates that there is indeed a difference in the distribution and exposure to POPs which is dependent on where in the world one resides. These conclusions are more easily made when comparing the different spatial distribution diagrams, e.g., Fig. 3. In general, it was found that DDT/DDE pesticides were reported in higher concentrations in mothers' milk from the regions of Africa, Asia, and Central America, with a propensity for agricultural economies and lower degree of   (2007) Sweden 120 Kärrman et al. (2006) Sweden 75 Sundström et al. (2011) The Americas USA 130 110 Tao et al. (2008a) When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a dash B-^(for more than two concentrations) industrialization. On the other hand, PCBs and dioxins were found to be reported to a higher degree in more industrialized regions, such as parts of Asia, Europe, and North America. A good example of this can be seen by comparing Figs. 3 and 4a-c, where the DDT/DDE concentrations clearly are lower, in general, in Europe and North America compared to the rest of the world (Fig. 3). Similarly, it can be seen from Fig. 4a-c that the PCB concentrations are higher in industrialized regions compared to the rest of the world. This pattern is also observed for HCHs, although there are a few observations of high concentrations in mothers' milk samples from Eastern Europe, i.e., Russia, Romania, and Ukraine ( Fig. 6a-d).This pattern is not surprising since PCBs and dioxins, not shown in spatial distribution diagrams, are related to a degree of industrialization, either as chemical products or impurities there within. DDT as well as HCHs has been used as pesticides in SC and has been used in the equatorial and subequatorial regions, which in general are less industrialized and more dependent on agriculture. Furthermore, it is clear that mothers' milk from the USA contains more PBDEs than the rest of the world (Fig. 7). BDE-47 is a biomarker of PBDE exposure and the lowest reported concentration is around five times as big as the highest concentration in a sample from outside of the USA. This can be explained by the stricter flame retardant policy enforced within the USA, a policy which calls for a greater use of flame retardant substances such as PBDEs primarily in upholstery (GSPI 2013;State of California 2000. A new fire safety regulation has recently been adopted, January 1st 2014, which does not call for the use of flame retardant chemicals and perhaps this will lead to a decrease of PBDEs in mothers' milk in the USA (GSPI 2013;State of California 2000. For the substances not mentioned, there are no observed, clear spatial distribution trends that can be explained by traditional/ historical use of the substances in question. This could be since there are too few reported concentrations available or that the differences in use or emissions are too small to observe.

Temporal trends
Two distinct objectives can be identified concerning temporal trend monitoring of contaminants. One is to quantitatively estimate the rate of changes in contaminant concentration, e.g., as a change in percent per year or as half-or doubling time in number of years. An example of this could be to estimate the response of measures taken to reduce the discharges of various contaminants. Dissimilarities in comparisons between the rate of change in contaminant concentration in mothers' milk and other environmental biological matrices (e.g., fish) can give information about the exposure patterns, i.e., if the mothers are exposed to contaminants not only from local food but also from imported food and the indoor environment, including a variety of man-made technical products. Another objective of temporal trend monitoring is to study emerging new substances and to detect renewed use of banned contaminants. In order to estimate the rate quantitatively with a high statistical power, it is essential to keep the random variation between years as low as possible. Compared to other matrices, When concentrations from more than one sampling site in the same country and study have been reported, the concentrations are given as two values separated by a semicolon B;^(two concentrations) or a dash B-^(for more than two concentrations) a Recalculated from fresh weight, assuming 4 % fat content b Nonspecified isomer/s mothers' milk seems to show a relatively low random variation (UNEP 2004). Inclusion of a time series in this review article requires a minimum of five reported data points. Only approximately half of the substance groups from only two countries, Sweden and Japan, fulfilled the described criteria. The temporal trends present data from 1972 to 2011.
In the graphs below, log-linear regression and a smoother test have been carried out. The smoother test checks if the smoother explains significantly more of the variation in concentration, than the regression line (Nicholson et al. 1998).The regression line and/or the smoother are plotted when significant (α=0.05).
DDT and DDT-related compounds In Fig. 11, the concentrations of DDE (ng/g fat) in the samples from Japan, 1972(Konishi et al. 2001), and Sweden, 1972(Athanasiadou and Bergman 2008Bergman et al. 2010;Lundén and Norén 1998), show significant decreasing trends over the whole time period of −9.1 % (p<0.001) and −8.5 % (p<0.001), respectively. The Japanese samples also show significant decreasing concentrations for the last 10 years of −13 % (p<0.001), while no trend is indicated in the Swedish samples for the last 10 years (estimated during a decade later than for the Japanese samples). The temporal trends for DDT in mothers' milk from Japan and Sweden are of similar magnitude as for DDE. In addition, the trends observed in Swedish mothers' milk for DDE coincide with the trends seen in Swedish freshwater (Nyberg et al. 2011) and marine (Bignert et al. 2012) biota.
The ratio of DDT/DDE (Fig. 12) shows similar log-linear trends in the samples from Japan, 1972(Konishi et al. 2001), and Sweden, 1972(Athanasiadou and Bergman 2008Bergman et al. 2010;Lundén and Norén 1998). The Japanese samples also show a significant nonlinear trend for the ratio of DDT/DDE, which might indicate that a new release of DDT has occurred during the monitoring period.
Only one temporal trend study on congener basis was found for PCBs within this review (from Sweden). In Fig. 14, the temporal trend for CB-153 (ng/g fat) in the Swedish samples, 1972-2010 (Bergman et al. 2010;Lundén and Norén 1998), is presented. CB-153 shows a significant decreasing trend over the whole time period and for the last 10 years of −4.9 % (p<0.001) and −5.9 % (p<0.042), respectively. Fig. 11 Temporal trends of DDE (ng/g fat) from Japan (Konishi et al. 2001) and Sweden (Lundén and Norén 1998;Athanasiadou and Bergman 2008;Bergman et al. 2010) Fig. 12 Temporal trends of the ratio between DDT and DDE (ng/g fat) from Japan (Konishi et al. 2001) and Sweden (Lundén and Norén 1998;Athanasiadou and Bergman 2008;Bergman et al. 2010) HCB and HCHs In Fig. 15, the concentrations of β-HCH (ng/g fat) in the samples from Japan, 1972(Konishi et al. 2001, show a significant decreasing trend over the whole time period as well as for the last 10 years of −12 % (p<0.001) and −11 % (p<0.001), respectively.

PCDDs, PCDFs, and DL-PCBs
In Fig. 16, the concentrations of PCDD (pg WHO 2005 -TEQ/g fat) from Stockholm, 1972-1997(Norén and Meironyte 2000 and 1972-2011 (Fång et al. 2013), show significant decreasing trends over the whole time period of −3.6 % (p<0.005) and −6.0 % (p<0.001), respectively. However, for the last  (Konishi et al. 2001) and Sweden (Lundén and Norén 1998;Athanasiadou and Bergman 2008;Bergman et al. 2010) Fig. 15 Temporal trends of β-HCH (ng/g fat) from Japan (Konishi et al. 2001) 10 years, a significant decreasing trend of −10 % (p<0.001) is only seen for the time series from 1972 to 2011 (Fång et al. 2013), covering the last decade in contrast to the study from 1972 to 1997 (Norén and Meironyte 2000), which might have too few samples during the last 10 years to detect a trend. It should be noted that the two studies by Norén and Meironyte and the study by Fång et al. are analyzing the same pooled mothers' milk sample during 1972-1997.
In Fig. 17, the concentrations of PCDF (pg WHO 2005 -TEQ/ g fat) from Stockholm, 1972-1997(Konishi et al. 2001Norén and Meironyte 2000) and 1972(Fång et al. 2013, show significant decreasing trends over the whole time period of −5.2 % (p<0.003) and −6.2 % (p<0.001), respectively. However, for the last 10 years, a significant decreasing trend of −7.3 % (p<0.001) is only seen for the time series from 1972 to 2011 (Fång et al. 2013), covering the last decade in contrast to the study from 1972 to 1997 (Norén and Meironyte 2000), which might have too few samples during the last 10 years (only 4) to detect a trend.
The trends observed in Swedish mothers' milk for DL-PCBs during the whole time period coincide with the trends seen in Swedish freshwater (Nyberg et al. 2011) and marine (Bignert et al. 2012) biota for the dioxin-like PCB congener CB-118 from the end of the 1970s to the beginning of the 1990s. However, during the last decade, the levels are decreasing at a higher rate in Swedish human milk than in marine and freshwater biota from Sweden.

PBDEs
In Fig. 19, the concentrations of BDE-47 (ng/g fat) in the samples from Japan, 1977(Akutsu et al. 2003, indicate an increasing trend over the whole time period of 9.1 % (p < 0.081). In contrast, a decreasing trend of −5.7 % (p < 0.093) is indicated in the Swedish samples from Stockholm (Athanasiadou and Bergman 2008;Bergman et al. 2010;Fängström et al. 2008) for the last 10 years of the study. This decreasing trend of BDE-47 coincides with the trends seen in Swedish marine (Bignert et al. 2012) and freshwater (Nyberg et al. 2011) biota over the last decade. PentaBDE was first phased out voluntarily by the industry in Germany in 1986 and end in Sweden in 1999 (Alcock and Busby 2006). Subsequently, BDE-47 was partially banned within the EU countries in 2004 and the declining concentrations in human milk and biota could to some extent be explained by these events. No trend could be observed in the Swedish time trend from Uppsala during 1996-2001 (Lind et al. 2003).  , 1986, (Konishi et al. 2001, show significant decreasing trends over the whole time period and for the last 10 years of −9.7 % (p<0.001) and −4.9 % (p<0.049), respectively.

Dieldrin, endrin, and aldrin
In Fig. 21, the concentrations of dieldrin (ng/g fat) in the samples from Japan, 1972-1982(Konishi et al. 2001, show a significant decreasing trend over the whole time period of −14 % (p<0.001).

HBCDD
The concentrations in Fig. 22 of HBCDD (ng/g fat) in the samples from Japan, 1987(Kakimoto et al. 2008) and Sweden, 1987(Athanasiadou and Bergman 2008Bergman et al. 2010;Fängström et al. 2008), show increasing trends over the whole time period of 5.4 % per year (p<0.061) and 7.6 % per year (p<0.001), respectively. The increasing trend of HBCDD seen in the Swedish milk samples coincides with trends in Swedish marine biota from the Baltic Sea (Bignert et al. 2012). HBCDD is still in use within the EU but is listed in REACHs authorization list as substance of very high concern (SVHC) and, since November 2014, included in the SC. Fig. 19 Temporal trends of BDE-47 (ng/g fat) from Japan (Akutsu et al. 2003) and from Stockholm (Athanasiadou and Bergman 2008;Fängström et al. 2008;Bergman et al. 2010) and Uppsala (Lind et al. 2003), Sweden Fig. 20 Temporal trend of heptachlorepoxide (ng/g fat) from Japan (Konishi et al. 2001) Fig. 21 Temporal trend of dieldrin (ng/g fat) from Japan (Konishi et al. 2001) PFOS In Fig. 23, the concentrations of PFOS (pg/ml) in the samples from Stockholm, 1972(Sundström et al. 2011, show a significant increasing trend over the whole time period of 3.3 % (p<0.012). In contrast, a significant decreasing trend of −11 % (p<0.002) is observed for the last 10 years. In the study with samples from Uppsala, Gothenburg, Lund, and Lycksele, 1996(Kärrman et al. 2007), a decreasing trend is indicated (p<0.059) for the whole time period. The trends observed in Swedish mothers' milk for PFOS coincide with the trends seen in Swedish marine (Bignert et al. 2012) and freshwater (Nyberg et al. 2011) biota over the last decade.

Critical remarks and conclusions
Some of the legacy POPs are the most well-researched environment pollutants among all. This is due to their global distribution and occurrence in humans and wildlife, being classified as CMRs, having endocrine disruption effects and/or having other toxic effects. Some early efforts focused on the transfer of these chemically stable and bioaccumulative compounds to nursing children. The transfer of POPs via mothers' milk initiated a still ongoing debate on risks for the newborn babies. In contrast to the strict recommendations to nursing mothers to avoid smoking and drinking alcohol, it is not possible to change the mothers' body burdens of POPs as their levels of these compounds have been built up over the mothers' whole lifetime. However, in some countries, there are at least dietary recommendations on how to limit the intake of POPs from some major food sources, in particular fatty fish. These recommendations are primarily targeting young women and women in child-bearing ages. Due to the many positive effects of nursing, WHO recommends mothers to breast feed their newborns for a minimum of 6 months (WHO and UNICEF 2014).
WHO also initiated a monitoring program for POPs in mothers ' milk in 1976' milk in (WHO 2009), but at this point, researchers had already started to do exposure studies of nursing infants, as reviewed by Norén and Meironyte (2000). In the present review, it becomes clear how abundant mothers' milk is a matrix for POP analysis, although only a limited number of POPs are assessed, i.e., 7 of the 24 (HCHs counted individually) POPs in Table 1 contribute with 80 % of the exposure studies discussed herein. There are no studies covering all, or even a majority of the POPs in the same study. The most comprehensive results for POP exposure analyses are when the same milk samples are utilized for assessments of as many POPs as possible. In reality, it is much more common with scattered studies globally, regionally and country-wise. This means that the cohorts are defined by different means and the objectives vary. This leads to the conclusion that there are very few studies that allow reliable comparisons, e.g., the temporal trend studies from Japan and Sweden. However, the temporal trend studies from Japan and Sweden presented did not cover POP analyses in an optimal manner, i.e., several of the POPs were not included in these studies. Another issue regarding Fig. 22 Temporal trends of HBCDD (ng/g fat) from Japan (Kakimoto et al. 2008) and Sweden (Athanasiadou and Bergman 2008;Fängström et al. 2008;Bergman et al. 2010) Fig. 23 Two temporal trends of PFOS (pg/mL) from Stockholm (Sundström et al. 2011) and Uppsala, Gothenburg, Lund and Lycksele (Kärrman et al. 2007), Sweden comprehensive time trend analysis is the need to establish environmental specimen banks whose goal is to collect and store environmental relevant samples, including mothers' milk, in a structured manner. In time, samples collected will allow a high qualitative time trend analysis on current POPs and currently unknown pollutants. Further, this is only the start of the problem comparing POPs in mothers' milk from around the world. Below, we list and shortly discuss some of the major shortcomings in some studies which, if rectified, would allow proper comparisons of POPs in mothers' milk.
Reporting base for POP concentrations in mothers' milk This is relevant for all POPs that are produced, used, and/or occur as mixtures of halogenated homologues and isomers-congeners. These POPs are PCBs, PCDFs, PCDDs, PBDEs, PBBs, chlordanes, toxaphen, and CPs. HCHs may be included even though only isomers of hexachlorinated cyclohexane is included. Similarly, HBCDD has three diastereomers (α-, β-, and γ-HBCDD), which are commonly discussed and often reported individually. DDTs and endosulfan, on the other hand, are both produced as two main isomers. However, both 4,4′-DDT and 2,4′-DDT are transformed to and occur in the environment, including mothers' milk, as the corresponding DDD and DDE compounds. This has led to a similar handling of the DDTs as of true congeners of, e.g., PCBs. For all of these compounds/compound classes, it is common to report concentration sums (e.g., sPCB or ∑PCB), sometimes indicating how many PCB congeners are included in the sum and presented as ∑PCB (7), indicating that seven congeners were included. However, highly variable sums are reported for the POPs. For PCBs, for example, a number of different sum values have been found in the review including 3,4,6,7,8,12,15,16,19,32, and 35 PCB congeners. Sometimes, the reported sum concentrations are referred to as Btotal PCBsŵ ithout any further specification. However, it is clear that it is still a summation of a defined number of PCB congeners, i.e., those quantified. To further complicate the issue, PCBs can also be reported as DL-PCBs and non-DL-PCBs or as non-ortho-PCBs, mono-ortho-PCBs, and di-ortho-PCBs. Hence, it is realistically not possible to compare ∑PCB concentrations unless they are reported in a similar manner. In this review article, we have used ∑PCB data but these are the weakest, the most unreliable, while the ∑PCB(6) and CB-153 concentration data are comparable if reported similarly by other means (mean, median, fat or fresh weight basis, cf. below).
The concentration reports for PBDEs follow a similar pattern as for the PCBs, often reporting ∑PBDEs but with differences in number and identity of PBDE congeners. Still, we have chosen to show ∑PBDE levels in Table 7, but for the purpose of comparison, BDE-47 and BDE-209 are recommended for use. The reporting variability applies for all POPs that have isomers and congeners and for which individual reference standards for analysis are available. The latter allows congener-, or isomer-, specific analysis, but the complexity of data generated call for simplifications.
Two classes of POPs, DDTs and dioxins, require further attention regarding concentration reports. The DDTs are commonly reported as the sum of 4,4′-DDT and its metabolites, 4, 4′-DDE (major transformation product) and 4,4′-DDD. However, the ∑DDT may also include the true isomers, 2,4′-DDT and 4,4′-DDT or even the two isomers plus their metabolites. Concentrations of the abundant compounds 4,4′-DDT, 4,4′-DDE, and 4,4′-DDD are however quite frequently reported. This allows proper comparisons of the concentrations and to calculate comparable ratios of DDT versus either DDE or DDE and DDD, which can be done using different methods. In Table 2, we have applied the ratio 4,4′-DDT/4,4′-DDE due to the abundance of individual concentration data for these two pollutants.
The reporting of dioxin concentrations is another problem, even though individual concentration data are generated from the chemical analyses. The actual concentrations of dioxins (PCDDs, PCDFs, and DL-PCBs) are commonly reported as sum of their TEQs, after recalculation of the concentrations utilizing the TEF values-the most commonly used nowadays are the WHO TEFs from 1998 and 2005 (Van den Berg et al. 1998Berg et al. , 2006. However, if the ∑PCDDs, ∑PCDFs, and ∑DL-PCBs, or worse ∑dioxins, are presented, it is strongly limiting any comparisons, unless the actual concentration data of the individual congeners are presented as well.
In conclusion, POPs in mothers' milk, as well as in other matrices, must be reported on a congener-or isomer-specific basis to promote proper trend studies. Unfortunately, this is not done in a structured manner today, which is strongly hampering the comparisons in the present data set on POPs in mothers' milk.
Concentration base for POPs in mothers' milk The most common way of reporting concentrations of POPs in human matrices and wildlife is on weight basis, i.e., microgram, nanogram, picogram per weight of the matrix (gram or kilogram), or on volume (e.g., mL) of the matrix, which relates to a fresh weight, volume, or extracted lipids/fats. The concentrations are rarely reported on a molar base (e.g., nmol/g or pmol/g). Despite the fact that this is the correct way of assessing exposures used for risk assessments and for correct comparisons, this means reporting is only found in very few studies of POPs in mothers' milk. This problem is particularly evident for the polybrominated pollutants, where the molecular weight varies greatly between the different congeners. The implications are shown in Fig. 24, where it is clear that the Bnumber of molecules^(molar base) of CB-153 is 2.5 times the number of BDE-209, although the masses of the two are equal. Since there is such an extensive span in molecular masses among the POPs, it is crucial that this must be considered for future studies/reports on POPs.
Another obstacle for comparisons of POPs in mothers' milk is how to handle more water-soluble POPs, e.g., PFOS, other perfluorinated compounds, and organohalogen phenols. PFOS is commonly reported on a fresh weight basis, i.e., picograms per milliliter, while other POPs are reported on a fat weight basis (e.g., ng/g fat). To allow conversions, it is necessary to know the fat content of the matrix analyzed.
In conclusion, a change from weight-to molar-based reporting on POPs is needed in order to avoid unnecessary errors in exposure assessments and to allow accurate trend analyses. Furthermore, the fresh weight of mothers' milk samples as well as concentrations on fat weight basis, and vice versa, should be mandatory since this would facilitate comparisons between studies.
Reporting The reported measure of central tendency of the concentrations of POPs is not consistent, i.e., the arithmetic mean, the geometric mean, or median values, and sometimes, only a range is given without a mean or a median. This hampers the possibilities to compare data. However, if a lognormal distribution can be assumed, which is common for contaminant data (see, e.g., Esmen and Hammad 1977), the geometric mean and the median can be considered equal. The arithmetic mean is, with the same assumption of log-normality, always higher than the median. Some guidance of how to adjust for this bias is given by Caudill (2010Caudill ( , 2012. In conclusion, commonly agreed upon guidelines for this part in reporting exposure data are also required.
Overall conclusions Unfortunately, reporting of POPs in mothers' milk differs greatly between the studies. This has limited the comparisons for both spatial and temporal trend studies.