Skip to main content
PMC home page
PLOS ONE logoLink to PLOS ONE
. 2021 Feb 22;16(2):e0247138. doi: 10.1371/journal.pone.0247138

Preterm disparities between foreign and Swedish born mothers depend on the method used to estimate gestational age. A Swedish population-based register study

Sol P Juárez 1,2,*,#, Marcelo L Urquia 3,4,#, Eleonora Mussino 5, Can Liu 1,2,6, Yao Qiao 4,¤, Anders Hjern 1,2,6
Editor: Kelli K Ryckman7
PMCID: PMC7899337  PMID: 33617565

Abstract

This study aims to examine whether disparities in gestational age outcomes between foreign and Swedish-born mothers are contingent on the measure used to estimate gestational age and, if so, to identify which maternal factors are associated with the discrepancy. Using population register data, we studied all singleton live births in Sweden from 1992–2012 (n = 1,317,265). Multinomial logistic regression was performed to compare gestational age outcomes classified into very (<32 weeks) and late preterm (32–36 weeks), term and post-term derived from the last menstrual period (LMP) and ultrasound estimates in foreign- and Swedish-born women. Compared to Swedish-born women, foreign-born women had similar odds of very preterm birth (OR: 0.98 [95% CI: 0.98, 1.01]) and lower odds of moderately preterm birth (OR: 0.95 [95% CI: 0.92, 0.98]) based on ultrasound, while higher risks based on LMP (respectively, OR: 1.10 [95% CI: 1.07, 1.14] and 1.09 [95% CI: 1.06, 1.13]). Conclusions on disparities in gestational age-related outcomes by mother’s country of origin depend on the method used to estimate gestational age. Except for very preterm, foreign-born women had a health advantage when gestational age is based on ultrasound, but a health disadvantage when based on LMP. Studies assessing disparities in very preterm birth by migration status are not affected by the estimation method but caution should be taken when interpreting disparities in moderately preterm and preterm birth rates.

Introduction

Preterm birth is an important indicator for perinatal surveillance; it accounts for more than 80% of neonatal deaths [1] and is a risk factor for a number of adverse outcomes across the life course [2, 3]. Large differences in preterm rates exist between immigrant populations residing in high-income countries [4, 5]. Although heterogeneity has been linked to study designs and definitions of migrants and reference groups [5], the extent to which preterm differences between groups and contexts is affected by varying methods of estimation of the gestational age is still unknown.

Ultrasound (UL) assessment and maternal recall of the date of the last menstrual period (LMP) are two commonly used methods for determining gestational age. The information available in the registers varies between countries and over time. For example, in the United States, vital statistics information on gestational age derived from LMP was the only available measure until 1989, when it was replaced by a clinical estimate definition (though not reported from California). The US birth certification was revised in 2003, and the best obstetric estimate of gestational age was incorporated [6], including ultrasound estimation but excluding postnatal assessment [7]. These changes can affect regional and international comparisons as well as national trends. International studies suggest that the progressive use of early-ultrasound estimates could partially explain the stabilization and rise in preterm birth (accompanied by the fall in post-term births) in some countries in the world [8].

Although UL is usually considered to be the gold standard for gestational age estimation [7], it is known that the accuracy of the estimation of gestational age is strongly dependent on when the ultrasound is performed, which in turn is conditioned on when the mother has her first prenatal care visit [9]. For this reason, countries such as Sweden use the ‘best estimate method’ for clinical and public health surveillance, although studies are needed to evaluate the impact of this strategy [10]. Reliable information on gestational age is important in general but particularly when comparing outcomes between foreign- and native-born women, as any systematic difference in the estimation of gestational age between these groups could induce a biased association with no biological basis [11]. Furthermore, if the factors proposed to explain health differences in preterm birth between foreign and native mothers (e.g., socioeconomic variables) are also associated with the misclassification of the outcome, it would be unclear to what extent adjusting for such variables might explain health differences or reduce measurement error.

This study aims to examine whether disparities in gestational age outcomes between migrants and natives are contingent on the measure used to estimate gestational age and, if so, to identify which birth and maternal factors are associated with the discrepancy between LMP and UL. This study will contribute to a better understanding of the consequences of using different methods of estimation of gestational age for preterm and post-term surveillance among immigrants.

Materials and methods

Data

In this population-based register study, we used the Swedish Medical Birth Register (MBR), which contains maternal and child information on almost all births in Sweden [12]. Although the MBR started in 1973, we had access to data for the period 1992–2012. The MBR was linked to the Multi-Generational Register [13] to identify adopted mothers, and the Longitudinal Integration Database for Health Insurance and Labour Market Studies (LISA) [14] to obtain socio-economic indicators via an exclusive identification number [15].

Study population

From a total of 2,239,806 observations, we excluded: 64,466 multiple births; 55,569 births with major congenital malformations; 1,115 adoptee women; and 55,788 women from other countries represented by less than 2,000 births. To ensure that we compared the same individuals throughout the study, we also excluded 723,253 observations with missing information in any of the following covariates: birthweight, maternal age, height, body mass index (BMI), smoking, country of birth, income, marital status and newborn’s sex. Thereafter we excluded observations with biologically implausible combinations of birthweight and gestational age (n = 24,696) [16]. All subjects in the final sample (n = 1,317,265) had complete information on all gestational age measures available (Fig 1).

Fig 1. Selection flow and study population.

Fig 1

Open in a new tab

We also performed sensitivity analyses in a sub-sample of 765,030 uncomplicated pregnancies with spontaneous onset of delivery, in order to improve the accuracy of the ultrasound-based measures. Given that the estimation of gestational age using ultrasound depends on the size of the fetus, the accuracy of the measure is sensitive to factors that could influence fetal growth variations [17]. Hence, gestational age is systematically underestimated in smaller fetuses compared to normal-sized fetuses, while the opposite is true for those with rapid fetal growth [18–20]. Prior studies have indicated that both fixed (newborn’s sex, maternal height) and modifiable (smoking, obesity) factors influence gestational age estimation [21, 22], and are thus implicated in the misclassification of preterm and post-term births [21]. For this reason, we conducted a sensitivity analysis in a sample of uncomplicated mothers with the aim of removing the potential effects of factors known to compromise the accuracy of the ultrasound-based estimates of gestational age. We defined uncomplicated pregnancies following the criteria used by INTERGROWTH-21st [23]. We excluded 765,030 observations corresponding to women with one or more of these characteristics: being younger than 18 or older than 34 at childbirth, with a pre-pregnancy BMI below 18.5 or above 30kg/m2, a height below 153cm, smokers in early pregnancy who suffered from diabetes and hypertension before or during pregnancy and whose pregnancies ended in a cesarean section (Fig 1).

Outcome variables

Two variables of gestational age were compared in this study: 1) date of the mother’s last menstrual period (LMP-based); 2) expected date of birth based on ultrasound assessment (UL-based), which has replaced LMP dating for clinical decision making since 1990 in Sweden. Each of the available gestational age variables was divided into: 1) preterm (<37 weeks), term (37–41 weeks) and post-term (≥42 weeks); 2) very preterm (<32 weeks), moderately preterm (32–36 weeks), term (37–41) and post-term birth (≥42 weeks).

Maternal birthplace

Migrants were defined according to their country of birth and grouped into seven geographical regions to warrant sufficient statistical power: Nordic countries, Europe & United States, Eastern Europe & Russia, Middle East, Africa, Asia and Latin America. The largest countries for each region were also selected (>2,000 observations) for specific country of birth analyses, which were conducted to assess heterogeneity within regional groupings.

Covariates

Available covariates were: year of birth (categorized as: 1992–1995, 1996–1999, 2000–2004, 2005–2008 and 2009–2012); newborn’s sex, parity (primiparous and multiparous); maternal height (linear and quadratic); maternal age (24 years or younger, 25–29, 30–34 and 35 years of age and above); maternal BMI in early pregnancy, categorized as normal weight (18.50–24.99 kg /m2), underweight (<18.50 kg/m2), overweight (25.00–29.99 kg/m2), and obese (>30.00 kg/m2); maternal self-reported smoking in early pregnancy, between gestational weeks 10 and 12 (non-smokers, 1 to 9 cigarettes/day, 10 or more cigarettes/day); mother’s educational attainment (low, middle and high); household disposable income (including salary, parental leave and other benefits, reported in quintiles); and maternal family situation (cohabiting with the father, single or other situation). All socio-demographic variables were considered for the year before birth.

Statistical analysis

We first computed preterm and post-term birth by UL and LMP and estimated p-values using McNemar test for paired data to compare preterm and post-term rates between methods of gestational age estimation (UL-LMP) by mother’s birthplace. Additionally, we estimated the Kappa coefficient (and their 95% Confidence Intervals, 95% CI) to assess the overall level of agreement between LMP and ultrasound categorized into preterm, term and post-term. The level of agreement in the Kappa coefficient is interpreted as follows: 0.20 (poor), 0.21–0.40 (fair), 0.41–0.60 (moderate), 0.61–0.80 (good) and 0.81–1.0 (very good) [24].

Second, we assessed the effect of using different measures of gestational age on disparities in the gestational age outcomes between the foreign-and Swedish-born. For this purpose, we used multinomial logistic regressions to derive odds ratios (OR) and 95% CI. We ran a total of eight models. First, we performed a set of two models (one based on ultrasound and the other on LMP), considering three categories of gestational age (preterm, term [reference outcome] and post-term) and using mother’s birthplace as foreign vs Swedish-born. Second, we performed another set of two models with the same description but now disaggregating the preterm category into very preterm and moderately preterm. Third, we replicated the above strategy running the additional four models considering mother’s birthplace into specific regions to see whether the main results were driven by a specific group of immigrants. The results are presented in forest plots and the estimates are included in tables as supporting material (S1 Table), together with additional models by mother’s country of birth (S2 and S3 Tables). Additionally, we performed sensitivity analyses to assess the extent to which our results are affected by exclusions due to missing values in several covariates (S4 Table), and we also replicated the analyses in a sample of uncomplicated pregnancies (S5 Table), to see whether the results were consistent.

Finally, we used binary logistic regression models to identify variables associated with the discrepancy in the gestational age categories between UL and LMP. Discrepancy is defined as those observations placed in different gestational age categories defined as very preterm, moderately preterm, term and post-term (Table 1). Models were run separately for foreign and Swedish-born women. To evaluate the impact of the associated variables in the comparison between foreign and Swedish-born, we present unadjusted and adjusted estimates from multinomial logistic models considering gestational age outcomes as: very preterm, moderately preterm, term [reference outcome] and post-term.

Table 1. Discrepancies in gestational age outcomes between methods of estimation of gestational age (Ultrasound [UL], Last Menstrual Period [LMP]) by mother’s birthplace.

UL LMP Swedish-born Foreign-born
Agreement (ref)   914,398 182,446
Term Term 770,432 157,218
Very preterm Very preterm 4,983 1,184
Moderately preterm Moderately preterm 20,257 3,882
Post-term Post-term 118,726 20,162
Disagreement   183,601 (100%) 36,820 (100%)
Underestimation 32,514 (17.8%) 7,707 (20.9%)
Moderately preterm Very preterm 0.21% 0.24%
Term Very preterm <0.01% 6 (0.02)
Term Moderately preterm 2.37% 4.39%
Post-term Very preterm ND ND
Post-term Moderately preterm 0.02% 0.04%
Post-term Term 15.31% 16.48%
Overestimation 149702 (82.2%) 28,823 (78.3%)
Very preterm Moderately preterm 0.54% 0.55%
Very preterm Term 0.01% 0.01%
Very preterm Post-term <0.01% ND
Moderately preterm Term 4.48% 4.32%
Moderately preterm Post-term 0.06% 0.07%
Term Post-term 76.98% 73.89%
Total No of births   1,097,999 219,266
Open in a new tab

ND = No Data.

All models were clustered by maternal identification number in order to obtain more accurate standard errors due to the existence of siblings in the dataset. All analyses were performed using Stata Statistical Software: Release 13, software (StatCorp, LP, College Station, Texas).

This study was approved by the Central Ethical Review Board of Stockholm in 2013 (decision no 2013/1058-32)

Results

Discrepancies in gestational age outcomes by method of estimation (UL and LMP) are observed in foreign and Swedish-born women. Relative to UL, LMP classifies 16.7% of records in different categories. LMP has a wider gestational age distribution than UL, resulting in 17.8% and 20.9% of discrepant records classified as shorter gestations and 82.2% and 78.3% as longer gestation, among Swedish and foreign-born, respectively (Table 1).

Differences in prevalence

Differences in preterm and post-term birth rates by gestational age estimation methods are shown for all women (Table 2). Preterm rates vary between methods among mothers born in Sweden (UL: 3.19%; LMP: 2.83%, p-value <0.001), Africa (UL: 2.46%; LMP: 3.22%, p-value <0.001) and Latin America (UL: 3.64%; LMP: 3.29, p-value <0.01). Larger differences were observed when comparing post-term births across methods of estimation of gestational age (~10 points estimate) in all groups (p-value <0.001). The overall level of agreement between LMP and UL is moderately good regardless of mother’s region of birthplace (Kappa coefficient ranging between 0.47 and 0.54 for Asia and Western Europe & USA, respectively) (Table 2).

Table 2. Preterm and post-term rates by method of estimation of gestational age and region of origin.

Preterm rates Post-term rates Level of agreement
UL LMP UL LMP Agreement
Live births % % P value % % P value Kappa Coef. 95%CI
Swedish-born 1,097,999 3.19 2.83 <0.001 13.38 23.70 <0.001 0.532 [0.530.0.533]
Foreign-born 219,266 3.18 3.19 0.6868 11.97 21.61 <0.001 0.530 [0.499,0.506]
By region                
Nordic 26,735 3.03 2.89 0.0463 12.93 22.19 <0.001 0.536 [0.529,0.542]
Western Europe & USA 8,619 2.38 2.20 0.1193 12.86 21.95 <0.001 0.540 [0.520,0.553]
Eastern Europe & Russia 48,331 3.19 3.14 0.3792 12.47 21.68 <0.001 0.513 [0.501,0.520]
Middle East 72,809 2.97 2.94 0.4297 10.02 20.70 <0.001 0.480 [0.471,0.487]
Africa 23,061 2.46 3.22 <0.001 23.04 28.94 <0.001 0.509 [0.502,0.519]
Asia 28,845 4.47 4.42 0.6062 7.25 17.94 <0.001 0.472 [0.464,0.841]
Latin America 10,866 3.64 3.29 0.0042 8.80 19.99 <0.001 0.477 [0.460,0.791]
Open in a new tab

UL = Ultrasound, LMP = Last Menstrual Period.

P-values indicate comparisons for preterm and post-term rates between UL and LMP within each mother’s origin.

The level of agreement in the Kappa coefficient is interpreted as follows: 0.20 (poor), 0.21–0.40 (fair), 0.41–0.60 (moderate), 0.61–0.80 (good) and 0.81–1.0 (very good).

Gestational age disparities by mother’s birthplace and method of gestational age estimation

The comparison of gestational age outcomes (preterm and post-term) between the offspring of foreign and Swedish-born mothers is contingent on the method used to estimate gestational age. The results from ultrasound showed that overall, compared to natives, foreign-born mothers had equal odds of having preterm births (OR: 0.98, 95% CI 0.96,1.01), lower odds of having moderately preterm and post-term births (respectively, OR: 0.95, 95%CI 0.92,0.98; OR:0.88, 95%CI 0.87,0.89), and higher odds of having very preterm births (OR: 1.13, 95%CI 1.07;1.20). However, the results derived from the LMP-based estimation led to consistently higher odds across all gestational age categories compared to term, except for post-term (Fig 2 and S1 Table). These inconsistencies were primarily driven by particular regional groups in each gestational age category (Fig 3 and S2 and S3 Tables). For example, in relation to preterm birth, the differences were mainly driven by mothers from Africa (ORUL: 0.87, 95%CI 0.80,0.95 and ORLMP: 1.23, 95%CI 1.15,1.34) and, to a lesser extent, mothers from Eastern Europe & Russia (ORUL: 0.99, 95%CI 0.94, 1.05 and ORLMP: 1.09, 95%CI: 1.03, 1.15), the Middle East (ORUL: 0.90, 95%CI 0.86, 0.94 and ORLMP: 1.00, 95%CI 0.96,1.05) and Latin America (ORUL: 1.09, 95%CI 0.98, 1.21 and ORLMP: 1.12, 95%CI 1.00,1.25).

Fig 2. Gestational age outcomes according to ultrasound and LMP estimates for the offspring of foreign-born women compared to Swedish-born women.

Fig 2

Open in a new tab

The stars indicate the level of significance comparing foreign-born and Swedish-born women for each model * p<0.05, ** p<0.01, ***p<0.001.

Fig 3. Gestational age outcomes according to LMP and ultrasound estimates for the offspring of foreign-born women (by region of origin) compared to Swedish-born women.

Fig 3

Open in a new tab

The stars indicate the level of significance comparing foreign-born and Swedish-born women for each model * p<0.05, ** p<0.01, ***p<0.001.

The results obtained for all foreign-born mothers were consistent with results obtained from a sample of uncomplicated pregnancies (S4 Table), and also when observations with missing in other covariates were included (S5 Table).

Variables associated with the differences between estimation methods of gestational age

The discrepancy in gestational age outcomes between methods of estimation of gestational age were, overall, similar for migrants and Swedes, with some exceptions: year of birth was associated with the difference between methods only among Swedes, showing consistently better agreement after 1995 compared to 1991–1995. Multiparous mothers (as opposed to primiparous) were associated with a lower odd (OR: 0.95, 95%CI 0.94,0.96) for discrepancy if they were Swedish-born but with higher odds if they were foreign-born (OR: 1.04, 95%CI 1.00,1.08). Similarly, while no differences were found in relation to marital status among Swedish mothers, a higher odd was found among non-married or cohabiting foreign-mothers (OR: 1.11, 95%CI 1.06,1.15) compared to married ones. Conversely, as compared to non-smoking, light smoking (9 cigarettes or less per day) during pregnancy was associated with a higher odd of discrepancy between methods only among Swedish mothers (OR: 1.07, 95%CI 1.05,1.09). Father’s origin (being a foreign vs Swedish-born) was associated with a lower odd of misclassification only among Swedes (OR: 0.96 95%CI 0.94,0.98) (Table 3).

Table 3. Odds of the association between discrepant ultrasound-LMP information affecting the determination of gestational age outcomes and birth characteristics among foreign-born and Swedish-born mothers.

  Foreign-born   Swedish-born 
  OR 95%CI p value OR 95%CI p value
Year of birth
1991–1995 (ref) 1 1
1996–1999 0.99 [0.95,1.04] 0.742 0.98 [0.96,1.00] 0.016
2000–2004 0.98 [0.94,1.02] 0.372 0.95 [0.94,0.97] <0.001
2005–2008 0.99 [0.95,1.03] 0.601 0.95 [0.93,0.97] <0.001
2009–2012 0.98 [0.94,1.02] 0.254 0.92 [0.90,0.94] <0.001
Newborn’s sex
Male (ref) 1 1
Female 1.11 [1.08,1.13] <0.001 1.12 [1.11,1.13] <0.001
Parity
Primiparous (ref) 1 1
Multiparous 1.04 [1.00,1.08] 0.007 0.95 [0.94,0.96] <0.001
Maternal Height (linear) 1.04 [1.01,1.06] 0.043 0.99 [0.95,1.03] 0.655
Mother’s BMI
Normal weight (ref) 1 1
Underweight 0.95 [0.90,1.01] 0.094 0.98 [0.95,1.01] 0.214
Overweight 1.13 [1.10,1.16] <0.001 1.08 [1.07,1.09] <0.001
Obese 1.33 [1.28,1.38] <0.001 1.26 [1.24,1.28] <0.001
Maternal age            
25–29 (ref) 1     1    
< = 24 1.12 [1.08,1.15] <0.001 1.19 [1.17,1.20] <0.001
30–34 0.88 [0.85,0.90] <0.001 0.87 [0.86,0.88] <0.001
> = 35 0.75 [0.73,0.78] <0.001 0.74 [0.73,0.76] <0.001
Mother’s education attaintment            
High (ref) 1     1    
Low 1.19 [1.15,1.23] <0.001 1.11 [1.09,1.13] <0.001
Middle 1.09 [1.06,1.13] <0.001 1.04 [1.03,1.05] <0.001
Household disposable income            
Botton (ref) 1     1    
2 0.99 [0.96,1.02] 0.351 0.96 [0.94,0.98] <0.001
3 0.95 [0.92,0.99] 0.014 0.92 [0.91,0.94] <0.001
4 0.96 [0.92,1.01] 0.103 0.90 [0.88,0.91] <0.001
Top 0.94 [0.90,0.98] 0.009 0.89 [0.87,0.90] <0.001
Marital status            
Married/cohabiting 1     1    
Other 1.11 [1.06,1.15] <0.001 0.99 [0.97,1.02] 0.262
Father’s origin            
Swedish-born (ref) 1     1    
Foreign-born 1.00 [0.97,1.03] 0.988 0.96 [0.94,0.98] <0.001
Smoking during pregnancy            
No (ref) 1     1    
1 to 9 cig/per day 1.01 [0.96,1.05] 0.710 1.07 [1.05,1.09] <0.001
10 + cig/per day 0.96 [0.89,1.03] 0.235 0.98 [0.96,1.01] 0.262
Open in a new tab

OR = Odd Ratios; CI = Confidence Intervals.

Models adjusted for the variables associated with the discrepancy between ultrasound and LMP generally contribute to reducing differences in gestational age outcomes between foreign and Swedish-born mothers. The reduction affects gestational age outcomes estimated with ultrasound and LMP (Fig 4). The exceptions are for ultrasound-estimated preterm births and for LMP-estimated post-term.

Fig 4. Gestational age outcomes according to LMP and ultrasound estimates for the offspring of foreign-born women (by region of origin) compared to Swedish-born women.

Fig 4

Open in a new tab

Differences between unadjusted and adjusted models. Model adjusted for year of birth, newborn’s sex, parity, maternal height, maternal BMI, maternal age, mother’s educational attainment, household disposable income, marital status, father’s origin, smoking during pregnancy. The stars indicate the level of significance comparing foreign-born and Swedish-born women for each model * p<0.05, ** p<0.01, ***p<0.001.

Discussion

Summary of findings

Our results indicate that disparities in gestational age outcomes by mother’s birthplace depend on the method used to estimate gestational age. While a health advantage is observed in almost all categories of gestational age when using ultrasound-based estimates (except for very preterm), foreign-born mothers exhibit worse outcomes when LMP dating is used. However, these differences do not affect all migrant groups equally; the largest differences are found in mothers coming from Africa and, to a lesser extent, those from Eastern Europe and Russia, and the Middle East. Year of birth, multiplicity, marital status, father’s origin and smoking were differentially associated with the disagreement between ultrasound and LMP estimates among foreign- and Swedish-born mothers. This finding suggests that controlling for these variables (to reduce confounding or to evaluate mediation) could be problematic because they might reduce measurement error in the outcome rather than explain health differences between immigrants and natives.

Links with previous research

In our study immigrants show lower odds of preterm births using ultrasound-based estimates. Although this result is in line with international evidence of a health advantage among immigrants compared to natives (healthy migrant paradox) [25], this finding is to some extent surprising in the Swedish context for at least two reasons. First, Sweden has some of the lowest rates of preterm birth in Europe (5.9% in 2009, including multiple births) [26] and the immigrant health advantage is observed among women from countries in conflict, who are likely to be refugees and who may have been exposed to adverse conditions before, during, and after migration. This advantage demonstrated among refugees contrasts with prior evidence from other national contexts [27]. Second, a recent study using the same population as our current study has shown that foreign-born mothers in Sweden have a higher risk of low birthweight deliveries [28], suggesting that migrants simultaneously have lower odds of preterm birth. These combined findings are puzzling since birthweight and gestational age are strongly correlated measures (i.e., fetal growth over time). Unlike gestational age, birthweight does not have differential measurement error by mother’s origin (i.e., all births are weighed in the same way). As such, although ultrasound has been regarded as the most accurate method to estimate gestational age, results on preterm birth based on LMP estimations are surprisingly more consistent with the results that show higher risks of low birthweight.

Our findings are in line with previous studies which have shown that the LMP estimate is consistently associated with higher post-term birth, while mixed results have been seen in relation to preterm [7, 29, 30]. Prior studies have also demonstrated that the two gestational age methods differ according to maternal-level characteristics, such as obesity and smoking during pregnancy [21]. Our findings indicate that region and country of origin are also important factors to take into consideration.

Year of birth, multiplicity, marital status, father’s origin and smoking were also differentially associated with the discrepancy between LMP and ultrasound depending on whether the mother is Swedish or a foreign-born. These variables do not explain the health advantage observed with ultrasound since consistent results were found in a sub-sample of uncomplicated pregnancies that exclude multiple births and smoker mothers. Furthermore, the results remain the same when single women and women with a Swedish partner are additionally excluded (results upon request). However, the association between these variables and the discrepancy between methods is of importance as this may lead to wrong conclusions when trying to use this data to identify risk factors for a gestational age outcome [17, 21]. For example, we cannot exclude the possibility that an association between smoking and preterm (estimated with ultrasound) may reflect the existence of measurement error in the outcome (gestational ages of small fetuses are underestimated with ultrasound and smoking might lead to small fetuses) rather than a risk factor for preterm.

Strengths and limitations

To the best of our knowledge, this is the first study to assess the implications of using different methods of gestational age estimation to study disparities in birth outcomes between immigrants and natives. This study is conducted using high quality and large population registers in a country context with a wide diversity of migrant origins. This provides enough statistical power to explore disparities by world regions and specific countries. Both ultrasound and LMP-based estimates of gestational age recorded in the MBR are measured in days, which allowed for a more precise evaluation of the discrepancy between the two dating methods than assessments based on weeks. The detailed information collected in the MBR also allows us to compare the results using a sub-sample of uncomplicated pregnancies. These analyses show that the differences observed between migrant and native mothers are not driven by uneven distributions of the observed risk factors, adverse environmental conditions, pathological process or obstetric interventions. However, other important variables are missing, including the date when the ultrasound was conducted or delays on the first antenatal visit, which might explain why larger differences are seen for some regions between estimation methods than others.

Implications for future research and policy

Our findings may also be useful in the interpretation of national comparisons of disparities in preterm birth by migration status and other sociodemographic characteristics. For example, year of birth was associated with the difference between modes of estimation of gestational age, but only among Swedes. This suggests that the quality of the information could be responsible for artefactual differences in gestational age outcomes in some earlier studies conducted.

From our findings, we recommend focusing on very preterm in studies on migration, as this category not only is of greater clinical relevance but it shows higher risks among migrants. We also observed that the smallest discrepancy between methods of gestational age estimation is for very preterm. Furthermore, studies have shown that the accuracy of gestational age is higher among very preterm than among the broad category of preterm in validation studies that compared hospital and vital statistics records [31].

There is an ongoing debate on whether ultrasound should be the first choice for dating pregnancies and/or exclusively early ultrasound [8]. Although we cannot disentangle this concern, our findings are relevant for those contexts (such as Sweden) where the ‘best estimate’ method is used for clinical and public health surveillance. Our study shows that the results obtained with the ‘best estimate’ method could also be affected depending on the extent to which ultrasound or LMP is used across groups since they lead to different results.

Future research should strive to understand the measurement properties of the gestational age estimate being used. Although our sensitivity analysis using a sample of uncomplicated pregnancies strengthens the accuracy of the ultrasound-based estimate (since under optimal conditions fetal growth should be similar worldwide), it does not rule out the possibility that other factors may influence the accuracy of the ultrasound estimation. For example, future studies may examine the extent to which the discrepancy between UL and LMP observed in this study is associated to the fact that foreign-born expectant mothers tend to arrive late to their first prenatal visit [32]. It is known that up to week 13 6/7 weeks of gestation, the ultrasound estimation of gestational age based on measurement of crown–rump length (CRL) has an error of ± 5–7 days while, after week 14, measurements with lower accuracy (e.g., fetal biparietal diameter, femur length, abdominal circumference or length of the humerus) have an estimated error of approximately 14 days between weeks 22 0/7 and 27 6/7 [33]. Prior perinatal research has demonstrated that discrepant information between UL and LMP per se is associated with adverse perinatal outcomes (such as infant mortality and morbidity) [34, 35], which might well be a consequence of poor early prenatal monitoring.

Future research on migrant disparities in preterm birth should exercise caution when interpreting the international literature. Given that preterm disparities between foreign and Swedish-born is contingent on the method used to estimate gestational age (lower OR of preterm using ultrasound but higher odds using LMP), our study suggests that the healthy migrant paradox [36–38] might to some extent depend on the method used to estimate gestational age, which varies between countries and within countries over time.

In this study we have focused on the implications of the UL-LMP discrepancy for gestational age surveillance in the context of migration and perinatal health research. However, the discrepancy observed between methods could also have clinical implications that are beyond the scope of this study but which could nevertheless influence public health surveillance of preterm outcomes (e.g. through selection bias). Gestational age is not only used to derive perinatal health outcomes (such as those examined in this study), but also to decide whether intrauterine death should be defined as stillbirth or fetal death, and to determine when a pregnancy reaches the viability threshold. This is necessary, for example, in order to determine the legality of any abortions that are performed [39]. Furthermore, gestational age is used to decide when induction of labor should start (generally after 41 weeks) [40].

Conclusions

The direction and strength of disparities in gestational age outcomes are affected by the method of estimation used, as do some of the risk factors associated with the timing of birth outcomes. We recommend focusing on very preterm in studies on migration, since this is the gestational age outcome that shows consistently higher risks among migrants across methods of gestational age estimation.

Supporting information

S1 Table. Gestational age outcomes by method of estimation of gestational age by mother’s birthplace (corresponding to Figs 2 and 3).

(DOCX)

Click here for additional data file. (23KB, docx)
S2 Table. Preterm according to LMP and ultrasound estimates by mother’s country of birth.

(DOCX)

Click here for additional data file. (23.8KB, docx)
S3 Table. Post-term according to LMP and ultrasound estimates by mother’s country of birth.

(DOCX)

Click here for additional data file. (22.7KB, docx)
S4 Table. Sensitivity analyses.

Gestational age outcomes according to LMP and ultrasound estimates in a subsample of 718,191 of uncomplicated pregnancies.

(DOCX)

Click here for additional data file. (18.5KB, docx)
S5 Table. Sensitivity analyses.

Gestational age outcomes according to LMP and ultrasound estimates including 723,253 missing in different covariates.

(DOCX)

Click here for additional data file. (18.5KB, docx)
S6 Table. Gestational age outcomes by method of estimation of gestational age by region of origin adjusted for maternal variables (corresponding to Fig 4).

(DOCX)

Click here for additional data file. (18.8KB, docx)

Data Availability

Data cannot be shared publicly because of confidentiality. In order to access Swedish register data, researchers must contact the agencies responsible for data protection, conditional on ethical vetting (contacts below). The National Board of Health and Welfare (Socialstyrelsen) - Postal address: Socialstyrelsen SE-106 30, Stockholm, Sweden; Visiting address: Rålambsvägen 3, Stockholm; Phone: +46 (0)75 247 30 00; Fax: +46 (0)75 247 32 52; E-mail: socialstyrelsen@socialstyrelsen.se. Statistics Sweden (SCB) - Postal address: Statistics Sweden, Solna strandväg 86, SE-171 54 Solna, Sweden; Visiting address: Solna strandväg 86; Phone: +46 (0) 10 479 40 00; E-mail: scb@scb.se. Swedish Ethical Review Authority (Etikprövningsmyndigheten) - Postal address: Box 2110, SE-750 02 Uppsala, Sweden; E-mail: registrator@etikprovning.se; Phone: +46 (0)10 475 08 00.

Funding Statement

Sol P. Juárez and Eleonora Mussino were supported by the Strategic Research Council of the Academy of Finland (TITA, Grant # 293103) and the Swedish Research Council (VR# #2018-01825); Sol P. Juárez and Anders Hjern were supported by the Swedish Research Council for Health, Working Life and Welfare (FORTE grant #2012-01190 and #2016-07128). Marcelo L. Urquia holds a Canada Research Chair in Applied Population Health. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

References

  • 1.Lawn JE, Blencowe H, Oza S, et al. Every Newborn: progress, priorities, and potential beyond survival. Lancet. 2014;12(384(9938)):189–205. [DOI] [PubMed] [Google Scholar]
  • 2.Bertagnolli M, Luu TM, Lewandowski AJ, et al. Preterm Birth and Hypertension: Is There a Link? Curr Hypertens Rep. 2016;18(4):28 10.1007/s11906-016-0637-6 [DOI] [PubMed] [Google Scholar]
  • 3.Bensley JG, De Matteo R, Harding R, et al. The effects of preterm birth and its antecedents on the cardiovascular system. Acta Obstet Gynecol Scand. 2016. 95(6):652–63. 10.1111/aogs.12880 [DOI] [PubMed] [Google Scholar]
  • 4.Sørbye IK, Wanigaratne S, Urquia ML. Variations in gestational length and preterm delivery by race, ethnicity and migration. Best Pract Res Clin Obstet Gynaecol. 2016;32:60–8. 10.1016/j.bpobgyn.2015.08.017 [DOI] [PubMed] [Google Scholar]
  • 5.Gagnon A, Zimbeck M, Zeitlin J, et al. Migration to western industrialised countries and perinatal health: A systematic review. Soc Sci Med. 2009;69:934–46. 10.1016/j.socscimed.2009.06.027 [DOI] [PubMed] [Google Scholar]
  • 6.Martin JA, Osterman MJK, Kirmeyer SE, et al. Measuring gestational age in vital statistics data: transitioning to the obstetric estimate. Hyattsville: Centers for Disease Control and Prevention; 2015. Contract No.: 5. [PubMed] [Google Scholar]
  • 7.Dietz PM, England LJ, Callaghan WM, et al. A comparison of LMP-based and ultrasound-based estimates of gestational age using linked California livebirth and prenatal screening records. Paediatric and Perinatal Epidemiology. 2007;21(2):62–71. 10.1111/j.1365-3016.2007.00862.x [DOI] [PubMed] [Google Scholar]
  • 8.Blondel B, Morin I, Platt RW, et al. Algorithms for combining menstrual and ultrasound estimates of gestational age: consequences for rates of preterm and postterm birth. BJOG. 2002;109(6):718–20. 10.1111/j.1471-0528.2002.01068.x [DOI] [PubMed] [Google Scholar]
  • 9.De la Fuente P. Diagnóstico del Embarazo In: Usandizaga JA, De la Fuente P, editors. Tratado de Obstetricia y Ginecología. 1 Madrid: McGrawHill; 2005. p. 153–92. [Google Scholar]
  • 10.Lack N, Zeitlin J, Krebs L, et al. Methodological difficulties of comparisons across Europe. Eur J Obstet Gynecol Reprod Biol 2003;111(1):33–44. 10.1016/s0301-2115(03)00164-7 [DOI] [PubMed] [Google Scholar]
  • 11.Jukic AM. The impact of systematic errors on gestational age estimation. BJOG. 2015. 122(6):(6):842 10.1111/1471-0528.13169 [DOI] [PubMed] [Google Scholar]
  • 12.Cnattingius S, Ericson A, Gunnarskog J, et al. A quality study of a medical birth registry. Scand J Soc Med. 1990;18:143–8. 10.1177/140349489001800209 [DOI] [PubMed] [Google Scholar]
  • 13.Ekbom A. The Swedish Multi-generation Register. Methods Mol Biol. 2011;675:215–20. 10.1007/978-1-59745-423-0_10 [DOI] [PubMed] [Google Scholar]
  • 14.Ludvigsson JF, Svedberg P, Olén O, et al. The longitudinal integrated database for health insurance and labour market Studies (LISA) and its use in medical research. Eur J Epidemiol. 2019;34:423–37. 10.1007/s10654-019-00511-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Ludvigsson JF, Otterblad-Olausson P, Pettersson BU, et al. The Swedish personal identity number: possibilities and pitfalls in healthcare and medical research. Eur J Epidemiol. 2009;24:659–67. 10.1007/s10654-009-9350-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Källén B. A birth weight for gestational age standard based on data in the Swedish Medical Birth Registry, 1985–1989. European Journal of Epidemiology. 1995;11(5):601–6. 10.1007/BF01719316 [DOI] [PubMed] [Google Scholar]
  • 17.Lynch CD, Zhang J. The research implications of the selection of gestational estimation methods. Paediatr Perinat Epidemiol. 2007;21(2):86–96. [DOI] [PubMed] [Google Scholar]
  • 18.Hadlock FP, Deter RL, Harrist RB, et al. Fetal Abdominal Circumference as a predictor of Menstrual Age. AJR. 1982;139(August):367–71. 10.2214/ajr.139.2.367 [DOI] [PubMed] [Google Scholar]
  • 19.Hadlock FP, Deter RL, Harrist RB, et al. Fetal Head Circumference: Relation to menstrual age. AJR. 1982;138:649–53. 10.2214/ajr.138.4.649 [DOI] [PubMed] [Google Scholar]
  • 20.Hadlock FP, Deter RL, Harrist RB, et al. Fetal Bieparietal Diameter: rational choice of plane of section for sonographic measurement. AJR. 1983;138:871–4. [DOI] [PubMed] [Google Scholar]
  • 21.Henriksen TB, Wilcox AJ, Hedegaard M, et al. Bias in studies of preterm and postterm delivery due to ultrasound assessment of gestational age. Epidemiology. 1995;6(5): 533–7. 10.1097/00001648-199509000-00012 [DOI] [PubMed] [Google Scholar]
  • 22.Kullinger M, Wesström J, Kieler H, et al. Maternal and fetal characteristics affect discrepancies between pregnancy-dating methods: a population-based cross-sectional register study. Acta Obstet Gynecol Scand. 2017;96(1):86–95. 10.1111/aogs.13034 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.International Fetal and Newborn Growth Consortium. The International Fetal and Newborn Growth Standards for the 21st Century (INTERGROWTH-21st) Study Protocol.; 2009.
  • 24.Altman DG. Practical Statistics for Medical Research. London England: Chapman and Hall; 1991. [Google Scholar]
  • 25.Urquia ML, Gagnon A. Glossary: migration and health. JECH. 2011;65(5):467–72. 10.1136/jech.2010.109405 [DOI] [PubMed] [Google Scholar]
  • 26.Zeitlin J, Szamotulska K, Drewniak N, et al. Preterm birth time trends in Europe: a study of 19 countries. BJOG. 2013;120(11):1356–65. 10.1111/1471-0528.12281 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Wanigaratne S, Cole DC, Bassil K, et al. The influence of refugee status and secondary migration on preterm birth. J Epidemiol Community Health. 2016;70(6):622–8. 10.1136/jech-2015-206529 [DOI] [PubMed] [Google Scholar]
  • 28.Juárez S, Hjern A. The weight of inequalities: duration of residence and offspring’s birthweight among migrants in Sweden. Soc Sci Med. 2017;175:81–90. 10.1016/j.socscimed.2016.12.045 [DOI] [PubMed] [Google Scholar]
  • 29.Yang H, Kramer MS, Platt RW, et al. How does early ultrasound scan estimate of gestationalage lead to higher rates of preterm birth?. American Journal ofObstetrics and Gynecology 2002;186:433–7. [DOI] [PubMed] [Google Scholar]
  • 30.Wilcox AJ, Dunson D, Baird DD. The timing of the "fertile window" in the menstrual cycle: day specific estimates from a prospective study. BMJ. 2000;18(321(7271)):1259–62. 10.1136/bmj.321.7271.1259 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Juárez S, Alonso Ortíz T, Ramiro Fariñas D, et al. The quality of vital statistics for studying perinatal health: the Spanish case. Paediatrics and Perinatal Epidemiology. 2012;26(4):310–5. 10.1111/j.1365-3016.2012.01287.x [DOI] [PubMed] [Google Scholar]
  • 32.Ny P, Dykes AK, Molin J, et al. Utilisation of antenatal care by country of birth in a multi-ethnic population: a four-year community-based study in Malmö, Sweden. Acta Obstet Gynecol Scand. 2007;86(7):805–13. 10.1080/00016340701415095 [DOI] [PubMed] [Google Scholar]
  • 33.Methods for estimating the due date. Comittee Opinion No. 700. American Colege of Obstetricians and Gynecologists. 2017;129:e150–4. [Google Scholar]
  • 34.Morken NH, Skjaerven R, Richards JL, et al. Adverse infant outcomes associated with discordant gestational age estimates. Paediatr Perinat Epidemiol. 2016;30(6):541–9. 10.1111/ppe.12311 [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Kullinger M, Granfors M, Kieler H, et al. Discrepancy between pregnancy dating methods affects obstetric and neonatal outcomes: a population-based register cohort study. Sci Rep. 2018;2(8(1)):6936 10.1038/s41598-018-24894-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Juarez SP, Revuelta-Eugercios BA. Too heavy, too late: investigating perinatal health outcomes in immigrants residing in Spain. A cross-sectional study (2009–2011). J Epidemiol Community Health. 2014;68(9):863–8. 10.1136/jech-2013-202917 [DOI] [PubMed] [Google Scholar]
  • 37.Speciale AM, Regidor E. Understanding the universality of the immigrant health paradox: the Spanish perspective. J Immigr Minor Health. 2011;13(3):518–25. 10.1007/s10903-010-9365-1 [DOI] [PubMed] [Google Scholar]
  • 38.Juárez S, Mussino E, Hjern A. Being a refugee or having a refugee status? Birthweight and gestational age outcomes among offspring of immigrant mothers in Sweden. Scand J Public Health. 2019;47(7):730–4. 10.1177/1403494818777432 [DOI] [PubMed] [Google Scholar]
  • 39.Gourbin C, Masuy-Stroobant G. Registration of vital data: are live birth and stillbirth comparable all over Europe? Bulletin of the World Health Organization. 1995;73(4):449–60. [PMC free article] [PubMed] [Google Scholar]
  • 40.Obstetrics ACoPB. ACOG Practice Bulletin No. 107: Induction of labor. Obstet Gynecol. 2009;114(2(Pt 1)):386–97. 10.1097/AOG.0b013e3181b48ef5 [DOI] [PubMed] [Google Scholar]

Decision Letter 0

Kelli K Ryckman

Transfer Alert

This paper was transferred from another journal. As a result, its full editorial history (including decision letters, peer reviews and author responses) may not be present.

12 Nov 2020

PONE-D-20-31406

Preterm disparities between migrants and native Swedes depend on the method used to estimate gestational age. A Swedish population-based register study

PLOS ONE

Dear Dr. Juárez,

Thank you for submitting your manuscript to PLOS ONE. After careful consideration, we feel that it has merit but does not fully meet PLOS ONE’s publication criteria as it currently stands. Therefore, we invite you to submit a revised version of the manuscript that addresses the points raised during the review process.

Please submit your revised manuscript by Dec 27 2020 11:59PM. If you will need more time than this to complete your revisions, please reply to this message or contact the journal office at plosone@plos.org. When you're ready to submit your revision, log on to https://www.editorialmanager.com/pone/ and select the 'Submissions Needing Revision' folder to locate your manuscript file.

Please include the following items when submitting your revised manuscript:

  • A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

  • A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

  • An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: http://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols

We look forward to receiving your revised manuscript.

Kind regards,

Kelli K Ryckman

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

2. We note that you have indicated that data from this study are available upon request. PLOS only allows data to be available upon request if there are legal or ethical restrictions on sharing data publicly. For information on unacceptable data access restrictions, please see http://journals.plos.org/plosone/s/data-availability#loc-unacceptable-data-access-restrictions.

In your revised cover letter, please address the following prompts:

a) If there are ethical or legal restrictions on sharing a de-identified data set, please explain them in detail (e.g., data contain potentially identifying or sensitive patient information) and who has imposed them (e.g., an ethics committee). Please also provide contact information for a data access committee, ethics committee, or other institutional body to which data requests may be sent.

b) If there are no restrictions, please upload the minimal anonymized data set necessary to replicate your study findings as either Supporting Information files or to a stable, public repository and provide us with the relevant URLs, DOIs, or accession numbers. Please see http://www.bmj.com/content/340/bmj.c181.long for guidelines on how to de-identify and prepare clinical data for publication. For a list of acceptable repositories, please see http://journals.plos.org/plosone/s/data-availability#loc-recommended-repositories.

We will update your Data Availability statement on your behalf to reflect the information you provide.

3. Please include captions for your Supporting Information files at the end of your manuscript, and update any in-text citations to match accordingly. Please see our Supporting Information guidelines for more information: http://journals.plos.org/plosone/s/supporting-information.

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Partly

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: No

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Preterm disparities between migrants and native Swedes depend on the method used to estimate gestational age. A Swedish population-based register study

Juarez et al use a Swedish population register to investigate the disparity in classifying births as preterm or post-term between menstrual period and ultrasound in foreign and Swedish-born woman.

The idea is novel: While it is well known that there is a discrepancy between those 2 methods, the clinical implications are not well studied.

I have a few major concerns/suggestions:

First, as the authors mention in the introduction of their paper, the accuracy of the US methods depends on when it is done during gestation and as a consequence depends on prenatal care. The authors do not present any data on prenatal care. Do the authors have any data on this? It is often collected in administrative databases. Could prenatal care be assessed as a mediator in the difference between the methods?

Table 2 would best be converted into a forest plot. It is very difficult to read. I am also worried about multiple comparison. Did the authors account for that? To that point, it would be nice to present the p-value in addition to the OR to see to what degree these results are sensitive to a Bonferroni (or similar) correction.

Table 3 is interesting but only shows half of the story. It shows that the discrepancy is higher for a lot of the socioeconomic variables, again this reviewer is wondering why there is a discrepancy, maybe it is all related to late initiation of prenatal care? If that is the case, then the discrepancy is not that important but rather the timing of initiation of prenatal care.

Generally, I think the manuscript could benefit from some more “user friendly” explanations of the finding. It took this reviewer a while to understand the data and the results. I think the manuscript is well written though and the analysis is interesting, it is just difficult to grasp and maybe an additional figure or example would help.

Reviewer #2: Thanks you for the opportunity to review this interesting manuscript based on a large Scandinavian population cohort.

Page 6, line 163 “Thereafter we excluded observations with biologically implausible combinations of birthweight and gestational age” please provide details and/or reference for the method by which you determined the parameters for biological impausibility for exclusion.

Although I understand why for clarify and simplicity it was tempting for the authors to use logistic regression to break GA into categories and use logistic regression, this results in a huge loss of information when in fact the authors had access to very granular GA estimates in days and could have looked at discrepancies in more detail rather that just as black and white yes/no agreement of clinical categories with problematic edge effects (ie 36 weeks and 6 days is preterm while 37 weeks and 0 days is term despite a discrepancy of one day--and this discrepancy would be treated equivalently to a discrepancy of 100 days by a dichotomous model)

Page 8, Line 231: We considered as ‘discrepant’ those observations placed in different gestational age categories, that is, preterm according to ultrasound-based, and term according to LMP-based estimates. Does this imply that the analysis presented in Table 3 only includes observations for which gestational category was discrepant? If so, please clarify this in the text, and also provide details of the sample size for this analysis. Or does this imply that discrepant vs. concordant was used at the dichotomous outcome for this analysis? This is not clear. This analytical strategy may also be flawed as it does not account for directionality (ie it treats misclassification from term to preterm the same as preterm to term—please clarify how the outcome was coded for this analysis, and what cohort was included.).

Page 9, line 253 and elsewhere, you have not provided a formal comparison to support your assertion that preterm rates were more consistent in foreign vs native mothers (granted the confidence intervals are very tight and non-overlapping, but for cases where the differences are not as clear, a formal difference test would be necessary to support these assertions.

Page 9 Line 255:

“With the exception of African mothers, preterm rates overlapped between methods of estimation in all groups”. This statement isn’t meaningful from a statistical standpoint. As above with respect to consistency, you haven’t provided a formal statistical comparison. Confidence intervals for two comparitors can overlap substantially and still be significantly different. If you want to make claims of similarity or differences please formalize your comparisons with statistical tests-but note that comparing

Page 11 Line 288-298

“Lower odds vs. higher odds” these are all relative so “lower” odds in one group merely means misclassification is higher in one group vs the other. Please present CIs for these in the text if presenting them. I find this analysis a little concerning/problematic as misclassification can be in either direction and has very asymmetric consequences.

Table 1, and elsewhere For all significance tests/p-values: Please do not use asterices to signify levels of significance. Provide exact p-values unless p-value is <0.0001.

Discussion/Conclusions: It’s unclear how the impact of the factors studied on misclassification by LMP vs. US when it doesn’t appear that directionality of the bias was considered. The results presented and discussed from Table 3 appear to be factors that were associated with discrepant measurements, ignoring the direction or severity of the discrepancy. This makes interpretation in a broader context difficult. If I am interpreting the construction of the analysis wrong, please make it more clear in the methods section and presentation of results.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Steven Hawken

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

PLoS One. 2021 Feb 22;16(2):e0247138. doi: 10.1371/journal.pone.0247138.r002

Author response to Decision Letter 0

14 Jan 2021

Thank you very much for your constructive comments. Below is the point-by-point response to your concerns.

Reviewer 1 #1: Preterm disparities between migrants and native Swedes depend on the method used to estimate gestational age. A Swedish population-based register study

Juarez et al use a Swedish population register to investigate the disparity in classifying births as preterm or post-term between menstrual period and ultrasound in foreign and Swedish-born woman.

The idea is novel: While it is well known that there is a discrepancy between those 2 methods, the clinical implications are not well studied.

I have a few major concerns/suggestions:

First, as the authors mention in the introduction of their paper, the accuracy of the US methods depends on when it is done during gestation and as a consequence depends on prenatal care. The authors do not present any data on prenatal care. Do the authors have any data on this? It is often collected in administrative databases. Could prenatal care be assessed as a mediator in the difference between the methods?

AUTHORS: Very unfortunately, information on timing for the prenatal care visits is not available in the Swedish Medical Birth Register. It is therefore impossible for us to determine the extent to which the discrepancy between methods of estimation of gestational age is due to a delay on the first prenatal visit. However, we have explicitly mentioned the need to evaluate this mechanism (see pages 22-23, lines 433-443) and also highlighted this lack of information as a limitation (see page 21, lines 403-406).

Reviewer 1 #2: Table 2 would best be converted into a forest plot. It is very difficult to read. I am also worried about multiple comparison. Did the authors account for that? To that point, it would be nice to present the p-value in addition to the OR to see to what degree these results are sensitive to a Bonferroni (or similar) correction.

AUTHORS: Thank you for your suggestion. We now include forest plots to present the main results and include the former table 2 (including p-values) in the appendix (for the relevance of future meta-analyses).

In the revised version we include p-values in all tables, as requested. However, we believe that multiple comparison is not a concern in our case for two main reasons. First, we use well-established categories in the literature of migration and health. Migration status (foreign vs natives), and categorizations by geographic proximity or countries of origin are very common in the literature. Membership to countries of origin is not a random process and we are not using a sample but the whole population. Therefore, it is very unlikely that our observed associations are due to chance. Second, we examine foreign-born mothers in a stepwise fashion trying to disentangle whether the overall results (from the foreign vs Swedish-born comparison) are driven by specific groups of immigrants. In relation to this last point, we would like to clarify that the main results come from eight independent multinomial models and some of them are quite parsimonious. We realized, however, that the description of the model strategy was not very clear in the previous version of our manuscript so we have now clarified this point in the revised version. See pages 8-9 lines, 192-205.

Reviewer 1 #3: Table 3 is interesting but only shows half of the story. It shows that the discrepancy is higher for a lot of the socioeconomic variables, again this reviewer is wondering why there is a discrepancy, maybe it is all related to late initiation of prenatal care? If that is the case, then the discrepancy is not that important but rather the timing of initiation of prenatal care.

AUTHORS: We would like to clarify that it is beyond the scope of this paper to explain the differences between methods of gestational age estimation. We identify the birth and maternal characteristics associated with the discrepancies with the purpose of further discussing the consequences of using different methods in the study of migration and perinatal health, not to explain the discrepancies between methods. Specifically, we want to examine whether variables commonly used as confounders or mediators in studies on preterm disparities by mother’s birthplace are also associated with the misclassification of the outcome. This is important as the inclusion of these variables in the model could reduce measurement error rather than explain health differences between groups. In the revised version we try to better explain the purpose of table 3. To that effect, we introduced the rationale in the introduction (see page 4, lines 99-103), included additional analyses showing how the adjustments for these maternal variables modify the differences between foreign-born and Swedish-born women (see figure 4; page 17, lines 316-320) and we interpreted the results in the discussion section (see page 19 lines 346-349).

Although we do not offer an explanation of the observed discrepancies, we believe that our findings could inspire further research in this direction. In this regard, we explicitly elaborate on how timing for prenatal care between groups could largely explain the observed discrepancy (see pages 22-23, lines 433-443).

Reviewer 1 #4: Generally, I think the manuscript could benefit from some more “user friendly” explanations of the finding. It took this reviewer a while to understand the data and the results. I think the manuscript is well written though and the analysis is interesting, it is just difficult to grasp and maybe an additional figure or example would help.

AUTHORS: In the revised version, we try to better guide the reader, especially through the method and result sections. We made our model strategy more explicit (see page 8-9), clarified the definition of some outcomes (see table 1), and we tried to be more upfront with our rationale throughout (see page 4, lines 99-103; page 9 lines, 207-214; page 10 lines 234-238; page 17, lines 317-320, figure 4, page 19, lines 346-349). All in all, we hope the final version makes for more pleasant reading.

Reviewer #2:

Reviewer 2 #1: Thanks you for the opportunity to review this interesting manuscript based on a large Scandinavian population cohort.

Page 6, line 163 “Thereafter we excluded observations with biologically implausible combinations of birthweight and gestational age” please provide details and/or reference for the method by which you determined the parameters for biological impausibility for exclusion.

AUTHORS: For the identification of implausible combinations of birthweight and gestational age we have used the thresholds based on fetal growth references published for Sweden. We have now included the corresponding reference. See reference #16 (page 5, line 128)

Källén B. A birth weight for gestational age standard based on data in the Swedish Medical Birth Registry, 1985-1989. European Journal of Epidemiology. 1995;11(5):601-6.

Reviewer 2 #2: Although I understand why for clarify and simplicity it was tempting for the authors to use logistic regression to break GA into categories and use logistic regression, this results in a huge loss of information when in fact the authors had access to very granular GA estimates in days and could have looked at discrepancies in more detail rather that just as black and white yes/no agreement of clinical categories with problematic edge effects (ie 36 weeks and 6 days is preterm while 37 weeks and 0 days is term despite a discrepancy of one day--and this discrepancy would be treated equivalently to a discrepancy of 100 days by a dichotomous model)

Page 8, Line 231: We considered as ‘discrepant’ those observations placed in different gestational age categories, that is, preterm according to ultrasound-based, and term according to LMP-based estimates. Does this imply that the analysis presented in Table 3 only includes observations for which gestational category was discrepant? If so, please clarify this in the text, and also provide details of the sample size for this analysis. Or does this imply that discrepant vs. concordant was used at the dichotomous outcome for this analysis? This is not clear. This analytical strategy may also be flawed as it does not account for directionality (ie it treats misclassification from term to preterm the same as preterm to term—please clarify how the outcome was coded for this analysis, and what cohort was included.).

AUTHORS: The analysis presented in table 3 shows the variables associated with the discrepancy between methods of estimation of gestational age (LMP-UL). The outcome is coded as agreement between ultrasound and LMP (reference) vs disagreement. We only considered ‘discrepancies’ when the differences between methods of estimation led to substantial changes in well-defined health outcomes. In the previous version, we used only three categories (preterm, term and post-term) to determine discrepancies between methods. However, considering the reviewer’s comment, in the revised version, we also considered discrepancies using more categories (very preterm, moderately preterm, preterm, term and postterm), though the results were identical. To help clarify how the outcome was coded, we included a table with all the possible combinations including sample size and indicating the direction of the discrepancies (see table 1), in addition to expanding the description in the text (see page: 9 lines 207-214).

The decision to examine discrepancies between methods (UL-LMP), using categories of gestational age (rather than days or weeks) while ‘ignoring’ the directionality of the discrepancies, is motivated by the purpose of this analysis. The aim of this analysis is to examine whether variables commonly used as confounders or mediators in studies on preterm disparities by mother’s birthplace are also associated with the misclassification of the outcome. As such, we believe that it makes more sense to define the discrepancies in alignment with how we classify the outcomes in the main analysis. The results suggested that both the exposure and the common explanatory factors (i.e. confounders and mediators) are associated with outcome misclassification. The misclassification of outcome is not trivial as the direction of the overall association can flip (as shown in forest plots). We understand that in the previous version of the manuscript the purpose of this secondary analysis was not clear enough. For this reason, in the revised version, we implemented a number of changes throughout: we introduced the aim (page 4, lines 99-103), included additional analyses to evaluate the impact of these variables in the comparison of gestational age outcomes between foreign and Swedish-born women (see figure 4 and page 17, lines 316-320), and interpreted the findings in the discussion section (page 19, lines 346-349).

Reviewer 2 #3: The analysis were designed in order to be aligned with the main results. As such, we considered as ‘discrepant’ observations that were placed in different categories of gestational age (here as preterm vs term) between different methods of estimation. As such,

Page 9, line 253 and elsewhere, you have not provided a formal comparison to support your assertion that preterm rates were more consistent in foreign vs native mothers (granted the confidence intervals are very tight and non-overlapping, but for cases where the differences are not as clear, a formal difference test would be necessary to support these assertions.

Page 9 Line 255:

“With the exception of African mothers, preterm rates overlapped between methods of estimation in all groups”. This statement isn’t meaningful from a statistical standpoint. As above with respect to consistency, you haven’t provided a formal statistical comparison. Confidence intervals for two comparitors can overlap substantially and still be significantly different. If you want to make claims of similarity or differences please formalize your comparisons with statistical tests-but note that comparing

AUTHORS: Thanks for pointing at this imprecise statement. We have rephrased it in the revised version. In order to formally compare preterm and postterm birth between ultrasound and LMP within origins, we implemented two changes in table 2. First, we included a column with p-values using McNemar test for paired data. Second, we included a column with a measure of overall agreement (Kappa coefficient) between ultrasound and LMP. See page 8, lines 182-187 in methods; page 11, lines, 244-251 in the results.

Although the reviewer’s comment refers to table 2, we would like to take the opportunity to clarify why we did not calculate p-values to formally compare the estimates derived from different multinomial models (using UL and LMP, see forest plots). In this case we did not consider it necessary to establish a formal comparison since we want to illustrate how, in practice, we can infer different conclusions from the comparison between immigrants and natives depending on whether we estimate gestational age outcomes from UL and LMP. Please, note that in most cases researchers do not have access to both UL and LMP to see how these results differ. This is the reason why the column “consistent” in the tables presented in the appendix is not supported by a formal test.

Reviewer 2 #4: Page 11 Line 288-298

“Lower odds vs. higher odds” these are all relative so “lower” odds in one group merely means misclassification is higher in one group vs the other. Please present CIs for these in the text if presenting them. I find this analysis a little concerning/problematic as misclassification can be in either direction and has very asymmetric consequences.

AUTHORS: In the revised version, we included confidence intervals when reporting the findings in the text and also clarified the comparison category (see page 13, lines 262-276). In a previous answer (see the response to your question #2) we hope we justified the decision not to include the directionality in the main analysis. However, we have included detailed information for the reader to see the most common type of misclassification (see table 1) and directionality (overestimation). See page 10, lines 234-238.

Reviewer 2 #4: Table 1, and elsewhere For all significance tests/p-values: Please do not use asterices to signify levels of significance. Provide exact p-values unless p-value is <0.0001.

AUTHORS: We have now added the corresponding p-values with four decimals in all tables.

Reviewer 2 #5: Discussion/Conclusions: It’s unclear how the impact of the factors studied on misclassification by LMP vs. US when it doesn’t appear that directionality of the bias was considered. The results presented and discussed from Table 3 appear to be factors that were associated with discrepant measurements, ignoring the direction or severity of the discrepancy. This makes interpretation in a broader context difficult. If I am interpreting the construction of the analysis wrong, please make it more clear in the methods section and presentation of results.

AUTHORS: please see our answer to your question #2 above.

Attachment

Submitted filename: Responses to reviewers_.docx

Click here for additional data file. (30.9KB, docx)

Decision Letter 1

Kelli K Ryckman

2 Feb 2021

Preterm disparities between foreign and Swedish born mothers depend on the method used to estimate gestational age. A Swedish population-based register study

PONE-D-20-31406R1

Dear Dr. Juárez,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Kelli K Ryckman

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #2: Yes

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #2: Thank you to the authors for their revised manuscript and thoughtful response to reviewer comments, I am satisfied that my concerns have been addressed in the revised version.

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #2: Yes: Steven Hawken

Acceptance letter

Kelli K Ryckman

11 Feb 2021

PONE-D-20-31406R1

Preterm disparities between foreign and Swedish born mothers depend on the method used to estimate gestational age. A Swedish population-based register study

Dear Dr. Juárez:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Kelli K Ryckman

Academic Editor

PLOS ONE

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Supplementary Materials

    S1 Table. Gestational age outcomes by method of estimation of gestational age by mother’s birthplace (corresponding to Figs 2 and 3).

    (DOCX)

    Click here for additional data file. (23KB, docx)
    S2 Table. Preterm according to LMP and ultrasound estimates by mother’s country of birth.

    (DOCX)

    Click here for additional data file. (23.8KB, docx)
    S3 Table. Post-term according to LMP and ultrasound estimates by mother’s country of birth.

    (DOCX)

    Click here for additional data file. (22.7KB, docx)
    S4 Table. Sensitivity analyses.

    Gestational age outcomes according to LMP and ultrasound estimates in a subsample of 718,191 of uncomplicated pregnancies.

    (DOCX)

    Click here for additional data file. (18.5KB, docx)
    S5 Table. Sensitivity analyses.

    Gestational age outcomes according to LMP and ultrasound estimates including 723,253 missing in different covariates.

    (DOCX)

    Click here for additional data file. (18.5KB, docx)
    S6 Table. Gestational age outcomes by method of estimation of gestational age by region of origin adjusted for maternal variables (corresponding to Fig 4).

    (DOCX)

    Click here for additional data file. (18.8KB, docx)
    Attachment

    Submitted filename: Responses to reviewers_.docx

    Click here for additional data file. (30.9KB, docx)

    Data Availability Statement

    Data cannot be shared publicly because of confidentiality. In order to access Swedish register data, researchers must contact the agencies responsible for data protection, conditional on ethical vetting (contacts below). The National Board of Health and Welfare (Socialstyrelsen) - Postal address: Socialstyrelsen SE-106 30, Stockholm, Sweden; Visiting address: Rålambsvägen 3, Stockholm; Phone: +46 (0)75 247 30 00; Fax: +46 (0)75 247 32 52; E-mail: socialstyrelsen@socialstyrelsen.se. Statistics Sweden (SCB) - Postal address: Statistics Sweden, Solna strandväg 86, SE-171 54 Solna, Sweden; Visiting address: Solna strandväg 86; Phone: +46 (0) 10 479 40 00; E-mail: scb@scb.se. Swedish Ethical Review Authority (Etikprövningsmyndigheten) - Postal address: Box 2110, SE-750 02 Uppsala, Sweden; E-mail: registrator@etikprovning.se; Phone: +46 (0)10 475 08 00.

    Articles from PLoS ONE are provided here courtesy of PLOS

    RESOURCES