Length of residence and caesarean section in migrant women in Sweden: a population-based study

Abstract

Background

Prior studies have reported substantial differences in caesarean rates between migrant and non-migrant women. In this study we investigate whether the association between maternal country of birth and caesarean section is modified by length of residence in Sweden.

Methods

Population-based register study. A total of 106 760 migrant and 473 881 Swedish-born women having singleton, first births were studied using multinomial multiple regression models to estimate odds ratios (OR) and 95% confidence intervals for mode of birth. Random effect meta-analyses were conducted to assess true heterogeneity between categories of length of residence.

Results

Longer duration of residence was associated with an increased overall risk of both unplanned and planned caesarean section among migrant women. This pattern was more pronounced among countries grouped as having higher prevalence (compared to Swedes) of unplanned: OR_≤1=1.41 (1.32–1.50); OR_>1–<6=1.49 (1.42–1.57); OR_6–<10=1.61 (1.50–1.72); OR_≥10=1.71 (1.64–1.79) and planned caesarean section [OR_≤1=1.14 (0.95–1.36); OR_>1–<6=1.30 (1.13–1.51); OR_6–<10=1.97 (1.64–2.37]; OR_≥10=1.82 (1.67–1.98)]. The results were robust to social, obstetric and health adjustments. There were some country-of-origin-specific findings.

Conclusions

The fact that the risk of unplanned and planned caesarean section tended to increase with length of residence, even with adjustment for social, obstetric and health factors, suggests that receiving country-specific factors are playing an important role in caesarean section.

Introduction

Disparities in caesarean section rates between migrant and receiving-country-born women have been reported.¹ Suggested explanations for higher rates among migrants include poorer general health and malnutrition,² gestational diabetes, high body mass index (BMI), feto-pelvic disproportion,¹ infectious diseases,³ less access to health care² or sub-optimal care.⁴ Possible explanations for lower rates include migrants being younger mothers, exposure to fewer interventions in labour and the ‘healthy migrant effect’, i.e. that migrants are healthier on arrival but their health declines over time. Non-medical factors seem also to be important, including language barriers, low social-economic status (higher rates) and cultural preferences (lower and higher rates).^3,5,6

Migrants’ length of residence in the receiving country is one factor that may modify the effect of origin-specific determinants, through changes in specific risk or protective factors, mode of birth preferences and responsiveness of care providers. Studies investigating possible associations between length of residence and caesarean section are few. In large nationally representative cohorts in the UK⁷ and Canada,⁸ residency of five years or more in the receiving country increased the risk of an unplanned caesarean section in multiparous women, but not in primiparous women, while a smaller Canadian study (n = 1025)⁹ found no increased risk of unplanned caesarean section among migrant women resident in Canada for two years or more.

A recent national population-based study from Norway¹⁰ studied the effect of length of residence in the receiving country in more detail. Irrespective of length of residence, the risk of an unplanned caesarean was 51–75% higher in migrant women than in Norwegian-born women. The risk of planned caesarean increased with increasing length of residence, but only in women from low incidence groups, i.e. migrant women from countries with an overall lower prevalence of caesareans in Norway compared with Norwegian-born women.

In this Swedish national population-based study, we had a sufficiently large sample to further investigate the effect of length of residence on the association between mother’s origin and caesarean section taking into consideration important covariates that might also explain changes over time.

Methods

Study population

Our study utilized records collected by the Medical Birth Register (MBR) between 1 January 1999 and 31 December 2012. The MBR includes ≈99% of all births occurring in Sweden,¹¹ which were linked to three other national registers using a unique personal identification number: the Total Population Register for information on the country of origin for both the mother and the father and length of residence in Sweden; the Swedish Income and Enumeration Survey for socio-economic information (maternal employment status, disposable household income); and the Multi-Generational Register to identify mothers born outside Sweden and adopted by Swedish parents (adoptee mothers).

From a total of 1 403 033 births (852 582 mothers), we excluded births of multiparous mothers (n = 787 616) to rule out the effect of previous mode of birth (figure 1). We also excluded the following: multiple births (n = 9107), observations where the mother was adopted (n = 484) or if the infant birth weight was biologically unlikely for its gestational age (n = 6782).¹² We also excluded observations with missing information on maternal country of birth (n = 163), length of residence in Sweden (n = 863), maternal age (n = 3) or type of caesarean (n = 17 374). The final sample consisted of 580 641 births.

Figure 1

Selection flow

Outcome, exposure and explanatory variables of interest

The outcome variable was mode of birth classified into three categories: vaginal (including instrumental vaginal), unplanned and planned caesarean, i.e. after and before the onset of labour, respectively. Maternal origin was one of two main exposure variables and we report findings for specific maternal countries of birth contributing 5000 births or more (before exclusions) during the study period. Length of residence was estimated by subtracting the year of immigration from the date of giving birth, and then categorized into ≤1 year, >1 to <6 years, 6 to <10 and ≥10 years.

Explanatory variables available for investigation were: year of birth (categorized as: 1992–99, 2000–08 and 2009–12); type of hospital (university/other); family disposable income estimated for the year before birth (salary, parental leave and other benefits) and categorized in quintiles; maternal employment the year before birth (yes/no/missing) and maternal marital status (cohabiting with the father/single/other status/missing). Maternal age was categorized as ≤24, 25–29, 30–34 and ≥35 years of age. Father’s origin was defined as foreign or Swedish-born. We included 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), obese (≥30.00 kg/m²) and missing.¹³ Maternal self-reported smoking in early pregnancy was classified as non-smoker, 1–9 cigarettes/day, 10 or more cigarettes/day and missing; hypertension and diabetes (gestational diabetes, pre-eclampsia and eclampsia) before and during pregnancy (yes/no); induction of labour (yes/no) and epidural analgesia (yes/no); breech presentation was defined by the ICD-9 codes (652.1) and ICD-10 (032.1), oligohydramnios (ICD-9: 658.0 and ICD-041.0) and placental disorders (ICD-9641 and ICD-10 O43). We also included infant birth weight (low birth weight <2500, normal weight 2500–3999 and macrosomia ≥4000 g) and gestational age (pre-term <37, term 37–41 and post-term ≥42 weeks).

All the above variables have previously been shown to be associated with caesarean section, or are hypothesized to be associated with differences in caesarean section by maternal country of birth.

Maternal employment and whether or not a woman’s partner was Swedish or foreign-born were included as proxies for increasing language fluency over time, as both are likely to serve to assist women with language acquisition.

Statistical analysis

Multinomial regression analyses were used to model simultaneously the different outcome categories (using vaginal birth as the reference category). We ran models adjusted by year of birth and maternal age and the available range of relevant social, obstetric and health covariates.

Using the prevalence of caesarean section (see figure 2), we classified maternal countries of origin into two groups: (i) those with a statistically significantly higher prevalence of unplanned/planned caesarean section in Sweden compared with Swedish-born women and (ii) those with a lower or similar prevalence. We excluded women from the Northern European countries (Denmark, Finland, Norway and Germany) as, with the exception of women from Finland, they did not show statistically significant differences compared to Swedes for either unplanned or planned caesarean section. Women from Turkey were not different from their Swedish peers regarding unplanned caesarean section but were kept in the analyses for consistency with the model for planned caesarean. We ran models to estimate the effect in these two groups by length of residence. The specific relative Odds Ratios (OR) for each length of residence category (years) is reported with the corresponding subscript in the text: OR_≤1, OR_>1–<6, OR_6–<10 and OR_≥10. We also present the information by maternal country of origin in order to discuss whether there is heterogeneity within these low and high prevalence groups. We estimated heterogeneity (I²) to formally test the effect modification by length of residence, i.e. to estimate the proportion of between-length of residence true heterogeneity from the total observed variation.

Figure 2

Prevalence of unplanned and planned caesarean section by maternal country of birth (%) Note: All maternal countries of birth show statistically significantly different prevalence compared with Sweden, with the exception of Denmark, Norway, Germany and Poland for planned caesarean section and women from Norway, Denmark, Germany and Turkey for unplanned caesarean section

The study was approved by the Central Ethical Review Board of Stockholm (decision no. 2014/4: 6). All analyses were performed using Stata software, version 14 (StatCorp, LP, College Station, Texas).

Results

Table 1 shows the characteristics of the study population. Overall, foreign-born mothers were more likely to have a foreign-born partner (66 vs. 10% among Swedish-born women), to be single (9 vs. 7%) and to have a lower income (41 vs. 14% in the lowest quintile), and were less likely to smoke in early pregnancy (6 vs. 8%), and to suffer from physical health conditions before and during pregnancy (6 vs. 7%). However, the figures for these maternal characteristics tended to increase for foreign-born women with length of residence.

Table 1

Characteristics of the sample (n = 580 641)

		Swedish	Foreign-born	Foreign-born by length of residence in Sweden
				≤1	2–5	6–9	≥10
		n = 473 881	n = 106 760	n = 30 203	n = 30 847	n = 14 539	n = 31 171
Variable	Categories	%	%	%	%	%	%
Background characteristics
Maternal age	<24	22.1	31.8	41.7	27.7	30.2	26.9
(years, age)	25–29	37.2	32.9	32.0	34.8	27.3	34.5
	30–34	29.8	24.2	18.5	26.1	28.8	25.8
	>35	10.8	11.1	7.8	11.3	13.7	12.8
Marital status	Cohabit-parent	88.3	86.2	89.7	86.8	82.8	83.8
	Single/other	6.9	8.9	5.4	8.6	12.4	11.0
	Missing	4.9	4.9	5.0	4.6	4.9	5.2
Father's origin	Born in Sweden	90.2	34.0	20.6	29.5	33.0	52.0
	Foreign-born	9.8	66.0	79.4	70.6	67.0	48.1
Maternal	Yes	85.1	43.9	5.6	40.7	63.0	75.1
employment	No	14.6	47.8	65.2	59.2	36.8	24.8
	Missing	0.4	8.3	29.2	0.1	0.2	0.1
Family	1 Lowest	13.6	41.0	59.0	39.5	35.1	27.6
disposable	2	11.9	19.7	23.5	21.0	17.0	16.0
income	3	17.3	12.3	7.9	12.8	13.8	15.3
(quintiles)	4	26.5	12.3	4.5	11.6	14.8	19.2
	5 Higher	30.7	14.8	5.1	15.0	19.4	21.8
	Missing	0.1	0.0	–	–	–	–
Year of birth	1999–03	31.9	25.8	26.6	23.1	32.0	24.9
	2004–08	37.1	36.9	38.0	35.0	33.2	39.6
	2009–12	31.0	37.2	35.5	41.9	34.8	35.5
BMI	Normal weight	59.5	59.9	59.1	60.9	61.2	59.0
	Underweight	2.3	4.7	5.4	5.1	4.6	3.7
	Overweight	20.1	19.1	19.1	18.8	19.3	19.5
	Obese	8.6	6.2	4.8	5.8	5.8	8.1
	Missing	9.6	10.1	11.7	9.4	9.2	9.7
Smoking during	Non-smoking	87.1	88.4	89.8	90.4	87.78	85.4
pregnancy	1–9	6.2	5.2	3.8	3.8	5.8	7.8
(cigarette/day)	>10	1.5	1.1	0.9	0.8	1.3	1.6
	Missing	5.2	5.3	5.3	5.0	5.1	5.3
PHPa	Yes/no	6.8	5.6	4.5	5.8	6.2	6.3
Indicationsb	Yes/no	45.2	43.9	41.3	45.0	46.0	44.6
Newborn's	<2500	3.5	4.3	4.1	4.0	4.4	4.9
Birth weight	2500–3900	81.0	85.6	87.4	85.6	84.0	84.6
(grams)	>4000	15.5	10.1	8.5	10.4	11.6	10.5
	Missing	0.0	0.0	0.0	0.0	0.0	0.0
Gestational	<37	5.6	5.3	4.8	4.9	5.2	6.1
age (weeks)	37–41	85.5	86.9	87.3	86.9	87.0	86.6
	≥42	8.8	7.8	7.9	8.2	7.8	7.2
	Missing	0.1	0.1	0.1	0.0	0.0	0.1
Epidural	Yes/no	44.8	43.6	40.3	43.2	44.3	46.9
induction of labour	Yes/no	12.4	12.7	11.7	13.0	13.6	13.1

a

MHP = Maternal health problems includes pre-pregnancy diabetes and hypertension, gestational diabetes, eclampsia and pre-eclampsia.

b

Indications includes breech presentation, oligohydramnios and placental disorders.

Figure 2 shows that unplanned caesareans were somewhat more prevalent among migrant women than among those born in Sweden: 12 and 11%, respectively. The prevalence of caesareans varied between women from different countries; between 8 and 21% for unplanned and between 3 and 10% for planned caesareans.

Figure 3A–D shows the effect of length of residence on the odds of experiencing an unplanned or planned caesarean section. The graphs represent women from countries contributing with higher and lower rates of caesareans in Sweden. Since most covariates have the potential to change with length of residence, the figure presents the adjusted results. The OR of having a caesarean section increased with the time spent in Sweden in all groups, independently of changes in individual social, obstetric and health factors over time.

Figure 3

(A–D) Risk for an unplanned (A–B) or a planned (C–D) caesarean section in women originating from countries with higher and lower prevalence of caesarean section compared with Swedish-born women, based on the information provided in figure 2, stratified by length of residence. A: Iran, Ethiopia, Somalia, India, Thailand, Korea Chile and rest of countries combined (higher prevalence of unplanned caesarean section). B: Poland, Iraq, Syria, Lebanon, Turkey and former Yugoslavia (lower prevalence of unplanned caesarean). C: Iran, Ethiopia, India, Korea, Thailand and Chile (higher prevalence for planned caesarean section). D: Poland, Syria, Iraq, Lebanon, Turkey, former Yugoslavia, Somalia and rest of countries (lower prevalence of planned caesarean section) Note: Odds ratios adjusted for birth year; maternal age; type of hospital; family disposable income; employment; father’s origin; hypertension and diabetes before and during pregnancy; BMI; smoking during pregnancy; birth weight; gestational age and indications for caesarean section (breech presentation, oligohydramnios and placental disorders). Lower and higher prevalence are defined according to figure  2

For unplanned caesarean section, the OR increased among those with overall higher prevalence than Swedish-born women [OR_≤1 = 1.41 (1.32–1.50); OR _>1–<6 = 1.49 (1.42–1.57); OR_6–<10 = 1.61 (1.50–1.72); OR _≥10 = 1.71 (1.64–1.79)] (figure 3A). The pattern was less apparent in the group with lower prevalence [OR_<1 = 0.96 (0.88–1.04); OR_>1–<6 = 1.12 (1.04–1.20); OR_6–<10 = 1.15 (1.04–1.27); OR_>10 = 1.18 (1.10–1.26)] (figure 3B).

For planned caesarean section, the OR also increased in a linear fashion in the first 10 years of residence in migrants with an overall higher prevalence [OR_≤1 = 1.14 (0.95–1.36); OR_>1–<6 = 1.30 (1.13–1.51); OR_6–<10 = 1.97 (1.64–2.37); OR_≥10 = 1.82 (1.67–1.98)] (figure 3C). In migrants with an overall lower prevalence than Swedish-born women the OR were initially lower [OR_≤1 = 0.69 (0.63–0.76); OR_>1–<6 = 0.89 (0.83–0.96)], but converged with the Swedish population by six years in the country [OR_6–<10 = 1.06 (0.97–1.17)] and exceeded it after 10 years [OR_>10 = 1.21 (1.13–1.30)] (figure 3D). The test for heterogeneity confirms the existence of effect modification by length of residence in all groups (P < 0.05). Supplementary tables S1 and S2 (Supplementary data) show that the results disaggregated by countries of birth are consistent overall with the general pattern shown in figure 3 for each category of duration of residence.

The OR for unplanned caesareans did not change with length of residence among women from the Northern European countries, while they show a moderate increase for planned caesareans (results upon request).

Discussion

Main findings

Longer duration of residence was associated with an increased overall risk of both unplanned and planned caesarean birth among primiparous migrant women. This was observed both for maternal countries of birth with lower and higher overall prevalence of caesarean sections in Sweden compared with Swedish-born women. These findings remained after adjusting for maternal social, obstetric and health-related characteristics. There were some country-of-origin-specific findings.

Interpretation

Studies to date have shown mixed findings in relation to length of residence and unplanned caesarean section with some migrant groups at increased risk with longer duration of residence and others at greatest risk in the first years after arrival; for still others, length of residence appears not to alter risk.¹⁴

Our findings are broadly consistent with those from the study conducted in Norway, especially for unplanned caesarean section; and this remained the case when adopting the same length of residence intervals as were used in the Norwegian study (<1 year, 1, 2–5 and ≥6 years; results available on request). The similarities between these studies are perhaps unsurprising, since in Norway, as in Sweden, antenatal and intrapartum care is free of charge, caesarean section rates and other infant outcomes are relatively similar and migrant women contribute around the same proportion of births annually.¹⁵

The fact that the risk of unplanned and planned caesarean sections tended to increase with length of residence, even with adjustment for social, obstetric and health factors, suggests that receiving-country-specific factors are playing an important role in caesarean section disparities between migrant and Swedish-born women. Explanations commonly offered for disparities found between migrants and receiving country populations may or may not be illuminating in relation to these length of residence findings.

While communication problems are thought to contribute to higher caesarean section rates in some migrants,¹ these problems might be expected to diminish in importance with longer duration of residence, as migrant women become more fluent in the receiving country language. Without access to data on Swedish language fluency however, it is not possible to rule out communication problems as a continuing contributing factor. Father’s origin might serve as a proxy for it and for social integration more generally, with an impact on the risk of caesarean section. In fact, in a previous study we showed an increased risk of unplanned and planned caesarean section when both parents were foreign-born.¹⁶ Similarly, a study in Sweden found an association between father’s origin and use of epidural,¹⁷ and the latter is associated with caesarean birth. In this study, we only found an interaction for women with high prevalence of unplanned caesareans (data not shown). Compared to Swedish-born couples, all foreign-born women had higher odds of unplanned caesareans regardless of the father’s origin and these increased with duration of residence.

It is unclear whether the increased risk of planned caesarean section with longer duration of residence might be explained by an increase in maternal request for caesarean birth among migrant women as a result of better Swedish fluency allowing women to express such a preference. Better communication between women and their caregivers should also enable more appropriate discussion and counselling about the risks of caesarean section in the absence of medical indications, with potential therefore to decrease the number of caesareans done for maternal request among migrant women. No information about maternal request was available in our dataset and caesarean on maternal request without a medical indication is not common in Sweden, especially among women having a first birth. A population-based Swedish study (n = 3065) showed that just 9% of migrant women and 8% of Swedish-born women said they would prefer a caesarean section.¹⁸ Women’s preferences have in fact remained relatively stable in Sweden over time and the rising caesarean rate has largely been associated with factors other than maternal request.¹⁹ A more recent study showed even lower rates for preferring a caesarean section in a Swedish sample of 920 women (primiparous 4.2% and multiparous 6.4%), but a slightly higher likelihood of such a preference in migrant women than in Swedish-born women.²⁰

It might be that health deteriorates with length of residence, as studies from other national contexts have suggested.²¹ Increasing obstetric risks and poorer health might contribute to more caesarean sections. We included in our models obstetric complications that are common indications for planned caesarean section (breech presentation, oligohydramnios and placental disorders)²² as well as health characteristics, such as BMI²³ and smoking.²⁴ The fact that being foreign-born remained statistically significant after the inclusion of these variables suggests that the differences found are not readily explained by worsening health among migrant women.

A recent paper from Canada has identified that being a refugee or asylum seeker with less than two years residence was a risk for unplanned caesarean section in low risk women, compared with women in other migrant categories.¹⁴ Unfortunately, we did not have migration status in our dataset to compare the results.

Alongside the general pattern in our data of increasing caesarean section with longer duration of residence were some country-of-origin-specific findings, a reminder that grouping migrants across countries of origin is not without problems. We observed e.g. that women from Turkey, Iraq and Somalia had a lower or similar risk of experiencing a planned section than Swedish-born women. Women coming from Somalia, however, had a higher probability of an unplanned section than Swedish-born women. Caesareans in women from Somalia, Turkey and Iraq mostly continued to increase over time. Some of these findings differ from those of the published meta-analysis of studies of caesarean section in international migrant.¹ For example, a systematically lower risk of overall caesareans among mothers from Thailand has been reported, while we found that Thais had almost twice the risk of unplanned caesareans than Swedes before and after adjustment, regardless of the time spent in Sweden. Few studies of Thai migrant women have previously adjusted for the range of factors adjusted for in this study, and this is a likely reason for the difference in findings with the earlier meta-analysis.

Strengths and limitations

This study had a number of strengths. These included the use of a large, linked, national population dataset with reliable information on the variables of interest, enabling investigation of the association between caesarean section and length of residence in migrant women, with the possibility of adjustment for a wide range of relevant social, obstetric and health factors.

Although we adjusted for relevant obstetric complications that have been shown to increase the risk of caesarean section (e.g.²⁵), the lack of information on the specific medical indication, means that we cannot fully rule out the possibility that the differences found are due to health-related differences between foreign and Swedish-born women. Hence, the purpose of this study has been to describe any disparities apparent between populations and the changes in prevalence of caesarean section with increasing length of residence among migrant women in Sweden. Length of residence data may be less precise for migrants with asylum seeking backgrounds, as year of arrival is defined in the registry by the year a residency permit is issued in Sweden. Therefore, length of residence for some may be under-estimated. Also, migrants without a Swedish personal identification number at the time of giving birth could not be included in this study, as this number is needed when linking information from different registers.

This is one of few studies to investigate caesarean section and length of residence in migrant populations. Further studies are needed to disentangle possible explanations for the persistent higher risks of unplanned caesarean section found among most migrant groups in Sweden and for the increase in planned caesarean with longer duration of residence. The inclusion of migration-related variables at the individual level in routine perinatal data collections—in particular, receiving country language fluency and migration status (refugee/non-refugee)—would enhance future research into explanatory factors. Finally, studies which investigate caesarean section decision-making and auditing of intrapartum care records of at-risk migrant groups compared with Swedish-born women might also shed light on factors that explain why caesarean section increases with longer duration of residence.

Funding

Centre for Clinical Research Dalarna and the Swedish Research Council for Health, Working Life and Welfare (FORTE) (#2012-01190, #2016-07128). The funder had no role in study design, analysis or preparation of the manuscript.

Conflicts of interest: None declared.

Key points

• The risk of unplanned and planned caesarean section increases with migrant length of residence.

• This modification effect persists after the inclusion of social, obstetric and health factors.

• The effect of length of residence is observed among women from countries with overall lower and higher prevalence of caesareans in Sweden.

• Receiving country-specific factors appear to play an important role in caesarean section disparities.

• Migrants should be considered a heterogeneous group and length of residence a valuable instrument for health surveillance.