Epidemiology/Health Services Research

Socioeconomic status, maternal risk factors, and gestational diabetes mellitus across reproductive years: a Finnish register-based study

Abstract

Introduction To evaluate the degree to which socioeconomic differences in gestational diabetes mellitus (GDM) are accounted for by differences in maternal risk factors, to assess whether age-related risks of GDM differ across socioeconomic groups, and to identify priority populations for future interventions.

Research design and methods We performed a register-based study using data from the Finnish Medical Birth Register and Statistics Finland on the 474 166 women who gave birth in Finland from 2008 to 2015. We collected information on GDM based on the International Classification of Diseases 10th Revision codes O24.4 and O24.9. We used multivariable models to examine the association between socioeconomic status, maternal risk factors, and GDM. We further tested interaction on multiplicative and additive scales.

Results The incidence of GDM was 8.7% in 2008–2011 and 12.5% in 2012–2015. Lower socioeconomic levels than upper level employees were associated with an increased risk of GDM. Up to 64.0% of socioeconomic differences in GDM were attributed to body mass index and 5.5% to smoking. There was evidence for effect modification. Relative to women in the upper level category who were aged less than 19 years, GDM adjusted ORs (95% CIs) for women 35 years or older in upper level versus long-term unemployed groups were 3.28 (2.08–5.18) and 5.29 (3.35–8.35), respectively.

Conclusions There is a paradox that socioeconomic advantage increases the incidence of GDM at the population level while reducing the incidence of GDM within the population. Nevertheless, socioeconomic differences in GDM persist and widen with increasing maternal age, even after accounting for maternal risk factors.

What is already known on this topic

  • Previous studies have found socioeconomic differences in type 2 diabetes mellitus. However, the relationship between socioeconomic status, maternal risk factors, and gestational diabetes mellitus (GDM) based on analyzing population-based data has not been adequately investigated.

What this study adds

  • Low socioeconomic status was constantly associated with increased adjusted ORs of GDM, and risk factors that contributed most to socioeconomic differences in GDM were overweight and obesity, and smoking.

  • We found postponing pregnancy to be more detrimental to women with lower socioeconomic status, even with equal access to health resources and after accounting for maternal obesity and smoking.

How this study might affect research, practice or policy

  • Policies addressing maternal risk factors and low socioeconomic status from the beginning of reproductive years might be a successful approach for reducing GDM risk at the population level; simultaneously, targeting priority populations is effective in narrowing socioeconomic inequalities in GDM within the population.

Introduction

Gestational diabetes mellitus (GDM) has become a challenge in maternity care.1 According to the International Diabetes Federation, an estimated 16.7% (21.1 million) of live births in 2021 were affected by some form of hyperglycemia, 80.3% of which were due to GDM.2 Studies indicated a consistent relationship between the socioeconomic status of women and type 2 diabetes mellitus (T2DM) in high-income countries, whether measured by educational level, income, or occupation.3 However, the mechanisms through which poor socioeconomic status and maternal risk factors contribute to the higher incidence of GDM have not been adequately investigated. There is evidence that GDM incidence increases along with the increase in living standards.4 5 At the same time, studies show that GDM incidence is inversely associated with socioeconomic status within the population; women with high socioeconomic status have a lower risk of GDM than women with low socioeconomic status. In other studies, GDM did not differ across socioeconomic groups once confounding factors were fully taken into account.6 7

The apparent discrepancy between the standard of living and the incidence of GDM at the ecosystem level and within the population can be explained by several mechanisms. Epidemiological transitions and changes in screening policies for GDM could be relevant. Furthermore, the relationship between maternal risk factors and GDM potentially may differ by socioeconomic status within the population. For instance, Geronimus and colleagues raised the possibility that advanced maternal age could be more harmful among women of lower socioeconomic status for adverse pregnancy outcomes.8 A similar mechanism could explain socioeconomic differences in GDM.

Understanding the socioeconomic paradox of GDM has public health importance. Therefore, we assessed the extent to which socioeconomic differences in GDM are accounted for by differences in maternal risk factors, to determine whether age-related risks of GDM differ across socioeconomic groups, and to identify priority populations for future public health interventions.

Methods

Data and study population

In this register-based study, data were derived from the Finnish Medical Birth Register (MBR) and Statistics Finland on the 474 166 women who gave birth in Finland from 2008 to 2015. The Finnish Institute for Health and Welfare maintains the MBR, which contains data for all live births and stillbirths with a gestational age of 22 weeks or more or birth weight of at least 500 g. We obtained data on women’s age, body mass index (BMI), smoking during pregnancy, parity, and marital status from MBR.9 Since 2004, the MBR has included information on pre-pregnancy maternal weight and height as well as on results of the oral glucose tolerance test (OGTT) for the screening and all pregnancy and delivery diagnoses, including GDM. We used the register of Statistics Finland to obtain information on the socioeconomic status of women. The data of MBR and Statistics Finland were linked by unique personal identification numbers and encrypted. In Finland, maternity care is regulated by law and is mostly publicly funded. All women have equal access to high-quality maternity care and health information, regardless of their societal position.

Outcome variable: GDM

The information on GDM was based on the International Classification of Diseases 10th Revision (ICD-10) classification. We included ICD-10 codes O24.4 (gestational diabetes) and O24.9 (unspecified gestational diabetes in pregnancy) to identify the presence of GDM. These ICD-10 codes were compiled together and dichotomized (yes/no). In Finland, GDM is diagnosed by risk-based screening (age ≥40 years, BMI ≥25 kg/m2, fetal macrosomia in current or previous pregnancy, GDM in a previous pregnancy, or family history of T2DM) before 2008. Since 2008, screening has been done using a 75 g 2-hour OGTT between 24 and 28 gestation weeks for all pregnant women, except for primiparous women aged less than 25 years without a first-degree family history of T2DM and with normal BMI (18.5–24.9 kg/m2). In Finland, the diagnostic thresholds for GDM are fasting plasma glucose ≥5.3 mmol/L, 1-hour ≥10.0 mmol/L, and 2-hour ≥8.6 mmol/L.10

Exposures: socioeconomic status and maternal age

We classified socioeconomic status according to the Classification of Socio-economic Groups 1989 from Statistics Finland, a widely used classification of social and economic position in research and welfare surveys. The Classification of Socio-economic Groups was first published in 1983, based on the Classification of Occupations 1980, and the classification was reviewed in 1989. The classification is based on international recommendations. The United Nations Economic Commission for Europe has given general statistical recommendations for population censuses to form the Classification of Socio-economic Groups. This classification is comparable to the Nordic classification of socioeconomic groups.11

The classification is likely to capture the person’s stage in life (family member, student, etc) and for economically active individuals, occupation, and occupational status (self-employed, wage or salary earner, assisting family member).11 We focused on the following categories: self-employed persons, upper level employees, lower level employees, manual workers, students, long-term unemployed, and others. Others include women of all unclassified occupations and unknown socioeconomic status. Upper level employees include employees working in management tasks of public administration, enterprises or organizations, employees working in planning, research and presentation, those working in education and other employees generally with higher university degrees. Lower level employees include employees in management and employees in clerical, sales, care and other tasks. Long-term unemployment refers to women who have been unemployed for at least 6 months, while women who have been unemployed for under 6 months are classified according to the occupation and workplace before unemployment. Unemployed women who have not yet been at work at all are classified in the group of others.11 Women in the category of upper level employees were selected as the reference group, which corresponds to the number of women with varying levels of university degrees (online supplemental table S1).

Moreover, we categorized maternal age as <19, 20–24, 25–29, 30–34, and >35 years according to most of the maternal health disparities literature. The age group of women below 20 years of age was considered as the reference group since the risk of GDM was the lowest in this age group.

Covariates

We calculated BMI from self-reported height and weight, checked and recorded at the first antenatal visit between 8 and 12 weeks of gestation by midwives and public health nurses, categorizing as underweight (BMI <18.5 kg/m2), normal weight (18.5–24.9 kg/m2), overweight (25–29.9 kg/m2), and obese (≥30.0 kg/m2). Smoking status was divided into the following categories: non-smoker, stopped smoking during the first trimester, continued to smoke after the first trimester, and missing information. Parity was categorized as nulliparity with no previous birth and multiparity with one or more previous births. Marital status was categorized as married or cohabiting and single.

Statistical analysis

The statistical analyses were performed using IBM SPSS Statistics (V.28.0; IBM) and Stata (V.18.0; StataCorp, College Station, Texas, USA). The significance level for hypothesis testing was 0.05.

Baseline characteristics of women were evaluated by GDM using Pearson’s χ2 to test for associations and to investigate whether the distribution of women in categorical variables differs across the categories of outcomes.

Multivariable logistic regression analyses were conducted to estimate the ORs, adjusted ORs (aORs), and 95% CIs of GDM with socioeconomic status and other covariates (maternal age, BMI, smoking during pregnancy, parity, and marital status). In the first analysis (model 1), we assessed the crude association between socioeconomic status and GDM. Model 2 for the association between socioeconomic status and GDM included maternal age and parity. We estimated the aORs by including other covariates in model 2. We determined the contribution of each covariate to the association between socioeconomic status and GDM in model 2 by calculating the percentage reductions in aORs using the formula (OR Model 2−OR Model ×)/(OR Model 2−1). The procedure explained above was repeated across time periods of 2008–2011 and 2012–2015 to take into account changes in risk factor prevalence for GDM.

To assess effect modification by socioeconomic status, we assessed interactions on both multiplicative and additive scales, as recommended by VanderWeele and Knol.12 To test the interaction hypothesis on the multiplicative scale, we categorized the sample based on both socioeconomic status and maternal age, and the stratum with the lowest risk was selected as the reference category. We then compared the aORs against this sole reference category. To assess additive interaction, we calculated the relative excess risk due to interaction (RERI) and the corresponding 95% CIs, as suggested by VanderWeele and Knol.12

Results

Of the 474 166 women who gave birth in Finland from 2008 to 2015, altogether 50 173 (10.6%) were diagnosed with GDM (table 1). The incidence of GDM was 8.7% in 2008–2011, and 12.5% in 2012–2015 (online supplemental table S2).

Table 1
|
Characteristics of the study population stratified by GDM in Finland from 2004 to 2015

Relative to women without GDM, women with GDM were more likely to be long-term unemployed, in manual occupations, 30–34 years of age or 35 years or older, smokers (first trimester and during pregnancy), multiparous, and overweight or obese (table 1). The main characteristics of women stratified by socioeconomic status were summarized in online supplemental table S3, illustrating that long-term unemployed women and women in the category of others were more likely to smoke during pregnancy, and BMI >30.0 kg/m2 was more prevalent among long-term unemployed women.

The crude ORs in table 2 and figure 1 revealed that GDM occurred more frequently among women with lower socioeconomic levels than upper level employees. After adding maternal age and parity to the crude model (model 1, table 2, online supplemental table S4), the strength of the associations between socioeconomic groups and GDM increased across all time periods. From 2008 to 2015, the aOR of GDM was 1.91-fold among long-term unemployed women compared with upper level employees (table 2/model 2). In model 3, after including BMI, the gap in GDM across socioeconomic groups decreased during this period; BMI alone explained 41.6–64.0% of the excess risk associated with socioeconomic differences in GDM from 2008 to 2015. When the variable smoking was added to model 2, smoking alone explained 2.6–5.5% of the associations in model 4 from 2008 to 2015. However, in model 5, the contribution of marital status to the socioeconomic differences in GDM was small. From 2008 to 2015, the contribution of BMI and smoking to the socioeconomic differences in GDM increased (table 2). By contrast, the socioeconomic differences in GDM narrowed from 2012 to 2015 (table 2).

Figure 1
Figure 1

Proportion of women with gestational diabetes mellitus (GDM) in percentage by age and socioeconomic groups from 2008 to 2015.

Table 2
|
Contribution of socioeconomic status to GDM from 2008 to 2015

Among all covariates, the strongest association was observed between maternal obesity and GDM (OR 11.69, 95% CI 10.70–12.77), overweight and GDM (OR 4.77, 95% CI 4.37–5.21), and maternal age 35 years or older (OR 3.32, 95% CI 3.01–3.66) and GDM (online supplemental table S4). The association between socioeconomic status and GDM remained when we adjusted for maternal age or other covariates (online supplemental table S4).

In table 3, compared with upper level groups aged less than 20 years of age, the aOR for GDM among women 35 years or older in similar socioeconomic groups was 3.28 (95% CI 2.08–5.18) but rose to 5.29 (95% CI 3.35–8.35) for long-term unemployed women aged 35 years or older. A positive multiplicative interaction was observed between socioeconomic status and maternal age on GDM (p value for interaction <0.001; aOR 1.00, 95% CI 1.00–1.001; table 3).

Table 3
|
Multivariable logistic regression models of the association between socioeconomic status, maternal age and GDM from 2008 to 2015

We found positive additive interaction between socioeconomic status and maternal age on GDM for subgroups of women with RERI larger than 0. The additive effect of socioeconomic status and maternal age was greater than the sum of their individual effects; being 30–34 years of age or 35 years or older and in manual occupation, students, or long-term unemployed led to a greater risk of GDM (table 4). The results of attributable proportion due to interaction and the synergy index are presented in online supplemental table S5.

Table 4
|
Relative excess risk due to interaction (RERI) and 95% CI for GDM

Discussion

In our population, GDM incidence increased from 8.7% in 2008–2011 to 12.5% in 2012–2015. At the same time, socioeconomic differences in GDM reduced but nonetheless persisting even after adjusting for maternal risk factors; the greatest difference for GDM across socioeconomic groups was for BMI. Our findings showed women with lower socioeconomic levels than upper level employees, particularly long-term unemployed women, experienced higher aORs of GDM with increasing maternal age. This disproportionate socioeconomic-related risk of GDM across age groups was not accounted for by differences in BMI or smoking status.

At the population level, during a period of advances in well-being and welfare policies, the incidence of GDM increased,13 14 while socioeconomic differences in GDM narrowed within the population. This paradoxical situation occurred alongside epidemiological transitions, including increases in rates of maternal obesity and maternal age.14 15

Our results show that women of lower socioeconomic status had a higher risk of GDM, in line with other studies that used education as a proxy for socioeconomic status.7 15 We performed an appropriately conservative statistical adjustment and found that socioeconomic differences in GDM in our population partly result from inequalities in smoking during pregnancy, despite strong tobacco control efforts, and maternal obesity.16 17 In addition, persistent socioeconomic differences in GDM, regardless of well-developed screening and welfare policies, indicate that women’s knowledge or other elements associated with socioeconomic status predict health-related behavior.14 Therefore, factors associated with socioeconomic status can be used to prevent GDM or reduce the incidence of GDM, as women of higher socioeconomic groups with higher levels of education and income experienced in our studies. Mackenbach suggested that during the period of epidemiological changes, health improvement depends on behavior changes such as healthy diet and quitting smoking.14 These changes occur first among the higher socioeconomic groups; this can partly explain the socioeconomic differences in GDM.14

Our results indicate that consequences of postponing pregnancy intentionally or unintentionally (eg, subfecundity related to obesity) were more harmful for GDM among those with lower socioeconomic levels than upper level employees, as also reported in other settings.18 19 However, interaction between socioeconomic status and maternal age regarding GDM was not tested. Our findings of the vulnerability of long-term unemployed women and women in manual occupations could arise through the clustering of risk factors and as the result of cumulative effects of psychosocial stresses and disadvantages associated with low socioeconomic status over the reproductive years.20 Furthermore, socioeconomic differences in maternal obesity and smoking have been evidenced from the beginning of reproductive years.16 17 Earlier onset of smoking and obesity among women with lower socioeconomic status means a longer duration of exposure and more severe consequences on pregnancy outcomes and the future health of mothers.21–23 Therefore, a more detrimental effect of risk factors among women with lower socioeconomic status could contribute to the socioeconomic differences in the incidence of GDM within the population. Unlike other studies, barriers to health resources do not explain the GDM paradox in Finland, where all women have the same access to health information and services.

Using the MBR with a high degree of coverage and a methodology free from selection bias enabled us to evaluate the association between socioeconomic status, maternal risk factors, and GDM.24 Finnish register data contain detailed information about maternal health and socioeconomic status over the life course and have been considered a potential gold mine for health inequality research.25 The large sample size allowed us to assess effect modification across subgroups, constituting a major strength of this study.

Using occupation and occupational status as the proxy of socioeconomic status was both a strength and a weakness. In Finland, education, occupation, and income are inter-related; women of higher socioeconomic status have higher education and income and can afford to live in higher socioeconomic areas. However, we acknowledge that indicators of socioeconomic status are not interchangeable, and each indicator has its own dimensions. Also, self-reported height and weight are known to be subjected to misreporting,26 but as midwives or public health nurses measured and checked height and weight at the first antenatal visit, the risk of measurement errors is small. As our data were based on routinely collected information for administrative purposes, we had no information on other factors that could affect the association between socioeconomic status and GDM, including lifestyle-related factors and gestational weight gain. Prior studies have found that excessive gestational weight gain was associated with an increased risk of GDM.27 In this study, we had no information on whether excessive gestational weight gain could affect the strength of the associations; this topic warrants further investigations. Our results are relevant in countries where women of lower socioeconomic groups have a higher risk of GDM, which cannot be explained by maternal risk factors or interventions.

In our study, younger mothers showed lower risks of GDM than other age groups; however, pregnancy in adolescents and young mothers is already considered a high-risk status because of an increased risk of adverse pregnancy outcomes28; particularly pregnancies occurring in young mothers with GDM would be expected to be at high risk.29 Additionally, studies have indicated that GDM in young mothers is becoming increasingly common, and it seems GDM cannot be only attributed to women 35 years or older.30 It is therefore fundamental to develop strategies for the prevention and intervention of GDM from the beginning of the reproductive years.

Moreover, addressing only unhealthy behaviors at the population level benefits mainly women with higher socioeconomic status and further increases inequalities.14 31 Based on proportion universalism, public health interventions should be universal, at the same time applying targeted interventions to narrow the maternal health inequality gap. To apply this strategy, we assessed interaction on the additive scale to identify priority populations who benefit the most from targeted interventions.32 Based on our results, maternal age of 30 years and older among long-term unemployed women, women in manual occupations and students generate more cases of GDM. Thus, the increasing trend in postponing pregnancy could increase GDM among women with lower socioeconomic status at the ecosystem level. In this situation, interventions directed at low socioeconomic status while considering maternal age would be able to prevent most cases of GDM, reduce the burden of GDM at the population level, and decrease inequalities in GDM within the population.32 This approach could be tested in designated prospective clinical trials. For instance, a prospective randomized controlled trial indicated that even simple lifestyle interventions among high-risk mothers reduced the incidence of GDM by 39%.33

Conclusions

In this study, we found socioeconomic differences in GDM, even after adjusting for maternal risk factors. The association between socioeconomic status and GDM differed across socioeconomic groups, being more pronounced among long-term unemployed women. BMI explained a substantial part of the socioeconomic differences in GDM. Interventions to target and reduce major maternal risk factors, particularly by targeting obesity and smoking at the population level, have been insufficient to counteract the increasing incidence of GDM or to reduce health inequalities within the population. Lower socioeconomic levels than upper level employees, particularly long-term unemployment, seem to exert a stronger influence among women 30 years or older, increasing their already higher risk of GDM. These findings suggest that addressing socioeconomic circumstances along with maternal risk factors from the beginning of the reproductive years could reduce the GDM burden at the ecosystem level. In addition, targeted intervention among high-risk mothers would yield a greater reduction in GDM and could narrow the socioeconomic inequality in GDM within the population.