Background: Fetal growth is highly socially patterned and is related to health across the life course, but how the social patterns of fetal growth change over time remains understudied. The time trends in maternal social disadvantage in relation to fetal growth were examined in the context of a universal welfare state under changing macroeconomic conditions over a 24-year period.
Methods: All births in Denmark from 1981 to 2004 were included, and the association between maternal social disadvantage and birthweight was examined for gestational age z-scores over time using linear regression.
Results: All measures of social disadvantage were associated with decreased fetal growth (p<0.001), but with considerable differences in the magnitude of the associations. The association was strongest for non-Western ethnicity (−0.28 z-score), low education (−0.19), teenage motherhood (−0.14), single motherhood (−0.13) and poverty (−0.12) and weakest for unemployment (−0.04). The deficit in fetal growth increased over time for all associations except for unemployment. Also, the measures of social adversity increasingly clustered within individuals over time.
Conclusion: Maternal social disadvantage is associated with decreased fetal growth in a welfare state. Social disadvantage is increasingly clustered so that fewer pregnancies are exposed, but those exposed suffer a greater disadvantage in fetal growth. The economic upturn in the last decade did not appear to weaken the association between maternal social disadvantage and decreased fetal growth.
Statistics from Altmetric.com
Fetal growth is related to health in both the long and the short term. Infants who are born light are at increased risk of neonatal and infant death. Low weight at birth has been associated with increases in adult life morbidity, including cardiovascular disease and diabetes.1 2 Low offspring birthweight has also been reported to be associated with increased paternal and maternal morbidity.3–5
Fetal growth is related to social circumstances. Indicators of maternal disadvantage such as low education and poverty are known to be associated with decreased fetal growth.6 7 As with other social gradients, the association between maternal social disadvantage and fetal growth is unlikely to be static over time. This is because the effect of social disadvantage is mediated by factors such as maternal smoking and body mass index (BMI), which are not related to disadvantage in any fixed or deterministic fashion. There might be important period effects in the association because the mothers’ health and health behaviour changes over time.
The use of Denmark as a study setting provides the opportunity to examine time trends and two hypotheses of interest. Denmark is an example of a Scandinavian universal welfare state, which provides extensive services to its citizens, eg unemployment benefits, sickness benefits, cash benefits, universal healthcare and prenatal care. Some markers of social disadvantage such as unemployment and poverty are directly associated with access to the benefits that might buffer the effects on fetal growth. If these benefits buffer the effects of social disadvantage, we would expect the effect of social disadvantage on fetal growth to be smaller in social environments that provide higher levels of benefits. Denmark also provides a setting for examining how economic growth affects the association between social disadvantage and fetal growth: Denmark experienced a prolonged economic recession from the early 1980s to the early 1990s, with the unemployment rate reaching 12% in the mid-1990s. From the mid-1990s until the end of the study period, Denmark experienced economic growth, with decreasing unemployment rates, growth in gross domestic product (GDP), surplus on the government budget and decreasing public debt.8 9 This provides an opportunity to see whether the prevalence of maternal social disadvantage and its inverse association with fetal growth are diminished during an economic boom.
The aim of this paper is to examine the temporal trends in social adversity and the associations between markers of social adversity and fetal growth over time.
MATERIAL AND METHODS
Data were obtained from the Medical Birth Registry in Denmark, which contains records on all births. This information was supplemented with health and socioeconomic data on the mother, father and child through linkage with other national registries (the Medical Birth Registry, the National Hospital Registry, the Integrated Database for Labour Market Research, the Population Registry and the Register of Transfer Payments). We included a total of 1 409 339 births from 1981 to 2004.
We excluded infants with missing values for gestational age or birthweight, infants with gestational ages of less than 28 or more than 44 completed weeks, and infants with implausible values of birthweight by gestational age according to the method described by Alexander et al.10 The percentage excluded was 1.8%.
Birthweight was recorded in grams. Gestational age was recorded in completed weeks. Birthweights for gestational age z-scores were calculated according to birthweight within strata of gestational age and sex. Parity was included as a dichotomous variable (primiparous vs multiparous)
Teenage motherhood was defined as a maternal age of less than 20 years at the time of birth of the offspring. Poverty was based on the income of the household in which the mother resided in the year before birth. Poverty was defined as residing in a household whose total taxable income after adjustment for household size fell below 60% of the mean household size-adjusted taxable income for all households where a birth occurred in a given year, which is a measure of relative poverty. It should be noted that Denmark does not have an official definition of poverty, and that the definition used here is specific to our population of mothers and does not pertain to the whole Danish population. Low education was defined as a recorded education of primary schooling or less. Primary school (9 years) was mandatory in Denmark during the period of investigation. Non-Western maternal ethnicity was defined as a maternal place of birth other than the Organization for Economic Cooperation and Development (OECD) countries. Single motherhood was defined as failure to meet any of the following criteria: being married or in a registered partnership, residing with the father of the child, residing with an unrelated person of the opposite sex with a maximum age difference of 15 years.
The intercorrelations between the markers of social adversity were calculated using polychoric correlations. Linear regression was used to model the relationship between the dependent variable z-scores for birthweight for gestational age. This was done in three steps. First, we regressed birthweight for gestational age z-scores on each marker of social adversity, parity and year of birth. This was done to estimate the effect of each marker while adjusting for parity and birth year. Then, we regressed the z-scores on each marker of social adversity, parity, year of birth and the interaction between year of birth (as a continuous variable) and the marker of social adversity in question. This was done to evaluate whether there was a (linear) trend in the effect of marker of social adversity over time. Finally, we regressed the z-scores on each marker of social adversity, parity, year of birth and the interaction between year of birth as a categorical variable to allow for non-linear time trends. Estimates from this model were plotted against year of birth to produce figure 2. Levene’s test was used to assess heterogeneity of variance.
The prevalence of teenage motherhood declined from 5% at the beginning of the period under study to 1% at the end. Similarly, the prevalence of low education—defined as mandatory schooling only—dropped from 50% to 22%. Single motherhood, unemployment and poverty remained stable with some fluctuations over time. The proportion of mothers with non-Western ethnicity rose from 2% in 1981 to 8% in 2004 (see figure 1).
The different indicators of social adversity were intercorrelated in this population (see table 1). Ranked by size of the tetrachoric correlation, the association was strongest between low education and teenage motherhood, poverty and single motherhood, poverty and teenage motherhood, poverty and non-Western ethnicity, poverty and low education, single motherhood and teenage motherhood, low education and non-Western ethnicity, and low education and single motherhood. Weaker associations were observed between unemployment and low education, poverty and unemployment, non-Western ethnicity and teenage motherhood, unemployment and single motherhood, and unemployment and teenage motherhood. Inverse associations were observed between non-Western ethnicity and single motherhood as well as between unemployment and non-Western ethnicity. The overall pattern of intercorrelation shown in table 1 hides considerable variation over time in the pattern of association. The correlations between unemployment and low education and between unemployment and non-Western ethnicity fell during the period, while all other correlations rose. The means of the 15 off-diagonal correlations displayed a monotone pattern of increase from r = 0.20 in 1981 to r = 0.38 in 2004 (data not shown). The most dramatic change in the prevalence of the markers of social adversity was the decrease in low education. When low education was excluded from the calculations, the mean intercorrelation increased from r = 0.09 to r = 0.21. The increase was thus not solely attributable to the secular trend in maternal education.
The mean birthweight for gestational age z-scores increased substantially from 1981 to 2004 (see figure 2). The associations between the different measures of social adversity and birthweight for gestational age z-scores over time are shown in figure 3. With the exception of unemployment, all measures were consistently associated with decreased fetal growth. Among the measures of social disadvantage, non-Western ethnicity (β = −0.28) was associated with the largest disadvantage in fetal growth, followed by low education (−0.19), teenage motherhood (−0.14), single motherhood (−0.13) and poverty (−0.12) and weakest for unemployment (−0.04). The association with birthweight for gestational age z-scores did not weaken for any of the measures of social disadvantage except unemployment, which was not associated with fetal growth at conventional levels of significance. Increases in the deficit in birthweight for gestational age z-scores were observed for all measures of social disparity except unemployment. The strongest absolute increase was observed for non-Western ethnicity, which was associated with a −0.14 deficit in 1981. In 2004, the corresponding estimate was−0.29. A change of a similar magnitude was seen for household poverty. In 1981, the difference was −0.01, but in 2004, the small effect due to poverty in 1981 had turned into a −0.14 deficit. Teenage motherhood (β = −0. 08 to β = 0. 15), low education (β = −0.13 to β = −0. 20) and single motherhood (β = −0.09 to β = −0.15) also became more strongly associated with decreased fetal growth from 1981 to 2004. Over time, there was a significant change in the magnitude of the association between unemployment and fetal growth (β = −0.08 to β = −0.02). The linear trends over time were all statistically significant at conventional levels (data not shown). When the six measures of maternal social adversity were mutually adjusted, the associations attenuated, but the associations and trends over time remained significant (data not shown). The increase in the deficit over time was thus not solely attributable to the increased clustering, but reflects a combination of increased clustering and an increased deficit in fetal growth associated with the measures of social disadvantage.
We conclude that non-Western ethnicity, low education, teenage motherhood, single motherhood and poverty were associated with decreased fetal growth. Unemployment was only weakly associated with fetal growth. This might be attributable to the existence of state benefit programmes such as social security and state-subsidised unemployment insurance that buffer its impact. Except for unemployment, all measures of maternal social disadvantage had an increasing inverse association with fetal growth over time. Over the period of observation, the measures of social disadvantage increasingly clustered, so that fewer mothers were disadvantaged, but those that were became more likely to experience multiple disadvantages.
This study is based on the whole Danish population and, as such, is not subject to biases due to selection. As all information was gathered from registries, information bias arising from self-reports was not present. A weakness of information gathering from administrative registries is that breaks in the time series might occur as a consequence of changes in administrative routines. In our data, the drop in relative poverty from 1994 to 1995 is likely to be an artefact caused by a change in legislation that meant that income from several types of social benefits was changed from net sum into a taxable gross sum as part of a tax reform.11 The artefactual change in registration does not appear to have changed poverty’s association with fetal growth in any major fashion, however. Another weakness of this study is that we do not have access to information on the relevant intermediary variables, perhaps most importantly maternal smoking and maternal anthropometrics, because previous research has shown these two factors account for a sizeable part of the variation in fetal growth.
In our interpretation of the results, we assume that a high z-score is “better” (ie healthier) than a low z-score. This is a reasonable assumption for most of the birthweight distribution, but it should be noted that very high birthweights are also associated with increased morbidity and mortality.7–9 12 This problem is perhaps particularly important in this study population where the mean birthweight increased substantially in the period under study as a result of changes in factors such as maternal anthropometrics and maternal smoking. The problems associated with the use of birthweight and z-scores have been reviewed in detail elsewhere.12 13
Some studies have examined time trends in maternal social disadvantage in relation to offspring fetal growth (eg 14 15), but only a few have done so within the context of the universal welfare state. Two studies of maternal educational attainment in relation to preterm birth and fetal growth in Denmark, Finland and Norway and Sweden from 1981 to 2000 show that inequalities persist over time.16 17 The present study makes it possible to compare directly the associations between fetal growth and teenage motherhood, non-Western ethnicity, single parenthood, low education, unemployment and poverty. It is perhaps surprising that socially patterned differences in fetal growth continue to exist and even increase. A possible explanation is that the fall in the prevalence of smoking and the increase in the prevalence of overweight and obesity occur in a way that increases the association between social disadvantage and fetal growth: For example, if mothers of non-Western ethnicity over time were less likely to quit smoking during pregnancy and/or less likely to have a high pre-pregnancy BMI compared with mothers of Western ethnic origin, this could explain the increased difference in the deficit in fetal growth experienced by this group. There is, however, not much data available to examine the time series in smoking and pre-pregnancy BMI in the period under observation. Increased clustering would also increase the association between single measures of social disadvantage, but an increase over time was observed for most measures even under mutual adjustment. It is possible that genetic differences account for parts of the association between non-Western ethnicity and decreased fetal growth, but it should be noted that the non-Western ethnicity group is genetically very heterogeneous and that non-genetic factors are perhaps more likely to explain the deficit. Although antenatal care is offered to all women in Denmark, the socially disadvantaged receive less care.18 The guidelines for antenatal care stress the importance of reaching the socially disadvantaged,19 and this service has (in theory) become increasingly needs-based and targeted towards the socially disadvantaged over the period of observation, but this does not appear to have removed the deficit in fetal growth. Fetal growth is not only affected by maternal factors at the time of pregnancy, but also by the mother’s environment, growth and development throughout her life course.20 From a life course perspective, it is conceivable that some of the changes in the association between social disadvantage and fetal growth reflect changes in exposures that occurred long before the pregnancy under study.
In conclusion, our findings suggest that the association between social disadvantage and fetal growth is not going away, even in a universal welfare state in a period of economic growth. In fact, it appears to be getting stronger. In addition, social disadvantage is increasingly clustered, so that fewer mothers are affected, but those who are disadvantaged with regards to one characteristic are more likely to be disadvantaged with regards to other characteristics.
What is already known on this subject
Fetal growth is important for short- and long-term health.
Maternal social disadvantage affects fetal growth.
What this study adds
Maternal social disadvantage is associated with decreased fetal growth in a universal welfare state.
Over time, social disadvantage increasingly clustered, so that fewer pregnancies were exposed, but those exposed suffer a greater disadvantage in fetal growth.
The economic upturn in the last decade did not appear to weaken the association between maternal social disadvantage and decreased fetal growth.
Competing interests: None.
Funding: NorCHASE was funded by the Research Program in Longitudinal Epidemiology, which is supported by the Nordic Council of Ministers and administered by NordForsk (the Nordic Research Board).
See Commentary, p 267
Ethics approval: The study is based on registry data. According to Danish legislation, ethics committee approval is not warranted.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.