This long-term population-based study from a relatively small conurbation with a stable population reports a clear socioeconomic gradient for continuous birthweight, LBW and preterm birth, with the higher risk of reduced birthweight, LBW and preterm birth in more deprived groups compared to the least deprived group using an area level deprivation measure. Over the four decades, mean birthweight increased and the percentage of LBW decreased across all deprivation groups. However, the percentage of preterm birth decreased in the least deprived groups and increased for the most deprived groups, thereby widening the gap between the most affluent and most deprived groups. There was a significant interaction between area deprivation and decade of birth for birthweight and preterm birth, while for LBW the SES gradients remained relatively smaller. In each SES group mean birthweight adjusted for gestational age (in all and term births) increased over time with a simultaneous decrease in the proportion of LBW, suggesting a possible increase in fetal growth across all TDS groups.
The findings of this study are consistent with those in similar studies across larger geographic areas in the United States, Canada, Europe, which report socioeconomic disparity for birthweight outcomes [15, 18, 25] and preterm birth [18, 20, 26, 27, 33, 34]. Socioeconomic inequality in perinatal outcomes has been reported in both developing and developed countries [18, 19, 34–37], where socioeconomic gradients are lower and universal access to high quality prenatal and other medical care exists. Socioeconomic differences in birthweight and preterm birth have been observed in other British locations [7, 13, 20, 36, 38] as well as in other European [34, 37, 39] and non-European societies with different ethnic structure [18, 20, 26, 40]. The inequalities in perinatal outcomes, particularly in fetal growth restriction, between more and less affluent socioeconomic groups are consistent across different societies irrespective of the measures of SES employed: individual, e.g. parental education , social class based on paternal occupation , or area-based measures [40, 41]. Racial differences [15, 40] or maternal demographic characteristics  cannot entirely explain the SES inequalities, although their contribution is substantial in locations with multiethnic populations. Despite an established association of SES with birthweight and/or preterm birth, only few population-based studies from European countries investigated temporal trends in the association of SES with these outcomes [34, 36, 37, 39]. These studies, exploring trends during the last 20 years of the 20th century, showed that the SES gap was either relatively stable or, similar to other health outcomes , was widening with time, despite temporal changes in society structure, maternal demographic characteristics and efforts to equalise access to medical care for all social groups. This tendency did not improve for very preterm birth (22–32 weeks) rate during 1994–2003 in the Trent health region, UK . Our Newcastle study over the four decades, which were characterised by economy transformation in parallel with temporal changes in social structure of the population, demographic characteristics (maternal age and parity) and improvement in people’s living conditions, nutrition and prenatal care, found that the gap between most and least deprived groups did not narrow for birthweight and widened for preterm birth.
Socioeconomic inequality in birthweight is shown to be mediated by factors which can directly affect fetal growth, such as fetal exposure to maternal smoking . The interaction between measures of socioeconomic and physical environment (i.e. traffic-related exposure) has been also demonstrated for birthweight, but less so for preterm birth . The factors mediating the impact of socioeconomic inequality on birthweight may differ between developing and developed countries. Differences in nutrition and access to medical care may be more important in developing countries , whereas differences in maternal anthropometric characteristics, such as pre-pregnancy weight, BMI , lifestyle , and environmental exposure, such as air pollution , may be leading contributors in industrialised countries. The adjustment for such individual-level factors as gestational age, maternal age, parity and infant sex attenuated our results of reduced birthweight in all TDS quartiles compared to the reference least deprived quartile. The analysis of the association between mean maternal age by TDS quartile for each decade showed that in all four decades, mean maternal age was significantly higher for the least deprived group compared to the more deprived groups, gradually decreasing from the least deprived to the most deprived group in the last three decades (Additional file 1: Table S1). These differences in mean maternal age may have contributed to the socioeconomic inequality in birthweight, as younger maternal age is known to be associated with lower birthweight. However, the socioeconomic gradients in birthweight were evident after adjustment for maternal age in the model, suggesting the independent effect of SES on birthweight.
The above mediators of the association of SES with fetal growth are less important for preterm birth. A systematic review showed that maternal anthropometric features such as BMI, pre-pregnancy weight and maternal height were poor predictors of preterm delivery . Cigarette smoking, a powerful mediator of the association between SES and fetal growth, may also mediate the relationship between SES and preterm birth, but to a lesser extent [2, 44]. Other interrelated factors that may play a role in preterm birth, such as infection, e.g. bacterial vaginosis, and psychosocial stress [17, 45], were suggested to be also linked with socioeconomic deprivation but the evidence for this is inconsistent .
The recent trend in increasing proportion of preterm birth observed in many countries may have resulted in part from more frequent obstetric interventions in compromised pregnancies at an earlier gestation as a result of improved survival of preterm babies. In this study, there was an increase in caesarean section from 1961 to 2000 , but the increase was similar across all four TDS groups (data not shown). Hence, the increase in the prevalence of preterm birth in the most deprived groups cannot be attributed to the higher increase in caesarean section compared to the least deprived groups in this population. A higher proportion of singleton pregnancies due to assisted conceptions may have contributed to the overall increase in the prevalence of preterm birth in the last decade of the study period , but it is more likely that their proportion was higher among women from advantaged SES groups and therefore cannot explain the rise in the prevalence of preterm birth in the more deprived groups.
This study has a number of strengths which have been described in detail elsewhere . The validity of the long-term trends in birthweight and preterm birth with respect to area deprivation depends on the completeness and accuracy of the data across the study period, including gestational age estimation, which is one of the major strengths of the dataset . Briefly, the accuracy of the data was ensured by multiple checking and internal and external cross-validation of the data. Estimation of gestational age was made as consistent as possible being mainly based on calculated values using the recorded estimated date of delivery (i.e. LMP based) throughout the study period rather than on the gestational weeks recorded in the neonatal records. The recorded gestational age may have been based on the ultrasound data for the later years if there was uncertainty in the LMP date or a significant discrepancy between the two estimates. Thereby we limited potential bias due to differences in sources for gestational age estimation over 40 years. Furthermore, we also demonstrated that temporal trends in birth numbers were similar to national and regional ones , which adds to the data validity. Data on gestational age and birthweight were missing for the majority of home births in the first decade of the study. Due to the lack of information from birth ledgers, it was not possible to assess whether there was a clear pattern in the distribution of home births by area deprivation in the early 1960s. As these births were not included in this analysis, data on birthweight and gestational age outcomes used in this study were compared across hospital births by decade and by TDS quartile (the percentage of home births in the last three decades of the study was very low, 1–2%). Perinatal mortality rates were lower among intended home births compared to hospital births in Newcastle upon Tyne during 1960–69 , suggesting that there was a lower percentage of LBW and preterm babies among them, but we were not able to prove this using our data.
In this study we used an area-based composite scoring method to identify the neighbourhood SES of the postcode area and assigned the same status to all mother/infant pairs residing within that postcode. Previous UK studies of birthweight outcomes showed that area-based TDS, as a proxy measure of SES, was a reliable alternative to individual measures . Area deprivation measures are considered a better estimator of social gradient associated with birthweight [38, 41]. A study from Newcastle found that TDS at the ED level was a good proxy for individual level deprivation to predict self-reported health . Nevertheless, we cannot completely rule out misclassification of SES, whereby relatively deprived families may reside within affluent neighbourhoods or vice versa, but this effect is likely to be random. TDS calculation for 1997–2000 was based on data at ward level, a larger area than ED, which could have resulted in misclassification of births into TDS quartiles. However, the misclassification could occur in both directions and is unlikely to have a major impact on LBW and preterm birth rates in the last decade. The temporal trend of TDS for specific locations, and therefore the inequalities in TDS, may have been influenced by the implementation of local and national policies. An example for such trend is the implementation of the Housing Act 1980, which gave five million council house tenants in England and Wales the ‘Right to Buy’ their house from their local authority at a discount price. This policy may have favourably affected the TDS for a number of households in Newcastle upon Tyne for the two last decades, 1981–90 and 1991–2000, as home ownership is one of the variables for TDS calculation. However, a previous paper from Newcastle upon Tyne  reported that the gap between the most affluent and the most deprived groups of the population widened over the study period, which does not indicate that this policy substantially improved inequalities. We are not able to quantify the effect of the ‘Right to Buy’ policy on the TDS and the observed inequalities, but we can speculate that they would have been more pronounced if it was not implemented.
One of the study limitations was lack of data on maternal smoking, an important determinant of fetal growth, which did not allow us to examine the direct contribution of smoking to the association of SES with LBW and preterm birth. Information on some maternal anthropometric characteristics such as height, BMI or pre-pregnancy weight, which may have mediated the association between area-based deprivation and birthweight, but not preterm birth , was also not available in the routine neonatal records, a major data source of the study. We did not consider lack of another possible modifier of SES on birthweight or preterm birth, such as ethnic disparity [15, 40], to be an important limitation of our study as Newcastle had a low percentage of ethnic minority groups (2%) .