This study contributes to the existing body of research exploring risk factors for SMM in high-income countries. The findings showed a linear relationship between neighbourhood deprivation and SMM at the time of birth, in primiparous women in England. Minoritised ethnic groups also experienced greater odds of SMM, with black of black British African having nearly two times the odds compared with white women. No ethnic group appeared to be more affected by the higher rates of SMM associated with deprivation, and the both deprivation and ethnicity appear to have independent associations with SMM rather than one being a surrogate marker for the other.

The strengths and limitations of using the HES APC for research purposes are discussed in online supplemental material.

Using a composite outcome helps avoid the problem of individual severe morbidities being relatively rare as well as conditions not being correctly coded in routinely collected health data, leading to false negatives. Although the EMMOI captures a range of outcomes that do not all share a direct a causal pathway, it is useful marker both for the overall preconception and pregnancy health and for the quality of care given during pregnancy and birth.

However, a key limitation of this study is the misclassification of the exposure. The association between deprived individuals and their risk of SMM may be attenuated as not everyone who lives in a deprived area is individually deprived.16 In addition, there are also some important missing dimensions to the IMD index such as social well-being and environmental quality,17 and, thus, it is not possible in this study to fully elucidate causal pathways that link deprivation to maternal morbidity without incorporating such factors.18

Additionally, some of the confounding factors included in the multivariable regression analysis may also be on the causal pathway. The confounding factors of pre-exisiting medical conditions and mental health history are individual factors which are included in calculating the IMD score of each area. There is, therefore, a risk of over adjustment, as the adjustments include both potential mediators and individual markers of deprivation. Thus, the results of the association between the IMD and SMM in the above-adjusted models need to be interpreted taking these factors into consideration.19

To our knowledge this is the first study to look at the relationship between neighbourhood deprivation and a composite measure of SMM in England. However, this association has been shown similar high-income countries including Canada,3 Australia5 and New Zealand.20 There are many possible reasons for the higher rate of SMM in more deprived neighbourhoods and the relationship between the confounding factors and causal pathways are complex. For example, having a pre-existing medical condition can be both a cause and consequence of living in a deprived area. It has been hypothesised that the increased risk of SMM for the most deprived women could be driven by a difference in individual health behaviours such as smoking and substance misuse, psychosocial factors such as chronic stress, material factors such as low income affecting the quality of nutrition and obesity, and environmental factors such as air pollution and poor housing.3 5 18 20 21 Indeed, in this study, it appears that some of the individual or ‘compositional’ disadvantage factors, which affect prepregnancy health, partially account for the relationship between deprivation and SMM, as the association is reduced after adjusting for individual pre-existing medical and mental health conditions, smoking, substance use and obesity. One explanation for the remaining risk after accounting for these individual factors is the contextual effects of living in a deprived area beyond the risks of being individually disadvantaged. It could also be explained by differences between care quality and access between different social groups. In the UK, secondary analysis of a National Maternity Survey22 showed that women who lived in the most deprived IMD quintile were less likely to have antenatal care and were more likely to report being treated disrespectfully or spoken to in a way they could not understand by doctors. It is possible that this difference in treatment of women by healthcare providers and access to maternity services may contribute to the differences in SMM for women living in the most deprived areas in England.

Studies conducted both in the UK and similar high-income countries have demonstrated that being from a minoritised ethnic group is a risk factor for maternal morbidity,6 7 which is also seen in our study. Indeed, being of a minoritised ethnicity appeared to have the greatest impact on risk of SMM compared with the other risk factors. One hypothesis previously suggested for what could be driving this trend is that ethnicity may be a surrogate marker for socioeconomic deprivation.23 However, a case–control study in the UK showed that the risk of morbidity is 43%–83% higher in women from a minoritised ethnic group compared with white women, and this was not confounded by occupation, which was used as a marker for socioeconomic status.24 This, alongside the results of our study, suggests that socioeconomic status (or deprivation) and ethnicity appear to be independently associated with increasing the risk of maternal morbidity, rather than one being mostly a marker for the other.

It has been argued25 that systemic racism, rather than genetic or biological phenomena, drives the increased risk in childbirth for black, brown and mixed ethnicity women. Racism can affect health outcomes throughout the life course, in multiple forms and through multiple upstream, midstream and downstream pathways.26 However, in this study, the increased risk of SMM in women from a minoritised ethnic group is not strongly related to the medical, behavioural or psychological factors, which were included and adjusted for in this study. A report reviewed in-depth testimony from over 300 people and found that women from minoritised ethnicities felt unsafe, were ignored and disbelieved, were subjected to racism by caregivers, were not given a proper choice or the means to give informed consent, reported being subjected to coercion, regularly dehumanised and were disproportionately affected by structural barriers to care. This report, alongside the results from our study, suggests that coexisting medical problems or health behaviours, which were able to be captured by the methods of this study, do not appear to account for the increased risk in childbirth for women from minoritised ethnic groups.

Finally, we were only able to adjust for pre-existing health conditions diagnosed before the time of birth. The long-term effects of socioeconomic deprivation and systemic racism may not have manifested in the form of chronic disease diagnosis at this point in the life course. However, these long-term effects may be an important underlying contributing factor to poorer preconception and pregnancy health and therefore risk of SMM.27