Neonatal and child mortality by mothers’ ethnic group and socio-economic status

This article explains findings from ONS research into inequalities in neonatal and child mortality in England and Wales, using information from birth registrations, 2011 Census data and death registrations.

If you are affected by the issues in this article, you can find information about where to seek help in the Help with issues raised in this article section.

Child mortality is a measure of the proportion of babies who die before reaching later childhood. It is commonly used as an indicator of population health.

Overall, child mortality in England and Wales has been reducing since the 1980s. In 2022, around 3.9 in every 1,000 infants died before their first birthday.

However, improvements to the child mortality rate have slowed in recent years. The UK has some of the highest child mortality rates in Europe, so it’s important to understand as much as we can about current trends.

In this article, we have followed a cohort of children born between 2011 and 2016, until December 2021, when the oldest children in the cohort were aged 10 years.

First, we looked at how neonatal and child mortality rates differed depending on the birth mothers’ ethnic group or the socio-economic status of adults in the household, using an occupation-based measure called National Statistics Socio-economic Classification (NS-SEC).

We then estimated risk of death in our cohort, using statistical modelling.

We have looked at neonatal mortality (from birth up to 28 days) and child mortality (after 28 days) separately, as the risk of death and causes of death are very different for these two groups.

Our analysis included 3 million babies born between Census Day 2011 (27 March 2011) and 31 December 2016. Further information on our study population is available in the Measuring the data section.

In total, there were 9,955 deaths in our follow-up period 4,750 (48%) of which occurred in the neonatal period (within 28 days of birth).

We measure mortality rates using a standard measure that compares the number of deaths to the years of life lived, called deaths per 1,000 child-years. This takes into account both the number of children in the cohort, and the follow-up time.

Babies born to Pakistani mothers had the highest rate of neonatal and child mortality, at 45.8 per 1,000 child-years and 0.8 per 1,000 child-years, respectively. This is an estimate that falls within a range; confidence intervals for all estimates are available in the associated data tables.

Meanwhile, neonatal and child mortality rates were lowest in the White Other group (15.9 per 1,000 child-years and 0.2 per 1,000 child-years, respectively).

When comparing rates by socio-economic status, households where the main earner had never worked or was long-term unemployed had the highest rates of neonatal and child deaths (31.9 per 1,000 child-years and 0.6 per 1,000 child-years, respectively).

We used statistical techniques to estimate inequalities in child mortality and understand which factors are most closely associated with the differences between ethnic groups and people of different socio-economic status.

First, we used a statistical model to estimate the difference in risk for babies born to mothers of different ethnic groups and socio-economic status (Model 1). This accounted for differences in the average age of mothers between groups and the babies’ sex. This is because we know that these factors play an important role in the risk of death.

A second model helped us understand the extent to which the differences in mortality between ethnic groups and socio-economic statuses found in Model 1 were related to other characteristics.

Our second model (Model 2) accounted for differences in household characteristics like mother’s country of birth, and babies’ (gestational) characteristics, including premature birth and suspected congenital abnormalities.

When looking at ethnicity, we also accounted for socio-economic status, and when looking at socio-economic status, we adjusted for ethnicity. We’ve provided further information about the models in the Measuring the data section.

If there is a big difference in the estimated relative risks between models, this means that the factors accounted for in Model 2 are influencing the estimated risk of death seen in that group.

It’s important to understand that the following figures reflect estimated risk based on a statistical model, accounting for multiple factors simultaneously, rather than actual mortality for children in our cohort.

These estimates tell us about statistical relationships between neonatal and child mortality and belonging to a certain ethnic or socio-economic group, but we cannot say what is causing these differences.

We estimated that babies and children born to mothers in the Pakistani ethnic group (Asian/Asian British: Pakistani) saw the biggest risk compared to those born to White British mothers (White: (English/Welsh/Scottish/Northern Irish/British). This reflects the trends seen in the overall mortality rates.

This was the case even after accounting for a range of other factors.

The chart below shows how the risk of neonatal death varies by ethnic group compared to the risk for those born to White British mothers.

In both models, babies born to Pakistani mothers saw the highest risk of neonatal death compared to that of White British mothers

Estimated risk of neonatal death, by mothers’ ethnic group, compared to mothers of White British ethnic group, England and Wales, 2011-2017

Download the data for risk of neonatal death by mothers' ethnic group [XSLX, 11KB]

Even after accounting for many other environmental, household, and gestational characteristics, the estimated risk remained almost two times higher than that of White British mothers.

In Model 1, which accounted for the babies’ sex and mothers’ age, babies born to mothers of Bangladeshi, Indian and Pakistani ethnic groups and Black / Black British ethnic groups had statistically significantly higher risk than babies born to White British mothers.

In Model 2, which accounted for many more socio-economic and gestational differences, risk of death was still significantly higher for babies born to mothers in the following ethnic groups:

• Black African (Black/African/Caribbean/Black British: African)
• Pakistani (Asian/Asian British: Pakistani)

However, for all other groups, there was no statistical evidence of differences in risk compared with the White British group. This means the factors accounted for in Model 2, like maternal and gestational characteristics (premature births and congenital abnormalities), could be contributing to the difference in risk seen by these ethnic groups.

Importantly, we still see differences in mortality risk between ethnic groups in both models. This indicates that other unmeasured factors not accounted for in our analysis, such as access to healthcare, could also be contributing to this difference in risk.

We define child mortality as a death from 28 days of age up to 10 years of age. For child deaths, the estimated risk was highest for children born to mothers in the following ethnic groups:

• Pakistani (Asian/Asian British: Pakistani)
• Black African (Black/African/Caribbean/Black British: African)
• Bangladeshi (Asian/Asian British: Bangladeshi)

Unlike the neonatal model, the risk of child death for different ethnic groups was similar in both models.

This suggests that the characteristics accounted for in Model 2 (such as mothers’ socio-economic status and babies’ birth weight and other gestational characteristics) did not play a substantial role in the estimated differences seen between ethnic groups.

Children of Pakistani, Black African and Bangladeshi mothers saw the highest estimated risk of death before and after accounting for other household and gestational factors

Estimated risk child death by mothers’ ethnic group, compared to mothers of White British ethnic group, England and Wales, 2011-2021

We also compared mortality risk for babies and children based on the socio-economic status of the adults in their household. To do this, we used the National Statistics Socio-economic Classification (NS-SEC), which groups people based on their occupation.

The groups range from those who have never worked or are long-term unemployed, to those in high-paid occupations such as chief executives and company directors.

In this analysis, we have compared mortality risk for all socio-economic groups to those in NS-SEC Class 1, which includes higher managerial and professional jobs. We’ve grouped our cohort based on the NS-SEC classification of the main earner in the household.

Babies born into households where the main earner was not in a managerial profession had a higher risk of neonatal death than those in households where the main earner was in a higher managerial or professional job.

However, once we accounted for a range of other household and gestational factors (in Model 2), there was no statistical evidence of a higher risk of neonatal death for any socio-economic group.

This suggests that the relationship between socio-economic status and neonatal mortality can largely be explained by other factors.

After accounting for gestational characteristics, the estimated risk of neonatal death was similar for all socio-economic groups

Estimated risk of neonatal death, by household NS-SEC classification, compared to those in higher managerial and professional jobs, England and Wales, 2011 to 2017

Download the data for risk of death by household NS-SEC classification [XLSX, 11KB]

When looking at the risk of death for children, those in all other socio-economic groups saw a higher risk than those in households with a higher managerial or professional worker.

In households where the main earner had never worked or was long-term unemployed, the risk in Model 1 was more than 2.5 times that of a child who lived with a higher managerial or professional worker.

The estimated differences between groups reduced in the second model, where more personal and environmental factors were accounted for.

However, children from households in all but two socio-economic groups (those in households with intermediate occupation workers and those in households with students) still saw a higher estimated risk of death than those in households with a higher managerial or professional worker.

This suggests socio-economic factors play a bigger role in risk of child death than neonatal death.

Children in households where there was long-term unemployment saw the highest risk of mortality

Estimated risk of child death by household NS-SEC classification, compared to those in managerial and employer jobs, England and Wales, 2011 to 2021

Download the data for risk of child death by household NS-SEC [XLSX, 11KB]

Return to the top of the page

Confidence interval

Confidence intervals are the range in which an estimated value is likely to lie. For more information, see Uncertainty and how we measure it.

Ethnicity

In this article we have used ethnic groups as reported on the 2011 Census. There were 18 response options on the census, which we combined into 9 groups. In some cases, we have shortened the full group name, for brevity. These abbreviations, and their full groupings, are as follows:

Bangladeshi

Asian/Asian British: Bangladeshi

Indian

Asian/Asian British: Indian

Pakistani

Asian/Asian British: Pakistani

Black African

Black/African/Caribbean/Black British: African

Black Caribbean

Black/African/Caribbean/Black British: Caribbean

Mixed / Multiple

Mixed/Multiple ethnic groups, Mixed / Multiple: write-in

Other

Other ethnic group (including Asian / Asian British: Chinese, and any other Asian background; any other Black/African/Caribbean/Black British background)

White British

White: English/Welsh/Scottish/Northern Irish/British

White Other

White: Other

Hazard ratio

A hazard ratio is a measure of the relative differences in the instantaneous rate of mortality between groups. In short, it is a measure of how often a particular event occur in one group compared to another (reference) group at a given point in time. A hazard ratio greater than one indicates the rate of mortality is higher, and likewise less than one lower, in the population group under study compared with a reference group.

Mortality rate per 1000 child-years

This is the rate based on deaths and the number of people, for a specified time period. Child-years takes into account both the number of people and follow-up time. This allows us to estimate the number of neonatal and child deaths that would occur among 1,000 individuals per year over our full study period.

NS-SEC

The National Statistics Socio-economic classification, or NS-SEC, is a classification system of socio-economic status based on occupations. Further information is available in our NS-SEC methodology page.

In this analysis, we assigned babies and children to a socio-economic group based on the main earner in their household, or household reference person. This could be the person who owns or rents the house they are living in, or the oldest adult in the household, or the highest earner.

This analysis used population level data, linking data from linked birth notifications (2011-2016) to Census 2011 information on mothers’ socio-economic and ethnicity characteristics.

The cohort included in this study included singleton live births born after 24 weeks of pregnancy, to mothers aged at least 12 years at the time of birth.

Babies whose mother was not a usual resident in England or Wales at the time of the 2011 Census were excluded. Stillbirths were also excluded from our cohort.

These babies were followed-up until December 2021, to a maximum of 10 years of age.

The number of deaths reported in this article and the associated data tables are rounded to the nearest five.

The statistical model used to estimate mortality risk by ethnicity and socio-economic status is called Cox proportional hazards, and these results were expressed in hazard ratios.

The full analysis included 5 models which is detailed in the Data.