This article deals with themes of suicide – if you are struggling to cope, please call Samaritans for free on 116 123 (UK and the Republic of Ireland). More information and resources are listed at the end of the article.
A death by suicide is devastating for those affected. While rates of suicide are highest in middle age, and particularly in men, there have been concerning increases among children and young people over recent years. The UK government has committed to reducing the number of deaths by suicide, with children and young people identified as a priority group for support in the national suicide prevention strategy for England.
We have used mortality data and information from the 2011 Census to build a picture of the groups of children and young people who experience a higher risk of dying by suicide. Our population includes nearly 8 million children and young people in England. This is the first time an analysis of this scale focusing on children and young people has been carried out.
In our cohort, suicide rates were higher for males than females, and for children and young people in households where the household reference person held formal qualifications, compared with those where they held no formal qualifications.
In a second piece of analysis, we used additional education data to quantify the risk of suicide for children and young people in England with Special Educational Needs (SEN) compared with their peers. This study followed nearly 4 million children and young people aged 10 to 18 years on Census Day 2011.
Males who had SEN support without a statement (such as School Action plans, Statutory Assessment or Early Years Intervention) had the highest risk of suicide compared with those who had no recorded SEN provision, with the risk in the former group around 1.5 times higher than in the latter.
We have worked with charities focused on suicide prevention and supporting young people to understand how our findings relate to people’s lived experiences. We have included some of their insights throughout the article.
If your gut is telling you something’s not right, then start a conversation about it. Finding the words to start the conversation is where expert organisations come in. We need to make it easier for people to find useful, accurate and trustworthy support.
Suicide rates were higher for males than females
Suicide rates for females in our cohort were around two thirds that of males. We see similar trends in suicide deaths in adults.
We studied a cohort of almost 8 million young people aged between 10 and 17 between Census day 2011 and 31 December 2022.
We used a statistical model to identify characteristics associated with a higher risk of dying by suicide while adjusting for other factors. Our results compare rates for each group to a reference group: Rates of suicide were significantly lower for females compared with males.
Suicide rates were higher for males than females
Incidence rate ratios for suicide for people aged 10 to 17 years, by sex, England, 2011 to 2022
Embed code
Notes
- Model has been adjusted for: age, ethnic group, main language, carer status, disability status, National Statisticians Socio-economic classification (NS-SEC) of the household reference person (HRP), highest level of qualification of the HRP, family composition, housing tenure and region
Download the data for incidence rate ratios for suicides by sex, (XLSX, 18KB)
I've found that males I know who struggle with suicidal ideation seem to feel that they should be strong enough to cope. Mental health issues are a reality for both males and females, and regardless of gender, I encourage everyone to talk about how they really feel.
Further results (including suicide rates per 100,000 person-years and minimally adjusted model outputs, which adjust for age and sex only) are provided in the associated data tables. We have provided more information about these methods in the How we carried out this research section.
Suicide rates were highest in households where an adult held a degree-level qualification
The suicide rates among children in households where the household reference person (HRP) had a formal qualification was significantly higher than those in households where the HRP had no formal qualifications.
The biggest difference was between children in households where the HRP had no formal qualifications and those where the HRP had a degree-level or above qualification, with suicide rates around 1.7 times higher in households with degree-level qualifications.
For rates of non-suicide deaths, we saw the opposite: rates were significantly lower for children in households where the HRP held formal qualifications compared with those where they did not. Rates of non-suicide deaths in households with a degree-level or above qualification were about a third lower than in households with no formal qualifications.
We cannot be certain of the causes of these differences, as a range of factors which we have not been able to control for in our modelling might be at play. These may include educational factors, familial relationships or breakdowns, and bereavements.
Households where the main adult had no formal qualifications experienced lower suicide rates
Incidence rate ratios for suicide, for people aged 10 to 17 years by qualification level of household reference person (HRP), England, 2011 - 2022
Embed code
Notes:
- Model has been adjusted for: age, sex, ethnic group, main language, carer status, disability status, NS-SEC of the HRP, family composition, housing tenure and region
Download the data for incidence rate ratios for suicide by qualification level of HRP (XLSX, 18KB)
Parents we spoke with said it can be common for children to appear like they're doing well, when in reality they are struggling.
If you ask someone how they’re doing, and they say, ‘I’m fine’, then ask them a second time. Because you might get a different answer. My daughter can seem great, but she often wears a mask, so everybody would say that she’s the life and soul. She’s doing well, but I know to ask her twice.
Children and young people in Muslim and Christian households had lower suicide rates
Children and young people from households where the household reference person (HRP) was Muslim or Christian had significantly lower suicide rates compared with children in households where the HRP had no religious affiliation.
The suicide rate for children in Muslim households was half that of those in households where the HRP reported no religion. For children in households where the HRP was Christian, the suicide rate was about a fifth lower. This is similar to trends seen among adults.
However, the rate of non-suicide deaths was significantly higher among children from Muslim households.
Children and young people in Muslim households had the lowest suicide rate, but the highest rate of non-suicide deaths
Incidence rate ratios for suicide, for people aged 10 to 17 years, by religion of household reference person, England, 2011 - 2022
Embed code
Notes:
- Model has been adjusted for: age, sex, ethnic group, main language, carer status, disability status, NS-SEC of the HRP, highest level of qualification of the HRP, family composition, housing tenure and region
Download the data for incidence rate ratios for suicide, by religion of HRP, (XLSX, 18KB)
In my experience, if a family has a faith, whatever that faith may be, it does give [...] a structure to help them steer through life. I could see that if youngsters lean into that it can provide support, strength and community belonging. But at the same time, I’ve had people ask me ‘if I take my life, will I go to hell?’ There can be an element of fear, too.
Suicide rates also varied by ethnicity, with rates for children and young people of Mixed or Multiple ethnic groups significantly higher (around 1.5 times) than for White ethnic groups.
Meanwhile, those of Asian or Asian British and Black or Black British ethnic groups had significantly lower suicide rates than White ethnic groups.
Rates for Black or Black British ethnic groups were just over half as high as for White ethnic groups, and rates for Asian or Asian British ethnic groups were about two thirds of the rate. Again, this is a similar trend to that seen in suicides among adults.
For all non-suicide deaths, children and young people of Black or Black British and Asian or Asian British ethnic groups had significantly higher rates than those of White ethnic groups.
Children and young people from a Mixed or Multiple ethnic background had higher rates of suicide compared with other ethnic groups
Incidence rate ratios for suicide, for people aged 10 to 17 years by ethnicity, England, 2011 – 2022
Embed code
Notes:
- Model has been adjusted for: age, sex, main language, carer status, disability status, NS-SEC of the HRP, highest level of qualification of the HRP, family composition, housing tenure and region
Download the data for incidence rate ratios for suicide by ethnicity (XLSX, 18KB)
[Research like this] tells us who we need to be reaching out to. We can look at our data and see who we are engaging with the most, but it’s really useful to then compare that with where there’s the greatest need – who should we be reaching?
Differences in suicide rates by other characteristics are available in the associated data tables.
Males with special educational needs saw a higher risk of suicide than those without
The risk of suicide for males with special educational needs (SEN) support (without a SEN statement) was around 1.5 times higher than for males with no recorded SEN. These findings are from secondary analysis we carried out linking education data from the Growing up in England (GUiE) dataset (which links administrative education data to 2011 Census) to mortality data.
As in our previous analysis, we used a statistical model to adjust for other characteristics to ensure that any association between suicide and SEN was independent of other factors. Individuals with either a SEN statement (meaning the local authority had to provide the support specified) or other SEN support without a SEN statement (this includes School Action plans, Statutory Assessment or Early Years Intervention), were compared with their peers who had no recorded SEN. More information is available in How we carried out this research.
While the risk for males with a SEN statement and females with SEN support but no SEN statement also appeared numerically higher than for their peers without recorded SEN, these differences were not statistically significant.
Males with SEN support but no SEN statement saw the highest risk of suicide
Hazard ratios for suicide, for people aged 10 to 25 years by SEN status, England, 2011 – 2022
Embed code
Notes:
- Model has been adjusted for: age, ethnicity, health and NS-SEC
Download the data for hazard ratios for suicide by SEN status (XLSX, 18KB)
About a quarter of our users [at The Mix] have a special educational need. When we include mental health issues too, it’s closer to a third.
The cohort for this study included nearly 4 million children and young people who attended school between 2001/02 and 2010/11, and were aged between 10 and 18 on Census Day 2011. We identified whether these people had special educational needs support recorded and followed up with them between Census day 2011 and 31 December 2022. More information about how we did this analysis is available in How we carried out this research. Further breakdowns are available in the associated data tables.
Among children who had a SEN statement, we assessed differences in suicide rates between different kinds of need. We did not run a model to compare risks for this group, because the group sizes were too small. Instead, we calculated overall rates, with no adjustments for other characteristics.
Those whose needs were related to social, emotional or mental health needs had the highest suicide rates, at 15.5 per 100,000 person years.
Children with a SEN statement with social, emotional and mental health needs had higher suicide rates than other SEN types
Rates of suicide per 100,000 person years, for people aged 10 to 25 years by SEN type, England, 2011 – 2022
Embed code
Notes:
- Results for people with sensory and physical needs have been suppressed because of low counts
- Rate per 100,000 person-years is a measure accounting for both the number of people in our cohort, and the length of time they were in the study; it represents the number of suicides that would be expected if 100,000 people were each followed-up for one year
Download the data for rates of suicide per 100,000 person-years by SEN type (XLSX, 18KB)
Something I often hear is [suicide or suicidal ideation] is a cry for attention. You should always take someone disclosing thoughts of suicide seriously, because if they’re trying to get your attention, they probably need your attention
Definitions
95% confidence intervals
A confidence interval is a measure of the uncertainty around a specific estimate. If a confidence interval is calculated at the 95% level, it is expected that the interval will contain the true value on 95 occasions if repeated 100 times. Wider confidence intervals reflect higher uncertainty. The size of the interval around the estimate is strongly related to the number of deaths, and the size of the underlying population. More information is available in Uncertainty and how we measure it for our surveys.
Suicide
This release is based on the National Statistics definition of suicide. This includes all deaths from intentional self-harm for persons aged 10 years and over, and deaths caused by injury or poisoning where the intent was undetermined for those aged 15 years and over. We used codes corresponding to intentional self-harm (X60-X84), injury (including, but not limited to, poisoning) of undetermined intent (Y10-Y34) from the World Health Organisation's (WHO) International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10).
Incidence Rate Ratio and Hazard Ratio
The incidence rate ratio (IRR) and the hazard ratio (HR) are both measures of the relative frequency of an outcome in one population compared with a different population. The IRR or HR for the reference group is always equal to 1. IRRs or HRs greater than one indicate that the outcome occurs more frequently compared with the reference group, while IRRs or HRs less than one indicate that the outcome occurs less frequently compared with the reference group.
How we carried out this research
Sociodemographic analysis
We linked the 2011 Census and death registrations data by NHS number for children in England.
Our final sample comprised 7,747,345 people who were aged between 10 and 17 years on 2011 Census Day (27 March 2011) or turned 10 before the end of the study period (31 December 2022). Children were followed up from study start if aged 10 on or before 2011 Census Day, else follow-up started when children turned 10. This cohort was followed until their 18th birthday, their death, or the end of the study (31 December 2022), whichever came first.
To reduce the potential bias caused by delays in death registrations, we used mortality data with date of death from 28 March 2011 to 31 December 2022 and registration date up to one year later (to 31 December 2023).
We fitted Poisson generalised linear models with death by suicide being the outcome of interest. An offset was added to the model to account for the different time-at-risk periods between individuals.
First, for each predictor of interest, to estimate the difference in the rate of suicide, we fitted models adjusted for age and sex.
Finally, to assess how each factor is independently associated with the risk of suicide, we fitted models which were adjusted for sex, age, ethnicity, main language, carer status, disability status, highest level of qualification (of the HRP), National Statistics Socio-economic classification (NS-SEC; of the HRP), tenure, family type and region.
Special educational needs analysis
We linked the Growing Up in England dataset (GUIE) (which links administrative education data to 2011 Census) to death registrations. Our final population included 3,959,390 10 to 18 years olds (age on 2011 Census day) who were in GUiE.
To identify if a child was receiving SEN support (or had a record of a SEN statement), data from academic years 2001/2 to 2010/11 was used to identify need. Our cohort was followed up until the end of 2022, their 25th birthday or their death, whichever was earliest.
Our exposure group (children with SEN provision) was categorised into three groups:
Individuals who received a higher level of provision with a SEN statement (meaning the local authority had to provide the support specified, now known as an Educational Health and Care Plan, or ECHP).
Individuals who received SEN provision without a SEN statement (this includes School Action (SA) plans, Statutory Assessment or Early Years Intervention).
Individuals with no recorded SEN provision
The outcome of interest was death by suicide in any time during the follow up period. This cohort was followed from 28 March 2011 until their 25th birthday, their death or the end of the study (31 December 2022), whichever came first.
The relative risk of suicide for children with SEN, compared with their peers, was assessed using a Cox Proportional Hazards model. The models were stratified by sex and adjusted for age (using a natural spline with three knots), ethnicity, health and NS-SEC. All covariates were sourced from Census 2011.
Descriptive counts and suicide rates were calculated for different groups of children with SEN. Children with a SEN statement were grouped into one of four SEN types, based on their statement from the most recently available academic year.
If multiple SEN types were present in the most recent recorded year, we then examined the child’s SEN history to determine the most frequently recorded type. If a child had multiple SEN types in the most recent year and their historical records showed an equal frequency of different types, we classified them as "Unresolved." We also categorised a child as "Unresolved" if the SEN type was either not recorded or listed as "Other." For data on suicide risk in this group, see the accompanying dataset
Acknowledgements
This project is funded by the National Institute for Health Research (NIHR) Policy Research Programme (NIHR205990). The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care.
We would like to thank the participants of our two public and patient advisory groups. Our groups comprised of young adults with experiences of suicidality or supporting others with suicidality, and secondly adults who had lost a child or family member to suicide, or who were supporting children or family members with suicidality either in a personal or professional capacity. Members of our public and patient advisory groups reviewed the results of the analysis and provided feedback. We would like to thank Papyrus UK and The Mix for facilitating these workshops.
We would also like to thank members of the Mutual Support for Mental Health Research (MS4MH-R) Patient and Public Involvement and Engagement group affiliated with the NIHR Greater Manchester Patient Safety Research Collaboration and Centre for Mental Health and Safety at The University of Manchester for their contribution to the interpretation of results and reviewing of publications.
Resources
The following support services are available to people in the UK who are struggling to cope, or concerned about someone they know.
Shout is the UK’s only 24/7 text messaging support service, powered by Mental Health Innovations. Anyone in the UK can text ‘SHOUT’ to 85258 for free, 24/7 mental health support.
The Mix is the UK’s digital safety net for young people. We provide free, anonymous advice and crisis support designed especially for under 25s. Open 24 hours a day, The Mix gives young people life skills and connections in a supportive digital community through content, counselling and crisis support.
Papyrus, Prevention of Young Suicide, is a national charity working to reduce the number of young people who take their own lives by breaking down the stigma around suicide and equipping people with the skills to recognise and respond to suicidal behaviour. Their HOPELINE247 service is confidential and available via telephone, text and email.
The Samaritans listening services is available via phone, email and online chat.