In this study, we found substantial variation in rates of general hospital admission for self-harm between different ethnic groups in London. Black and Asian people had lower rates of admission than the White population in both sexes and across all age groups. These groups were also less likely to be readmitted within a year. These differences were not explained by deprivation: standardisation for IMD had only a modest impact on rates which increased the differences in rates between the Black and Asian groups and the White group. The picture for the Mixed group was more complicated: overall rates were lower than the White group at all ages for both sexes, once standardised for deprivation. However, rates of readmission in the mixed group were higher than in the White group. There were also differences by ethnicity in the age at which self-harm peaked for men, with Black and Mixed ethnicity men having higher rates in the 25–49 age group whilst all other groups saw rates peak in people below 25.
The extent to which rates of admission following self-harm differed from those in the White British population varied considerably for more detailed ethnic groups within the broad categories. Within the White group, the Irish population had higher rates than the British population, while the White Other group had much lower rates. There were also differences between the Black groups: in both sexes Black Africans had lower rates relative to the White British group than the Black Caribbean group, who in turn had lower rates than the Black Other group. These findings highlight the importance of disaggregating data within broad ethnic categories. The Black African population is the fastest-growing minority group in the UK, doubling between 1991 and 2001, and again by 2011 [23] so that it is now the largest Black group. Likewise, the White Other group has grown rapidly in recent decades and is the largest minority group in London and the UK as a whole. However, there is little research evidence available to understand self-harm rates in either of these populations. The last community survey that reported suicidal behaviours by ethnicity in the UK, the EMPIRIC study from the late 1990s, looked specifically at the Black Caribbean and White Irish groups but had insufficient numbers to show differences in odds of self-harm [24]. More recent studies using service use data have combined the Black African group with Black Caribbean and Black Other, a heterogenous group, about half of whom identified their ethnicity as Black British in the 2011 census. Similarly, all White ethnicities tend to be combined together and used a reference group. This study suggests the experiences of these separate groups may be quite different, and that the White Other group may have more in common with other ethnic minority groups than the White British group for this outcome.
Comparison with previous studies
Our findings of lower rates of self-harm in South Asians than White British people, for both men and women, confirms similar findings in a study using ED data from Manchester, Derby and Oxford for 2001–2006 [9], in contrast to earlier studies which found higher rates of self-harm in South Asian women [7]. Cooper et al. [9] suggested this difference may be due to different South Asian populations being included in different studies. This study found similar, lower rates of self-harm in Indian, Pakistani, Bangladeshi and Other Asian groups suggesting a pattern across all South Asian populations. However, the migration status and socio-economic position of these populations may be changing over time. It may also be that greater attention and service provision for these populations based on previously high rates has been beneficial.
This study also found lower rates of self-harm in Black women than White women at all ages, in contrast to previous studies’ findings of raised rates in young Black women [9, 10]. The authors of these studies suggested that higher rates may be due to greater socio-economic adversity affecting this group. The Black population of London in this study is disproportionately concentrated in more deprived areas so it seems unlikely that the difference could be explained by the Black women included in this study being less deprived than in previous studies. Indeed, adjusting for area socio-economic deprivation lowered rates further in comparison to White women. Another possibility lies in the differences seen between different Black groups described above. For women, the Black Other group’s rate of self-harm admission was similar to the White British group before standardisation for deprivation [RR 0.94 (0.86–1.02)]. It may be the Black group in previous studies, which did not adjust for deprivation, contained different proportions of different Black ethnicities and this study partly reflects the changing make-up of the Black population in London. An additional consideration is the role of ethnic density in protecting the mental health of individuals from ethnic minorities. Living in areas with a higher proportion of people of the same ethnicity has been found to be associated with lower rates of self-harm [25] and suicide [26] for individuals from ethnic minorities. The lower rates of self-harm in ethnic minority groups in this study may partly reflect this.
It is important to also note that the previous studies referenced used data on ED attendances while this study was restricted to admissions. It may be that the likelihood of admission following an ED presentation with self-harm varies by ethnicity. This could be either because of differential treatment of people from different ethnic groups within the same hospitals or because of differences in admission practices between hospitals which serve different populations. For example, work in South East London has found substantial differences in admission practices between hospitals, with the hospital least likely to admit following self-harm serving the most deprived and ethnically diverse areas [27]. The sensitivity analysis using ED data found lower rates of attendance following self-harm in the non-British White, Black, Asian and Mixed groups compared to the White British group, mirroring the findings from the admissions data. However, this only reflects the experience in a small part of the total study area and does not rule out a role for admission practices.
Some of our findings confirm those of previous research. The White Irish population in the UK have previously been found to experience worse health outcomes, including for psychological distress [28], which have persisted across generations, and also have higher suicide rates in men [29]. Previous studies have suggested that much of this health inequality can be explained by experiences of material deprivation in childhood [28]. In this study, the current deprivation of the area that people lived did not explain higher rates of self-harm in the Irish population, however, this is not an individual level measure, nor does it necessarily capture the conditions someone lived in growing up, hence different exposure to deprivation over the life course may still explain differences in rates.
A broader question is why there appear to be lower rates of admission for self-harm amongst most of the ethnic groups investigated when compared to the White British population, despite their greater exposure to socio-economic stressors. In their cultural model of suicide, Chu et al. [5] suggest several points at which an individual’s cultural context, particularly their ethnicity, may impact on their risk of self-harm and suicide. The stressors most associated with the risk of self-harm and suicidality may vary between groups, so the understanding of risk factors developed from suicide and self-harm research that has overwhelmingly been based on White populations may not generalise well to minority groups. Religious sanctions around suicide and self-harm may also make self-harm a less acceptable response to stressors for some groups, either making them less likely to self-harm or more likely to hide having done so. However, the pattern of lower rates in virtually all minority groups, despite their heterogeneity, should be a caution against locating explanations solely within their cultures [5]. The common experience amongst these groups is a position of a structural disadvantage as a minority within the UK [30], which could also impact the ways in which people feel able to express distress and the likelihood of their accessing help [31].
Rates of self-harm may appear lower in some ethnic groups due to differing “idioms of distress”, resulting in distress being expressed in different ways [4]. This study, like most others examining rates of self-harm by ethnicity in the UK, uses a definition of self-harm as intentional self-injury or poisoning that excludes other risk taking or harmful behaviours that could be more common responses to distress in some ethnic groups. Hospital data also relies on people presenting to services following self-harm and clinicians recognising and recording a presentation to services as self-harm. There is evidence from community surveys that people of the Black Caribbean and South Asian ethnicities were less likely to seek professional help following self-harm [24]. Studies from the USA have also found that African Americans are less likely to present clinicians with the expected “classic signs” when suicidal [4]. Research in both the UK [32] and USA [33] has suggested that suicides in ethnic minorities are less likely to be identified as such and more likely to be found of undetermined intent or misattributed to accidents, raising the possibility that similar misclassification occurs for non-fatal acts of self-harm seen in hospitals.
Strengths and limitations
This study is strengthened by being based on a large dataset containing self-harm admissions for nearly 60,000 individuals from an underlying population of over 8 million with a very high level of completeness for the ethnicity variable at 94%. London is the most ethnically diverse region of England, with only 46% of its population identifying as White British at the 2011 census compared to 81% for England and Wales overall. There are likely to be similarities to other large urban centres in England: Birmingham and Manchester for example have similar proportions of their populations identifying as White, although there are differences, in particular, all the Black ethnic groups and White Other form a larger proportion of London’s population than other British cities [34]. The ethnic diversity within London allowed us to examine rates for more specific ethnic groups than previous studies and in doing so reveal important variations in rates within broad ethnic categories. However, the experiences of individuals from ethnic minority groups within the city cannot be assumed to be representative for the country as a whole.
An important limitation of this study is that we had to rely on data from the census in 2011 to provide denominator populations by ethnicity when calculating rates of self-harm. This was the best source of data available to provide the level of detail required to standardise rates of self-harm admission. However, the proportion of London’s population who are from ethnic minorities increased substantially between the 2001 and 2011 censuses and is likely to have continued increasing in the years since [23]. Given this, in the later years of the study period, the denominator used to calculate rates of self-harm admission for some ethnic minority groups may have been underestimated, making the rates calculated appear higher. The effect sizes for almost all ethnic minority groups, which show lower rates of self-harm admission than the White British population, may then be an underestimate of the difference.
The findings of the study are also limited by our dependence on service use data. The rates calculated only represent differences in admissions to hospital following self-harm. While admission may represent episodes of self-harm with more severe physical health consequences, they will also be affected by differences in help-seeking and admission practices between different ethnic groups. Using routine data also means that our definition of ethnicity comes from medical records. Ideally, ethnicity would be based on individuals’ self-identification. We cannot know whether the individuals in the study were always asked what they considered their ethnicity to be, for some it may have been assigned by hospital staff. The categories of ethnicity, while more specific than previous studies, are still heterogenous and may mask important variation within categories. The Other and Mixed ethnic groups (the largest subgroup of which is Mixed Other) in particular may not represent coherent groups with similar racialised experiences and are difficult to draw conclusions about from this data.
There are other individual-level variables that are likely to impact the risk of self-harm admission and readmission and may do so differently for different ethnic groups, for example, the presence of a psychiatric history or substance misuse. We were not able to adjust for these because they are not recorded in HES data. We adjusted for socio-economic position using the deprivation of individuals’ place of residence. This will not be an accurate reflection of individual deprivation for all people. An analysis using individual-level variables for socio-economic position, as well as migration status would allow greater exploration of the role of these factors in self-harm.