This retrospective case series study included CYP experiencing suicidal crisis who had attended an ED at a local pediatric hospital between March 2019 and March 2021 (n = 240). Access to the anonymised data was approved by the hospital’s research department.

Clinical records at the hospital were reviewed between March 2019 and March 2021. Inclusion criteria included any patients aged 16 or younger (the hospital advises anyone age over 16 to attend an adult ED) who presented to ED in suicide crisis (with and without self-harm) during the study period. Data on CYP that visited the hospital in suicide crisis were provided to the researcher by the hospital data team. An electronic inspection of the clinical notes was performed through the Meditech system (Medical Information Technology Inc., Westwood, MA, United States). All patient notes under potentially relevant codes (e.g., low mood, suicide thoughts, social problems, overdose) were audited, and those indicating suicidal crisis or ideation were extracted, collated, and anonymised. Each patient’s clinical record was inspected and included in the study only if suicide crisis or ideation was clearly reported in the clinical notes.

Variables examined included sex, ethnicity, SEN, presence of suspected ASC traits, mental health conditions, suicide ideation with or without self-harm, history of self-harm, clinician determined risk (in terms of Pierce score), and parental socio-demographics. It was also investigated whether the children were previously known to CAMHS or were under CAMHS at the time of the ED presentation. These data were either collected from the family using a standard pro forma completed by the clinician when triaging the patient, or they were already available on the hospital system if the patient had previous been under Community Pediatrics and/or was currently known to CAMHS. Further details regarding the demographics of the patients are presented in the results section below.

Our sample size was predetermined based on the number of CYP attending ED. This was an exploratory analysis, whereby descriptive statistics were conducted to identify the socio-demographics of the sample and the factors characteristic of CYP presenting in suicidal crisis. Chi-squared analyses, regressions, and independent samples t-tests were also conducted to establish statistically significant associations and differences in the dataset. We chose not to conduct Bonferroni corrections for multiple comparisons based on recommendations from Armstrong (40) and Rothman (41) that corrections for multiple comparisons in exploratory studies are not required, due to the increased likelihood of Type 2 errors. Analyses were conducted in IBM SPSS 26.

While the researchers had access to all records, the dataset only captured entries made in clinical records; unrecorded clinical activity or missing information from ED documents was therefore unavailable. For the purposes of this study, only the presence of each factor within each person’s clinical records was used for the analysis. It is possible this strategy may have led to underestimation of some factors: for example, sexual orientation.

The majority of CYP presenting with suicidal crisis reported separation or loss of a parent (68%; n = 162), and most were living with a single parent or a single parent and siblings (44%; n = 106). Parental mental health issues (44%; n = 105), parental drug misuse (17%; n = 41), and parental criminality (16%; n = 38) were also recorded among a sizeable minority of CYP, although a large proportion of data were missing for these variables. Approximately one-quarter of CYP had experienced neglect (24%; n = 57) or domestic violence (24%; n = 57), and one-third experienced some form of physical, emotional and/or sexual abuse (32%; n = 77). 23% were known to a social worker (see Table 2).

Further examination of parental factors revealed a number of significant associations with CYP’s mental health diagnosis (if they had received one), as displayed in Table 3. A statistically significant association was found between parental mental health (reported by parents) and CYP’s previous mental health diagnoses [X(4) = 15.30, p = 0.004]; as such a young person was significantly more likely to have been diagnosed with mental health difficulties if their parent also had mental health difficulties. A significant association was also found between parental drug misuse [X(4) = 13.92, p = 0.008], parental criminality [X(4) = 13.62, p = 0.009] and CYP mental health diagnosis. Results suggested those CYP whose parents had misused drugs or reported criminality were more likely to have diagnosed mental health difficulties. Furthermore, those CYP with diagnosed mental health difficulties were also statistically significantly more likely to have experienced neglect [X(4) = 16.87, p = 0.002], have witnessed domestic violence [X(4) = 12.68, p = 0.013], abuse [X(4) = 14.84, p = 0.005], and have experienced parental separation or loss [X(4) = 12.23, p = 0.016].

Based on descriptive statistics, a number of differences were identified between March 2019–2020 and March 2020–2021. As the first COVID-19 lockdown occurred in March 2020 in the United Kingdom, this allows for a comparison pre and post the COVID-19 pandemic beginning. The mean age of the sample was broadly similar between the 2 years (13.48 vs 13.56). The number of females attending slightly increased from 64% in 2019–2020 to 69% in 2020/2021, whereas the number of males slightly reduced (36 to 31%). More CYP were previously known to CAMHS in the year following the pandemic beginning (69 vs 60%), and more were currently under CAMHS (29 vs 17%).

A series of independent t-tests were conducted to establish if there were any significant differences in attendees’ characteristics before and after the pandemic beginning. There were significant differences in the number of patients who had a social worker [t(227) = −2.0, p = 0.048], and were currently under CAMHS [t(219) = −0.2.27, p = 0.024], whereby more CYP were known to a social worker (28.1 vs 18.3%) and were under CAMHS (28.9 vs 16.7%) in the year after the pandemic began. There was also a significant difference in sub-diagnosis [t(228) = −2.61, p = 0.010], with significantly more CYP presenting with suicide ideation with deliberate self-harm after the pandemic (43%) than before (27%).

In terms of referral pathways, there were significant differences between the 2 years regarding the number of CYP who were followed-up by CAMHS [t(218) = −2.15, p = 0.033], and who were referred to other services [t(219) = 2.89, p = 0.004]. Specifically, a higher number of CYP were followed-up by CAMHS after the pandemic (26%) compared to the year before (15%), whereas a higher number of CYP were referred to other services or specialties in the year before the pandemic (21%) than afterward (8%). No other differences were statistically significant.

Chi-square analyses also indicated a significant association between year of attendance and code used [X(7) = 38.59, p < 0.001, φ = 0.40], with the code “social problems” being used more in the year before the pandemic (31%) than afterward (12%). This is illustrated in Figure 2.

Our findings are consistent with existing evidence regarding the individual factors associated with suicidality in CYP, including previous self-harm and a history of mental health difficulties (42, 43). Furthermore, the mean age of attendees (13.5 year old) is in line with the notion that early adolescence is a critical period for the onset of mental health conditions, with current statistics suggesting that 50% of lifetime difficulties are first experienced by 14 years of age (44). However, while the higher rate of girls relative to boys presenting in the ED in this study is counter-intuitive based on existing evidence highlighting that men, particularly young men, are more likely to die by suicide than women (45), this may be explained by the gender paradox in suicidal behavior. For example, it is thought that while young men have low rates of suicidal behavior relative to women, they have higher rates of suicide mortality (46); in other words, men are more likely to use violent means and have more serious intentions to die when attempting suicide (47). Thus, they may be less likely to present in the ED looking for help, and their attempts may be more likely to result in death before they can receive support. Women are also more likely to seek help for mental health conditions generally (48, 49) and are more likely to experience internalizing difficulties (e.g., depression/anxiety) commonly associated with suicidality (50, 51), which may also help to explain the gender discrepancy identified here.

An unanticipated and particularly noteworthy finding in this sample is the relatively large proportion [37 vs 1.76% in the wider population; (52)] of CYP who were diagnosed with ASC or suspected to have ASC traits, and specifically those who were autistic girls [61 vs 33% of autistic individuals in the wider population are girls/women; (53)]. There is some evidence to suggest that autistic adults are at greater risk of suicidal ideation and behaviors, including deaths, relative to the rest of the population [0.31 vs 0.04% premature death by suicide; (17, 54–57)]. Indeed, a Swedish mortality study showed a sevenfold increased risk of premature death by suicide in people with ASC compared to the general population (58). However, one study by Hannon and Taylor (59) examining suicidal behavior among autistic young people found that rates were similar in comparison to the general population, with overlapping risk factors. While they initially suggested that ASC traits might be risk factors (e.g., social and communication difficulties may lead to interpersonal problems and social isolation), another study with adolescents and young adults found that it was those with less “severe” autistic traits that were at heightened risk, potentially due to their better emotional insight and more contact with others putting them at increased risk of distress and peer victimization (59, 60).

This hypothesis may lend support for the high presentation rates of autistic girls in the current study, as they are more likely to adopt masking behaviors and engage in social activities than autistic boys (61), which may put them at heightened risk of suicidality. However, one study by Hedley et al. (62) found that while social support and loneliness predicted suicide ideation in autistic individuals, the pattern of relationships in their path analyses was nearly identical for males and females. Thus, this casts doubt on the social interaction hypothesis. Conversely, diagnosis stage may provide an alternative explanation. Girls are more likely to be diagnosed with autism at a later age than boys (19, 63), so there is potential that the delayed diagnosis and subsequent lack of attribution for their experiences may be causing girls heightened levels of distress (60, 64), although more research is first needed before this hypothesis can be confirmed. However, based on the findings identified here, autistic CYP, particularly girls, are a vulnerable group at an increased risk for suicidal crisis, and there is a clear need for greater support and urgency for further research into risk detection and prevention of suicide in autistic people. Increased acceptance of autistic CYP in schools and social groups, and greater awareness and flexibility for autistic CYP, will benefit everyone in society and may also lead to a decrease in feelings of rejection and suicidal thinking for autistic individuals (57). Schools, primary care, and CAMHS services need to be aware of this risk, and should be delivering targeted and effective interventions and services for autistic CYP, in order to prevent them from reaching the point of crisis.

In terms of familial characteristics, high rates of CYP in this sample reported parental or familial difficulties, many of which have previously been identified in the international literature as common risk factors for suicidal ideation or behaviors. For instance, these findings are in keeping with reports from Taliaferro and Muehlenkamp (29) that factors associated with abuse, parental substance misuse, and familial dysfunction or violence are related to adolescent suicide attempts. Furthermore, Perkins and Hartless (27) found a clear association between abuse and frequent suicidal thoughts and attempts for all adolescents, regardless of gender or ethnicity, although factors such as parental substance abuse and family structure were not significant once other risk factors were accounted for. However, another study of adolescents hospitalized for suicidal ideation and/or behavior did find that poorer perceptions of family functioning by adolescents were positively associated with suicide ideation and history of suicide attempts (65). Similarly, a study by Thompson et al. (66) in the United States, found that the presence of familial adverse childhood experiences (ACEs), including abuse, parental incarceration, and family history of suicidality in childhood increased the odds of later suicide attempts or ideation, with the accumulation of multiple ACEs further increasing the odds. Thus, while our findings are in keeping with previous international literature, this is one of the first times that familial factors associated with suicidal crisis in a clinical sample of CYP have been examined in a United Kingdom setting. The characteristics identified here can be used to help inform early identification and intervention with CYP who may be at heightened risk of experiencing suicidal crisis, although further studies with larger samples are still needed.

Eight different codes were used by clinicians when recording patients who presented in the ED with suicidal crisis, the most common of which was “social problems.” Perhaps unsurprisingly, codes of “overdose” and “deliberate self-harm” were used more frequently for CYP who had presented with suicidal ideation and self-harm. However, few other significant predictors of code use were identified. The heterogeneity in coding practices found here is a cause for concern that has been flagged in several other reports recently, as current data may be misrepresentative of the true volume of suicidal crisis presentations [e.g., (33, 67)]. Given that EDs are often the first point of contact for people experiencing distress relating to suicidality (68), it is important that we have a system that can provide an accurate indication of the number and characteristics of individuals who are presenting in crisis, in order to be able to offer tailored and effective levels of support.

A “suicidal thoughts” code is available for use (R45.81) in the current coding system (International Classification of Diseases 10^th Edition; ICD-10); however, in the present sample, it was only used 10% of the time, despite all patients in the study having attended ED with “suicide ideation” as their primary diagnosis. It appears that this is not a unique issue, with other international studies reporting similar findings. For instance, research conducted by Sveticic et al. (67) in Australia also found great variation in code use for suicide ideation, with 38 different codes used, and the R45.81 code being used less than half the time. Another study by Anderson et al. (69) in America reported the same issue, with only 3% of patients with an indication of suicidal ideation in the notes field having a corresponding ICD-9 code. While the exact reason for the coding issue in the current study is unknown, it may be due to the ICD-10 guidelines stating that this code should only be used if the clinician is certain there is no underlying mental disorder, meaning that other codes are chosen. Further research is needed in this area, to determine clinicians’ reasoning for code choices, and there is a clear need for improved coding practices relating to suicidal crisis (33).

This study has several strengths and provides significant contributions to the extant evidence base, by helping to paint a picture of the CYP who are seeking help from EDs for suicidal crisis in North-West England. This information can help to inform early identification strategies to determine those who may be at-risk, meaning effective intervention and support strategies can be offered, and subsequent distress prevented. However, these findings should be interpreted within the context of some methodological limitations. Firstly, the relatively small sample size limits the statistical power, and effect sizes are also generally small, meaning that any significant findings should be interpreted with caution, and analytic techniques are limited in their complexity. The sample is also limited to CYP who are seeking help, and thus likely only reflects a small minority of CYP who are experiencing suicidal crisis. Furthermore, as a case series study, there was no control group, meaning that true risk factors cannot be identified. Comparative data would highlight the socio-demographic characteristics that puts some CYP at heightened risk of suicidal crisis, relative to the rest of the population. In addition, data were only collected from a single hospital in one region of England over a 2-year period. Thus, care must be taken when attempting to generalize these findings to other geographical regions. It is also important to note that the variables available for analysis were limited to those self-reported by attendees, their parents, and/or clinicians. Finally, due to its retrospective nature, some patient data have been inevitably missing.

Notwithstanding these limitations, this was a highly exploratory study, aiming to gain insight into and understanding of patient profiles and pathway management. Thus, this provides a basis for further work in this area, utilizing larger samples, comparison groups, and multiple NHS Trusts across various regions of the United Kingdom.

Our results provide the first clear indication in the United Kingdom of the common socio-demographic characteristics of CYP in North-West England who present in EDs seeking help for suicidal crisis, and the resultant care pathways that they follow. The findings highlight the high proportion of attendees who were White and female, and who were diagnosed with or suspected to have autism, as well as inconsistencies in the coding practices used when recording patient attendances. Thus, our findings have implications for early identification and intervention with children who may be at a heightened risk for suicidal crisis, and for the development of guidance surrounding the coding practices that need to be used. In turn, this will help to ensure an accurate understanding of the rates of suicidal crisis among CYP in the United Kingdom, and will support the development of effective services for CYP. However, we acknowledge the highly explorative nature of this study, and call for similar research to be conducted on a larger scale in order to confirm our findings.