Large-scale population-level crises, such as the coronavirus disease 2019 (COVID-19) and social unrest, are increasingly common in today's global society (Marani, Katul, Pan, & Parolari, Reference Marani, Katul, Pan and Parolari2021; Ni et al., Reference Ni, Yao, Leung, Yau, Leung, Lun and Leung2020). Apart from their consequences on mental health (Torales, O'Higgins, Castaldelli-Maia, & Ventriglio, Reference Torales, O'Higgins, Castaldelli-Maia and Ventriglio2020; Wong et al., Reference Wong, Hui, Suen, Wong, Chan, Lee and Chen2021a), concerns have been raised regarding their impacts on suicidal risks (Gunnell et al., Reference Gunnell, Appleby, Arensman, Hawton, John and Kapur2020). Most existing studies, however, tended to focus on intrinsic vulnerabilities and personal life stressors and rarely accounted for the influences of ongoing societal circumstances. Notably, suicide remains to be one of the leading causes of mortality and disability, with young people being particularly vulnerable to suicidal thoughts and behaviours (O'Connor & Nock, Reference O'Connor and Nock2014; World Health Organization, 2021). Self-disclosure of suicidal behaviours and help-seeking are, nonetheless, disproportionately suboptimal due to stigma (Calear, Batterham, & Christensen, Reference Calear, Batterham and Christensen2014; Sheehan, Corrigan, & Al-Khouja, Reference Sheehan, Corrigan and Al-Khouja2017). With the persisting challenges in predicting suicidal risks (Turecki & Brent, Reference Turecki and Brent2016), the identification of novel markers that are less stigmatising is crucial for facilitating early risk detection and intervention.
Suicidal ideation, plan, and attempt are among the three most commonly studied non-fatal suicidal behaviours that vary on a continuum of severity (Nock et al., Reference Nock, Green, Hwang, McLaughlin, Sampson, Zaslavsky and Kessler2013; Turecki & Brent, Reference Turecki and Brent2016). While suicidal ideation is more common (Nock et al., Reference Nock, Borges, Bromet, Alonso, Angermeyer, Beautrais and Williams2008), this milder form of suicidal behaviour is often considered to precede suicide planning and attempt (Hintikka et al., Reference Hintikka, Koivumaa-Honkanen, Lehto, Tolmunen, Honkalampi, Haatainen and Viinamäki2009). Each of these outcomes thus represents important intervention targets (Turecki & Brent, Reference Turecki and Brent2016). Few studies, however, have examined how a range of personal and environmental risk factors may be similarly or differentially associated with all three outcomes in epidemiological youth samples.
Various individual-level risk factors for suicidal behaviours have been discussed, including younger age, higher levels of hopelessness and neuroticism, self-harm, as well as pre-existing psychiatric conditions, particularly depressive disorder and excessive substance use (Cha et al., Reference Cha, Franz, Guzmán, Glenn, Kleiman, Nock and Nock2018; Mars et al., Reference Mars, Heron, Klonsky, Moran, O'Connor, Tilling and Gunnell2019). Cognitive deficits, such as updating, have also been suggested to be associated with greater suicidal risks (Bredemeier & Miller, Reference Bredemeier and Miller2015). These factors are, nonetheless, either difficult to modify or are non-specific to suicidal outcomes. While self-harm and psychiatric conditions may be strong indicators of risk, young people are typically also less willing to disclose such experiences and seek help (Gulliver, Griffiths, & Christensen, Reference Gulliver, Griffiths and Christensen2010; Michelmore & Hindley, Reference Michelmore and Hindley2012; Rowe et al., Reference Rowe, French, Henderson, Ougrin, Slade and Moran2014).
One potentially important yet seldomly investigated risk factor is suicide-related rumination. Prior studies that examined rumination in mental health and suicidal outcomes have typically emphasised the “self-focused” depressive-type rumination, in which the thought content surrounds one's distress symptoms and their causes and consequences (Miranda & Nolen-Hoeksema, Reference Miranda and Nolen-Hoeksema2007; Rogers & Joiner, Reference Rogers and Joiner2017). Suicide-related rumination, in contrast, describes repetitive thoughts about suicide, which may hold greater relevance to the determination of suicidal risks. Indeed, a few initial studies have offered support for suicide-related rumination as an indicator of suicidal risks (Höller, Teismann, & Forkmann, Reference Höller, Teismann and Forkmann2021; Rogers & Joiner, Reference Rogers and Joiner2018; Rogers, Gallyer, & Joiner, Reference Rogers, Gallyer and Joiner2021). Notably, a recent study has found depressive rumination to be no longer predictive of lifetime ideation and attempt in both student and community adult samples when suicide-related rumination was accounted for (Rogers, Law, et al., Reference Rogers, Law, Houtsma, Tucker, Anestis and Joiner2021), highlighting its potential relevance.
In addition, considering the roles of both proximal and distal environmental stressors can be important. For instance, the impact of exposure to personal stressful life events (SLEs) on suicide has generally been studied (Stewart et al., Reference Stewart, Shields, Esposito, Cosby, Allen, Slavich and Auerbach2019; Turecki and Brent, Reference Turecki and Brent2016). Family functioning has also been raised to be a key factor of youth suicide (Law & Shek, Reference Law and Shek2016; Leung, Kwok, & Ling, Reference Leung, Kwok and Ling2016), which is particularly relevant to Hong Kong since most young people tend to live with their family due to limited living space and high cost of living. How population-level environmental stressors, such as social unrest and COVID-19 – which have been associated with increased mental health risks (Ni et al., Reference Ni, Yao, Leung, Yau, Leung, Lun and Leung2020; Wong et al., Reference Wong, Hui, Wong, Suen, Chan, Lee and Chen2021b) – may additionally contribute to different suicidal outcomes in epidemiological samples remains to be studied.
Findings from a limited number of studies on this topic have suggested the influence of epidemics on suicidal behaviours (Leaune, Samuel, Oh, Poulet, & Brunelin, Reference Leaune, Samuel, Oh, Poulet and Brunelin2020; Zortea et al., Reference Zortea, Brenna, Joyce, McClelland, Tippett, Tran and Platt2021). While mixed findings have been reported concerning suicidal risks before and during COVID-19 (Cheung, Lam, Lee, Xiang, & Yip, Reference Cheung, Lam, Lee, Xiang and Yip2021; Sher, Reference Sher2020), a recent study has found disproportionately higher rates of suicidal ideation during COVID-19 among young adults in Hong Kong compared to other populations (Schluter et al., Reference Schluter, Généreux, Hung, Landaverde, Law, Mok and Roy2021). This finding was partly attributed by the researchers to the cumulative effects of the series of social unrest since June 2019 (which had only gradually subsided since the local COVID-19 pandemic in January 2020) and the pandemic (Schluter et al., Reference Schluter, Généreux, Hung, Landaverde, Law, Mok and Roy2021). Indeed, the additive effects of social unrest-related traumatic events (TEs) and COVID-19 pandemic-related events (PEs) on psychopathological symptoms in young people have previously been reported (Wong et al., Reference Wong, Hui, Suen, Wong, Chan, Lee and Chen2021a).
The current study thus aimed to first provide an update on the prevalence of lifetime and 12-month suicidal ideation, plan, and attempt in a large representative epidemiological youth sample in Hong Kong. We also aimed to examine factors associated with these three key suicide-related outcomes using a comprehensive multifactorial model, covering aspects ranging from demographics to psychological and personality factors, interpersonal relationships, psychiatric conditions, as well as suicide-related rumination, cognitive ability, and both proximal and distal personal and population-level stressors.
Population and study design
Participants were consecutively recruited between September 2019 and 2021 as part of the ongoing Hong Kong Youth Epidemiological Study of Mental Health (HK-YES). The HK-YES is an ongoing territory-wide, household-based epidemiological study of youth mental health in Hong Kong that adopts a stratified multistage cluster sampling design as in previous epidemiological studies (Lam et al., Reference Lam, Wong, Wang, Chan, Chen, Ng and Bebbington2015). Young people between the ages of 15 and 24 at the time of recruitment were invited via mail sent to addresses obtained from the local government, which were randomly selected and clustered by districts and housing types. All assessments were conducted through face-to-face interviews, with the option of online video conferencing using the same procedures during COVID-19. Domains including suicide-related outcomes, psychiatric symptoms and disorders, personality and psychological factors, external stress exposure, and sociodemographic characteristics were assessed in this study. During the data collection period, it was estimated that our invitation letters were able to reach 5035 valid addresses (invalid addresses were defined as ‘age range not met’, ‘unit not occupied’, or ‘false address’). Of these valid addresses, 2956 participants (58.7%) were recruited, among which 2540 (85.9%) provided complete data on the variables of interest and were included in this study.
Seven participants (0.3%) of this sample reported being neither female nor male. Due to the small sample size, main analyses of this study focused on the biological sex of participants. Proportions of young people with suicidal ideation and behaviours in the past 12 months across each identified gender group are provided in the online Supplementary Material Table S1.
Suicidal ideation, plan, and attempt
Lifetime and 12-month suicidal ideation, plan, and attempt, respectively, were assessed using items adapted from the Columbia–Suicide Severity Rating Scale (C-SSRS; Posner et al., Reference Posner, Brown, Stanley, Brent, Yershova, Oquendo and Mann2011) as in previous large-scale suicide research (e.g. Nock et al. Reference Nock, Stein, Heeringa, Ursano, Colpe and Fullerton2014, Reference Nock, Millner, Joiner, Gutierrez, Han, Hwang and Kessler2018). Suicidal ideation was assessed using the items ‘Did you ever in your life wish you were dead or would go to sleep and never wake up?’ or ‘Did you ever in your life have thought of killing yourself?’. Those who reported ‘yes’ to either of these items were further asked if they have ever made a suicide plan [‘Did you ever think about how you might kill yourself (e.g. taking pills, shooting yourself) or work out a plan of how to kill yourself?’] and suicide attempt [‘Have you ever made a suicide attempt (i.e. purposefully hurt yourself with at least some intent to die)?’]. The scale has shown good validity for use in young people (Posner et al., Reference Posner, Brown, Stanley, Brent, Yershova, Oquendo and Mann2011).
Among those with lifetime suicidal ideation, participants were asked to rate the degree to which they have ruminated about their suicidal thoughts using a modified item also from the C-SSRS (Posner et al., Reference Posner, Brown, Stanley, Brent, Yershova, Oquendo and Mann2011) as in Nock et al. (Reference Nock, Millner, Joiner, Gutierrez, Han, Hwang and Kessler2018): ‘During that worst week, how easy was it for you to control those thoughts or push them out of your mind when you wanted to?’ on a 5-point Likert scale (‘easy’ to ‘impossible, unable to control the thoughts’).
Population-level and personal stressful events
Three types of external stressful events – social unrest-related TEs, COVID-19 PEs, and personal SLEs – were assessed. First, exposure to TEs since June 2019 was assessed using a three-item binary checklist capturing key experiences during the social unrest as has been previously reported (Wong et al., Reference Wong, Hui, Wong, Suen, Chan, Lee and Chen2021b), including ‘crowd dispersal by the use of force’, ‘arrest or detention’, and ‘media viewing of others being physically attacked’.
COVID-19 PEs were asked using four items capturing stressful experiences related to COVID-19, including ‘having sufficient protective gears’ (reversed coded), ‘increased personal and rest time due to remote work/school’ (reversed coded), ‘increased conflicts with family due to remote work/school’, and ‘increased work/studies hours due to remote work/school’. Each item was assessed using a 5-point Likert scale (‘completely disagree’ to ‘completely agree’) before November 2020 and in the form of a binary checklist from November 2020. For those who completed their assessments before November 2020, a rating of ‘completely disagree’ or ‘disagree’ for the items ‘having sufficient gears’ and ‘personal and rest time increased’ and a rating of ‘agree’ or ‘completely agree’ for the other two items were recoded into 1. A score of 0 was given to those whose data were collected prior to the first confirmed case in Hong Kong (23 January 2020).
Personal SLEs were assessed using an adapted version of the List of Threatening Experiences (Brugha, Bebbington, Tennant, & Hurry, Reference Brugha, Bebbington, Tennant and Hurry1985), which is a 12-item binary checklist that captures different personal life stressors during the past year, such as serious illness, death of a first-degree or second-degree relative, and loss of a steady relationship (Brugha et al., Reference Brugha, Bebbington, Tennant and Hurry1985; Rosmalen, Bos, & de Jonge, Reference Rosmalen, Bos and de Jonge2012). ‘Expelled from school’ and ‘dropped out of school’ were provided as alternatives to ‘sacked from job’ and ‘unemployment’, respectively, to adapt to the current youth population. An ‘others’ option was also added, yielding a final number of 13 SLEs. The items endorsed in each of the measures were independently summed to generate composite scores for TEs, PEs, and SLEs, respectively.
Hopelessness and neuroticism
Hopelessness during the past week was assessed using the 20-item Beck's Hopelessness Inventory (Beck, Weissman, Lester, & Trexler, Reference Beck, Weissman, Lester and Trexler1974). Items were rated on a binary checklist (true/false) and summed to generate a composite score. Neuroticism was assessed using the neuroticism subscale (eight items) of the 44-item Big Five Inventory (John & Srivastava, Reference John, Srivastava, Pervin and John1999). Items were rated on a 5-point Likert scale (‘disagree strongly’ to ‘agree strongly’) and summed to generate a composite score (α = 0.81).
Family functioning was assessed using the 16-item Brief Family Relationship Scale (Fok, Allen, & Henry, Reference Fok, Allen and Henry2014), which captures three key domains, including family cohesion (e.g. ‘help and support each other’), expressiveness (e.g. ‘can talk openly in our home’), and conflict (e.g. ‘we argue a lot’), with each rated on a 4-point Likert scale (‘a lot’ to ‘not at all’). A higher score suggests poorer family functioning (α = 0.90).
Working memory, as reflective of cognitive ability, was assessed using the Forward and Backward Digit Span Test (DS-F and DS-B, respectively) in the Wechsler Adult Intelligence Scale, third edition (Wechsler, Reference Wechsler1997). The DS-F requires participants to store the sequence of digits in their memory (more reflective of short-term memory), while the DS-B requires participants to process and reorganise information while storing the sequence of digits simultaneously (Grégoire & Van der Linden, Reference Grégoire and Van der Linden1997). A lower score on either of the tests suggests poorer working memory capability.
Non-suicidal self-harm, depressive disorder, and alcohol use
Lifetime non-suicidal self-harm was assessed using the item ‘Did you ever do something to hurt yourself on purpose, without wanting to die (e.g. cutting yourself, hitting yourself, or burning yourself)?’ from the Self-Injurious Thoughts and Behaviors Interview (Nock, Holmberg, Photos, & Michel, Reference Nock, Holmberg, Photos and Michel2007). Any major depressive episode (MDE) during the past 12 months based on the DSM-V criteria was determined using items from the Composite International Diagnostic Interview–Screening Scales (Kessler et al., Reference Kessler, Calabrese, Farley, Gruber, Jewell, Katon and Wittchen2013). The degree of alcohol use was assessed using the 10-item Alcohol Use Disorders Identification Test (Saunders, Aasland, Babor, de la Fuente, & Grant, Reference Saunders, Aasland, Babor, de la Fuente and Grant1993), which captures aspects of hazardous alcohol use, dependence symptoms, and harmful use of alcohol, with each item rated on a 5-point Likert scale. A higher total score reflects higher risks in excessive alcohol use (α = 0.80).
Past adversity through the age of 17 was assessed using items adapted from the Adverse Childhood Experiences International Questionnaire (World Health Organization, 2018) as in previous studies (e.g. Mall et al., Reference Mall, Mortier, Taljaard, Roos, Stein and Lochner2018), capturing aspects including abuse (physical, emotional, sexual), neglect, and parental psychopathology using a 5-point Likert scale (‘never’ to ‘very often’). Those who gave a rating of 3 (sometimes) to any of the experiences were categorised into a group denoting participants ‘with experience of childhood adversity’.
The prevalence of lifetime and 12-month suicidal ideation, ideation-only, plan, and attempt were each estimated with weighting adjustments according to age and sex data from the 2019 Hong Kong Census to improve representativeness.
Four separate multivariable logistic regression models were then applied to four key suicidal outcomes to examine their associated factors, namely 12-month suicidal ideation, ideation-only (i.e. excluding those without suicide plan or attempt), suicide plan, and suicide attempt. The variables included in each model included suicide-related rumination; exposure to TEs, PEs, and SLEs; hopelessness and neuroticism; family functioning; working memory (forward and backward digit span scores); lifetime non-suicidal self-harm (yes/no), 12-month MDE (yes/no), and 12-month alcohol use, and age, sex (female/male), and childhood adversity (yes/no). Online Supplementary analyses were performed to identify factors associated with 12-month ideation-only, plan, and attempt with age as a categorical variable (15–18/19–25 years) (see online Supplementary Material Table S3). Additionally, a two-stage design was applied to examine factors associated with 12-month suicide plan and attempt, respectively, among participants with suicidal ideation during the past 12 months. Adjusted odds ratios with 95% confidence intervals were reported, with statistical significance set at the p < 0.05 level. All analyses were performed using SPSS version 26.0.
Prevalence of lifetime and 12-month suicidal ideation, plan, and attempt, suicide-related rumination, and non-suicidal self-harm among young people in Hong Kong
Sample characteristics are presented in Table 1. After weighting adjustments, the lifetime prevalence of suicidal ideation, plan, and attempt in the Hong Kong youth population was 47.4, 14.6, and 5.5%, respectively. The weighted 12-month prevalence of suicidal ideation, plan, and attempt was 20.0, 4.6, and 1.3%, respectively. Suicidal ideation-only was reported in 15.4% of the local youth population (after weighting). A flowchart illustrating the 12-month prevalence of suicide plan and attempt among those with suicidal ideation are presented in Fig. 1.
MDE, major depressive episode; PEs, COVID-19 pandemic-related events; SLEs, personal stressful life events; TEs, social unrest-related traumatic events.
Note. Statistics are presented in the form of mean (s.d.), unless stated otherwise. Values significant at the p < 0.05 level are in boldface.
Regarding suicide-related rumination, 5.8 and 3.4% of the participants (after weighting) reported the thoughts being ‘very difficult’ and ‘impossible’ to control even when they wanted to, respectively. Meanwhile, the weighted prevalence of lifetime non-suicidal self-harm was 14.0%.
Factors associated with 12-month suicidal ideation, ideation-only, plan, and attempt in the epidemiological youth sample
Table 1 presents findings from the four respective multivariable logistic regression models for each of the 12-month suicidal outcomes. Some similarities and differences in the patterns of associations were observed. Summarised findings from the regression models are presented in online Supplementary Material Table S2 for ease of comparison.
Shared factors associated with 12-month suicidal outcomes
Among the variables, suicide-related rumination was among the only factor that was significantly associated with all four suicide-related outcomes, including suicidal ideation (OR 1.71, CI 1.58–1.86), ideation-only (OR 1.64, CI 1.51–1.79), plan (OR 1.55, CI 1.35–1.77), and attempt (OR 1.52, CI 1.18–1.95) (all p < 0.01).
Meanwhile, poorer family functioning was associated with elevated odds of 12-month suicidal ideation and ideation-only (OR 1.04, CI 1.02–1.06 for both) and attempt (OR 1.08 CI 1.02–1.13), but not with plan. Higher levels of hopelessness and having an MDE in the past 12 months were associated with elevated odds of suicidal ideation, ideation-only, and plan (all p < 0.05), but not with attempt. Female sex was associated with both suicidal ideation and plan (both p < 0.05). Meanwhile, self-harm was associated with suicide plan and attempt (both p < 0.05), but not with ideation (Table 1).
Unique contributing factors to 12-month suicidal outcomes
Apart from the shared risk factors, a number of other factors showed specificity in their associations with the respective suicide-related outcomes. Personal background factors, such as higher degrees of neuroticism and younger age, were associated with elevated odds of suicidal ideation (OR 1.07, CI 1.05–1.10 and OR 0.93, CI 0.89–0.97, respectively) and ideation-only (OR 1.08, CI 1.05–1.11 and OR 0.93, CI 0.88–0.97, but not with plan or attempt (Table 1). Poorer working memory capacity, as reflected by forward digit span score, was associated specifically with suicide plan (OR 0.82, CI 0.72–0.94). Meanwhile, COVID-19 PEs were specifically associated with elevated odds of suicide attempt (OR 1.61, CI 1.07–2.41).
Factors associated with 12-month suicide plan and attempt among those with ideation: a two-stage approach
Further analyses were performed to determine factors associated with 12-month suicide plan and attempt among young people with suicidal ideation in the past 12 months (n = 533).
Suicide-related rumination remained to be significantly associated with elevated odds of 12-month suicide plan (OR 1.24, CI 1.05–1.47) (Table 2). A non-significant positive trend was observed between suicide-related rumination and suicide attempt (OR 1.32, CI 0.98–1.78, p = 0.064). Poorer working memory capacity (OR 0.84, CI 0.73–0.97) and 12-month MDE (OR 2.04, CI 1.27–3.28) were specifically associated with suicide plan. Meanwhile, with all other factors accounted for, COVID-19 PEs (OR 1.72, CI 1.13–2.62), poorer family functioning (OR 1.06, CI 1.01–1.12), self-harm (OR 3.60, 1.56–8.33), as well as personal SLEs (OR 1.32, CI 1.00–1.74), were among specifically associated with elevated odds of suicide attempt (Table 2).
MDE, major depressive episode; PEs, COVID-19 pandemic-related events; SLEs, personal stressful life events; TEs, social unrest-related traumatic events.
Note. Values are presented in the form of mean (s.d.), unless stated otherwise.
Using data from a representative epidemiological sample of young people in Hong Kong, we found nearly half of the local youth population (47.3%) have had suicidal thoughts at least once or more in their lifetime, one in seven (14.6%) have made a plan, and one in 18 (5.5%) have ever made an attempt. The prevalence of 12-month suicidal ideation in the Hong Kong youth population was also as high as 20.0%, which is slightly higher than those reported in a previous local study among secondary school students (13.7% in Shek and Yu, Reference Shek and Yu2012) and in other meta-analysis studies (18.8% among high school students in the United States between 1991 and 2017 in Lindsey, Sheftall, Xiao, and Joe, Reference Lindsey, Sheftall, Xiao and Joe2019; 18.7% among left-behind children in China in Qu et al., Reference Qu, Shu, Zhang, Wu, Ma, Han and Sun2021).
Importantly, our findings showed that suicide-related rumination was among the only factor associated not only with suicidal ideation but also plan and attempt. This observation extended findings from a number of recent works (Rogers & Joiner, Reference Rogers and Joiner2018; Rogers, Gallyer, et al., Reference Rogers, Gallyer and Joiner2021; Rogers, Law, et al., Reference Rogers, Law, Houtsma, Tucker, Anestis and Joiner2021) to suggest its utility in identifying young people at greater suicidal risks. Notably, we provided novel evidence to show the specific associations between COVID-19 stressors and suicide attempt beyond other previously reported intrinsic and personal-level extrinsic risk factors of suicide. The importance of accounting for the influences of suicide-specific intrinsic factors, as well as both personal and population-level extrinsic factors, to guide more accurate and effective suicide prevention and intervention work is highlighted.
Significance of suicide-related rumination
Ruminative experiences related to suicidal thoughts may reflect a form of preservative cognition in which one's attention is disproportionately narrowed to suicide-related content (Whitmer & Gotlib, Reference Whitmer and Gotlib2013). This would be in line with previous findings in suggesting attentional biases to and fixating on suicide-related contents could act as cognitive warning signs for suicidal behaviours (Cha, Najmi, Park, Finn, & Nock, Reference Cha, Najmi, Park, Finn and Nock2010; Moscardini, Aboussouan, Bryan, & Tucker, Reference Moscardini, Aboussouan, Bryan and Tucker2020). The notion of ‘cognitive rigidity’ in the cognitive model of suicide (Ellis & Rutherford, Reference Ellis and Rutherford2008) may also be relevant. Notably, suicide-related rumination not only may be modifiable (Grierson, Hickie, Naismith, & Scott, Reference Grierson, Hickie, Naismith and Scott2016; Holdaway, Luebbe, & Becker, Reference Holdaway, Luebbe and Becker2018; Watkins et al., Reference Watkins, Mullan, Wingrove, Rimes, Steiner, Bathurst and Scott2011), but could also be understood as increased ‘stickiness’ of suicidal thoughts in one's mental representation (Joormann, Levens, & Gotlib, Reference Joormann, Levens and Gotlib2011), which may help to reduce the stigma associated with suicidal behaviours.
Significance of non-suicidal self-harm
The finding that non-suicidal self-harm was a relatively robust risk factor for suicide attempt is consistent with the literature (Castellví et al., Reference Castellví, Lucas-Romero, Miranda-Mendizábal, Parés-Badell, Almenara, Alonso and Alonso2017; O'Connor et al., Reference O'Connor, Wetherall, Cleare, Eschle, Drummond, Ferguson and O'Carroll2018). Thoughts about self-harm have also been suggested to indicate risks for psychiatric conditions and have long-term consequences on physical health and functioning (Bjureberg et al., Reference Bjureberg, Ohlis, Ljótsson, D'Onofrio, Hedman-Lagerlöf, Jokinen and Hellner2019; Turner, Chapman, & Layden, Reference Turner, Chapman and Layden2012). As with the tendency to conceal or deny suicidal behaviours, nonetheless, only a minority of young people would disclose their self-harm experiences and seek help due to the fear of being perceived as ‘attention-seeking’ (Michelmore & Hindley, Reference Michelmore and Hindley2012; Rowe et al., Reference Rowe, French, Henderson, Ougrin, Slade and Moran2014). Considering other indicators that are less stigmatising (e.g. ever had suicide-related rumination, poor family functioning, exposure to population-level stressors) would thus also be important to confirm the level of risk among those who did and did not report no self-harm experience.
Significance of personal and population-level external stressors
The observations that COVID-19 stressful events and poorer family functioning showed stronger associations with suicide attempt, rather than intrinsic vulnerabilities, are thus critical. As the common perceptions of suicidal behaviours tended to emphasise individual vulnerabilities (Batterham, Calear, & Christensen, Reference Batterham, Calear and Christensen2013; Carpiniello & Pinna, Reference Carpiniello and Pinna2017), placing greater weight on environmental influences may further help to reduce stigma and facilitate help-seeking. Importantly, in the case of not only Hong Kong but also many other societies, the effect of COVID-19 on suicidal behaviours could have been exacerbated by other prior or concurrent population-level stressors, such as social protests, financial crises, and environmental crises (Hou et al., Reference Hou, Hall, Liang, Li, Liu and Galea2021; Moutier, Reference Moutier2021; Wong et al., Reference Wong, Hui, Wong, Suen, Chan, Lee and Chen2021c). The resulting societal, lifestyle, and psychosocial changes can make meaning-making and adaptations difficult in young people, who are simultaneously undergoing critical developmental transitions (Arain et al., Reference Arain, Haque, Johal, Mathur, Nel, Rais and Sharma2013; Park, Reference Park2010). The role of the family in suicidal outcomes during youth may be even more pronounced in collectivistic cultures like Hong Kong, where family relationships are highly valued (Ho et al., Reference Ho, Mui, Wan, Ng, Stewart, Yew and Chan2016). Supporting parents or guardians in cultivating more open communications may help young people better cope with these circumstances and facilitate early risk detection and help-seeking.
Identifying those at greater suicidal risk using a two-stage approach
To further facilitate the identification of young people at greater suicidal risks in population and clinical settings, a two-stage design can be useful (i.e. selecting those with suicidal ideation). Based on our analyses, suicide-related rumination appeared to remain a strong indicator of suicidal risk (significant for suicide plan and positive trend for attempt). Interestingly, whereas other intrinsic factors (poorer cognitive ability and history of depressive disorder) showed stronger associations with suicide plan, the experience of external triggers (COVID-19 stressors, poorer family functioning, and personal SLEs), as well as behavioural indicators (lifetime non-suicidal self-harm), may play more critical roles in suicide attempt. These observations shed light on the potential interplay of intrinsic-extrinsic factors across the various suicidal outcomes, which have implications both on our understanding of suicidality and intervention design.
Notably, during the period of social unrest and COVID-19, the disruptions to interpersonal relationships, education, and employment could have added to the cognitive load of young people. Suicide and ‘self-sacrifice’ were indeed common sentiments among young people on online platforms at the time (Yip & Pinkney, Reference Yip and Pinkney2022). Young people with fewer cognitive resources (as reflected in general intellectual levels) could possibly be more susceptible to such influences (Bittár, Falkstedt, & Wallin, Reference Bittár, Falkstedt and Wallin2020). Coupled with difficulties in coping and identifying alternative solutions (Bredemeier & Miller, Reference Bredemeier and Miller2015), suicide may be viewed as the ‘only option’ (Na et al., Reference Na, Park, Kwon, Kim, Kim, Baik and Lee2020). The additional experiences of significant external stressors can possibly cause a shift to the ‘already fragile’ system and lead it to move past a ‘tipping point’ (de Beurs et al., Reference de Beurs, Bockting, Kerkhof, Scheepers, O'Connor, Penninx and van de Leemput2021; Kalisch et al., Reference Kalisch, Cramer, Binder, Fritz, Leertouwer, Lunansky and Van Harmelen2019), resulting in suicide attempts.
Strengths and limitations
To the best of our knowledge, this was to date the first youth-specific epidemiological study on suicidal behaviours conducted in Hong Kong and among the first to be conducted during COVID-19 globally. Apart from presenting an update on the local prevalence of youth suicidal behaviours, the current study also offered insights into the markers and risk factors associated with youth suicidal ideation, plan, and attempt, respectively. In the hope to improve future suicide screening in the population and clinical decisions, a two-stage approach was further applied to identify factors associated with more severe suicidal behaviours (plan and attempt) among those at increased risk (with suicidal ideation). The differences in risk factors and potential underlying mechanisms across these suicidal outcomes are also suggested.
Some limitations of this study were nonetheless acknowledged. First, the cross-sectional nature of this study prevented the determination of the direction of associations among the putative factors and outcomes. For instance, while memory impairments may prevent appropriate retrieval of past experiences for identifying solutions to current issues, which contributes to risk for suicide plan (Richard-Devantoy, Berlim, & Jollant, Reference Richard-Devantoy, Berlim and Jollant2015), it is also possible that prolonged engagement in suicide planning and continued ruminations could further deplete cognitive resources (Whitmer & Gotlib, Reference Whitmer and Gotlib2013). It would be helpful for future studies to adopt a longitudinal study design to test our study findings.
Furthermore, while some of our observations were in line with previous findings, some differences were also noted. For instance, we did not find excessive drinking to be associated with suicidal outcomes as in previous overseas studies (Darvishi, Farhadi, Haghtalab, & Poorolajal, Reference Darvishi, Farhadi, Haghtalab and Poorolajal2015). One reason may owe to the comparatively lower binge-drinking rate among young people in Hong Kong (Huang, Ho, Wang, Lo, & Lam, Reference Huang, Ho, Wang, Lo and Lam2016; Rotheray et al., Reference Rotheray, Cattermole, Liow, Machin, Graham and Rainer2011). We also did not find gender minority to be associated with elevated suicidal risks (Horwitz et al., Reference Horwitz, Berona, Busby, Eisenberg, Zheng, Pistorello and King2020; Jackman, Caceres, Kreuze, & Bockting, Reference Jackman, Caceres, Kreuze and Bockting2021), plausibly due to the small proportion of participants identifying as non-cis-gender in our epidemiological sample (n = 7; 0.3%). Nonetheless, early detection and intervention work specifically targeted at this group may be helpful to best support these young individuals. Greater attention should also be given to this population in future suicide-related research.
We also note that the associated odds in our multivariable models were generally small, which may partly relate to the use of a population-based epidemiological rather than clinical sample. Indeed, the values were similar to those reported in a previous large community-based study (Fairweather, Anstey, Rodgers, & Butterworth, Reference Fairweather, Anstey, Rodgers and Butterworth2006). The wide-ranging factors we accounted for, as opposed to a limited set of factors, might also have contributed to such differences. It would be worthwhile for future studies to apply a similar model to test the patterns of associations across cultures and in different populations and settings.
Finally, while the single-item measure of suicide-related rumination can be beneficial in low-resource and time-limited settings and in engaging young people (Wong et al., Reference Wong, Lam, Wong, Lee, Wong, Lui and Chen2021d), a more detailed assessment would be helpful to improve our understanding of this phenomenon. Capturing the daily moment-to-moment experiences of suicide-related thoughts and rumination through the experience sampling method may also help to establish a more fine-grained picture of these experiences and determine their associations with momentary mood states in real-world contexts (Kleiman & Nock, Reference Kleiman and Nock2018; Myin-Germeys et al., Reference Myin-Germeys, Kasanova, Vaessen, Vachon, Kirtley, Viechtbauer and Reininghaus2018).
In the current global context of ongoing stress, preparing clinicians and other service providers for the early detection of young people at heightened risk for suicidal behaviours is ever more important. Suicide-related rumination revealed to be a promising marker that is not only modifiable but also less stigmatising. Efforts in conveying the message about the significance of not just personal but also other circumstantial and population-level influences in suicidal outcomes to the public can also be critical in reducing stigma and improving help-seeking. Developing strategies for reducing suicide-related rumination, mitigating the impact of COVID-19 stresses, and improving family functioning can be important in future suicide prevention work.
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291722001519
The authors would like to thank all young people who participated in the HK-YES for their participation and contribution.
The study was funded by Phase I of the Commissioned Study on Mental Health Survey, Food and Health Bureau, HKSAR Government (Reference number: MHS-P1 (Part 2)).
Conflict of interest
E. Y. H. Chen has received speaker honoraria from Otsuka and DSK BioPharma, research funding from Otsuka, participated in paid advisory boards for Janssen and DSK BioPharma, and received funding to attend conferences from Otsuka and DSK BioPharma. All other authors declare no competing interests.