Helping actions given and received in response to suicide risk: Findings from an Australian nationally representative telephone survey

Abstract

We conducted a nationally representative telephone survey of 3000 Australian adults. We assessed helping responses toward people in severe distress and experiencing suicidal thoughts by asking whether respondents had undertaken 15 specified helping actions (10 recommended and 5 non-recommended actions) to support such a person. We also asked respondents who had experienced suicidal ideation in the last 12 months whether the most helpful person at that time had undertaken the 15 specified helping actions. We weighted the data to represent the Australian adult population, calculated percentage frequencies for the helping actions and used logistic regression to analyse whether sociodemographic and exposure variables related to helping actions taken. Recommended supportive actions consistent with best practice were most commonly undertaken (e.g., listened to their problems without judgement: 96.5%, 95% CI 94.6–97.7); however, some non-recommended actions (e.g., reminded the person what they have they going for them: 91.6%, 95% CI 89.1–93.5) were also very common. Suicide risk assessment actions such as asking if they had a plan for suicide (39.8%; 95% CI 35.0–44.9), and encouraging professional help (e.g., helped make an appointment with a health professional: 61.3%, 95% CI 57.4–65.2) were relatively uncommon. Age, gender, level of education, language spoken at home, own suicidal ideation and suicide prevention training were shown to affect the odds of carrying out various helping actions. Australian adults need to be educated to ask direct questions about suicide risk and to encourage others in distress to seek professional help. There may also be a place for discouraging certain behaviours that oppose best practice in suicide prevention. While most previous studies have assessed intentions to help a person at risk of suicide, this study makes a unique contribution to the literature by assessing ‘real-world’ helping behaviour, including the occurrence of helping actions undertaken that oppose best practice in suicide prevention.

Highlights

• •Surveys a nationally representative sample of Australian adults (n = 3000).
• •Identifies helping actions from non-professionals toward people at risk of suicide.
• •Finds that actions opposing suicide prevention best practice are common.
• •Finds that asking questions to assess suicide risk is uncommon.
• •Actions taken predicted by sociodemographics, suicidal ideation and training.

Background

Australia's suicide rate is currently 12.6 per 100 000 people (Australian Bureau of Statistics, 2018). More than half of the Australian adult population know someone who has died by suicide, and one third of these had a close relationship with that person (Maple, Sanford, Pirkis, Reavley, & Nicholas, 2019; Maple et al., 2016). Up to half of people who attempt, or die by, suicide express their suicidal thoughts or intentions close family and friends (Isometsä, 2001; Paraschakis, Michopoulos, Christodoulou, Koutsaftis, & Douzenis, 2015; SANE Australia and University of New England, 2015; Wasserman et al., 2008; Wolk-Wasserman, 1986). However, family and friends can find expressions of suicidal thoughts difficult to interpret and can be unsure how best to respond (Owen et al., 2012; Owens et al., 2009, 2011).

Studies of what people have done to help those at risk of suicide are rare (Rossetto, Jorm, & Reavley, 2016), and studies that examine what help is received are even rarer (Shand et al., 2015). Relevant studies published in the last decade have used largely qualitative methods to ask Australian adults how they have assisted a person close to them at risk of suicide. One Australian national survey study asked 548 people who had assisted a person experiencing depression with suicidal thoughts an open-ended question regarding how they had helped the person at risk of suicide (Rossetto et al., 2016). The other study involved interviews and focus group with 35 Australian men who had attempted suicide and 47 family members and friends (Shand et al., 2015). The former study focused purely on the actions recalled by the helper. In both studies, the open-ended data collection techniques used relied on participants to recall and state what their helping actions had been, increasing the likelihood of recall and social desirability biases.

To minimise these biases in the current study, we conducted a survey in which we asked Australian adults to indicate whether they had undertaken a series of specified helping actions. Some of these actions were consistent with best practice and some contrary to best practice. By doing this, we hoped to capture whether respondents had undertaken a variety of specific behaviours, rather than relying on recall of their dominant helping actions more broadly. We also wanted to assess whether they had undertaken actions that oppose best practice, but which may appear, to many, to be helpful. Such actions include trying to convince the person that suicide is wrong and telling them how much it would hurt their friends and family if they were to kill themselves. We also asked respondents who had been at risk of suicide to indicate the actions taken by the person who had been most helpful. We are not aware of any similar large-scale study that has assessed the viewpoint of people who has received assistance from others when experiencing suicidal thoughts. Understanding how Australian adults respond to those close to them who are at risk of suicide could help to identify behaviours and population sub-groups for targeted community suicide prevention campaigns and education.

Method

We undertook a computer-assisted telephone survey of Australian adults to identify Australian community members’ confidence, attitudes, intentions and behaviours toward helping someone in severe distress or at risk of suicide. We report here on results from those sections of the survey that included questions about the actions taken by respondents to help a person in severe distress or at risk of suicide, and the help received if the respondent had themselves been at risk of suicide. Findings from other sections of the survey are reported elsewhere (Jorm et al., 2018a, Jorm, Nicholas, Pirkis, Rossetto, & Reavely, 2018; Nicholas et al., 2019). This paper reports on the real-life helping actions taken in response to suicide risk by Australian community members. Our previous papers relating to this survey relate to confidence and intentions to help a person at risk (Nicholas et al., 2019), the effects of suicide prevention training on helping intentions and behaviors (Jorm et al., 2018a, 2018b) and the effects of exposure to suicide on own suicidal thoughts (Maple, Sandford et al., 2019). Further papers will report on the relationships between believing in common suicide myths and helping intentions and behaviors. The study was approved by the University of Melbourne Human Research Ethics Committee (Ethics ID: 1648060). Data collection took place from February 14 to March 29, 2017.

Sampling

To sample 3000 Australian adults, we used a dual frame (mobile and landline) random digit-dial method. Landline numbers were from all states and territories of Australia. Six contact attempts were made per telephone number and for landlines, interviewers asked to speak to the person associated with each number who was 18 years or over and had the last birthday. We commissioned Roy Morgan Research Ltd to conduct the survey.

Survey

To provide context for the survey questions, we randomly presented respondents with one of six vignettes. These vignettes described a person in severe distress following a series of stressful life events and were based on a vignette used by Jorm et al. in a study of intended mental health first aid behaviours of Australian adults (Jorm, Blewitt, Griffiths, Kitchener, & Parslow, 2005). All vignettes provided a description of a person showing signs of depression and impaired daily functioning. The six versions of the vignette varied by gender (John/Jenny) and directness of communication of suicide risk (distressing life events only, indirect verbal communication of suicide risk, direct verbal communication of suicide risk). The vignettes have been published in full elsewhere (Nicholas et al., 2019).

Following presentation of the vignette, interviewers asked respondents if someone in their family or close circle of friends had experienced distress like John's/Jenny's in the last 12 months. If the respondent answered ‘yes’, they were asked if they did anything to support that person. If they had, the interviewer asked if the respondent had performed the following 10 general helping actions for the person in distress (yes/no): (1) asked about how they are feeling; (2) listened to their problems without judgement; (3) asked how you could help; (4) helped make an appointment with a professional - for example, a GP [General Practitioner]; (5) went to an appointment with a professional with them – for example, a GP; (6) called a crisis line – for example, Lifeline; (7) asked if they had been thinking about killing themselves; (8) tried to solve their problems; (9) reminded them what they have going for them; and (10) reassured them you know exactly how badly they feel. Actions 1 to 7 are consistent with current best practice guidelines for assisting a person at risk of suicide and therefore recommended. These actions are recommended in the Australian Mental Health First Aid Guidelines for suicidal thoughts and behaviours and identified in our own previous expert consensus study (Mental Health First Aid Australia, 2014; Nicholas et al., 2018). Actions 8 to 10 are non-recommended actions, as stipulated in the Australian Mental Health First Aid Guidelines for suicidal thoughts and behaviours (Mental Health First Aid Australia, 2014). Respondents were then asked if, during the time they were supporting the person in distress, they found out or suspected the person was thinking about suicide. Those who said ‘yes’ where then asked if they performed the following five suicide-specific actions (yes/no): (1) asked if they had a plan for suicide – for example, a date or how they will die; (2) asked if they had a means to kill themselves – for example, pills or a weapon; (3) listened to why they wanted to die; (4) tell them how much is would hurt their friends and family if they were to kill themselves; and (5) tried to make them understand that suicide is wrong. Actions 1 to 3 are recommended, and actions 4 and 5 are non-recommended actions (Mental Health First Aid Australia, 2014; Nicholas, Rossetto, Jorm, Pirkis, & Reavley, 2018). The order of presentation of the actions was randomised.

All respondents were asked if, in the last 12 months, they had considered taking their own life (Paykel, Myers, Lindenthal, & Tanner, 1974). Those who had were asked if they had seriously considered suicide or made a plan for suicide (Paykel et al., 1974). If they said ‘yes’, they were then asked if they made a suicide attempt (Paykel et al., 1974). Any respondent who had thought about suicide in the last 12 months was asked if they had received either professional or non-professional help for their suicidal thoughts. Those who had received help from a non-professional were asked if that person had carried out the 15 helping actions described above. Respondents then rated the overall degree of helpfulness of the actions taken (1 = not at all helpful to 5 = extremely helpful).

All respondents were asked if they had ever had a job that involved providing treatment or services to a person ‘experiencing distress like John's/Jenny's?’ (yes/no), and if they had completed any suicide prevention training (yes/no). Respondents also answered sociodemographic questions relating to age, gender, marital status, language spoken at home, country of birth, identification as Aboriginal or Torres Strait Islander, highest level of education, and location of residence.

Statistical analysis

We conducted descriptive analyses and present the results as percentages with 95% confidence intervals. We conducted logistic regressions to determine if sociodemographic and exposure variables were associated with helping a person in distress and the types of actions taken. All predictors were entered simultaneously, and a significance level of p < 0.01 was used as the threshold to indicate evidence of association. All results were calculated using sample weights to account for non-response. Weights were calculated using gender, age, region, level of education and telephone status using raking (Lumley et al., 2010) and are described more fully elsewhere (Nicholas et al., 2019). These weights adjusted for under-sampling of men and older people and for over-sampling of respondents with an education of Bachelor degree or above. All analyses were carried out using STATA 15.0.

Results

Helping actions undertaken by respondents to support others

Fig. 1 shows the flow of participants through the survey questions related to helping a person close to them in distress. Each percentage is a proportion of the number of respondents who answered ‘yes’ to the previous question. The percentage of respondents who had known someone in the last 12 months experiencing a level of distress similar to the person in the vignette was not different across vignettes (directness of suicidal communication: P = 0.29; gender: P = 0.52). The majority of respondents who knew someone in distress in the last 12 months offered them support (87.6%, 95% CI 84.8–89.9). Sociodemographic and exposure characteristics of those who did and did not support the person in distress are shown in Table 1. Logistic regression (Table 2) showed that those who had been at risk of suicide in the last 12 months, compared with those who had not, had higher odds of having offered support to a person in distress (OR = 2.8; 95% CI, 1.3–5.9, p < 0.001).

Fig. 1.

Flow of respondents through questions relating to supporting someone experiencing distress similar to the person in the vignette

Note: We have not reported the small proportions of respondents who answered ‘don't know’ or who refused to answer the question, so the totals for each question do not add to 100%; percentages are a proportion of the number of people who answered ‘yes’ to the previous question.

Table 1.

Sociodemographic and exposure characteristics of population who knew someone in distress in the last 12 months and who did and did not offer them support.

Characteristics	Offered support (%)	Did not offer support (%)
Gender
Male	39.1	48.6
Female	60.9	51.4
Age group
18-30	25.9	28.8
31-44	29.8	21.5
45-59	26.3	22.4
60+	17.9	27.4
Marital status
Never married	28.8	27.6
Married/de facto	56.2	62.4
Separated/widowed/divorced	14.9	8.6
Language spoken at home
English	92.9	95.0
Language other than English	7.1	5.0
Country of Birth
Australia	82.4	82.5
Other	17.6	17.5
Aboriginal and/or Torres Strait Islander
No	95.3	95.5
Yes	4.6	4.5
Level of education
Below Bachelor degree	77.9	81.2
Bachelor degree or above	22.1	18.8
Location
Major city	56.8	53.1
Regional, rural or remote	43.2	46.9
Suicidal ideation
No	73.7	88.6
Yes	26.3	11.4
Professional experience in suicide prevention
No	76.5	79.7
Yes	23.5	20.3

Note. Weighted data are used. Percentage who refused to provide these details are not included, so totals across groups may not equal 100%.

Table 2.

Logistic regression using sociodemographic and exposure characteristics to predict likelihood of having offered assistance to a family member or close friend like the person in the vignette.

Predictor (Reference)	OR	95% CI (lb)	95% CI (ub)	p
Gender (Male)
Female	1.57	0.92	2.71	0.100
Age, years (18–30)
31-44	1.50	0.66	3.39	0.330
45-59	1.55	0.74	3.25	0.243
60+	0.84	0.41	1.71	0.630
Speak a language other than English at home (No)
Yes	1.32	0.44	3.92	0.617
Highest Education (Below Bachelor degree)
Bachelor degree or higher	1.12	0.64	1.96	0.689
Geographic location (Major city)
Regional, rural or remote	0.92	0.53	1.57	0.747
Vignette (Life events only)
Indirect communication of suicide risk	0.91	0.47	1.77	0.787
Direct communication of suicide risk	0.50	0.28	0.88	0.017
Gender of the person in the vignette (Male)
Female	0.82	0.48	1.39	0.456
Own suicidal ideation (No)
Yes	2.82	1.34	5.96	0.006
Professional experience (No)
Yes	1.04	0.60	1.81	0.904
Suicide prevention training (No)
Yes	2.23	0.41	12.08	0.352

General helping actions

Helping actions in decreasing order of frequency were as follows (non-recommended actions are marked with an asterisk*): listened to their problems without judgement (96.5%, 95% CI 94.6–97.7); asked about how they were feeling (95.3%, 95% CI 93.5–96.6); asked how you could help (91.6%, 95% CI 89.3–93.5); reminded them what they have going for them* (91.6%, 95% CI 89.1–93.5); helped make an appointment with a health professional (61.3% , 95% CI 57.4–65.2); reassured them you know exactly how they feel* (56.2%, 95% CI 52.2–60.1); asked if they had been thinking about killing themselves (48.5%, 95% CI 44.4–52.6); tried to solve their problems* (44.2%, 95% CI 40.3–48.3); went to an appointment with a professional with them (39.0%, 95% CI 35.1–43.0); and called a crisis line (19.2%, 95% CI 16.3–22.6).

Table 3 shows results of logistic regression analyses using sociodemographic and exposure variables to predict whether respondents had undertaken each of these 10 general helping actions. Those aged 60+, compared with those aged 18 to 30, had lower odds of having asked how you could help (OR = 0.32; 95% CI, 0.14–0.76; p = 0.009) and to have asked if they had been thinking about killing themselves (OR = 0.40; 95% CI, 0.24–0.67; p = 0.001). Female respondents had greater odds than male respondents of having helped make an appointment with a health professional (OR = 1.74; 95% CI, 1.23–2.48, p = 0.002). Those with a Bachelor degree or higher had lower odds than those with a lower level of education to having went to an appointment with a professional with them (OR = 0.60; 95% CI, 0.42–0.87; p = 0.002) and to have reassured them you know exactly how badly they feel (OR = 0.52; 95% CI, 0.37–0.75; p = 0.000). Those respondents who had experienced suicidal thoughts in the last 12 months, compared with those who had not, had higher odds of having told the person in distress ‘I know exactly how you feel’ (OR = 2.15; 95% CI, 1.40–3.28; p < 0.001).

Table 3.

Results of Logistic regression analysis using sociodemographic and exposure characteristics to predict whether respondents carried out specified actions.

Predictor (Comparator)	Ask how are feeling		Listened		Asked how they can help				Helped make an appointment		Went to an appointment
	OR	95% CI	OR	95% CI	OR		95% CI		OR	95% CI	OR		95% CI
Gender (Male)
Female	1.64	0.81, 3.32	1.14	0.46, 2.83	1.26		0.70, 2.27		1.74	1.23, 2.48	1.41		0.98, 2.03
Age, years (18–30)
31-44	0.45	0.12, 1.74	1.76	0.41, 7.50	0.86		0.32, 2.29		1.18	0.70, 1.98	1.31		0.77, 2.25
45-59	0.51	0.15, 1.69	1.30	0.30, 5.70	0.98		0.40, 2.37		1.04	0.63, 1.71	1.40		0.83, 2.36
60+	0.25	0.08, 0.76	0.94	0.23, 3.83	0.32		0.14, 0.76		0.97	0.58, 1.62	1.45		0.85, 2.47
Speak language other than English at home (No)
Yes	0.86	0.12, 6.02	3.91	0.47, 32.52	0.54		0.19, 1.54		1.38	0.65, 2.93	1.34		0.65, 2.74
Highest Education (Below Bachelor degree)
Bachelor degree or higher	0.67	0.28, 1.57	0.93	0.36, 2.41	0.80		0.45, 1.42		0.87	0.60, 1.27	0.60		0.42, 0.87
Geographic location (Major city)
Regional, rural or remote	0.79	0.39, 1.60	0.60	0.25, 1.44	1.12		0.64, 1.97		1.30	0.92, 1.84	1.31		0.92, 1.86
Own suicidal ideation (No)
Yes	1.08	0.44, 2.67	2.06	0.60, 7.04	0.91		0.46, 1.82		0.99	0.66, 1.49	1.33		0.89, 2.00
Professional experience (No)
Yes	1.14	0.47, 2.80	2.42	0.84, 6.95	1.09		0.55, 2.14		1.39	0.92, 2.10	1.03		0.68, 1.55
Suicide prevention training (No)a
Yes	2.48	0.47, 12.95	-a	-a	2.64		0.51, 13.50		2.44	0.93, 6.41	1.09		0.47, 2.52
	Called a crisis line		Reminded them what have going for them			Tried to solve their problems		Told them you know exactly how they feel				Asked if they had been thinking about suicide
Predictor (Comparator)	OR	95% CI	OR	95% CI		OR	95% CI	OR		95% CI		OR	95% CI
Gender (Male)
Female	1.50	0.95, 2.38	1.71	0.95, 3.06		1.07	0.75, 1.53	0.92		0.64, 1.33		1.00	0.71, 1.42
Age, years (18–30)
31-44	1.56	0.77, 3.18	0.68	0.25, 1.83		0.73	0.44, 1.22	1.10		0.65, 1.87		0.68	0.41, 1.13
45-59	1.71	0.85, 3.44	0.83	0.31, 2.22		0.54	0.33, 0.88	1.61		0.98, 2.66		0.84	0.52, 1.38
60+	1.39	0.67, 2.87	0.40	0.16, 1.03		0.55	0.33, 0.92	1.54		0.92, 2.58		0.40	0.24, 0.67
Speak language other than English at home (No)
Yes	0.68	0.26, 1.79	0.56	0.19, 1.59		2.33	1.09, 4.99	2.06		0.95, 4.45		0.99	0.48, 2.02
Highest Education (Below Bachelor degree)
Bachelor degree or higher	0.57	0.36, 0.92	0.87	0.46, 1.67		0.85	0.58, 1.23	0.52		0.37, 0.75		0.71	0.49, 1.02
Geographic location (Major city)
Regional, rural or remote	0.93	0.60, 1.45	2.21	1.19, 4.09		1.48	1.05, 2.09	1.11		0.79, 1.58		1.06	0.75, 1.51
Own suicidal ideation (No)
Yes	1.04	0.62, 1.74	1.64	0.76, 3.57		1.03	0.69, 1.54	2.14		1.40, 3.28		1.56	1.05, 2.34
Professional experience (No)
Yes	1.34	0.79, 2.26	1.37	0.70, 2.67		0.69	0.46, 1.04	0.66		0.44, 0.99		1.38	0.92, 2.08
Suicide prevention training (No)a
Yes	0.92	0.37, 2.27	6.38	0.78, 52.23		0.94	0.35, 2.48	0.92		0.38, 2.20		2.68	1.09, 6.62

Note: significant predictors are shown in bold type: P < 0.01 ^aThese cells are empty as no respondent with suicide prevention training answered ‘no’.

Note: significant predictors are shown in bold type: P < 0.01.

Suicide-specific actions

Respondents who found out or suspected that the person they were supporting was thinking about suicide (63.4%, 95%CI 59.4–67.1) were asked to indicate of they had undertaken five suicide-specific helping actions. The frequency of these actions in decreasing order was as follows (non-recommended actions are marked with an asterisk*): listened to why they wanted to die (91.8%; 95% CI 89.1–93.9); told them how much it would hurt their family and friends if they were to kill themselves* (78.0%, 95% CI 73.7–81.8); tried to make them understand that suicide is wrong* (70.4%, 95% CI 65.9–74.5); asked if they had a means to kill themselves (43.0%; 95% CI 38.0–48.1); and asked if they had a plan for suicide (39.8%; 95% CI 35.0–44.9).

Table 4 shows the results of logistic regression analyses using sociodemographic and exposure variables to predict whether respondents had undertaken five suicide-specific actions. Those with a Bachelor degree or higher, compared with those with a lower level of education, had lower odds of having asked if they had a plan for suicide (OR = 0.48, 95% CI 0.29–0.78; p = 0.003); having told them how much it would hurt their family and friends if they were to kill themselves (OR = 0.35; 95% CI 0.21–0.59; p = 0.000); and having tried to make them understand that suicide is wrong (OR = 0.31; 95% CI 0.19–0.50; p = 0.004). Having undertaken suicide prevention training was associated with higher odds of having asked if they had a plan for suicide (OR = 13.44; 95% CI 4.23–42.72, p = 0.000) and if they had a means to kill themselves (OR = 7.39; 95% CI, 2.34–22.27, p = 0.001). Those aged 60+ had higher odds than those aged 18 to 30 of having listened to why they wanted to die (OR = 0.22; 95% CI 0.08–0.62; p = 0.000) and of having tried to make them understand that suicide is wrong (OR = 0.30; 95% CI 0.14–0.64; p = 0.002). Those aged 45 to 59 had lower odds than those aged 18 to 30 of having tried to make them understand that suicide is wrong (OR = 0.38; 95% CI 0.19–0.75; p = 0.000). Those who spoke a language other than English at home had higher odds then those who spoke English at home of having tried to make them understand that suicide is wrong (OR = 5.22; 95% CI 0.19–0.50; p = 0.003).

Table 4.

Results of Logistic regression analysis using sociodemographic and exposure characteristics to predict whether respondents carried out suicide-specified actions when the person in distress had been thinking about suicide.

Predictor (Comparator)	Asked about a plan		Asked about means		Listened to why want to die		Tell them it would hurt their friends and family		Try to convince them suicide is wrong
	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
Gender (Male)
Female	1.81	1.12, 2.89	1.55	0.97, 2.47	1.38	0.65, 2.90	1.14	0.67, 1.92	0.69	0.43, 1.11
Age, years (18–30)
31-44	0.81	0.43, 1.55	0.57	0.30, 1.11	0.55	0.14, 1.99	0.54	0.25, 1.18	0.40	0.19, 0.85
45-59	0.79	0.41, 1.51	0.85	0.44, 1.62	0.49	0.18, 1.54	0.54	0.26, 1.13	0.38	0.19, 0.75
60+	0.55	0.28, 1.08	0.53	0.26, 1.05	0.22	0.08, 0.62	0.40	0.19, 0.83	0.30	0.14, 0.64
Speak language other than English at home (No)
Yes	2.32	0.95, 5.68	3.27	1.28, 8.35	2.28	0.51, 10.22	1.45	0.40, 5.19	5.22	1.72, 15.87
Highest Education (Below Bachelor degree)
Bachelor degree or higher	0.48	0.29, 0.78	0.59	0.37, 0.95	0.39	0.18, 0.82	0.35	0.21, 0.59	0.31	0.19, 0.50
Geographic location (Major city)
Regional, rural or remote	0.98	0.62, 1.54	1.17	0.74, 1.83	1.52	0.77, 3.00	1.40	0.84, 2.32	1.39	0.91, 2.14
Own suicidal ideation (No)
Yes	1.53	0.92, 2.54	1.80	1.09, 2.96	0.97	0.41, 2.28	0.55	0.33, 0.94	0.95	0.55, 1.62
Professional experience (No)
Yes	1.16	0.69, 1.97	1.22	0.73, 2.05	1.01	0.47, 2.19	0.85	0.49, 1.47	1.14	0.67, 1.95
Suicide prevention training (No)a
Yes	13.44	4.23, 42.72	7.39	2.34, 22.27	7.64	0.91, 63.84	1.68	0.59, 4.75	0.46	0.16, 1.38

Note: significant predictors are shown in bold type: P < 0.0.

Helping actions respondents received from a non-professional when at risk of suicide

Sources of help

Fig. 2 shows the flow of respondents through the survey questions about receiving help when at risk of suicide. Percentages are a proportion of the entire population, drawn from the total weighted sample. Of those who had experienced suicidal thoughts in the last 12 months (15.9%, 95% CI 14.2–17.7), more than half (56.4%, 95% CI 50.5–62.2) had received help from a non-professional for their suicidal thoughts.

Fig. 2.

Flow of respondents through questions relating to receiving help for suicidal thoughts

Note: We have not reported the small proportions of respondents who answered ‘don't know’ or who refused to answer the question, so the totals for each.

Helping actions received from a non-professional

Fig. 3 shows the percentage of people who had received the 15 specified helping actions from a non-professional. Non-recommended actions are shown with an asterisk*. The most commonly received helping actions were that the helper asked about how they are feeling (96.8%; 95% CI, 92.7–98.7), listened to their problems without judgement (94.5%; 95% CI, 89.3–97.2), reminded them what they have going for them* (92.4%, 95% CI, 86.8–95.7), and asked how [they] could help (91.9%, 95% CI, 86.1–95.3). The least common action was that the helper called a crisis line (12.3%, 95% CI, 95% CI 7.7–19.2), followed by having the helper ask if they had a plan for suicide (28.4%; 95% CI, 21.4–36.6), and having asked if they had a means to kill themselves (32.1%, 95% CI, 24.7–40.6).

Fig. 3.

Percentage (95% CI) of population who had experienced suicidal thoughts in the last 12 months who received 15 specified types of helping action from a non-professional.

Similarities and differences between helping actions undertaken to support another person at risk of suicide and helping actions received when at risk of suicide

The proportion of respondents who reported having both given support to someone in severe distress and having received support for their suicidal thoughts from a non-professional was 56.0% (95% CI, 50.1–61.8; see Fig. 2). For the whole sample, the proportion of respondents reporting having received each helping action were similar to those reported as having been undertaken by respondents who had helped another person. The greatest discrepancies were for the suicide-specific actions tried to make them understand that suicide is wrong (helped another = 70.4% vs helped me = 44.0%), tell them it would hurt their friends and family if they were to kill themselves (helped another = 78.0% vs helped me = 59.9%), and listened to why they wanted to die (helped another = 91.9% vs helped me = 75.4%).

Helpfulness of actions by a non-professional

Almost 90% of respondents indicated they found the actions taken by a non-professional when they were considering suicide helpful (34.4%, 95% CI 26.9–42.5) or extremely helpful (55.3%, 46.9–63.5). The remaining 10% found the help provided neither helpful nor unhelpful (7.4%, 95% CI 3.9–13.7), not helpful (1.5%, 95% CI 0.4–5.9), or not at all helpful (1.5%, 95% CI, 0.4–6.0).

Discussion

Our telephone survey of 3000 Australian adults showed that nearly 40% of people had known someone close to them experiencing severe distress in the last 12 months. Almost 90% did something to support that person. Supportive actions consistent with best practice were most commonly undertaken (e.g., listened to their problems without judgement), although some non-recommended actions (e.g., reminded the person what they have they going for them) were very common. Of those who had assisted a person in distress, 63.4% found out or suspected that person was thinking about suicide. The most common suicide-specific action was listening to why the person wanted to die and the least common were asking about plans and means for suicide. Those who had undertaken suicide prevention training had had far greater odds of asking about a plan and means for suicide. Sixteen percent of respondents had thought about suicide themselves in the last 12 months, 6% had seriously considered suicide, and 1% had attempted suicide. Of those who had thought about suicide, half received help from a health professional and 56% from a non-professional. Responses regarding types of helping actions received from a non-professional were very similar to those actions reported by those who had helped others, and the actions from a non-professional were rated as being helpful.

The relatively high rate of suicidal thoughts among the population and the high rate of help-seeking from informal sources emphasise the likelihood of Australians having someone close to them being at risk of suicide at some time in their lives. The finding that informal supports were used as commonly as formal supports echoes previous research (Houston, Hawton, & Shepperd, 2001; Rich, Fowler, Fogarty, & Young, 1988). It is therefore imperative that community members know how to identify suicide risk and to take appropriate actions to keep the person at risk safe, and to encourage them to seek professional help.

This study was unique in including the perspective of both those who had supported others at risk of suicide and those who had received support from others. Helping behaviours reported by these two groups were very similar, providing evidence for the validity of the self-report methodology. There was also significant overlap between those who had supported others and those who had received support (more than half reported both), and the similarities between helping behaviours given and received also suggests that this group may undertake actions when supporting others that they have found helpful themselves. Our findings are also in line with previous studies of intended and actual helping behaviour in demonstrating that Australian adults are most likely to carry out general supportive behaviours like asking how they can help, listening and asking how the person is feeling (Jorm et al., 2005; Rossetto et al., 2014, 2016). These actions are consistent with current best practice guidelines for assisting a person at risk of suicide (Mental Health First Aid Australia, 2014). In addition, most respondents who received support from a non-professional for suicidal thoughts rated it as helpful. There is also previous research to suggest that help provided by family and friends can indeed be helpful to many people when experiencing mental health difficulties (Reavley, Morgan, & Jorm, 2017; Riedel-Heller, Angermeyer, & Matschinger, 2001).

The findings of our study are positive in that Australian adults were largely willing to help others at risk of suicide, and often undertook appropriate helping actions. However, we also found that a significant proportion of Australian adults took no steps to ensure that the person in distress got professional help (Jorm et al., 2005; Rossetto et al., 2014). They were also unlikely to ask about plans and means for suicide, even when they were aware the person was experiencing suicidal thoughts (Jorm et al., 2005; Rossetto et al., 2014). Encouraging professional help, risk assessment, and the appropriate follow-up steps, are central to best practice in suicide prevention (Mental Health First Aid Australia, 2014). These actions therefore need to be further encouraged. Conversely, several behaviours outlined in the ‘what not to do’ section of the Mental Health First Aid Guidelines for Suicidal Thoughts and Behaviours (Mental Health First Aid Australia, 2014) were commonly reported by respondents to our study. Reminding the person ‘what they've got going for them’ was reported by more than 90% of respondent groups as an action both undertaken and received. Telling the person that their suicide would hurt their friends and family, and trying to convince them that suicide is wrong were also common. These actions need to be discouraged.

Implications for practice

Our findings, and those of others, show that gatekeeper suicide prevention training, such as Mental Health First Aid, has a significant effect on the likelihood of asking questions to assess suicide risk (Cimini et al., 2014; Jones, Walker, Miles, De Silva, & Zimitat, 2015; Tompkins & Witt, 2009). Such training could therefore be encouraged among more Australian community members, and particularly among those who might be natural ‘gatekeepers’. Such gatekeepers include community members who are not health or mental health professionals but who hold some position of authority who are likely to observe or be sought out by people in extreme distress. These gatekeepers could include teachers, coaches, managers, human resources professionals, volunteers and others.

In addition to specific training of targeted groups, our study findings indicate there is potential to improve the skills of Australian adults in general to take appropriate suicide prevention actions towards their family members and friends. In particular, community members need to be educated to ask direct questions about suicide risk and to encourage others in distress to seek professional help. There may also be a place for discouraging certain behaviours that oppose best practice in suicide prevention. There is substantial evidence in other areas of public health, such as road safety, to suggest that media-based public education campaigns can modify behaviour in desirable ways at a population level (Wakefield, Loken, & Hornik, 2010). It is therefore reasonable to propose that a media campaign aimed at improving suicide prevention behaviours among Australian adults might have similar success. Based on this premise, the findings of our research have been used to develop an Australian public education campaign aimed at increasing the likelihood that Australian adults will identify when family members and friends are at risk of suicide and to improve their skills in providing them with appropriate support. The #YouCanTalk campaign (https://www.lifeinmindaustralia.com.au/youcantalk/) was launched in 2018 through a collaboration between several of Australia's major not-for-profit mental health organisations (beyondblue, Black Dog Institute, Everymind, headspace, Lifeline, ReachOut and R U OK?). Each partner organisation uses its social media presence to promote resources on their websites that provide information about what to do and what not to do to support someone close to them who they think might be considering suicide, in line with our research findings.

Limitations

It is possible that we oversampled people with a specific interest in the survey topic, which we introduced as ‘helping people in severe distress’. Respondents' interest could have arisen from personal experience as someone who had helped another in distress or because they had received help from others at a time when they were particularly distressed themselves. Therefore, our results may overestimate the proportion of the population who had recognised a family member or friend in distress in the last 12 months and who had offered them support. Our results might also overestimate the rate of suicidal ideation among the general population. The proportion of the population in our study who had seriously considered suicide (6%) is substantially higher than in a previous Australian population study (Slade et al., 2009). There are some differences in the wording of the suicidal ideation questions asked that might account for this variation. But it also cannot be ruled out that the rate of suicidal ideation among Australian adults has actually increased, in line with the increased suicide rate (Australian Bureau of Statistics, 2018).

We also had a relatively low response rate (12%). Low response rates to telephone surveys are now commonly reported (Hu, Balluz, Battaglia, & Frankel, 2011; Wallander, Tikkanen, Mannheimer, Östergren, & Plantin, 2015), but could also increase the likelihood of sampling bias. However, our use of weighted data in our sample corrected some sampling biases in that we ensured that the data was representative of the population in terms of age, gender, level of education, geographic location and telephone status.

We relied on self-report of helping behaviours undertaken and received, and therefore the responses are prone to recall and social desirability biases. However, we attempted to minimise these biases by asking respondents if they had undertaken or received 15 specific helping actions, rather than relying on purely open-ended responses, as in previous studies (Rossetto et al., 2016; Shand et al., 2015). To counter social desirability biases, we also asked if respondents had undertaken or received five non-recommended actions, and a number of these were commonly indicated. This suggests that social desirability bias was minimised, in part because some of these actions that are not recommended may seem to many to be helpful (e.g., trying to convince the person that suicide is wrong). Asking respondents to answer ‘yes or no’ to whether they undertook the 15 helping actions has the potential to elicit a higher number of ‘yes’ responses. However, before asking respondents to respond to these yes/no questions, we asked them an open-ended question ‘what did you do [to support the person at risk]?‘. The results from this open-ended question closely replicate the findings from the ‘yes/no’ responses to the 15 helping actions (unpublished report available from the authors upon request).

Conclusion

A third of Australians will have someone close to them die by suicide (Maple et al., 2016). Those people at risk who receive support for their suicidal thoughts are most likely to receive that support from a non-professional. Many will not receive any professional support (Houston et al., 2001; Rich et al., 1988). Therefore, it is a national imperative that Australians develop the skills to know when a family member or friend might be at risk of suicide and to take action to ensure that person stays safe and gets the professional support they need. While Australians report being confident and willing to help a person close to them at risk of suicide (Nicholas et al., 2019), this study identified two significant deficits in the helping behaviours of Australian adults when compared with current suicide prevention best practice (Mental Health First Aid Australia, 2014). First, assisting the person to get professional help and asking questions to assess suicide risk were relatively uncommon. Second, some non-recommended actions, such as telling the person at risk ‘what they have going for them’ were very common. There is evidence to suggest that existing gatekeeper training, such as Mental Health First Aid, can have significant positive effects on such behaviours (Cimini et al., 2014; Jones et al., 2015; Tompkins & Witt, 2009; Wakefield et al., 2010). These behavioural deficits also provide specific targets for community-level suicide prevention education and public health campaign messages. Equipping Australian community members to intervene with family members and friends at risk of suicide is a crucial component of our national suite of suicide prevention strategies.

Conflicts of interest

Declarations of interest: none.

Ethics approval

The study was approved by the University of Melbourne Human Research Ethics Committee (Ethics ID: 1648060).