The effects of a celebrity suicide on suicide rates in Hong Kong.

BACKGROUND: Deaths of celebrities, especially by suicide, can be followed by an increase in population suicide rates, particularly where there is extensive media reporting. We have examined the impact on suicides following the death of a famous Hong Kong pop singer whose death from suicide by jumping from a height, occurred on 1st April 2003, and resulted in extensive and often dramatic media coverage.
METHODS: Data on suicides were obtained from the Hong Kong Census and Statistics Department and the Coroner's Court. The numbers of suicides in 2003 before and after the death of celebrity were compared to the same period in 1998-2002. The case files and suicide notes of people who died by suicide in 2003 were also studied qualitatively.
RESULTS: There was a significant increase in suicides following the celebrity death, compared with the average over the preceding three months as well as the corresponding monthly average during 1998-2002. It was particularly marked in a subgroup comprising males, aged 25-39 years, many of whom died by jumping. The name of the celebrity was often mentioned in case files and suicide notes. LIMITATIONS: The statistical results in showing the excess of suicides were based on aggregated data only.
CONCLUSIONS: This study provides further confirmation of the potential harmful consequences of sensational and excessive reporting of celebrity deaths.

Introduction

Evidence has been found of an increase in overall suicide rates following reporting of celebrity suicides (Jobes et al., 1996, Phillips, 1974, Stack, 1987, Takahashi, 1998, Tousignant et al., 2005) and non-suicide deaths (Hawton et al., 2000). It appears that media reports on celebrity suicides are more likely to have such an effect than those of non-celebrity suicides (Stack, 2000, Wasserman, 1984). Furthermore, the impact of suicides of entertainment celebrities appears to have a larger impact than that of other types of celebrity (Stack, 1987, Wasserman, 1984).

Moreover, young people seem to be more vulnerable to media influences (Hawton and Williams, 2005, Zahl and Hawton, 2004). People come to learn the details of methods of suicide through media reporting of them, such as asphyxiation from the smoke of burning charcoal, especially in places like Hong Kong and Taiwan (Chan et al., 2005). Suicide deaths by charcoal burning have increased from 3% in 1997 to 24% in 2002 of the total suicide death (Yip et al., 2005). The relationship between media reporting of non-fictional suicides and subsequent suicides can reasonably be regarded as causal (Hawton and Williams, 2005, p. 303; Pirkis and Blood, 2001, p. 150).

However, few studies have investigated the effects of celebrity suicides and the media reporting of them in Asia (Takahashi, 1998, Au et al., 2004), despite extensive mass media exposure and traditional idolization of celebrities within the region.

Leslie Cheung, a Hong Kong pop singer and actor who became famous during in the 1980s and 1990s, suffered from depression and committed suicide by jumping from a height on 1st April 2003 in Hong Kong. His funeral was held on 8th April 2003 and there was extensive media coverage of his death. According to an electronic news clipping database, which covers all of Hong Kong's newspapers, there were 1243 news articles with his English name or Chinese name during the period between the 2nd April 2003 and 9th April 2003. This compared with only 12 news articles in the preceding week from 25th March to 1st April (WiserNews, 2003). Of the 1243 articles, 565 (46%) of them mentioned his full name or his nickname in the headline and 35 (3%) were on the front page of the newspapers. On six of the eight days between the 2nd and 9th April (except 4th April and 5th April) at least one newspaper reported news of Cheung's death on the front page. Most of the newspaper articles were published in the first 16 days after his death (1701 articles). During the subsequent 16 days there were 315 articles, and during the next 16 days only 102 articles.

The specific aims of this study were to answer the following questions: (i) Was there a significant increase in suicides following the death of Cheung? (ii) Was there a particularly vulnerable group in the community more likely to be affected by news of Cheung's suicide? and (iii) Was there any difference in the profile of vulnerable people compared to those reported from similar studies in western countries?

Methods

Mortality data (including suicides) in Hong Kong, SAR during the period 1998 to 2003 were made available by the Census and Statistics Department and the Coroner's Court. All suicide deaths, coded in the range of E950–E959 were classified and recorded according to the ICD-9 standard (World Health Organization, 1978). With the consent of the Coroner's Court we examined the case files of people who died by suicide in 2003. The files included information on socio-demographic variables, psychiatric illness, police investigation files and suicide notes (where available). For comparison, we selected the case files of people who died by suicide during the first six months of each year between 1998 and 2002.

The suicide rate in Hong Kong has increased markedly in recent years, rising from 13.2 per 100,000 in 1998 to 18.6 per 100,000 in 2003 (Center for Suicide Research and Prevention, 2005). Therefore to allow a robust comparison of the monthly distribution of suicides against the background of this increasing trend, we detrended the monthly suicide numbers by subtracting the mean of the first six months in the corresponding year from the monthly figures. (Kaiser and Maravall, 2001, Harvey and Proietti, 2005). The number of suicides in April 2003 was compared with the average for the first six months of the year, the average for the immediately preceding three months (January–March 2003), and with the average for April in the preceding 5 years (1998–2002). A chi-square test was used to examine the significance of differences in distribution. In addition, we also examined the Coroner's death reports as well suicide notes for suicide deaths in 2003, and identified those which mentioned Cheung's death.

Results

Monthly distribution of suicide deaths

Fig. 1 shows the mean monthly raw and detrended numbers of suicide deaths during January to June in 2003 and during 1998–2002. There was a clear peak in suicides in April 2003, when there were 134 deaths. This was approximately 35% (p  < 0.01) more than the monthly average for January–March 2003 (99 cases) and 56% (p  < 0.01) more than the April average for the preceding 5 years (85 cases in April 1998–2002).

Fig. 1

Average raw and detrended monthly numbers of suicides 1998–2002 and 2003 in both genders.

The detrended number of suicides in April 2003 was 24.5 cases greater than the average monthly number of suicides during the first six months of April 2003 (109.5 cases). The detrended suicide figure for April 2003 was about 6.2 times higher (i.e. 24.5 cases vs. 3.9 cases) in comparison with the number of suicides during April in the previous 5 years.

The excess number of suicide deaths in April 2003 was clearly evident in males (Fig. 2 ). There were 92 suicides in males in the month, compared with the monthly average of 65 suicides during the prior three-month period, a 64% increase (p  < 0.01). Also, the detrended number of suicides in April 2003 was 17.3 cases more than the half-year monthly average in 2003, and 4.3 times higher than that of the April average during the previous 5 years.

Fig. 2

Average raw and detrended monthly numbers of suicides 1998–2002 and 2003 in males.

Table 1a, Table 1b suggest that the increase in suicides in April 2003 was most marked for suicide by jumping from a height and among males aged 25–39 years. In that month, 42 male suicide deaths involved jumping from a height, which was 11 more than monthly average for the first six months of 2003. It was also 72% greater than the monthly average for the previous three months of 2003 (24.3 cases) (p  < 0.01). A significant but smaller increment significance was also found in May 2003. On average about 6 males in this age group committed suicide by this method each month during the months of January to March 2003 and there was about 10 cases in this subgroup in April each year during the period 1998–2002, whereas 15 males used this method in April 2003. Moreover, the corresponding detrended figure was about 3.5 times higher than that of the average detrended figure for April in the preceding 5 years.

Table 1a

Average raw and detrended monthly numbers of suicides 1998–2002 and 2003 in males by jumping and non-jumping methods

1998–02 vs 2003	Jan	Feb	Mar	Apr	May	Jun
1998–02 avg raw jumping	22.0	21.4	23.4	25.2	21.2	19.4
1998–02 avg raw non-jumping	31.2	26.8	30.0	30.8	33.2	27.6
2003 raw jumping	23	20	30	42	40	31
2003 raw non-jumping	38	44	41	50	41	48
1998–02 avg detrended jumping	− 0.1	− 0.7	1.3	3.1	− 0.9	− 2.7
1998–02 avg detrended non-jumping	1.3	− 3.1	0.1	0.9	3.3	− 2.3
2003 detrended jumping	− 8.0	− 11.0	− 1.0	11.0	9.0	0.0
2003 detrended non-jumping	− 5.7	0.3	− 2.7	6.3	− 2.7	4.3

Table 1b

Average raw and detrended monthly numbers of suicides 1998–2002 and 2003 in males aged 25–39 years by jumping and non-jumping methods

1998–02 vs 2003	Jan	Feb	Mar	Apr	May	Jun
1998–02 avg raw jumping	6.2	6.4	5.2	9.4	7.8	4.0
1998–02 avg raw non-jumping	10.4	7.0	8.8	10.2	10.4	7.4
2003 raw jumping	10	3	5	15	9	10
2003 raw non-jumping	12	11	17	11	12	12
1998–02 avg detrended jumping	− 0.3	− 0.1	− 1.3	2.9	1.3	− 2.5
1998–02 avg detrended non-jumping	1.4	− 2.0	− 0.2	1.2	1.4	− 1.6
2003 detrended jumping	1.3	− 5.7	− 3.7	6.3	0.3	1.3
2003 detrended non-jumping	− 0.5	− 1.5	4.5	− 1.5	− 0.5	− 0.5

While there was a small rise in suicides by females in April 2003 (Fig. 3 ), the number of suicides (N  = 42) was not significantly different from the monthly average for the preceding three months in 2003. There were 7.2 more suicide cases than the average detrended monthly figure for the April average in the previous five years. There was little change in the number of suicides by method in females during the study period (Table 2 ).

Fig. 3

Average in raw and detrended monthly numbers of suicides 1998–2002 and 2003 in females.

Table 2

Average raw and detrended monthly numbers of suicides 1998–2002 and 2003 in females by jumping and non-jumping methods

1998–02 vs 2003	Jan	Feb	Mar	Apr	May	Jun
1998–02 avg raw jumping	12.8	12.8	16.8	13.6	18.2	12.4
1998–02 avg raw non-jumping	16.6	11.8	19.2	16.2	15.8	12.8
2003 Raw Jumping	20	15	14	21	13	15
2003 raw non-jumping	22	16	14	21	23	15
1998–02 avg detrended jumping	− 1.6	− 1.6	2.4	− 0.8	3.8	− 2.0
1998–02 avg detrended non-jumping	1.2	− 3.6	3.8	0.8	0.4	− 2.6
2003 detrended jumping	3.7	− 1.3	− 2.3	4.7	− 3.3	− 1.3
2003 detrended non-jumping	3.5	− 2.5	− 4.5	2.5	4.5	− 3.5

Information from Coroner's death investigation files and suicide notes

Any unnatural death (including suicide) in Hong Kong is investigated by Coroners, who are normally assisted by a police investigation of the circumstance and cause of the death (Coroner Court, 1998). The police reports and suicide notes provide another source of information about factors contributing to suicide. There were 13 cases in 2003 in which the death of Cheung was mentioned in either a suicide note (5 cases) or records of the police investigation (8 cases). In one case there was mention in both sources. The cases occurred between April and October 2003. The contents of the deaths notes are summarized in Appendix A. Among these 13 cases, nine were male and six of these were aged 25–39 years. Four individuals expressed their suicidal intentions implicitly or explicitly by mentioning their desire either “to accompany Cheung” or “to follow his way to exit” (Cases 6, 10, 11 and 12). The remaining three individuals included people with life problems (Cases 8 and 9) and general distress (Case 13). It is unusual for a celebrity to be mentioned in a suicide note and only about 20% of suicides in Hong Kong are accompanied by suicide notes. The content of these notes commonly states the victim's specific instructions, including funeral arrangement, expresses affection for family members and loved ones, and gives details of the individual's own personal difficulties, health and/or financial problems (Ho et al., 1998).

Discussion

This study showed an unusual rise in suicide cases in April 2003 in males, aged 25–39, who committed suicide by jumping from a height following the suicide death by jumping of a Hong Kong pop star. However, the overall effect on suicides appears to have been relatively short lived. Other research on celebrity suicides or deaths has also shown relatively short-term effects on suicides, usually of between two and four weeks (Hawton et al., 2000, Phillips, 1974, Stack, 1987). It is important to note that there was no significant decrease in suicides in May or June 2003. This is consistent with previous studies which have rejected the hypothesis that the imitative effect caused suicides to occur earlier than would otherwise have been the case (Hawton et al., 2000, Phillips, 1974). However, the death investigation files from the Coroner's Court showed that some further individuals who died by suicide between May and October 2003 had also mentioned the name of Cheung in suicide notes. This suggests the possibility that celebrity suicide could have a longer-term effect, even though it might be insufficient to have a significant effect on the aggregated monthly suicide rate.

Studies of the effects of celebrity suicides or other death have shown specific effects on subsequent suicides in terms of gender and age (Hawton et al., 2000, Hawton and Williams, 2005, Schmidtke and Häfner, 1988). Cheung died at the age of 46 years. According to his fan club website (http://lesliecheung.cc), “A turning point in Cheung's career” was the release of his album “Monica” in 1984, when he was 28 years old. The cohort of teenagers who were more likely to be keen on pop music in 1984 and who were probably more able to identify with the behaviour of Cheung would have been in the 25–39 year age group in 2003.

The results of this study suggest not only a modelling effect, based on the characteristics of the model, but also an imitation of the actual suicide method. The effect of the suicide of Gaetan Girouard in Quebec was similar (Tousignant et al., 2005). Portrayal of a suicide method on television has been shown to be associated with an increase in suicidal behaviour by that specific method (Hawton et al., 1999, Shoval et al., 2005). The major increase in suicide deaths in Hong Kong was among males aged 25–39 years. This suggests that the characteristics of the model were crucial factors in the overall imitative effect. In western countries media effects have been found in both genders and in the very young (Schmidtke and Schaller, 2000).

Three other factors may have contributed to a rise in suicides in April 2003, namely high unemployment, the Severe Acute Respiratory Syndrome (SARS) epidemic, and a seasonal effect. If the high unemployment rate contributed to the rise in suicides the effect should have been seen not only in April but for an extended period. Also, the months of May–July period and not April had the highest unemployment rate (8.5%) in Hong Kong. Furthermore, it should have affected the community as a whole rather than a particular subgroup, and a particular method of suicide. Though the outbreak of SARS hits Hong Kong badly, there was no marked increase in suicides during May and June even though the community was still very much in the midst of SARS throughout the territory. In addition there was no increase in suicides during the SARS crisis in Singapore, which was also severely affected in 2003 (Chia, in preparation). Finally, research in Hong Kong has shown no evidence of significant seasonal variation in suicides in either gender and certainly no indication of an April peak (Yip and Yang, 2004).

While it is still not possible to be absolutely certain that the rise in suicide deaths in April 2003 was mainly due to media reporting of Cheung's death, the evidence strongly suggests that it was, especially because of the very marked increase in suicides in young adult males, the excess of deaths by jumping, and the evidence from suicide notes and police files. This conclusion is consistent with research from other countries which has shown that the death of a celebrity can ignite a contagious effect within a community (Hawton et al., 2000, Takahashi, 1998, Tousignant et al., 2005). For this reason, the World Health Organization has listed “toning down reports in the media” as one of its six key strategies on suicide prevention (World Health Organization, 1998). Improving media reporting and portrayal of suicidal behaviour is also one of the major initiatives included in the national suicide prevention strategies of several countries (Commonwealth of Australia, 2000, Department of Health and Human Services, 2001, Department of Health, 2002). Media professionals should be well informed about the potential risks associated with reporting celebrity suicides. Several media guidelines are available to assist media professionals on how to report suicide news in a responsible manner (American Foundation for Suicide Prevention, 2002, Center for Suicide Research and Prevention, 2004, Samaritans, 2002, World Health Organization, 2000). However, a systematic review of suicide prevention strategies has highlighted the need for evaluation of the effectiveness of such guidelines (Mann et al., 2005).

Limitations

The results of this study are largely based on aggregated statistical data. The level of exposure of individuals to the news of Cheung's death was not examined. We cannot be certain of the mechanisms involved in the increase in suicides following Cheung's death. It might have been due to bereavement effects of his death rather than a modeling effect of celebrity suicide. However, our analysis of suicide notes and police reports suggests that copycat suicidal behaviour is likely to have been the main mechanism.

Case no.	Month	M/F	Age	Suicide method	Explanations in suicide notes	Explanations in case notes
1	Apr	M	25–39	CO poisoning		“Have you heard that Cheung jumped off a building?” and “Yes, even a celebrity like Cheung can choose to die.” he said.
2	Apr	M	25–39	Jumping		“Cheung is so rich and has reputation and social status. But still he chooses to die.” he said.
3	Apr	M	40–59	Jumping		“A rich man like Cheung chooses to die when he is depressed, so a poor man like me can't live any longer.” he said.
4	Apr	F	40–59	CO poisoning	“A rich man like Cheung is so unhappy and chooses to die.”
5	Apr	M	40–59	CO poisoning	“A rich and depressed man like Cheung commits suicide. Why not a poor and depressed person like me?”	He was unhappy over his idol's death. He said he was depressed like Cheung and felt so uncomfortable and painful.
6	May	F	40–59	Hanging	“I think life is so meaningless and Cheung's act is so brave. Perhaps it is the way of myfinal exit …. ”
7	Jun	F	40–59	CO poisoning		She was discharged from hospital and was disturbed by SARS and Cheung's suicide. She said even a rich man like Cheung chose to die.
8	Jun	M	25–39	Hanging		The deceased mentioned the death of Cheung in his letter.
9	Jul	M	40–59	CO poisoning		The deceased had kept some newspaper clippings about Cheung's suicide.
10	Sep	M	40–59	Jumping	“I'll go to another world. Cheung, I will be there.”
11	Sep	M	25–39	Jumping		He had mentioned that he would choose the same way as Cheung.
12	Oct	M	25–39	Hanging	“Good bye, everyone. I am going to accompany Cheung. Don't be unhappy for me.”
13	Oct	F	25–39	CO poisoning		She was Cheung's fan. From the beginning of April she was sleepless and crying for no apparent reason.