Parent-reported suicidal behavior and correlates among adolescents in China

1. Introduction 

Suicide represents a major worldwide social and public health problem with about one million deaths and 10–20 million attempters in the world each year (National Center for Injury Prevention and Control, 2002). China has by far the world's largest number of reported suicides: about 300,000 each year, comprising 42% of all suicides worldwide and 56% of all suicide in women (Lee and Kleinman, 2000). A recent national estimation reported an annual suicide rate of 23 per 100,000; two times higher than the average global rate (Phillips et al., 2002a). Among young adults aged 15–34 years, suicide accounts for 19% of all deaths and is the leading cause of death. Unlike most other countries, suicide rates in women are much higher than in men, with 25% higher than that for men (Phillips et al., 2002a). Psychological autopsy studies from different parts of the world show that over 90% of suicide victims have current mental disorders (Hawton and van Heeringen, 2000). However, only about half of suicide victims in China are known to be associated with psychiatric disorders (Phillips et al., 2002b, Zhang et al., 2004). Despite high suicide rates and unique epidemiological characteristics of suicide in China and non-fatal suicidal behaviors being a major risk factor for completed suicide in the future, relatively little is known about epidemiological characteristics of non-fatal suicidal behavior and its risk and protective factors in Chinese youths.

Recently, our group has reported suicidal behavior and risk factors in a sample of 1362 adolescents in a rural prefecture of China (Liu, 2004, Liu and Tein, 2005, Liu et al., 2005a, Liu et al., 2005b). Results of the study indicated that 19% of the sample reported having suicidal ideation and 7% reported having suicide attempts during the past 6 months. Suicidal behavior in elder, female adolescents was very prevalent, with 33% and 12% of 18-year-youths reporting some suicidal ideation and attempts during the past 6 months, and about 50% and 17% of females reporting suicidal ideation and attempts, respectively (Liu et al., 2005b). As previous studies of youth suicidal behavior in Western countries (Borowsky et al., 2001, Lewinsohn et al., 1994, Shaffer et al., 1996), we found that suicidal behavior in Chinese adolescents is associated with multiple biological, psychological, and family factors, including age, sleep loss, nightmares, life stress, poor family relationships, depressive symptoms, and aggressive behavior (Liu, 2004, Liu and Tein, 2005, Liu et al., 2005a, Liu et al., 2005b).

There is evidence that reported behavioral problems in children may differ across sources of information (Achenbach et al., 1987, Deng et al., 2004). Research shows that parent-reported behavioral problems in Chinese children are less prevalent than in Western children (Liu et al., 1999), but self-reported problems in Chinese adolescents are comparable to Western samples (Liu et al., 2000b). Thus, the prevalence and risk factors of suicidal behavior as reported by parents may also be different from child self reports. For example, parent-reported suicidal behavior may represent serious suicidal thought and more severe suicide action that had caused attention of parents. However, no studies have examined parent-reported suicidal behavior and psychosocial correlates in Chinese adolescents. The current study reported here is sought to examine the prevalence of parent-reported suicidal behavior of Chinese adolescents and its associations with various child developmental, psychosocial, and family risk factors. Suicidal behavior in the current study was defined as suicidal talk or deliberate self-harm/suicide attempt as measured by the Child Behavior Checklist (Achenbach, 1991).

2. Methods 

3. Results 

3.1. Prevalence of suicidal behavior 

Overall, 2.4% of the sample had talked about suicide during the past 6 months, 3.2% had deliberately hurt or attempted to kill themselves, and 5.1% had either suicidal talk or self-harm (i.e., suicidal behavior). Of 62 adolescents who had self-harm, 19.4% talked about suicide, while among 47 of those who talked about suicide, 25.5% had self-harm/suicide attempt (OR = 12.50, 95%CI = 6.13–25.51).

Fig. 1 shows the prevalence rates of suicidal behaviors (suicidal talk, self-harm, and either suicidal talk or self-harm) by age and gender. For both boys and girls, the combined prevalence rates of either suicidal talk or self-harm began to increase at age 14 and were highest at age 16 years. Across all age groups, boys and girls had no significant differences in the prevalence rates of suicidal talk (2.9% vs. 2.0%, χ2 = 1.59, p > .05), self-harm (3.0% vs. 3.5%, χ2 = 0.26, p > .05), and either suicidal behavior (5.2% vs. 4.8%, χ2 = 0.15, p > .05). At age 13, however, girls were more likely to have self-harm than boys (5.4% vs. 0.6%, Fisher's exact test, p = .014).

View Large Image Download to PowerPoint

Fig. 1. Parent-reported suicidal behavior (%) among Chinese adolescents. Empty bars represent boys, dotted bars for girls, and netted bars for total.

3.2. Univariate logistic regression analyses 

Table 1 present the prevalence rates of suicidal behavior and odds ratios for each category of potential risk factors that were significant in univariate logistic regression models.

Table 1. Prevalence rates and odds ratios of suicidal behavior in adolescents aged 11–16, in univariate logistic regression models (n=1920)

		%	Odds ratio	95%CI	Wald χ2	p
	Age (years)
	11–12	3.6	1.00
	13–14	4.7	1.30	0.73–2.30	0.83	NS
	15–16	6.6	1.88	1.11–3.19	5.49	.019
	History of chronic diseases (yes/no)	9.0/4.6	2.03	1.16–3.55	6.10	.014
	Child–parent relationship
	Good with both parents	4.6	1.00
	Good with one parent	10.4	2.42	1.25–4.70	6.84	.009
	Poor with both parents	12.8	3.08	1.17–8.07	5.21	.022
	Academic performance
	Excellent/good	3.6	1.00
	Fair	5.7	1.60	1.02–2.51	4.21	.040
	Poor	11.2	3.35	1.67–6.73	11.58	.001
	Peer relationships
	Good	3.9	1.00
	Fair	6.5	1.71	1.13–2.58	6.38	.012
	Poor	7.7	2.04	0.47–8.91	0.90	NS
	Psychopathology (top 10%, yes/no)
	Withdrawn	12.6/3.9	3.51	2.24–5.51	29.97	<.001
	Somatic complaints	10.7/3.5	3.34	2.21–5.06	32.48	<.001
	Anxious/depressed	14.9/3.1	5.45	3.59–8.29	63.04	<.001
	Social problems	15.4/4.0	4.40	2.75–7.03	38.29	<.001
	Thought problems	10.3/2.9	3.88	2.55–5.90	40.32	<.001
	Attention problems	16.9/3.4	5.79	3.77–8.91	63.96	<.001
	Delinquent behavior	15.8/3.7	4.83	3.08–7.55	47.44	<.001
	Aggressive behavior	16.2/3.8	4.88	3.09–7.71	46.33	<.001
	Father's general health
	Good	4.5	1.00
	Fair	6.6	1.50	0.91–2.48	2.48	NS
	Poor	15.2	3.77	1.41–10.05	7.03	.008
	Father's psychiatric history (yes/no)	11.9/4.6	2.79	1.53–5.08	11.22	<.001
	Mother's general health
	Good	4.4	1.00
	Fair	6.0	1.39	0.86–2.25	1.80	.180
	Poor	19.1	5.20	2.41–11.22	17.69	<.001
	Mother's psychiatric history (yes/no)	12.3/4.8	2.76	1.21–6.26	5.90	.015
	Parental relationships
	Good	3.7	1.00
	Fair	5.7	1.55	0.98–2.46	3.54	NS
	Poor	23.6	7.97	4.30–14.74	43.65	<.001
	Family cohesiona
	High	2.6	1.00
	Middle	4.1	1.63	0.83–3.21	2.03	NS
	Low	7.5	3.07	1.67–5.67	12.92	<.001
	Family conflict
	Low	2.2	1.00
	Middle	3.9	1.78	0.93–3.43	3.00	NS
	High	10.1	4.94	2.92–8.37	35.26	<.001
	Discipline style
	Reasoning	3.4	1.00
	Scolding	8.5	2.67	1.47–4.86	10.42	.001
	Physical punishment	21.7	7.94	4.73–13.28	62.19	<.001
	Indifference/neglect	5.0	1.51	0.53–4.28	0.60	NS

Note: CI = confidence interval; NS = not significant.

a Three categories were divided on the basis of the 25th and 75th percentile.

As shown in Table 1, parent-reported suicidal behavior was associated with a wide range of child and family factors. Poor parent–child relationship, poor peer relationship, and poor academic performance were associated with elevated risk for suicidal behavior. Children with history of chronic diseases were 2 times more likely to have parent-reported suicidal behavior. All behavioral/emotional problems were related to more than 3 fold risk for suicidal behavior. The following family factors were associated with increased risk for suicidal behavior: poor general health and psychiatric history of parents, poor parental relationship, low family cohesion, high family conflict, and poor parenting styles (scolding and physical punishment). Other factors such as parental education and occupation, family size, and child gender were not found to be significantly associated with suicidal behavior.

3.3. Multivariate logistic regression analyses 

Multivariate stepwise logistic regression analyses were performed to determine major risk factors of suicidal behavior and to control potentially confounding effects. All variables that were significant in univariate models in Table 1 were initially forced to enter for stepwise regression. Similar results were obtained using forward and backward methods. Table 2 presents adjusted odds ratios and their 95% confidence intervals for significant risk factors in the final regression model (Nagelkerke R2=.205, Hosmer and Lemeshow χ2=4.82, df=8, p>.05).

Table 2. Adjusted odds ratios of suicidal behavior: stepwise logistic regression model (n=1920)

		Adjusted odds ratio	95%CI	Wald χ2	p
	Mother's general health
	Good	1.00
	Fair	0.87	0.48–1.56	0.01	NS
	Poor	3.59	1.42–9.10	7.17	.007
	Anxious/depressed (top 10%)	3.08	1.79–5.29	19.98	<.001
	Family conflicta
	Low	1.00
	Middle	1.68	0.79–3.58	2.21	NS
	High	2.88	1.49–5.55	11.15	.002
	Discipline style
	Reasoning	1.00
	Scolding	1.10	0.52–2.31	0.64	NS
	Physical punishment	3.42	1.81–6.43	19.37	<.001
	Indifference and neglect	1.17	0.34–4.01	0.29	NS

Note: CI = confidence interval; NS = not significant.

a Three categories were divided on the basis of the 25th and 75th percentile.

As indicated in Table 2, four factors were found to be significantly associated with suicidal behavior in the final model. Anxiety/depression was a significant risk factor. Children who were scored in the top 10% on the anxiety/depression scale were more than 3 times more likely to have suicidal behavior compared to children who scored below the 90th percentile. Physical punishment of parenting style and family conflict were also retained significantly. Physical punishment was related to 3.4 times risk for suicidal behavior compared to reasoning parenting style. Children in high conflicting families were at about 3 times risk for suicidal behavior compared to those in low conflicting families. Poor maternal general health was also a significant factor of suicidal behavior in the final model (OR=3.59, 96%CI=1.42–9.10).

4. Discussion 

This report represents the first investigation on the prevalence and risk factors of parent-reported suicidal behavior among adolescents from mainland China. Principal findings of this study include: 1) During the past six months, 2.4% of the sample had talked about suicide, 3.2% had deliberately hurt or attempted to kill themselves, and 5.1% had either talked of or actually attempted to harm themselves (i.e., suicidal behavior); 2) depression/anxiety was the only significant psychopathological domain that was associated with suicidal behavior after controlling for other psychosocial and family variables; and 3) poor maternal health, family conflict, and physical punishment of parental discipline style were three family variables that were significantly associated with elevated risk for suicidal behavior in the multivariate model.

In the current study, parents reported that 5.1% of Chinese adolescents had suicidal behavior. Our previous study of 1362 Chinese adolescents in the same area found that 21% reported having suicidal ideation and 9% reported having suicide attempts/deliberate self-harm during the past 6 months (Liu et al., 2005b). The prevalence rates of parent-reported suicidal behavior among Chinese adolescents are markedly lower than those reported by youths themselves. This may be due to that many adolescents with suicidal thoughts did not talk about suicide with their parents and many suicide attempts without serious consequence did not get attention from parents. However, the prevalence rates reported here may represent more serious suicidal thought and suicide attempts. Unfortunately, the two studies were conducted with different samples. It would be interesting to compare the similarities and differences between parent report and self report on suicidal behavior in a sample of Chinese youths in the future. Because either adolescent or parent may underreport suicidal behavior due to stigma, a combination of the two reports may be more complete than a single report. This would suggest that clinician should gather the data for suicide risk assessment not only from the adolescent but also from significant others and thus those adolescents with suicidal ideation/gesture as reported by self or significant others should be targeted for further psychiatric and suicide risk assessment in order to prevent the tragedy of youth suicide.

Univariate analyses indicated that all CBCL scales including withdrawn, somatic complaints, anxious/depressed, social problems, thought problems, attention problems, delinquent behavior, and aggressive behavior were significantly related to suicidal behavior, but only anxious/depressed remained significant after controlling for all other behavioral/emotional problems and family and individual variables. Our previous study of Chinese adolescents and a number of studies in Western countries (Gould et al., 2003, Liu et al., 2005a, Liu et al., 2005b, Shaffer et al., 1996) demonstrated that not only anxiety/depression but also aggressive behavior/conduct disorder is associated with elevated risks for self-reported suicidal behavior. There are two possible explanations for it. First, anxiety/depression may play more important and independent roles in parent-reported adolescent suicidal behaviors than other behavioral problems, including aggression and delinquent behavior. Second, in the current study we included a broader range of child and family variables, such as child illness history, school performance, peer relationships, parental health and history of mental disorders, parenting style, and family conflict and cohesion. When all of these variables were statistically controlled, the association between aggression and delinquency and suicidal behavior disappeared, possibly due to complete control of other variables. However, most previous studies did not control so many variables as the current study. Thus, the disappearance of the association may be explained by the different variables entered to the multivariate models across studies.

Consistent with most previous studies in Western countries (Gould et al., 2003, Lewinsohn et al., 1994), we found a number of parental and family variables were associated with risks for suicidal behavior in Chinese adolescents. These variables include poor perceived health and psychiatric history of parents, poor parental relationship, family conflict, poor family cohesion, and physical punishment of parenting style. Physical punishment and family conflict remained to be significant in the final multivariate model. Most Chinese parents, especially in rural China, think children are their own properties and children have to obey the rules made by parents, whether wrong or right. Chinese parents are less likely to communicate with their children but more likely to use physical discipline if their children break the rules made by parents. In our previous study, we found that 14% of Chinese rural adolescents had been beaten by parents during the past year. Adolescents who reported physical punishment by parents during the past year were at 2–3 fold risk for suicidal ideation or attempt (Liu and Tein, 2005). Inter-parental conflicts in rural China are very common due to rapid social and cultural change, women's disadvantaged status, mother-in-law's involvement, patriarchal values, and limited opportunities (Pearson, 1995). Research has shown family conflict or parental divorce to be a significant risk factor for children's behavioral and emotional problems in China (Dong et al., 2002). Family conflicts may exert a direct role on adolescent suicidal behavior, perhaps through the lack of support or a direct stressful effect. Suicidal behavior may also be a way of escaping from parental punishment and family conflicts. These findings highlight the importance of family intervention as one possible strategy to prevent adolescent suicide in China.

There are several limitations to these data that should be acknowledged when interpreting our findings. Our reliance on parents' reports to estimate suicidal behavior and all other measures is a major limitation of the study. Parents may underreport adolescents' suicidal behavior because suicidal adolescents may not expose their suicidal thoughts to their parents and some suicide actions may not cause serious harm that causes parents' attention. Another limitation is the identification of psychopathology by the CBCL (Achenbach, 1991), a screening instrument. It is unclear to what extent the CBCL syndromes cited here, on the basis of a predetermined cutoff score, would match clinical diagnoses. As most cross-sectional and retrospective studies, we cannot determine the causal relationships between psychosocial factors and adolescent suicidal behavior in the current study. Although the response rate is relatively high (93%), we cannot estimate the extent to which our estimations could be biased as a result of 7% non-responders. The prevalence rates of suicidal behavior may be underestimated if parents with psychopathology or problem children are less likely to participate in the survey. Furthermore, this study was conducted in one prefecture in China. It is unknown if these findings could be generalized to adolescent populations in others areas of China.

In summary, this school-based survey of adolescents in mainland China showed that 5.1% of Chinese adolescents had parent-reported suicidal behavior during the past 6 months. Regardless of cultural differences, one of the most robust suicidal risk factors in adolescents in China is depression and anxiety, like in most Western countries (Borowsky et al., 2001, Gould et al., 2003). Family conflict and physical punishment of parental discipline style are also associated with suicidal risk in Chinese adolescents. Longitudinal studies are warranted to examine the pathways of family environmental factors and depression/anxiety to suicidal behavior in Chinese youths.