Keywords: Adolescence, hopelessness, IRQ, PBI, suicidality
Allison, S, Pearce, C, Martin, C, Miller, K, & Long, R. Accepted 21 March 1995
Child and Adolescent Mental Health Service,
Flinders Medical Centre, Bedford Park,
South Australia, 5042.
Tel.: 08--204--5412. Fax: 618--204--5465
As part of a series of studies into the early detection of suicidal behaviour in adolescence, this study investigated relationships between adolescents' self-reports of perceived parental style, pessimism, and the spectrum of suicidal behaviour in a sample of Australian high school students (Mean age = 15.8). Three hundred and seven students completed a questionnaire that included demographic details, the Influential Relationships Questionnaire (IRQ), the Beck Hopelessness Scale, and the Adolescent Suicide Questionnaire. There was a high level of suicidal behaviour reported, with suicidal adolescents perceiving their parents to be significantly more critical, less caring and more overprotective. Multiple regression analyses examining a proposed pathway from perceived parental style to suicidality through the mediation of hopelessness found that both parental style and hopelessness made unique contributions to a composite suicidality score. The quality of parenting is relevant to the assessment of adolescent suicidality and the IRQ may be useful for the detection of perceived parenting difficulties in suicide prevention programmes.
Adolescent suicidal behaviour is increasingly recognized as a major public health problem (Goldney, 1987; Kosky, 1987). It is of particular concern in Australia which UNICEF (1993) reports has the highest rate of completed suicide among young people in the industrialized world (16.4 per 100.000), and where suicide is the second most common cause of death for 15--24 year olds ( Australian Bureau of Statistics, 1993). Whilst the medical seriousness of suicide attempts varies, considerable resources are needed for the medical and psychiatric treatment of complications (Goldney, 1987; Kosky, 1987). Awareness of these issues has led to an increased emphasis on the early detection of those at highest risk of suicide, investigations of the interpersonal environment of adolescents and, in particular, the influence parents have on the development and destiny of their offspring (Martin & Waite, 1994; Pfeffer, 1989). The current study into perceived parenting was one of a series of preliminary investigations into suicidal behaviour prior to the commencement of a longitudinal study of early detection and prevention of suicidal behaviour in South Australian high schools.
Connections between suicidality and the quality of parenting are most obvious where family life is repeatedly disrupted by parental discord, parental separation, changes in caretaker, and physical or sexual abuse (De Wilde et al., 1992; Kosky et al., 1990). In addition clinical experience suggests that, even in the absence of major disruption, suicidal adolescents are often exposed to high levels of blame and criticism in face-to-face contact with parents, not only during suicidal crises but on a longer term basis (Rotheram-Borus et al., 1994).
Effects of high levels of interpersonal criticism have been investigated in studies of expressed emotion (EE, conceptualised as critical hostility and emotional over-involvement; Kazarian, 1992, p.~51). The standard measurement tool, the Camberwell Family Interview (CFI), rates criticism, hostility, emotional over-involvement, warmth and positive comments (Vaughn & Leff, 1976). High levels of EE have been shown empirically to predict poor outcome in adults with schizophrenia (Kavanagh, 1992), unipolar depression (Hooley et al., 1986) and bipolar disorder (Miklowitz et al., 1988). Increased risk for childhood depressive disorder, substance abuse and conduct disorder has been reported with higher maternal EE (Schwartz et al., 1990). Asarnow and coworkers (1993) also found that EE had predictive validity for children hospital\-ized with depression where high EE homes resulted in more persistent disorders.
There are no published studies specifically investigating associations between EE and suicidal behaviour; links remain speculative. The paucity of published research may be due to the CFI. It is a demanding procedure, requiring special training to administer and score. Fortunately, simple self-report measures of criticism may be superior to observational techniques in assessing the impact of criticism on mood (Hooley & Teasdale, 1989). One promising measure is the Influential Relationship Questionnaire (IRQ; Baker et al., 1984; Kazarian & Baker, 1987) which has been employed with adults for the prediction of re-hospitalisation with schizophrenia (Baker et al., 1984, 1987).
The IRQ consists of a Criticism scale combined with an established measure, the Parental Bonding Instrument (PBI; Parker, Tupling, & Brown, 1979). Validity studies of the PBI show that self-report on the Care and Protection scales measure perceptions related to an observable parental style (Parker, 1989). PBI scores remain remarkably reliable over lengthy periods (Wilhelm & Parker, 1990), and seem to be independent of mood state. PBI scores showed no significant change after mood improved in studies of depressed adults (Parker, 1983; Plantes et al., 1988). These characteristics make the PBI a suitable measure of enduring perceptions of parenting style in the investigation of suicidal behaviour where associated mood changes are common.
In studies of young female suicide attempters (Goldney, 1985) and high school students with suicidal behaviours (Martin & Waite, 1994), parenting was rated as less caring and more overprotective among the suicidal groups. This combination of low care and high overprotection has been termed affectionless control (Parker, 1983). With addition of the Criticism scale, the IRQ can measure the level of perceived critical affectionless control. To date, the IRQ has not been reported in adolescent populations and one aim of this study was to determine whether perceptions of parental critical affectionless control were associated with suicidal behaviour in a non-clinical adolescent population.
During childhood development, there are links between social support, particularly parental support, and pessimism (Kashani et al., 1994). Beck (1976) has emphasised the central role of pessimistic cognitions in a pathway to suicidal behaviour and relates pessimism to experiences of loss or rejection. The current study investigated relationships between negative perceptions of the quality of close relationships (parents are distant, critical and intrusive), pessimism about the future, and adolescent suicidal behaviour, which may be an escape from relationships perceived as negative and unchangeable.
Pessimism has been measured as the cognitive dimension of hopelessness (Beck, 1967). In many previous studies, hopelessness has been associated with the spectrum of suicidal behaviour in clinical (Alloy et al., 1988; Asarnow & Guthrie, 1989; Beck et al., 1985; Beck et al., 1990; Dyer & Krietman, 1984) and community (Keinhorst et al., 1990) samples. Hopelessness can distinguish suicidal from nonsuicidal adolescents (Cole, 1989; Swedo et al., 1991). Hopelessness is a well defined concept with an established unitary \linebreak measure (Young et al., 1992). This contrasts with the depressive syndrome which includes a complexity of affective and motivational changes, negative cognitions, behavioural observations and physiological symptoms which leads to difficulties in definition and measurement (Hamilton & Shapiro, 1990). Although measures of depression are associated with suicidal behaviour in community studies (Garrison et al., 1991), it is uncertain whether measures of adolescent depression screen for the clinical disorder or measure a more general component of psychiatric distress (Garrison, et al., 1991; Martin & Waite, 1994).
To further understand the ways in which adolescents become involved in suicidal behaviour, it is informative to routinely explore relationships between risk factors as well as their direct association with suicidality. One suitable method is path analysis and the directions of the pathways between variables can be hypothesised if previous studies are available. This study examined a possible relationship between negative perceptions of parental style and adolescent suicidal behaviour being mediated through a sense of hopelessness. Perceptions of parental style were assumed to predate cognitions of hopelessness and suicidal behaviour. This assumption is consistent with available literature on parenting and negative mood (Parker, 1983; Tiggeman et al., 1992).
The sample consisted of 156 male and 151 female students aged between 13 and 17 years (Mean = 15.78, s.d. = 0.6) attending one randomly chosen metropolitan State high school. Located in the southern half of South Australia (population 1.5 million), the school serves a defined area population with predominantly middle-class residents.
Parenting style was assessed using the Influential Relationships Questionnaire (IRQ; Baker et al., 1984, 1987; Kazarian & Baker, 1987). This 37-item self-report instrument consists of Care and Protection scales of the Parental Bonding Instrument (PBI; Parker et al., 1979; Cubis et al., 1989), plus a 12-item Criticism scale. The Criticism scale of the IRQ uses the same reference format as the PBI and includes a series of items with face validity (e.g. 'Points out my weaknesses rather than praising me'). Its authors report moderate to high test-retest reliability ( r ranging from .53 to .85) and Internal Consistency (KR-20 coefficient alphas ranging from .76 to .91) for the three scales for ratings of the two most influential people in a respondent's life.
Hopelessness was measured using the Beck Hopelessness Scale (Beck et al., 1974). This 20-item self-report instrument assesses the degree to which an individual holds negative expectations towards their future. The underlying assumption is that hopelessness can be objectively measured by defining it as a system of cognitive schemas with a common denominator of negative expectations (Beck et al., 1974). The scale has been used extensively with adolescents, and has been shown to have high internal consistency (KR-20 coefficient alpha = 0.93) and a relatively high correlation with clinical ratings of hopelessness ( r = 0.74) in a population of 294 hospital patients with recent suicide attempts (Beck et al., 1974). It was chosen in preference to the Hopelessness Scale for Children (Kazdin et al., 1986) because its psychometric qualities are more impressive, and adult-level language seems more appropriate for an adolescent population.
Suicidality was assessed using the Adolescent Suicide Questionnaire (ASQ; Pearce & Martin, 1994). Derived from the concept of a spectrum of suicidal behaviours, the ASQ asks respondents whether they have ever engaged in a series of suicide behaviours (e.g. 'Have you ever tried to kill yourself?'). Unpublished data from a sample of high school students ( N = 70) indicates moderate to high test-retest reliability (Phi coefficients for nominal data ranging from 0.59 to 0.91; Pearson r ranging from 0.58 to 0.89) over a three week interval. A composite index derived from the yes/no responses to the spectrum of suicidal behaviours identified suicide attempters from non-attempters with 96% sensitivity and 79\ specificity (Pearce & Martin, op. cit). For this study, five items seeking yes/no responses to suicide ideation, plans, threats, deliberate self-harm, and suicide attempts were used in data analyses to form a composite suicidality score with sound internal consistency (standardised item alpha = 0.74). The questionnaire in this simple form does not measure the seriousness of suicidal experience.
Finally, a range of demographic details were collected, including age, gender, and year-level. Questions were all arranged in a single booklet.
With permission to conduct the study from the school principal, and approval from Flinders Medical Centre ethics committee, a letter was distributed to parents explaining the study and seeking assent. To assist parents in making this decision, copies of the questionnaire were made available for inspection at the school office. All students enrolled in years 10 and 11, present on the day of data collection were given both written and verbal explanations of the study, and questionnaires were completed during one normal class period. Subjects were assured of confidentiality, and asked not to write their names on any part of the booklet. Given the sensitive nature of some sections of the questionnaire, included on the inside cover of the booklet was an invitation to contact one of several community service organisations should they experience distress.
Forty-nine percent ( n = 150) of the sample reported ever having had suicide ideation; 14% ( n = 42) reported having planned a suicide attempt without carrying it out; 13% ( n = 38) reported having made suicide threats; 30% ( n = 91) reported having engaged in deliberate self-harm; and 9% ( n = 27) reported having made a suicide attempt. As previously noted the seriousness of suicidal behaviours was not measured.
Chi-square analyses indicated no statistically significant gender effects in reported prevalence of suicidal behaviours. A series of t-test analyses also showed no statistically significant gender difference in IRQ scale scores, hopelessness scores or the total ASQ suicidality scores. With regard to IRQ dimensions, 97% ( n = 295) of the sample indicated the female most responsible for their upbringing in the last five years was their biological mother; a further 1.5% ( n = 5) indicated a step or adoptive mother. In contrast, 89% ( n = 254) of the sample indicated the male most responsible for their upbringing in the last five years was their biological father; a further 7% ( n = 20) indicated a step or adoptive father. Hence, it was evident that we were measuring perceptions of parental figures, usually biological parents. There were significant differences in the IRQ scores if the sample was grouped according to the presence or absence of suicidal cognitions and behaviours reported in the ASQ.
Using the ASQ composite index of suicidality derived from the item
scores (Pearce & Martin, 1994), a series of Pearson correlation analyses portrayed
statistically significant relationships between low parental care and suicidality, high
parental overprotection and suicidality, and high parental criticism and suicidality
(ranging from 0.20 to 0.30, all p's <.01). In order to determine the relative contribution each made to the suicidality score a multiple regression analysis was performed. Whilst no single IRQ dimension made a statistically significant unique contribution to suicidality when controlling for the others, the predictors as a group made a statistically significant contribution ( F(6,267)="7.41," p < .001), and accounted for 12% of the variance. Pearson correlation analysis also portrayed a
statistically significant relationship).
Pearson correlation analysis also portrayed a statistically significant relationship).
All pages copyright ©Priory Lodge Education Ltd 1994-1999.