The Construction and Selection of Predictive Scales, with Special Reference to Parasuicide

Abstract

Issues concerning the use of predictive and screening instruments include the difficulties posed by temporal changes in the base rates for the behaviour of interest, the possible advantages of using three rather than the customary two risk categories, and the desirability of specifying in advance the proportional sizes of the risk subgroups. A method is proposed for constructing and assessing tripartite risk scales, and a new scale for predicting the repetition of parasuicide, with the results of a prospective validation study, is reported.

Distinguishing the relevant features of frequent suicide attempters

“…To date, most studies addressing the prediction of suicide re-attempts have used no measures of predictive accuracy (Bille-Brahe et al, 1997;Cedereke and Ojehagen, 2005;Colman et al, 2004;Morton, 1993;Sidley et al, 1999), limiting the clinical utility of their findings (Galfalvy et al, 2008). However, when the precision of the prediction is measured, the results are disappointing, finding roughly 39% sensitivity and 86% specificity in a study of socio-demographic risk factors of parasuicide (Kreitman and Foster, 1991), 28% sensitivity and 99% specificity in a sample of children after deliberate self-harm (Chitsabesan et al, 2003), or 85e94% sensitivity and 26e38% specificity when assessing self-harm in emergency settings (Cooper et al, 2007). Thus, the weak predictive value of risk factors leads to large numbers of false-positives or false-negatives (Maris, 2002).…”

“…This may be in part because there is no clear understanding of the hierarchy or relationship between a variety of factors that predict repeat attempts. These include socio-demographic factors such as age (Christiansen and Jensen, 2007;Corcoran et al, 2004), unemployment (Tejedor et al, 1999), not being married (Kreitman and Foster, 1991) or low educational level (Christiansen and Jensen, 2007) which have been associated with repetition of suicidal behaviour (Osvath et al, 2003). Clinical factors include lethality of the index attempt (Gibb et al, 2005), poor physical health (Colman et al, 2004), and mental illness or psychiatric comorbidity, which has been noted to be a risk factor for repetition in studies on parasuicides (Colman et al, 2004;Osvath et al, 2003), non-fatal suicidal behaviours (Kapur et al, 2006) and SAs among inpatients (Spiessl et al, 2002).…”

A risk index for 12-month suicide attempts in the National Comorbidity Survey Replication (NCS-R)

“…Previous research has shown that up to 20% of ideators in a given year make a suicide attempt during that year (Kessler et al, 2005a;Kuo et al, 2001), justifying this focus. We use an actuarial, data-driven approach similar to the approach used successfully to develop prediction models for violent behavior (Dawes et al, 1989;Harris et al, 2004;Harris et al, 2003;Hilton et al, 2004;Mossman, 1994), and less successfully to predict repetitive suicidal behavior (Buglass & Horton, 1974;Corcoran et al, 1997;Hawton & Fagg, 1995;Kreitman & Foster, 1991). We focus on potential risk factors that clinicians can assess quickly using fully structured assessments.…”

Parasuicide: features of repetition and the implications for intervention

“…Kreitman & Foster (1991) have also suggested that the factors which predict repetition may change as an individual's history progresses. Nevertheless, their own weighted prediction scale contains items similar or equivalent to those of Buglass & Horton, though Kreitman & Foster report that their own scale allows more efficient allocation of patients to low-, moderate-and high-risk groups.…”