Letter to the Editor: Multivariate Analysis in Suicide Attempts: A Useful Tool for the Clinician

Abstract

Suicide attempts are currently considered an important health care subject in European countries (Appleby, 1992;Bille-Brahe, Kerkhof, De Leo et al., 1996;Ruiz-Doblado, 1999); there are several epidemiological and psychopathological risk factors that clinicians consider in their daily decision-making process of hospitalization (Ruiz-Doblado, 2001). However, risk factors are not highly accurate. In order to establish a mathematical decision-making model which could be useful for the clinician in psychiatric emergences, clinical, psychopathological and sociodemographic variables are collected. Data of n ¼ 137 suicide attempts were analysed. The patients were taken from the Epidemiological Survey of Suicide Attempts of Osuna Hospital, in a rural area in southern Spain. The main aim of the study was to analyse the risk factors that determine the decision of hospitalization in a psychiatric unit.Initially, we collected 10 variables: gender, age, psychiatric previous history, previous suicide attempts, impulsive vs. planned behaviour of the suicidal act, comorbid chronic somatic disease, use of psychotropic vs. non-psychotropic drugs as a method of committing the attempt, possibility of help or rescue during the act, self-criticism after the attempt, and diagnosis of 'minor' pathology (adjustment, neurotic, non-melancholic mood disorders) vs. 'major' pathology (organic mental disorders, substance-related disorders, schizophrenia or mood psychosis). Initial results showed that, excepting comorbid somatic disease, all variables were considered as statistically significant (p < 0:01) in a non-multivariate, simple analysis model. When we introduce all the significant variables in a multivariate explicative, decision-making model (stepwise logistic regression), in order to obtain the final mathematic equation, psychiatric hospitalizations were mainly determined by three variables: First, female gender was a protective factor against hospitalization (Odds Ratio: 0.316). Second, patients that used psychotropic drugs as the method of committing the attempt were six times more frequently hospitalized than patients using non-psychoactive drugs (OR: 6.00). Finally, if the attempt was realized without possibility of help or rescue, clinicians determined the hospitalization 15.1 times more frequently (OR: 15.119). Results of the logistic regression are showed in Table 1.Indeed, male gender, the use of psychotropic drugs and non-possibility of help or rescue during the attempt were predictive, mathematic factors of hospitalization. These simple 'markers' of severity could assist the psychiatrist in his or her daily decision-making process at emergency departments, bringing a mathematic equation of high internal validity, and also