Assignment: Critically Appraising Quantitative Studies

The percentage of suicide attempters with risk factors, pro- tective factors, and diagnosis of mental illness are presented in Table 1. In addition to risk and protective factors, features of the attempt, such as prior planning (11.0% Yes, 89.0% No), attempt to hide (30.3% Yes, 69.7% No), and place of suicide attempt (80.1% Home, 2.2% Workplace, 9.3% Public place, 2.8% Friend’s house, .6% Public building) were also included in the analysis.

Direct logistic regression was performed to assess the impact of available variables, namely, risk factors, protective factors, and features of the suicide attempt on the likelihood that suicide attempters were diagnosed with mental illness. Logistic regression was used in similar studies for a large number of predictors [25, 26] and is typically used to develop a subset of variables useful for predicting the criterion, by eliminating superfluous variables. Our sample size is sufficiently large and representative for statistical regression [47]. The full model (see Table 2) containing all available predictors was statistically significant, 2 (23, N = 462) = 83.40, p < .001, indicating that the model was able to distinguish between attempters with and without diagnosis of mental illness.Themodel as awhole explained between 16.5% (Cox and Snell 2) and 24.4% (Nagelkerke 2) of the variance inmental illness and correctly classified 79.0% of the cases. As shown in Table 3, only six of the independent variables made a unique statistically significant contribution to the model (unemployment, mental illness or suicide in family, alcohol or drug abuse, habitual poor coping, willing to seek help, and positive future planning). The strongest predictor of mental illness was mental illness or suicide in family, with an odds ratio of 2.75. This indicated that attempters who had mental illness or suicide in family were 2.75 times more likely to have a diagnosis of mental illness than those without mental illness, controlling for all other predictors in the model. The second strongest predictor was unemployment with an odds ratio of 2.43. This indicated that attempters who were unemployed were 2.43 times more likely to have diagnosis of mental illness. The third strongest predictor was willing to seek help, with an odds ratio of 2.28. This indicated that attempters who were willing to seek help were 2.28 times