Table 2.

Multivariate analyses: separate logistic regression analyses of variables predicting suicide attempter status: model 1 ever-attempter (ever-A), compared with non-attempter (non-A) status, and model 2 multiple-attempter (multiple-A), compared with single-attempter status (single-A), controlling for covariates^a

	Model 1 Ever-A compared with non-A				Model 2 Multiple-A compared with single-A
Variable	B	Wald (df =1)	OR (CI 95%)	P	B	Wald (df =1)	OR (CI 95%)	P
Childhood sexual abuse
Intercourse	−0.45	0.65	0.64 (0.21–1.92)	0.42	−0.55	0.45	0.58 (0.12–2.85)	0.50
Parent figure	−0.66	1.41	0.52 (0.18–1.53)	0.24	2.51	6.20	12.27 (1.71–88.33)	0.01
Physical force	0.26	0.23	1.29 (0.45–3.68)	0.63	0.02	<0.001	1.02 (0.19–5.36)	0.98
Onset before age 6 years	−0.75	2.49	0.47 (0.19–1.20)	0.11	0.86	1.40	2.36 (0.57–9.74)	0.24

^aThe covariates included age, race or ethnic origin (White as the reference group), history of drug or alcohol dependence, chronic depression, adulthood sexual assault, and IPV.

Model 1. According to the Hosmer-Lemeshow²⁹ goodness-of-fit test, the overall model fit was acceptable (n = 106; χ² = 6.21, df = 8, P = 0.62). Overall classification was 70.8%; correction classification rates were 63.8% for no history of suicide attempts and 76.3% for ever-attempter status.

Model 2. Overall model fit was acceptable (8, n = 106; χ² = 18.65, df = 8, P = 0.02), and overall classification was 83.1%; 86.7% for single-attempter status and 79.3% for multiple-attempter status