. 2025 Mar 24;66(10):1449–1460. doi: 10.1111/jcpp.14144

Table 3.

Multinomial logistic regressions of criticism class membership on baseline variables (n = 831)

Baseline variable	High with small decrease			Sharply decreasing			Increasing
Baseline variable	Estimate	SE	p	Estimate	SE	p	Estimate	SE	p
Treatment as usual	−0.232	0.214	.278	−0.465	0.534	.384	−0.366	0.731	.616
Depression ^a	−0.008	0.010	.413	0.017	0.034	.622	−0.041	0.030	.175
Suicidal ideation ^a	0.015	0.015	.308	−0.045	0.056	.421	0.020	0.045	.655
Hopelessness ^a	0.015	0.031	.635	−0.106	0.095	.262	−0.036	0.084	.666
Parent mental distress ^b	0.151	0.025	<.001	0.072	0.070	.303	0.052	0.078	.507
Characteristics of index self‐injury episode
For emotional relief ^a	−0.191	0.080	.017	0.140	0.243	.563	−0.306	0.326	.349
For interpersonal influence ^b	0.139	0.084	.097	0.288	0.153	.060	−0.096	0.907	.916
Intent to die ^a	0.496	0.237	.036	0.330	0.562	.557	0.877	0.827	.289
Lethality ^a	0.592	0.256	.021	−0.089	0.896	.921	1.365	0.592	.021
Low intervention probability ^a	−0.299	0.298	.314	−0.119	0.682	.861	−0.157	1.193	.895

Bold values indicate a significant difference. Reference class is the low/stable class. Variables were entered in separate analyses along with control variables.

^{^a}

Analysis controlled for youth age, gender, and abuse history.

^{^b}

Analysis controlled for youth age and abuse history. Gender omitted due to issues of convergence/small cells.