. Author manuscript; available in PMC: 2022 Feb 1.

Published in final edited form as: Transcult Psychiatry. 2020 Apr 11;58(1):126–139. doi: 10.1177/1363461520916697

Table 2.

Multivariate Regression Models predicting Certainty of Diagnosis, Use of Coercive measures, and Service Use

Outcome:	Certainty of Diagnosis^a	Certainty of Diagnosis^b	Use of Coercive Measures	Use of Coercive Measures	IM medication	IM medication	Involuntary commitment	Involuntary commitment

Geographical Origin (ref=Spaniards)
Latinos	0.78	1.06	1.43	1.42	1.39	2.04	0.24^*	0.26^*
	[0.45,1.36]	[0.59,1.92]	[0.46,4.48]	[0.43,4.70]	[0.38,5.14]	[0.53,7.91]	[0.07,0.80]	[0.08,0.89]
North Africans	0.58	1.14	4.23^*	2.12	3.88	3.49	1.33	1.34
	[0.29,1.16]	[0.53,2.44]	[1.26,14.17]	[0.54,8.32]	[0.97,15.44]	[0.77,15.78]	[0.44,4.07]	[0.39,4.53]
Asians	0.32^*	0.92	1.5	0.31	2.03	1.20	0.39	Omitted
	[0.13,0.79]	[0.35,2.43]	[0.26,8.79]	[0.03,3.56]	[0.32,13.00]	[0.10,13.73]	[0.05,3.35]
Eastern Europeans	0.30^**	0.71	2.10	1.06	2.84	3.14	1.14	0.77
	[0.14,0.64]	[0.31,1.64]	[0.49,8.97]	[0.21,5.35]	[0.60,13.52]	[0.59,16.63]	[0.31,4.21]	[0.18,3.36]
Western Europeans	0.51	0.85	2.70	1.68	1.68	1.53	2.41	1.54
	[0.24,1.08]	[0.38,1.89]	[0.71,10.22]	[0.38,7.47]	[0.32,8.77]	[0.26,8.94]	[0.82,7.08]	[0.45,5.27]
Patient Gender (ref=Male)
Female		0.73		0.47^*		0.29^**		1.18
		[0.48,1.09]		[0.22,0.99]		[0.12,0.72]		[0.54,2.55]
Understand Spanish (ref=None/Little)
Quite/A lot		1.22		1.08		0.76		0.53
		[0.62,2.40]		[0.35,3.37]		[0.21,2.78]		[0.18,1.59]
Culture Influence (ref=None/Little)
Quite/A lot		0.61		0.98		0.37		0.69
		[0.37,1.01]		[0.41,2.37]		[0.11,1.23]		[0.24,1.99]
Clinician Comfort (ref=None/Little)
Quite/A lot		5.02^***		0.21^***		0.88		0.54
		[2.75,9.17]		[0.09,0.51]		[0.26,2.98]		[0.19,1.50]
N	368	365	369	366	369	366	347	344

Note. 95% Confidence Intervals of ORs are reported in brackets.

p<0.05,

^**

p<0.01,

^***

p<0.001. Two models are fitted for each outcome.

The first model is the null model that only adjusts for patient’s geographical origin.

The second model is the full model, which further adjusts for gender, language proficiency, level of the clinician’s perceptions of culture influence, and clinician’s comfort during the interview. Asians were omitted in the last regression because being Asians perfectly correlated with the outcome after controlling for the rest factors. An Ordered Logistic Regression model was used to predict “Certainty of Diagnosis” and a Logistic Regression model for the other outcomes.