Table 5.

Multivariable analysis: patient physician communication

Dependent variable	Pseudo R 2	Independent variable	Coefficient (β)	Odds ratio	95 % confidence interval	P value*
Copinga	0.14	Intercept	−2.00
		Age	−0.033	0.97	0.94–0.99	0.032
		PROMIS-Pain Interference CAT	0.069	1.1	1.0–1.1	0.013
		Marital status (ref: single)
		Living with partner	b
		Married	0.77	2.2	0.75–6.3	0.15
		Separated or divorced	−1.2	0.3	0.054–1.6	0.16
		Widowed	b
		Intervention(PROMIS feedback)	0.83	2.30	0.95–5.6	0.064
Circumstancesc	0.17	Intercept	−3.7
		PROMIS-Pain Interference CAT	0.063	1.1	1.0–1.1	0.042
		Prior surgery	−1.1	0.34	0.11–1.0	0.052
		Marital status (ref: single)
		Living with partner	b
		Married	1.0	2.8	1.1–7.3	0.039
		Separated or divorced	1.0	0.36	0.035–3.6	0.38
		Widowed	b
		Working status (ref: full-time)
		Part-time	1.9	6.8	1.1–39	0.032
		Homemaker	−0.098	0.38	0.040–3.5	0.39
		Retired	−1.1	0.35	0.12–0.98	0.046
		Unemployed, able to work	b
		Unemployed, unable to work	b

PROMIS Patient-Reported Outcomes Measurement Information System, CAT computer-adaptive testing

*P < 0.05, level of significance

^aHosmer-Lemeshow χ ² = 16, P = 0.046; the area under the receiver operating characteristic (ROC) curve was 0.72 (95 % CI 0.61–0.83)

^bCoefficient and OR could not be calculated because of low number of cases. Observations have been dropped and not used in order to not bias the remaining coefficients in the model

^cHosmer-Lemeshow χ ² = 5.3, P = 0.73; the area under the receiver operating characteristic (ROC) curve was 0.61 (95 % CI 0.51–0.71)