Table IV.

Binomial logit coefficients for managing at least one patient for heavy drinking or alcohol-related problems versus none during the past year.

		Coefficient	SE.	% change in Odds	p
Profession (0 = MD, 1 = NP)		−1.03	0.31	−64.0	0.01
Hours of post-graduate training or CME on alcohol and alcohol related problems	1=1–4 h	0.75	0.38	113.0	0.05
	1=4 h plus	0.05	0.37
Objective Knowledge Scale	0–6	0.07	0.13
My colleagues are somewhat sceptical about the efficacy of behavioural medicine	1=Agree	0.10	0.37
	1=Neutral	−0.14	0.33
It is easy to generalize the really bad cases of alcoholism to all patients with alcohol- related problems	1=Agree	−0.12	0.31
	1=Neutral	−0.58	0.39
Problem drinkers are more likely to be non-compliant patients	1=Agree	0.11	0.42
	1=Neutral	−0.08	0.51
There is not enough time to advise patients about drinking	1=Agree	0.24	0.32
	1=Neutral	−0.12	0.38
Alcohol is a factor in most of the medical conditions I see	1=Agree	1.44	0.48	323.0	0.01
	1=Neutral	0.99	0.40	169.0	0.01
I don’t know how to identify at-risk drinkers who have no obvious symptoms of excess consumption	1=Agree	−1.37	0.34	−75.0	0.01
	1=Neutral	−0.79	0.40	−55.0	0.05
I am not aware of a single problem drinker who ever cut back on his or her drinking upon medical advice	1=Agree	0.61	0.50
	1=Neutral	−1.00	0.42	−63.0	0.01

Log likelihood = −165.0, LR χ² = 75.0, Pseudo R² = 0.18, p = 0.00.