Skip to main content
. Author manuscript; available in PMC: 2007 Sep 25.
Published in final edited form as: Psychiatr Serv. 2007 Jan;58(1):27–33. doi: 10.1176/appi.ps.58.1.27-a

Table 3.

Logistic regression analysis of the likelihood that persons with bipolar I disorder in the Medstat database received treatments either recommended or advised against by guidelinesa

1994
1999
Variable z score SE z score SE
Lithium, valproic acid, or carbamazepine −1.56 .165 4.61** .885
Therapy −4.70** .062 −14.40** .002
Lithium, valproic acid, or carbamazepine plus therapy −2.83* .120 −3.93* .093
Antidepressant in the absence of lithium, valproic acid, or carbamazepineb 4.79** .757 .44 .293
a

The reference year was 1991. The following variables were controlled for in the model: gender, relationship to employee (that is, spouse, elderly dependent, or child), age, medical comorbidity, substance use disorder, and other psychiatric comorbidity.

b

Treatment advised against by guidelines

*

p<.01

**

p<.001