Data Analysis
Demographic variables were compared between groups using
chi square2 or independent t tests as indicated. Correlations between
patient and caregiver distress variables and treatment targets were
calculated using Pearson’s correlation analyses.
When more than one treatment target was significantly correlated
with a given distress variable, stepwise linear regression
analysis was performed to determine which of the 3 treatment
targets (BK, stigma, and therapeutic alliance) were predictors of the
distress variable. These analyses were conducted with the distress
variable as the dependent variable and either caregiver or patient
treatment targets as predictors.
A large percentage of bipolar families who participated in
intake assessments (50%) never provided data at the 3-month point.
Therefore we examined differences in demographics, distress variables,
and treatment targets between these nonadherent participants
and those that continued to participate to explore variables that may
improve retention. As there was no expectation that families who
presented for consultation alone would continue at our center, these
families (n 8) were excluded from analysis of adherence.
As this study was exploratory in nature and the sample size
was relatively small, we did not correct for multiple comparisons.