Basic HTML Version
214
MARK P. JENSEN ET AL.
for example, from a small to medium treatment response in all partici-
pants or from a large response in a very few participants), a responder
analysis was conducted to estimate the number of participants who
showed a clinically meaningful change in pain in both the intent-to-
treat and the efficacy analysis samples. A pretreatment to posttreatment
pain-intensity change of 30% was used as the cutoff for identifying a
clinically meaningful change, given previous research that has shown
that improvements of 30% or more are associated with patient reports
of meaningful change across a number of chronic pain conditions (Farrar,
Young, LaMoreaux, Werth, & Poole, 2001).
The first secondary study question, examining whether changes in
secondary outcome domains (pain unpleasantness, pain interference,
depressive symptoms, and perceived control over pain) occur with
hypnotic-analgesia treatment, was addressed using the same ANOVA
strategy used to address the primary study question, with the secondary
outcome variables as the dependent variables. Finally, the associations
between the potential predictors of treatment response and treatment
outcome were examined by using a repeated measures ANOVA for
the categorical predictor (diagnostic group) and correlational analyses
for the continuous predictors (hypnotizability, treatment-outcome
expectancy, concentration of treatment, and change in pain intensity
during the first treatment session). Only the efficacy sample was used
for these analyses to reduce the bias that would be introduced by the
lack of full participation in treatment or the development of a medical
problem that could impact pain. Pretreatment to posttreatment change
in pain intensity was the dependent/criterion variable in all of these
predictor analyses.
R
ESULTS
Primary Outcome: Average Pain Intensity
In both the intent-to-treat and the efficacy analysis samples, a signif-
icant time effect was found in the ANOVAs that used average pain
intensity as the dependent variable (Table 1). Subsequent univariate
analyses with the intent-to-treat sample indicated a statistically signifi-
cant (
p
< .05) decrease in pain intensity from pretreatment to posttreat-
ment but no significant changes in pain intensity over the baseline
period. The improvement in pain was maintained for 3 months after
treatment, as indicated by the lack of significant change in pain over
this time period. On average, there was a pretreatment to posttreat-
ment decrease of 0.79 (16% change) in pain intensity (on the 0–10
NRS), representing a “small” to “medium” effect size (
d
= .41, Cohen,
1988). Ten of the 30 intent-to-treat participants reported a pretreatment
to posttreatment decrease in pain intensity of 30% or more. Using the
conservative assumption that the 3 participants who did not provide