A Preliminary Double-Blind, Placebo-Controlled Randomized Study of Baclofen Effects In

Leggio et al.

A preliminary double-blind, placebo-controlled randomized study of baclofen effects in alcoholic smokers.

Supplemental Material

Moderator analyses

Alcohol Dependence, nicotine dependence, smoking goal, and drinking goal were tested as moderators of the two primary outcome measures, and five supplemental outcomes measures, i.e., % days smoking, % days drinking, cigarettes per day (CPD), drinks per day and heavy drinking days (HDD). Alcohol dependence severity (ADS) score moderated BACL effects on % days of abstinence from alcohol-tobacco co-use [F(1,194.4) = 17.39, p < .001; Figure 2S], such that BACL increased % days of abstinence from both substances among patients with higher alcohol dependence severity (22.2 ± 2.8) vs. those with lower severity (1.7 ± 2.8). Nicotine dependence severity (FTND score) also moderated % days of abstinence from alcohol-tobacco co-use, but with the less-severe nicotine dependent group having a greater response to BACL [F(1,196.4) = 15.00, p = .008; Figure 3S]. The continuous measures of ADS and FTND were not related [r(28) = .17, p = .38] nor were the dichotomized values of these terms (median split), used in the moderation analysis [X2(1,N=30) = 0.12, p = .73]. Cross-substance moderation was evidenced in that ADS interacted with the length of abstinence from tobacco [F(1,25) = 5.58, p = .03] with greater alcohol dependence and BACL having the longest abstinence from tobacco (8.10 ± 2.12), followed by PLA and low alcohol dependence (3.43 ± 1.98), BACL and low alcohol dependence (0.44 ± 2.23), and finally, PLA and high alcohol dependence (0.17 ± 2.36). By contrast, FTND did not interact with length of abstinence from alcohol [F(1,25) = 0.03, p = .87]. The average cell size was 7.5 in these moderation analyses.

ADS was also tested as moderator of the five supplemental outcomes. These results are presented in Table 1S. Means (Ms) and standard errors (SEs) for the significant interactions are reported in Table 2S.

Table 1S. P-values for Mixed Model Analyses of ADS and supplemental outcomes.

% days % days drinks

smoking drinking CPD per day %HDD

Medication < .001 < .001 n.s. .01 < .001

ADS .004 n.s. n.s. n.s. n.s.

Medication X ADS < .001 n.s. .006 < .001 .01

Baseline value of DV n.s. < .001 < .001 < .001 < .001

Race n.s. < .001 < .001 .001 < .001

Table 2S. M + SEs for significant interactions of Table 1S.

% days smoking CPD

Low ADS High ADS Low ADS High ADS

Baclofen 97.9 + 3.2 72.5 + 3.1 12.8 + 0.9 11.4 + 0.9

Placebo 95.0 + 2.8 100 + 4.2 9.3 + 0.9 14.0 + 1.3

Drinks per day %HDD

Low ADS High ADS Low ADS High ADS

Baclofen 4.57 + 0.44 2.71 + 0.42 43.3 + 4.0 35.5 + 3.7

Placebo 1.66 + 0.38 3.22 + 0.59 9.8 + 3.3 23.0 + 5.1

Similarly, FTND was also tested as moderator of the same five supplemental outcomes. These results are presented in Table 3S. Ms and SEs for the significant interactions are reported in Table 4S.

Table 3S. P-values for Mixed Model Analyses of FTND and supplemental outcomes.

% days % days drinks

smoking drinking CPD per day %HDD

Medication .006 < .001 .02 < .001 < .001

FTND .002 .04 < .001 .04 .03

Medication X FTND < .001 n.s. .02 n.s. n.s.

Baseline value of DV n.s. < .001 < .001 < .001 < .001

Race n.s. < .001 < .001 < .001 < .001

Table 4S. M + SEs for significant interactions of Table 3S.

% days smoking CPD

Low FTND High FTND Low FTND High FTND

Baclofen 76.1 + 2.8 99.5 + 3.6 9.6 + 0.8 16.2 + 1.1

Placebo 98.4 + 3.4 95.3 + 3.1 9.6 + 1.1 11.7 + 1.0

Treatment goal as a moderator

There was a smoking goal by medication interaction on % days abstinence from co-use. The nature of the interaction was as follows: those with an abstinence goal who received BACL and those with a reduction goal who received PLA had the best outcomes [13.5 ± 2.4; 13.0 ± 4.9, respectively] while those with a reduction goal on BACL and those with an abstinence goal on placebo had the poorest outcomes [2.3 ± 4.5, 1.4 ± 2.5, respectively, F(1,196.1) = 9.01, p = .003]. For co-use the placebo effect persisted [BACL: 50.5 ± 3.7, PLA 27.2 ± 4.0; F(1,195.1) = 17.96, p < .001], there was a smoking goal main effect on co-use [Abstinence goal 30.5 ± 2.5, Reduction goal 47.2 ± 4.8%; F(1,194) = 9.24, p = .003], a smoking goal X medication interaction [F(1,194.8) = 4.02, p = .046] and an alcohol goal by medication interaction [F(1,194.4) = 20.7, p < .001]. Given the distribution of these goals, we have more confidence in the findings regarding drinking goals, than smoking goals (closer to a 50/50 split). When the length of abstinence analyses (alcohol and tobacco separately) were re-analyzed with the additional treatment goal terms and interactions, there were no significant predictors in the entire models.

Treatment goals were also tested as moderators of the same five supplemental outcomes. These results are presented in Table 5S. Ms and SEs for the significant interactions are reported in Table 6S.

Table 5S. P-values for Mixed Model Analyses of treatment goals and supplemental outcomes.

% days % days drinks

smoking drinking CPD per day %HDD

Medication n.s. < .001 .027 < .001 < .001

Smoking Goal n.s. .008 n.s. < .001 .001

Drinking Goal n.s. .057 .007 n.s. n.s.

Med. X smoking goal .005 n.s. n.s. .031 n.s.

Med. X drinking goal n.s. < .001 n.s. .076 n.s.

Goal X Goal n.s. n.s. .016 n.s. n.s.

Baseline value of DV .001 < .001 < .001 < .001 .015

Race n.s. .002 < .001 .001 < .001

Table 6S. M + SEs for significant interactions of Table 5S.

% days smoking drinks per day

Cut down Quit Smoking Cut down Quit Smoking

Baclofen 63.5 + 6.5 41.2 + 3.4 6.1 + 0.6 2.9 + 0.4

Placebo 31.2 + 6.9 25.1 + 3.5 2.6 + 0.7 1.8 + 0.4

% days drinking CPD

Cut down Quit Drinking Cut down Quit Drinking

Baclofen 47.9 + 4.6 56.8 + 5.2 Cut down 14.5 + 1.4 8.2 + 1.6

Placebo 42.7 + 4.7 13.5 + 5.8 Quit Smoking 11.0 + 0.8 10.7 + 0.8

Adverse Events

Proportion of participants experiencing adverse events across 17 categories (and collapsed across these categories) were examined. There were no significant differences among the two groups (Table 7S). A patient assigned to Placebo presented with clinically significant high blood glucose levels. The medication was stopped at Week 06 and patient was kept in the study receiving only Medical Management (MM). Another patient assigned to Baclofen reported clinically significant sedation. The medication was kept at the dose of 40 mg/day but sedation persisted. The medication was stopped at Week 06 and the patient was kept in the study receiving only MM.

Table 7S. Adverse Events for Placebo and Baclofen.*

Placebo (%) Baclofen (%)

Sedation or Drowsiness 27 47

Skin Rash or Itching 0 0

Shortness of Breath 13 7

Problems Urinating 0 7

Constipation 7 0

Dizziness 0 13

Sensory Changes 13 20

Pains/Aches 27 20

Upper Respiratory Problems 33 47