The drug for 'mild alcoholics' - possibly slightly better a placebo

I've been looking at the evidence for nalmefene, the drug that is to be given to 'mild alcoholics' under the name Selincro (see previous post). I'm surprised by how weak it is. The London New Drugs Group/London Medicines Evaluation Network Review looked at "three multi-country European phase 3 clinical trials". In each case, drinkers were split into two groups, one of which was given nalemfene, the other was given a placebo. Both groups were given counselling.

Here are the results...

ESENSE 1 was a 6 month, randomised, double blind, placebo-controlled, parallel group study in 604 adults (placebo n=298, nalmefene n=306), mean age 51.6 years, (67% men). For the co-primary outcome, nalmefene was significantly superior to placebo in reducing the number of heavy drinking days (-2.3 days/month [95% CI: -3.8 to -0.8]; p=0.0021) and total alcohol consumption (-11.0 g/day [95% CI: -16.8 to -5.1]; p=0.0003) at month 6 when compared to placebo.

The mean number of heavy drinking days in the nalmefene group decreased from 19 to 8 days/month, and the mean total alcohol consumption decreased from 84g to 33g per day. In the placebo group, the mean number of heavy drinking days decreased from 20 to 11 days per month and the mean total alcohol consumption decreased from 85g to 45g per day at month 6.

In short, the drug was associated with a reduction in alcohol consumption of 61% while the placebo was associated with a reduction of 43%.

ESENSE 2 was identical in design to ESENSE 1, included 718 adults (placebo n=360, nalmefene n=358), mean age 44.8 years, (73% men). With regard to the two primary end points, at month 6, nalmefene was statistically superior to placebo in reducing heavy drinking days (group difference: -1.7 days/month [95% CI -3.1; -0.4]; p=0.012) but did not achieve a statistically significant reduction in total alcohol consumption (group difference: -5 days last month [95% CI -10.6; -0.7]; p=0.088). 

The mean number of heavy drinking days in the nalmefene group decreased from 20 to 7 days per month, and the mean total alcohol consumption decreased from 93g to 30g per day. In the placebo group, the mean number of heavy drinking days decreased from 18 to 8 days per month and the mean total alcohol consumption decreased from 89g to 33g per day at month. Adverse events were more common in the nalmefene group however; the incidence of adverse events leading to withdrawal was at the same level as placebo.

In this study, the drug was associated with a reduction in alcohol consumption of 68% while the placebo was associated with a reduction of 63%. The difference between the two outcomes is not statistically significant.

SENSE was a 12 month study in 675 adults (placebo n=166, nalmefene n=509), mean age 44.3 years, (77% men). 422 of 665 treated patients (63%) completed the study. The reasons for this high drop-out rate are not clear from the paper, which is not fully published. Nalmefene did reduce the number of heavy drinking days and the total alcohol consumption over the study period and at most time points this favoured nalmefene. However, it did not achieve a statistically significant effect when compared to placebo at the primary endpoint of reduction in heavy drinking days and total alcohol consumption at month 6.

At month 13 the mean number of heavy drinking days with nalmefene had decreased from 15 to 3 days per month (15 to 6 in the placebo group) and the mean total alcohol consumption from 75g to 16g per day (75g to 27g per day in the placebo group).

In this study, the drug was associated with a reduction in alcohol consumption of 79% while the placebo was associated with a reduction of 64%. Again, the difference between the two outcomes is not statistically significant.

Perhaps NICE thinks that something is better than nothing, but considering the cost of nalmefene (£3 per pill) and the apparent effectiveness of counselling, we shouldn't expect too much from a drug that made no statistically significant difference in two out of three studies and made only a fairly modest difference in the other.

In any case, the widely reported claim that nalmefene can "cut alcohol consumption by 61%" needs to be seen in the context of the placebo which "cut" alcohol consumption by 50%. And that's only if confine yourself to the more optimistic evidence.