Two current British government campaigns pull no punches about urging all smokers trying to quit to use drugs. One puts it bluntly: “Don’t go cold turkey.”

Another poster on display in the nation’s waiting rooms says: “There are some people who can go cold turkey and stop smoking. But there aren’t many of them.”

That statement is manifestly incorrect and an inquiry should be undertaken into how such nonsense was approved for publication. In 1986, just a few years after nicotine replacement therapies became available, the American Cancer Society stated: “Over 90% of the estimated 37 million people who have stopped smoking in this country since the Surgeon-General’s first report linking smoking to cancer [1964] have done so unaided.”  How did they possibly manage to do it without drugs?

We have long known that if you survey ex-smokers and ask them what strategy they used on their final, successful quit attempt, around two-thirds to three quarters answer “cold turkey”. This was the case in the early days of NRT, and it remains so today.

In a national US survey of 29,537 smokers, of  those who had quit in the past 12 months for more than four weeks, unassisted cessation produced more than double the number of successes than all other methods combined. Yet it continues to be denigrated by those promoting pharmaceuticals as having the worst success rate.

But the notion of the quitting attempt requires careful scrutiny. Millions around the world make quit attempts each year. Some are serious attempts, but others are half-hearted, brief and quickly forgotten. Mainly of these attempts barely deserve the name, so if they are entered into success estimates, unassisted cessation can appear to do badly.

The much-telegraphed claim that pharmaceutically assisted cessation doubles or triples your chances of quitting derives from a large bedrock of clinical trial data. But there are important differences between trying to quit when in a clinical trial and  being a smoker trying to quit out in the “real world”. Some examples:

  • Trialists have frequent contact with  researchers trained in cohort retention. This creates Hawthorne effects (effects caused by the attention paid to you when being researched);
  • Trial participants are unrepresentative of the general population
  • Cessation trials exclude many people, including  light smokers and those with mental health problems who are heavily over-represented among smokers. This removes many “hard cases”, flattering clinical trial effects.
  • Trialists complete their drug courses  far more than in real world use
  • NRT trials have poor blindness integrity. Over half of studies in one review showed trial participants were significantly more likely than chance to accurately guess that they were allocated to the placebo arm, meaning that their faith in the treatment they received was likely to be poor. This would tend to exaggerate the differences between placebo and active NRT.

All of these combine to produce inflated success rates in trials that are often not reflected in real world quitting.

An important illustration of this has just been  published in the British Medical Journal’s Tobacco Control. The study examined an important simple question likely to be on the minds of many wanting to quit: if you follow a group of smokers who have quit smoking using different methods for two years after they have stopped, which method  produced the best long-term quit rates?

The Massachusetts study found that after two years, those who used NRT to quit had relapsed at the same rate as those who quit on their own. This has caused a storm among smoking cessation leaders, many of whom have long histories of engagement with pharmaceutical companies. (One review showed that industry sponsored trials produce better outcomes than those conducted without ties).

But none of the reactions are so revealing as that from the Association for the Treatment of Tobacco Use and Dependence, representing nearly 450 tobacco treatment specialists. Their statement emphasised two arguments.

The first acknowledged that in population studies of cessation (as opposed to clinical trials): “Studies have shown that those who chose to use NRT have more past failures, more dependence, etc and thus should have lower quit rates. This bias, in which the more severely ill subjects receive a treatment and the less-ill do not is known as “indication bias”.

This is a clear admission that those who chose to use NRT in real world quit attempts often have poorer quit success than those who try to quit unaided.