Editorial Note

At the heart of The Antidepressant Web is a conundrum. When someone stops taking a drug they have used for some time, and then seem to get ill again, is it a withdrawal reaction or is it relapse? Has the user actually become ill again, or are drug withdrawal symptoms making them seem (and feel) ill?

Sometimes the difference is clear enough. If one took morphine for several weeks to control severe pain, the symptoms felt on withdrawal would be different from the feeling of pain. But what if you were prescribed a sleeping pill, when the most prominent effect of withdrawal was insomnia?

The confusion goes back a long way and is found with many drugs, especially sedatives and hypnotics. It partly explains the failure to spot dependence on benzodiazepines (eg Valium/diazepam), for many years. BDZs were usually prescribed for anxiety and stress - so if people felt anxious and stressed when they tried to stop taking them, it was easy to conclude that the drugs worked and were worth taking. World-wide, billions of prescriptions had been written before it was officially recognised that this compounded the dependence, worsening symptoms on withdrawal. The BDZs didn't really go on working at all.

Apart from this, the problem was that the powers-that-be didn't want to know, and went to some lengths to deny the problem could exist. Leading experts from government and industry produced large amounts of lack of evidence to show that dependence was non-existent or rare. Problems were seen as exceptions to the rule, and often attributed to patients with 'dependence-prone personalities', or to sensational reporting by the media. The first serious warnings about benzodiazepine dependence started to appear only after legal action began.

Plus ca change? Now the same seems to be happening with antidepressants; the drugs differ but the issues are much the same. Lessons from the past still haven't been learned: drugs prescribed for psychic distress often produce psychic distress on withdrawal. Users more than doctors seem familiar with the general idea that what goes up, must come down

This time, however, the problem is hidden in more complex ways - as described in The Antidepressant Web. Over and above the uncertainty about "withdrawal versus relapse", several developments dating from the early 1990s also play some part:

1. The concept of "dependence" was completely redefined - so that now, neither BDZs nor SSRI antidepressants are considered drugs of dependence, on the strength of withdrawal symptoms alone. The new, approved definition of 'dependence' approximates to frank drug abuse - and on this basis, doctors tell patients there is no dependence risk. This would also partly explain why, after 40 years of use, apparently most doctors do not appreciate that antidepressant withdrawal symptoms actually exist.

2. Treatment guidelines have been radically revised - again through processes in which commercial interests were never too far away. Ten years ago, prescribers were advised to treat someone who was depressed for perhaps four to six months, before reducing the dosage and tapering off treatment. Nowadays, the official recommendations are for increased dosage and long-term or indefinite use. Thus withdrawal symptoms might never be seen.

3. Mindless understandings have been widely promoted - notably the idea that "depression" (over 300 different varieties) is an organic disease, the result of too little serotonin in the brain. The implication is that people with depression need drugs that supply the missing molecules, just as people with diabetes need insulin. The general idea is that, if there is some sort of 'dependence', it is entirely benign.

Such developments hardly promote "evidence-based medicine", or is that part of the grand delusion too?

Neuroleptics as well? An important new study has examined the relapse-versus-withdrawal problem in drugs used to treat psychoses, mainly schizophrenic behaviours. Tranter & Healy (1998) find there is strong evidence that a discontinuation syndrome exists, though nothing like enough to be able to pinpoint how much of a problem it is.

They reached this conclusion after reviewing much evidence on different fronts, including a review of studies in which neuroleptic drugs had been used to treat non-psychiatric states. These are revealing because, if "psychiatric" symptoms appear when a drug is discontinued, they must be true withdrawal symptoms (as there cannot be relapse). The authors point, for example, to long-neglected evidence from the 1960s, when chlorpromazine (the archetypal neuroleptic drug) was used experimentally to treat TB:

"In the late 1950s, it was discovered that chlorpromazine was tuberculostatic, in vitro. Accordingly, it was given to patients with tuberculosis by Hollister and colleagues. Following six months of treatment with chlorpromazine daily in a placebo-controlled double-blind protocol that produced little effect on the tuberculosis, treatment was discontinued. This led to a withdrawal syndrome in five of 17 subjects. The syndrome was characterised by nausea, vomiting, restlessness and sleeplessness; it could be mitigated by restarting chlorpromazine. This study was the first to show withdrawal effects to an agent that had no abuse potential and also the first to show withdrawal at therapeutic doses"

Overall, this review has profound implications, many relating to issues discussed on this website. Its conclusions mean that the long-term effectiveness of neuroleptics may need to be thoroughly re-evaluated. If withdrawal symptoms have invariably been interpreted as evidence of relapse, the effectiveness of these drugs would have been greatly overestimated over the years.

These findings also suggest some need for more immediate reassessment of risks and benefits for individual patients. Antipsychotic drugs not infrequently produce ghastly side effects. If, for some users, they are also 'once on, never off' drugs, they need to be all the more selectively and sparingly used.

There is clearly some risk that Tranter & Healy's paper will promote a rush for the earplugs, because it threatens major change to established wisdoms and methods of treatment. It also threatens products. If these authors have got it wrong, there will surely soon be a rapid, well-orchestrated and crushing response. Otherwise, a fair measure of the relevance of this paper might be the depth of the silence, the completeness of the lack of any coherent critical response.  

Data sheet communication of risk This new paper also provides an opportunity to contemplate the activity and effectiveness of the regulators. How good are the official warnings that such risks for patients might be there?

A quick check reveals that, of 23 current data sheets for products to which the above advice would apply, 13 neither warn of the dangers of withdrawal, nor advise gradual discontinuation. These include the data sheet for Largactil, the original brand of chlorpromazine, on the market for over 40 years.

However, one would be hard put to say that some of the advice provided is better than none - notably because of the tendency in data sheets to promote the idea of low risk, not on the basis of any sensible estimates, but on the strength of lack of published reports. The net result would be the recycling of misunderstanding, but the MCA/CSM permit it because they prepare risk estimates in the same way.

"Abrupt discontinuation of high doses of antipsychotics has very rarely resulted in acute withdrawal symptoms, including nausea, vomiting and insomnia. Gradual withdrawal is advisable." (Orap/pimozide, Serenace/haloperidol)

"As with all major tranquillisers … there may be delay before recurrence of symptoms after stopping treatment. Gradual withdrawal from high-dose treatment is advisable." (Stelazine/trifluoperazine)

"Acute withdrawal symptoms including nausea, vomiting and insomnia have very rarely been described after abrupt cessation of high doses of antipsychotic drugs. Relapse may also occur and gradual withdrawal is advisable." (Haldol/haloperidol)

"If drugs are withdrawn, recurrence of symptoms may not become apparent for several weeks or months" (Modecate/fluphenazine; Clopixol/zuclopenthixol)

Such warnings leave much to be desired. In labelling language, the term "rare" is used to describe an event with a frequency of between 1/1000 and 1/10000 people affected. Such estimates are misleading, if the British National Formulary is anything to go by.

The BNF is the classic paperback source of drug information for doctors and, helpfully, it does spell out the dangers of abrupt withdrawal, and acknowledges the existence of a neuroleptic withdrawal syndrome. What is missing is the idea that there might be confusion between the two, and that "relapse" might be no such thing:

"Withdrawal of drug treatment requires careful surveillance because the patient who appears well on medication may suffer a disastrous relapse if treatment is withdrawn inappropriately. In addition, the need for continuation of treatment may not become immediately evident because relapse is often delayed for several weeks after cessation of treatment …

"Withdrawal of antipsychotic drugs after long-term therapy should always be gradual and closely monitored to avoid the risk of acute withdrawal syndromes or rapid relapse".

So far, the MCA/CSM have been cool to the idea that some joyless physical dependence might explain a good deal of psychiatric drug prescribing. It seems worth writing to them, even so.


R. Tranter, D. Healy, Neuroleptic discontinuation syndromes, J. Psychopharmacology, 1998, 12(3), 306-311


Charles Medawar
25 August 1998


Contents page