Note: This is the fourth in the Lasagna series of posts â Not So Bad Pharma,  April Fool and Tragedy. It will be followed by The Empire of Humbug: Not so Bad Pharma, Brand Fascism & Witty A: Report to President.
The First RCT
In 1956, two of the creators of the modern RCT, Lou Lasagna and Michael Shepherd, met. The randomization in randomized placebo controlled trials came from Bradford Hill in Britain and placebo controls from Beecher, Gold and Lasagna in the US. In 1956, Michael Shepherd, Bradford-Hillâs representative in all things psychiatric, came on sabbatical to stay with Lasagna at Hopkinsâ.  [Lasagna is on the left].
In 1947 Bradford-Hill ran the first randomized trial. In 1955, Shepherd added in Lasagna and Beecherâs placebo to the original randomized design and published in the Lancet the first ever RCT of the type used today in drug trials submitted to FDA for regulatory approval. His trial compared reserpine to placebo in a group of outpatients with mixed anxiety and depression â similar patients to those later given Prozac and other SSRIs. Reserpine was better than Prozac and other SSRIs later were, and indeed did better than imipramine did a few years later in a trial run by Lasagna. There was no sign in Shepherdâs trial of the suicide hazard for which reserpine is now best remembered.
The two articles that immediately preceded Shepherdâs study in this issue of the Lancet described hypertensive patients being treated with reserpine who became suicidal. When later asked why his trial had almost no impact, Shepherdâs off-the-cuff response was that doctors were not used to seeing results being presented in this way. But there was in fact no commercial incentive for any company to use his results to relegate the accounts of suicide on reserpine to the status of anecdotes.
Evaluating Drugs
Just as Shepherdâs paper was published in 1955, chlorpromazine was taking the American market by storm and looked set to change psychiatry as much as penicillin had changed medicine. This earthquake in therapeutics led to a landmark Evaluation in Psychopharmacology meeting convened by Jonathan Cole and Ralph Gerard in Washington in September 1956, at which Lasagna was a key note presenter. Shepherd attended.
Critically just 6 years before it became a legal requirement to use controlled trials and only controlled trials to evaluate drugs, there was no controlled trial section to these proceedings. Instead, there were notes of caution about clinical trial methods from the grandees in the field â positions that years later Lasagna endorsed.
Fritz Freyhan criticized a naĂŻve faith in technical objectivism:
âThe time, we hope, has not come when the clinician abdicates and the rating-scale marker takes over as a judgeâ. âWhile one can easily spot the potential and actual shortcomings of nonexperimental studies, it appears to be tremendously difficult for multidisciplinary groups to acknowledge limitations and sterility in rigid experimental designâ.
Nate Kline said a rating scale approach to clinical evaluation risked creating a rabbit-out-of-the-hat illusion. Pulling a rabbit out of the hat depends critically on getting the rabbit in there first â which was what rating scales did for drugs.
Ed Evarts noted that if chlorpromazine and controlled trials had been invented a few years earlier, chlorpromazine would have been shown to âworkâ for dementia paralytica (neurosyphilis) and its use would likely have inhibited recognition of the fact that penicillin was a cure.
Shepherd and Lasagnaâs faith in controlled trials was a minority taste within psychopharmacology.
The two men born in 1923 had contrasting styles. Shepherd was aloof, drew lines in the sand and made enemies. Lasagna was sociable, worked to find areas of agreement and retained the affection of many who took opposite points of view. Shepherdâs insistence that controlled trials were absolute embroiled him in a set of bitter disputes about lithium that derailed his career. Lasagna viewed controlled trials as a pragmatic option, but his pragmatic insertion of controlled trials into the 1962 amendments led to a bureaucratic absolutism that also derailed his career (see Tragedy).
Although not a psychiatrist, Lasagna ended up with a wider following within mental health than Shepherd for two reasons â his work on placebos and on rating scales.
The Humble Humbug
While Lasagna was not the creator of the placebo, Gold, Beecher and others did earlier work (see Tragedy), his 1954 paper with Beecher ended up in Medicineâs top 30 cited articles ever and put placebos into RCTs. The placebo for most people, lay or medical, at this time was a form of hypnosis â a bridge between psychodynamics and pharmacology.
The hypnosis myth was fostered by a 1954 editorial on the placebo in the Lancet entitled The Humble Humbug. The phrase stuck. At a time when magazines were emblazoned with titles about miracle drugs, a Readersâ Digest piece in Sept 1956, coinciding with the Washington meeting, ran a piece on âperhaps the most remarkable of the miracle drugsâ under the heading of âMedicineâs Humble Humbug: The powerful role in sickness of mental attitudeâ. By the time the Readersâ Digest piece came out, however, both Lasagna and Shepherd were reaching for a more unsettling understanding of what placebos might mean (see forthcoming Empire of Humbug: Not so Bad Pharma).
The second contribution came in demonstrating that rating scales could be used to test analgesics and hypnotics. This persuaded the field it was possible to objectively study internal mental states and in so doing Lasagna laid the basis for the insertion of controlled trials into psychiatry.
But thereâs a rub. When we think about effectiveness, what do we mean? Without rating scales we might have to show treatments saved lives or got people back to work.
Lasagna was one of the first to appreciate the ambiguities in what seems like a simple question. It could mean â how good is this drug? Alternately it could mean â have the benefits of this drug been proven to be greater than the harms?
In 1962, although Lasagna appreciated these ambiguities more than anyone else, with the thalidomide crisis the pragmatic need was to find some basis for inserting an effectiveness criterion into regulation. The Democrats whom he was advising wanted one. The Republicans and business didnât. The struggle came down to a battle over words â whether the preponderance of the evidence should support the application or whether the existence of a substantial amount of supporting evidence would suffice.
Lasagna came up with a winning formula â the application should be supported by evidence from adequate and well-controlled trials conducted by suitably qualified experts. This became two adequate and well-controlled RCTs, because ordinarily in science if a finding is replicated we can rely on it.
Based on Science
The 1962 amendments to the Food and Drugs Act and in particular the effectiveness criteria form the centerpiece of FDAâs proudest boast, repeated widely, that these amendments were the first example of regulation based on science.
The claim gives rise to a myth â that âscientistsâ from FDA are supposedly keeping the public safe. The myth is found everywhere from academic commentators on the FDA, as in Harvardâs Daniel Carpenterâs Reputation and Power, to views from the popular end of the spectrum, as in Philip Hiltâs Protecting Americaâs Health. There is no questioning of the mantra.
Not since the Wizard fooled the citizens of Oz into thinking that he was keeping them safe from the Witches of the West and the East and all dangers in between have we been so misled.
Controlled trials are based on a null hypothesis. Their early use was to debunk claims that treatments were effective. The trials we celebrate most, such as the Womenâs Health Initiative Study of Hormone Replacement Therapy, continue to be ones which show how powerful this debunking can be. They are not a natural means to test for effectiveness.
If the controlled trial formula had been restricted to rejecting applications for treatments that didnât work, companies who thought they had a good treatment but whose compounds failed in clinical trials might have had to go back and pick out a sub-group of patients the treatment clearly did benefit, for instance.
Testing for a negative is an unnatural frame of mind. The gestalt flips into viewing a disconfirmation of a null hypothesis as evidence that something positive has been demonstrated. When this happens, RCTs become a potentially invalid and certainly ambiguous metric.
The idea that what they were doing was based on science appears to have engendered a bureaucratic hubris. Regulators certainly donât act like they have let a drug on the market because they cannot say itâs not effective.
But what if there have been five studies â two positive and three negative? In both cases the science has been replicated. In this instance the drug is let on the market, as the regulatory criterion has been met. When this happens the regulator demonstrates he is solving a regulatory problem rather than engaging in science and that regulation is about the mindless application of a bureaucratic rule.
As regulating takes place against a background bristling with the legal resources a pharmaceutical company can bring to bear on decisions that are not congenial to their interests, while it was briefly different leading up to the Panalba case (see Tragedy), FDA staff now learn quickly that regulation is unquestionably about the mindless application of a bureaucratic rule.
The rule says that it takes two positive studies to approve a drug. In law, if there were only two positive studies out of 100, the regulator would approve the drug. They have in fact approved applications where only 1 in 5 studies were positive. A variety of justifications are brought to bear on the issue, never found in science, such as the notion of failed trials. The bottom line is the studies in which the null hypothesis is upheld and the claims for drug are not upheld, do not count.
One consequence of this is that there are now tens, perhaps hundreds, of thousands of trials in pharmacotherapeutics, while there may still be less than one thousand trials in surgery and as few in preventive medicine. But where the practice of surgery has made dramatic strides, in many areas of pharmacotherapeutics the field has gone backwards.
Catch 62
Aside from the fundamental confusion between the scientific process and regulation, there are a series of profound misunderstandings that have made the 1962 amendments close to a disaster.
The word effective is a key problem. The regulations call for demonstrations of effectiveness while a great deal of the talk in 1962 featured the word efficacious. Treatments meanwhile also have effects, both beneficial and adverse.
Since 1962, effectiveness has come to be defined as evidence that treatments work in the real world. Virtually no drugs brought on the market demonstrate effectiveness in this sense. Almost by definition effectiveness cannot have been shown to be present whenever a study uses a surrogate marker or a rating scale score as its outcome measure.
Efficacy is used when a treatment does something useful in controlled trials but benefits are not necessarily realized in the real world. Almost all industry trials have to fall under this heading, if only because they come with so many real world exclusions and involve samples of convenience rather real patients.
An effect is an action on a structure or function of the body that may be beneficial and is something that may or may not take a controlled trial to demonstrate.
The 1962 Act uses the word effectiveness. There are other definitions of these words but whatever definitions are used there is no way to semantically solve the issues we are tackling here. The 1962 amendments are fatally flawed.
A subsequent regulation (CF 310.3 (4)) is more ambiguous:
The newness of use of such drug in diagnosing, curing, mitigating, treating, or preventing a disease, or to affect a structure or function of the body, even though such drug is not a new drug when used in another disease or to affect another structure or function of the body.
Randomized placebo controlled trials have not shown any drug within the mental health domain is effective. If a treatment were effective virtually every RCT undertaken would show a positive result.
Some psychiatric drugs are extraordinarily effective, for instance benzodiazepines for catatonia or SSRIs for premature ejaculation. These treatments are so effective that controlled trials are an irrelevance. Every trial conducted would show a positive result. The point here, developed below, is not that it is impossible for a treatment to achieve effectiveness but rather that controlled trials have little useful to contribute to the issue of effectiveness.
The public and legislators have been baited with effectiveness and sometimes efficacy but sold something else that doesnât fit neatly under effectiveness, efficacy or effect. Some of these issues are now the subject of consumer fraud actions.
The Kefauver hearings centered on analgesics, hypoglycemics, tranquilizers and steroids. With each of these drugs there is an obvious effect but is there effectiveness? Hormone replacement therapy has an obvious effect but when a proper study was undertaken it turned out to be ineffective for the benefits claimed for it. For most doctors in the 1960s, Upjohnâs oral hypoglycemic, tolbutamide, clearly worked â it lowered blood sugars. But when a large controlled trial was undertaken in the late 1960s tolbutamide led to more deaths than other forms of treatment â was it effective?
The tolbutamide and HRT trials took 5 or more years to run. No one would ever suggest companies should go through trials like this to get on the market â certainly not Lasagna. This is what trials are for though. They are not for getting drugs on the market. The difference between WHI and NIH trials and company trials not about one set of trials being independent and the other not – itâs about the purpose of the trial.
If a short-term hypoglycemic or tranquilizing effect seems on the surface beneficial, it might be possible to talk of efficacy but this falls short of effectiveness, not because of something mysterious that happens when we move from clinical trials to the real world, as the definitions above suggest, but because while useful in both trials and the real world these effects are not effective in the sense of curative.
Analgesia and tranquilization are therapeutic principle rather than Magic Bullets.
Therapeutic Principle or Magic Bullet?
Part of the problem is that antibiotics have skewed our notions of what drugs should do. They have created a Magic Bullet template. It is all but assumed that if a treatment is approved it maps onto this template when in fact most treatments in medicine are helpful rather than curative â they are therapeutic principles rather than magic bullets. Analgesia, tranquilization, temperature reduction, or all sorts of other things can be helpful within a therapeutic relationship.
The idea of a Magic Bullet suggests a treatment that works regardless of setting or circumstance. Therapeutic principles in contrast have always been recognized to be dependent on constitutional or temperamental factors and on setting. Agreeing that a treatment âworksâ or âis effectiveâ (Magic Bullets: MBs) rather than this treatment has an effect that may be beneficial for you (Therapeutic Principle: TPs) risks condemning many to the wrong treatment for them.
Condemning is too mild a word for the forcible medication of many mental health patients with what were once called tranquilizers (TP) but are now called antipsychotics (MBs).
One consequence of Lasagnaâs 1962 wording has been to move us from a world of Evident Based Medicine, which he agreed with, to Evidence Based Medicine, which he didnât agree with. âEvidence of effectivenessâ is now used to persuade or all but force patients to take and doctors to give treatments even when they are evidently not doing you good (see False Friends).
If trials indicate treatments are effective, itâs in the interest of the State (as well as pharmaceutical companies) to âpersuadeâ or force as many of the citizenry to take treatment as possible, as clearing up diseases should make the population more efficient. To this end, a certain amount of casualties en route can be accepted as collateral damage. In the case of the antidepressants and suicide, regulators again and again made clear that they would not warn of this hazard, in case people might be deterred from seeking treatment, or doctors deterred from putting them on treatment.
Coming back to Lasagnaâs question â what does effectiveness mean? Even when controlled trials show a net loss of life on a treatment, regulators now almost always state that the risk benefit ratio remains favorable. This view can only be sustained on the basis that they believe if the treatment is effective, as for instance in the case of the antidepressants, even if there is a net loss of life from suicide in short term trials, that somehow in the longer run an effective treatment will lead to lives saved.
This is not an evidence based position. It can only begin to possibly make sense on the basis of some belief about an effectiveness that conveniently for regulators is rarely put to the test â the WHI HRT and NIH tolbutamide trials were monumentally inconvenient for regulators but even they didnât lead to the drugs being pulled. Thatâs a medical issue not a regulatory issue.
Truly effective treatments should be self-funding or even lead to a fall in healthcare costs. If lives are saved and people are put back to work, or become more efficient at work, then national economies will perform better and wealth will increase. But not only are healthcare costs increasing exponentially, morbidity and mortality from treatment is increasing so that treatment induced death is heading toward being, may even be, the normative way of dying.
None of this should be happening if our treatments are effective. RCTs, as used by companies, may be efficacious to get drugs on the market but for the most part they arenât effective in leading to better health.
Conversations in the Speakeasy
The irrationality of what has happened can be brought out by comparing antidepressants with alcohol. In the case of imipramine, a much more potent antidepressant than Prozac, Lasagna who ran the first RCT on imipramine recognized that it was difficult to even prove efficacy in an RCT.
In the case of the later SSRIs, there is distinct psychotropic âeffectâ â they emotionally numb. SSRIs âworkâ in the sense that alcohol âworksâ. Using the same testing approach that brought SSRIs to market that supposedly demonstrates their effectiveness we could bring alcohol to market as an antidepressant or anxiolytic if it could be patented so that it was worth our while. All it takes is a result in two of ten short 6-week trials, in which it would have almost certainly produce a âbenefitâ on the right rating scales and with fewer serious side effects than SSRIs â there would not be a doubling of suicidal acts, with up to 5% of patients having suicidal ideation as there is with SSRIs (see Shadow Dance).
Because everyone knows what alcohol does, they can see that such a demonstration for alcohol is not a demonstration of effectiveness, but they seem unable to see that this is also the case for SSRIs, for every other psychotropic drug, and for a majority of drugs in medicine.
If we were to try and bring alcohol on the market, we would do so on the basis of its effect on a structure or function of the body rather than for supposed effectiveness. What would the world have looked like if the same had been done for the SSRIs?
At the end of his career, Lasagna mused âThe days when a drug company [developing a drug] would go to skilled and sophisticated psychiatrists [like Fritz Freyhan] and give them a supply of a new drug and ask them to try it on some different patients seem gone forever. Is this a cause for celebration or depression?â
Before an effectiveness criterion was brought in, under the 1938 Act, several classes of antibiotics, diuretics, antihistamines, hypoglycemics, anticonvulsants, steroids, antihypertensives, major tranquilizers (antipsychotics), minor tranquilizers, analgesics, antidepressants, stimulants, oral contraceptives, the first chemotherapies for cancer and a range of other highly effective or efficacious treatments were brought to the market, all without the help of controlled trials.
Since 1962, it is doubtful if as many effective medications have been brought on the market. It is likely that more relatively ineffective but highly dangerous drugs have been marketed since 62 than before 62. This is an almost completely predictable consequence of what we have done.
(See Empire of Humbug: Not so Bad Pharma. Â âAndrew Wittyâ will also offer a view on what the right wording of Drug regulations should be in Witty A: Report to President).