This is part one of a two part post. The image is from the opening sequence of Raiders of the Lost Arc and shows Indiana Jones being chased by angry natives with the shakiest of means of escape – like trying to escape in a Ford Pinto which risked bursting into flames. A Fireser 747 Max 8 might have been more up to date option. Part two next week.
When vaccines and drugs are approved, these days, regulators state they have a favourable Benefit-Risk ratio. Regulators have only made statements like this for approximately two decades. Prior to that, treatments were approved on the basis of an established treatment effect with some evidence for a likely acceptable safety profile.
Benefit-Risk ratios are now, according to regulators, established in Randomized Controlled Trials (RCTs), also called clinical trials. RCTs are now, and have been since 1962, essential to approval, but there has been a shift from using them to establish a treatment effect to viewing them as laying the basis for a Benefit-Risk ratio.
Approval is not granted on the basis that this drug acts on the serotonin system or that vaccine produces Spike proteins. These biological effects may be true but the observed effects on people in RCTs in the real world are given primacy. The biology may help explain those observations but is not a basis for approval, other than in exceptional cases.
The approval process requires pharmaceutical companies to run RCTs, the results of which are submitted to regulators (FDA, HealthCanada, EMA, MHRA, and TGA). These bodies may regulate Food and Drugs, as FDA does, or medical drugs and devices but not food (MHRA).
The pivotal (key) RCTs are ordinarily published in medical journals, usually a prestigious journal if the treatment is noteworthy. These publications are the primary means of conveying key details of a drug to doctors who as learned intermediaries are entrusted with responding to queries from patients about their conditions and possible treatment benefits and risks.
For both drugs and vaccines, the RCTs that lead to an approval are viewed as offering a favourable Benefit-Risk ratio. Individuals clearly can opt not to consent when the taking of a particular drug like an SSRI. Regulators believe that across a population the benefits will outweigh the risks and since a set-piece pitching Company RCTs against convincing Case Reports of Treatment Induced Suicidality on SSRIs in 1991, they invoke a fear of deterring people from seeking a benefit that RCTs promise to explain their caution about warnings about hazards on the basis of Case Reports.
Companies rather than regulators decide whether to warn about a hazard. They may do so because the hazard is serious, or because it is one that can be avoided if warned about, or when it is one that induces a frisson that might increase sales. Companies write drug labels. Regulators review these for wording that is not supported by evidence, but do not write them.
RCTs have developed a reputation as a good way to manage all potential sources of bias or confounders that may compromise the evaluation of a treatment. This reputation is misplaced according to Sir Austin Bradford Hill, who developed and ran the first RCTs, and Dr. Louis Lasagna, who introduced RCTs to the regulatory apparatus in 1962 to help establish whether treatments are effective. From the start, companies have been to the forefront in encouraging doctors to practice in accordance with evidence that only they have the resources to produce.
RCTs can manage some confounders in very limited settings. This management, however, involves a trade-off. A drug or vaccine can have a hundred or more significant effects. Selective Serotonin Reuptake Inhibiting (SSRIs) antidepressants, for instance, have multiple actions other than just their action on the serotonin system. In addition, they affect every bodily system – eyes, ears, skin, bones, muscle, gut, blood, heart, sperm, hormones, allergic responses, urinary system, genitals and other aspects of sexual function, as well as the peripheral and central nervous systems, where they affect appetite, sleep, libido, anxiety levels, memory and mood.
Similarly, it was hoped the Spike proteins, that mRNA agents generate, might lead to immunity but in addition they affect most bodily systems in a manner that does not happen with traditional vaccines.
An RCT necessarily focuses on the effect of primary therapeutic interest, called the primary endpoint. Does this drug affect mood? Does this vaccine produce immunity? For the purposes of a trial, we are not interested in whether an SSRI also has antibiotic properties as several do or whether an antibiotic inhibits serotonin reuptake and might affect mood as some do.
Most (90%) examinations and measurements in an RCT centre on the primary endpoint, with investigators having little time, or space, to record other effects.
For more on RCTs, see
Ideally in this case the Indiana Jones image above would have some role reversal as in the following Professor Pi image to really hit the nail on the head.
From Model to Mandate?
A model in which companies are required to prove a treatment benefit through an RCT has some utility when the task is one of establishing whether a treatment has an effect of possible therapeutic value, where those effects are uncertain because of natural variation or placebo effects. The intention was to hold companies seeking to make money out of us at our most vulnerable to a higher standard than charlatans.
It was once thought that a positive result in one RCT of a medicine would invariably make for positive results in other RCTs of that medicine, but this has not been borne out.
The model however also has a bureaucratic function. It gives regulators a box to tick, or process to adhere to. They can approve a drug where there are two positive RCTs, even if the majority of RCTs done have been negative.
The Primary Endpoint Problem
The model when introduced was not viewed as providing a basis for a Benefit-Risk claim. It does not now provide such a basis.
The basis for a possible claim of a favourable Benefit-Risk hinges on the primary endpoint in a trial offering a significant benefit and being by far the commonest effect of that treatment. Other effects may occur, but they are viewed as rare or events that only appeared after the trial period was over and therefore not missed deliberately.
For instance, in the case of a parachute, saving a life is significant and the life-saving effect is the commonest effect of that device. Limbs might break in the process, but these injuries are less important and less common.
A positive Benefit-Risk balance in other words is obvious. RCTs are not needed. If the primary endpoint in a trial is less common than other effects and less beneficial than the harm these other effects entail, the basis for a positive Benefit-Risk claim disappears. In these cases, RCTs are deployed for their rhetorical value; evidence is used to persuade us something is evident when it’s not.
For instance, SSRIs have been licensed for their effect on mood. This effect is marginally more common than the effects of natural variation or response to placebos on mood. About 1 person in 10 has a benefit from an SSRI that they would not have had from natural variation or placebo factors and this effect may only be apparent after six to eight weeks of treatment.
This might offer the basis for a positive Benefit-Risk claim, provided there were no other more common significant effects. In the case of the SSRIs, there are many more common effects. The most common effect is a numbing of, or irritability effect on, genitals. This happens close to universally (9 out of 10 at least) within 30 minutes of a first pill but was missed in all RCTs of these drugs because of their focus on a primary mood endpoint.
This genital effect can significantly affect sexual functioning. Over time libido falls. These drugs also cause dependence, so a person may be trapped on them unable to make love for years, having to risk withdrawal induced suicidality in order to make love again.
Even if the person stops treatment, the ability to make love and their libido may never return. This is termed post-SSRI sexual dysfunction (PSSD).
These drugs also cause an emotional numbing in about 8 people out of 10. This change can be obvious enough within a day or two of starting treatment to be observable by others – nearly as obvious as the effects of alcohol. This is the effect through which SSRIs mediate the benefit for which they were licensed. But this effect was not detected in RCTs.
The failure to detect this emotional numbing, a ‘chill’ factor, introduces an incoherence into treatment. We give these drugs to treat a mood disorder but do not tell people how they will help bring about this change in mood and so people cannot monitor whether the right things are happening or not. Even if an emotional numbing is present, rather than the irritability or agitation which happens to 1 person in 5, this numbing may be unwelcome to the person on treatment who may opt to stop this treatment and turn to other options and could do so much earlier if properly informed.
SSRIs also cause nausea in up to 1 person in 3, more commonly that is than the hoped-for therapeutic endpoint. In the case of people who stay on treatment, SSRIs double the rates of miscarriage in pregnant women and double the rates of birth defects and behavioural problems in the children born from those pregnancies, and agonisingly for women cause a dependence that can be so severe that women aware of the hazards to a child are unable to stop treatment.
Regulators proclaim these SSRIs have a favourable Benefit-Risk ratio, despite the fact that they can make people suicidal and homicidal, and even though sexual dysfunction, emotional numbing, nausea and other effects are more common than a beneficial mood effect and some of these may be more significant to individuals than even a realized benefit.
Thalidomide offers a striking symbol of the Primary Endpoint problem. It was put into a placebo-controlled RCT, before marketing in the USA, perhaps the first drug to be assessed in this way. It proved efficacious at putting people to sleep and appeared to have fewer side effects than other sleeping pills then available. Regulators and companies, behaving as they do now, would point to the scientific grounds for stating thalidomide has a favourable Benefit-Risk ratio.
mRNA Primary Endpoint
In the case of mRNA vaccines, the primary endpoint of the trial was whether the person caught a Covid infection. If catching significant Covid infections was almost universal and the difference between being treated or not was clear, this might offer the basis for a positive Benefit-Risk assessment.
The commonest effect following dosing is an action of the mRNA agent on the cells. Apparently, the hope is that this will produce Spike proteins which will generate an antibody response which will bring about immunity and that in turn will prevent death or other serious injury.
MRNA may in its own right act as a drug rather than simply as an agent to generate Spike proteins. If mRNA drug effects could be optimised to produce helpful effects such as tackling Covid pneumonia without producing Spike proteins, such a treatment might have a clear Benefit Risk ratio.
(The effects of mRNA agents are more consistent with a treatment than a vaccine model. These agents do not induce immunity or stop transmission as a vaccine ordinarily would).
More speculatively, mRNA agents may change our genetic disposition so that enduring effects occur that may or may not be helpful or that may be transmitted to the next generation of children. Earlier this month an article appeared confirming transgenerational effects for the anticonvulsant valproate. Only time will tell whether such effects are realized but they are a significant possibility and Benefit-Risk ratios that discount such effects are incautious. In addition if these agents provide a treatment rather than a vaccine effect, many people like pregnant women might opt not to be treated.
MRNA agents do produce Spike proteins, in greater amounts and from more bodily locations than expected, with a wider range of actions than just generating an antibody response.
Generating an antibody response moreover is more complex than most of us once thought. The antibodies the Spike protein generates can be neutralising or enhancing antibodies and they are IgG antibodies rather than the IgA antibodies found in the lung in respiratory infections, which would on the face of it be a better bet.
Like SSRI induced emotional numbing, Spike proteins have significant effects, other than a hoped-for immune effect. They cause many deaths, and injuries like myocarditis, pericarditis, thromboses and neurological problems after vaccination. Our focus on a hoped-for immunity means that the full range of effects of these proteins, like the full range of effects of emotional numbing, remains unknown and commonly missed in clinical practice as a result. Because these effects were not reported in the clinical trials, they are not therefore on clinician’s radars.
In summary, there are many more common effects than the apparent benefit on the hoped-for primary endpoint. As with SSRIs, therefore, it is not possible to say that the RCTs for the current Covid vaccines offer a favourable Benefit-Risk ratio. There are grounds for concern that there are more common, or at least as common, serious effects than there are benefits but these have not been and are not being looked for.
There are more deaths in the active treatment arm of these trials than in the placebo arms.
In contrast, Pfizer are claiming an 89% reduction in deaths and hospitalisations for their protease inhibitor nirmatrelvir-ritonavir combination compared to placebo in a trial that only had to recruit 2,246 patients to demonstrate this benefit.
While there may be a treatment rather than a vaccine effect from mRNA agents that can be of benefit in cases that develop serious complications, there are also several times more reported deaths from these agents in one year than from all vaccines combined over a twenty-year period. These are reports to a system that heavily under-reports serious events like death – at best 1 death in 10. No-one is putting these mRNA deaths in a benefit-risk scale to see where the balance lies.
Besides death, the other key index of benefit for a ‘vaccine’ would be sustained immunity to infection and a greatly reduced risk of transmission to others. Even Messrs Gates and Fauci agree these agents do not give us these outcomes.
The Surrogate Outcome Problem
In SSRI trials, there is a higher death rate in the SSRI arm of trials than in the placebo arm and a higher suicidal event rate on active treatment than on placebo, when one might have expected a licensed treatment with a positive Benefit-Risk ratio to save lives and get people back to work.
The answer to this conundrum is that SSRIs were licensed on the basis of rating scale changes. Rating scale changes are surrogate outcomes, which we expect to reflect processes that will lead to the benefits we hope for such as lives saved, or people restored to functioning.
In similar fashion, we license statins on the basis of lowered cholesterol levels rather than any benefit on mortality. We license hypoglycemic agents on the basis of an ability to lower glucose, even though this commonly leads to more hospitalizations than diabetes. The drugs for osteoporosis are licensed on the basis of an ability to thicken bones rather than reduce fractures with the result that there are now more serious fractures of abnormally thickened bones in the elderly after falls than there ever were before these drugs were introduced.
We have the same situation with the mRNA trials. The ability to reduce infection rates would not ordinarily be thought of as a surrogate marker but reducing the risk of death or of infection severe enough to lead to hospitalization are the hoped-for outcomes. If these do not happen, then reducing apparent infection rates may even be a bad thing. Abolishing symptoms like cough may make infected younger people more likely to kill their elderly relatives.
The hoped-for outcomes on hard endpoints like death were not measured systematically because they are too rare even for a trial with 40,000 entrants to pick up differences, which is an ironic comment on the severity of the infection. Covid is not Ebola.
Between the primary endpoints and surrogate outcomes that trials need, as Bradford-Hill recognised RCTs are almost by definition not a good way to evaluate a drug or vaccine. They can establish that there is a potential for benefit. They do not provide a basis to establish a Benefit-Risk ratio. When there is a benefit in terms of lives saved as clear as there is with parachutes, there is no need to run RCTs.
Trials can be run for two reasons. One is to demonstrate that we know what we are doing, in which case, as with parachutes, we would expect to get the same result every time. There is no treatment that regulators approve that meets this criterion.
The other reason to run trials is when we do not know what we were doing. In this case, even with 40,000 patients recruited and trial designs that enable a lot of deaths to vanish, we still ended up with more dead bodies in the treatment arm of mRNA trials than in the placebo arm.
Not knowing what we are doing, and not getting a clearly favourable result, we nevertheless mandated the administration of these agents because of their favourable Benefit-Risk ratio.
The Conduct and Publication of Trials
As of the 1980s, companies began to outsource the running of their clinical trials to Contract Research Organizations (CROs) and the medical writing of reports of what those trials showed to medical writing companies. The medical writing side of the business is regularly called ghostwriting because the authorship line of these papers will list academic physicians but not the writers who wrote the drafts and even wrote the cover letter for the submission of the paper to a medical journal.
The upshot of this outsourcing was that in the interests of speedy completions CROs took trials to Eastern Europe, India, Africa and South America. The pressure to get trials done quickly meant that the patients didn’t always exist. North American academics nevertheless remained the named authors on these trials although they had never collected a scrap of data or met a single patient.
Trials also became larger and multinational. Larger does not mean better. Weaker medicines need larger trials to demonstrate an effect. It only takes a trial of 12 people to show that the genital numbing SSRIs cause can make them a treatment for premature ejaculation, whereas it takes several hundred people to have a hope of finding an antidepressant effect. Even snake oil can demonstrate a benefit with large enough numbers and the right outcome measures.
The running of trials required CRO personnel to shuttle between an increasing number of increasingly far-flung centres. This created a natural drift to store the master copy of all records in one location. An academic investigator, who was increasingly unlikely to have seen the patients anyway, no longer had the records from the whole trial that s/he could analyse. At best s/he had copies of a small subset of files.
The CRO transcribes the data from the master files into spreadsheets. For instance, in Study 329, where paroxetine (Paxil) was compared to placebo in depressed adolescents, there were 77,000 pages of records, Clinical Report Forms (CRFs), with each subject in the trial having 300 or so pages of trial records. The data from CRFs, along with various protocols took up 5000 pages of appendices, accompanying an 800 page Clinical Study Report (CSR). In many instances, data goes missing in the transcription process. Particular problems can end up effectively concealed under esoteric coding rubrics, such as emotional lability for suicidality.
The 800-page CSR gives a company view of what this trial has shown. The final medical journal publication from this trial may then amount to 10 pages.
Company submissions to regulators centre primarily on the Clinical Study Report. The 5000 pages of appendices will almost certainly accompany these but are rarely looked at by regulators. The 77,000 pages of Clinical Report Forms may be notionally available to regulators but are never looked at by them.
If a problem blows up about a drug, regulators will ask the company to review their material and report back to them. They will not go to their own archive and see what they make of what is there. Even if the 77,000 pages were there all names are redacted so the regulators can never contact the person behind the record to find out what in fact actually happened – or even establish if they actually existed.
Regulators never see the bedrock data from RCTs. They do not have access to either the figures in the CRFs behind the figures appearing in the appendices or the people from whom those figures come. The regulators of medicines have for three decades depended on the integrity of company submissions. The apparent success of this practice may underpin the recent turn of air safety regulators to depending on Boeing’s submissions on its planes as the sole basis for approval.
Prior to the 1980s, clinical trials of medicines had usually been run in a small number of university hospitals, and the clinicians involved knew the patients and had access to and analysed the data. When an RCT was brought into a legal setting as evidence, the Court could bring an expert into the proceeding to answer questions about the effects of a drug on real patients who could potentially also be called to testify. This understanding made it sensible to grant the evidence from RCTs a Hearsay exemption. It no longer makes sense.
The changes that were taking place from 1990 onwards led New York State to lodge a fraud charge against GlaxoSmithKline in respect of an RCT of paroxetine done in depressed teenagers. This trial, Study 329 was carried out in the mid-1990s and published in 2001. It claimed the drug worked well and was safe. Documents emerged that made it clear that GSK knew the drug did not work but the company intended to pick out the good bits of the study and publish them. This laid the basis for a fraud charge and later a $3 billion fine from the Department of Justice.
It transpired that the entire literature on the results of the use of antidepressants in RCTs for depressed children at that point was ghostwritten,
There was a complete mismatch between what the ghostwritten articles said and what the data showed when accessed.
GSK told FDA that Study 329 was negative, as were their other two studies. FDA acknowledged this and still issued an approvable letter for paroxetine for the treatment of depressed adolescents, and in addition agreed to GSK’s request not to mention that these studies were negative in the label of the drug. Other regulators almost certainly did exactly the same thing.
Study 329 was not a bad study. It was conducted to a higher standard than the recent vaccine trials. What this points to is that, quite aside from the limitations of RCTs, fraud is the de facto company modus operandi when it comes to their RCTs.
(What companies have been doing is not fraud in terms of business processes. No business rules are intentionally broken. In terms of the norms of science, however, the company trials are fraudulent. The practices adopted are designed to maximise commercial advantage rather than adhere to scientific norms, but the marketing rhetoric summons us to follow the science and as doctors to prescribe and as patients to take).
In the case of the RCTs for the Pfizer mRNA agent, this was run by a CRO – ICON. There is a published BMJ article on how shoddy some of the recruiting centres were. There is every chance this shoddiness, already put on the record in the BMJ, and uncontested, is nothing compared to what has happened in other centres in which this trial was run that have not yet been looked at.
The published articles reporting the trials of mRNA and related agents were ghostwritten. These published reports are patently misleading in terms of the conduct and findings of these trials. The harms that occurred in these trials have been egregiously written out of the script.
As things stand these agents cannot be given or taken with informed consent.
Mandates vitiate consent. If mandated to take a treatment we have to trust the authorities. In this case, the authorities never had the data and never will have it. In terms of what has been submitted to FDA and other regulators (company reports of what their trial has shown and correspondence between regulators and companies in respect of the trial), FDA have blocked access to this for 55 years and are trying to have that period extended.
Were the material that FDA is blocking from release available we still would not have the raw data. The bedrock information on which consent should be based is not available to anyone. The authorities have not been able to review this material before instituting mandates.
They have not mandated a recording of or assessment of the hazards of treatment following the requirement for all of us to be treated. See Mandates Yes and No.
Consent can be given on the basis of trust but with a track record of fraud and regulatory willingness to turn a blind eye to that fraud, there are no grounds for trust either.
Re Ghostwriting see
- Clinicians do not have access to clinical trial data on medicines or vaccines.
- Close to all of the medical literature reporting trial results for on-patent drugs and vaccines is ghostwritten, hyping the benefits and hiding the harms.
- Clinical trials of these treatments that are negative on their primary or their most common outcomes are often published in prestigious journals as positive.
- Clinical trials have their harms airbrushed out of ghostwritten publications.
- Regulators (FDA, Health Canada, MHRA, EMA) do not get to see the full trial data.
- Regulators approve treatments as working even when more people die on active treatment than on placebo.
- Regulators approve medicines on the basis of negative studies and agree not to let the wider world know about this.
- Regulators say nothing when companies publish negative studies as positive and make adverse effects of treatment, including death, vanish.
- For many trials there are more deaths on active treatment than on placebo, but this does not lead regulators to warn about hazards as to do so would in their stated view deter people from seeking a benefit (even when the benefit is better characterized as a commercial benefit to a company rather than a benefit to the individual in terms of a life saved or a restoration of function).
This pre-vaccine summary and the entire document, along with the material next week has been sent to
- Joe Biden – Love in a Time of Covid,
- the BMA,
- Sajid Javid, Minister for Health in Wales, threatening mandates
- Eluned Morgan, Minister for Health in Wales,
- Robin Swann, Minister for Health in N Ireland,
- Daniel Sokol, lawyer with an interest in ethical issues, who has written in favour of mandates,
- Jeff King, lawyer who has written in favour of mandates,
- the Canadian Health Minister,
- the Chief Public Health Officer for Ontario
- the Ontario Human Rights Commission
- the College of Physicians and Surgeons of Ontario (CPSO)
- the President of McMaster University
- the Dean of Medicine at McMaster University
There have been others. Some of the above have responded at various points (4,5,6,7,9). No-one has demurred on any of these points (except perhaps JK). The CPSO (13) did however make a marvellous contribution that will feature in Part 2 of this post.
There is an ironic background here. In 1979 Peter Medawar coined the term Conspiracy of Goodwill to indicate that pharmaceutical companies, governments, doctors and we who take medicines all want them to work. He was explaining that RCTs were our best antidote against being seduced into this conspiracy.
Charles Medawar and Social Audit resurrected the idea of a Conspiracy of Goodwill around 2004 but pointed to the ambiguities in the term that were more obvious by then. Even he however did not get to the idea that far from being an antidote, RCTs might be the key that locks the Iron Cage around us.