In the early 1980s, the pharmaceutical industry outsourced the running of its randomized clinical trials (RCTs) to Contract Research Organizations (CROs) and the writing up of RCT results to medical writing companies.
With this, the physicians and academics running trials of drugs and vaccines lost access to clinical trial data. Even the apparent authors on the manuscripts representing the results of those trials no longer had access to and could not interrogate the data or stand as guarantors for the accuracy of the published record of these trials.
Study 329, a study comparing paroxetine to placebo in depressed minors conducted in the mid-1990s, illustrates the changes then taking place. A 2001 article with a distinguished authorship line, in the journal with the highest impact factor in child psychiatry, claimed paroxetine worked well and was safe. It led to a considerable increase in the prescribing of paroxetine to minors.
Based on documents I provided, in 2004 New York State took a fraud action against the makers of paroxetine, GlaxoSmithKline (GSK), which was resolved with a promise to make company trial data more available. This high-profile action led to Black Box Warnings on antidepressants. It made clear that the entire published literature on RCTs of antidepressants in children at that time was ghostwritten – the names on the authorship lines for medical paper reporting the results of RCTs do not write those papers. It also laid bare a comprehensive mismatch between the published claims and the data when accessed.
Study 329 and paroxetine were also the centrepiece of an action the US Dept of Justice took against GSK that was resolved in 2012 with a payment of $3 billion then the largest such payment in corporate history.
Study 329 was not an aberration. It represents standard industry modus operandi as of the mid-1990s. It was performed in the best academic centres with good oversight, and to higher standards than most trials are now.
These legal actions made Study 329 trial data uniquely available. This permitted colleagues and I some years later, to Restore Study 329 and demonstrate the range of tricks companies use in hiding harms.
Most of the tricks companies now deploy to hide harms and fudge efficacy were being worked out at the time this trial took place. Company abilities to hoodwink us have improved greatly since then and there is likely more fraud now than then.
A recent BMJ publication on Pfizer’s vaccine trials illustrates that these trials are affected by the practices that were becoming standard in the 1990s.
Besides company tricks, even if done by angels, as pointed out by Austin Bradford Hill who ran the first randomized trial, RCTs can be helpful in evaluating one of the more than one hundred things every drug does but this, by definition, makes them a poor way to evaluate a drug or vaccine overall.
The need to supplement RCTs with other evaluative approaches can be brought out by a simple example. Before the selective serotonin reuptake inhibiting (SSRI) group of antidepressants were put into clinical trials, it was known these drugs affect sexual functioning in close to 100% of us within 30 minutes of a first pill. But in RCTs of these drugs undertaken for the purposes of establishing the existence of a much less common benefit for nervous problems, the focus on a primary endpoint that RCTs require meant these sexual effects essentially vanished.
The vanishing-effects problem is even greater in current vaccine trials, where participants have been presented with a prepopulated list of a small number of side effects that might happen to them as a result of the vaccine (see New England Journal of Misinformation and appendices 2 and 3). This is like a company claiming an improvement in its consumer satisfaction levels after eliminating the complaints department.
Two issues arise when treatment effects, other than those of commercial interest, vanish. Because of a premium now put on RCTs, companies claim that the only things that happen on a drug or vaccine are things that happen to a statistically significant effect in an RCT. Everything else is anecdotal or psychogenic.
Statistical significance testing, however, should only be applied to the primary endpoint of a trial. It should not be applied to effects that are not being investigated.
As a result, the consent forms for people taking Covid vaccines, in the UK for instance, now present people with a strong steer that the only things likely to happen them after injection are a sore arm, headaches or other aches, fatigue, stomach upset, or a mild fever.
Pfizer knew about the effects of SSRIs on sexual functioning and risk of triggering suicidality before launching sertraline. The company similarly knew about the risks of its vaccine to individuals with pre-existing neurological conditions such as multiple sclerosis, or the more general risk of effects on the nervous system such as Guillain-Barre Syndrome, Bell’s Palsy, demyelinating disorders or transverse myelitis, all of which people should be warned about prior to deciding whether to take the vaccine or not.
A second issue stems from the current misreading of what RCT data demonstrate, namely that it is not possible on the basis of RCTs to establish a Risk-Benefit ratio for a drug or a vaccine.
Regulators routinely claim that the Risk-Benefit ratio favours the approval of this drug or vaccine or not warning about harms, basing this claim on company RCT data and assuming the primary endpoint is by far the commonest effect of treatment with all other effects being rare or idiosyncratic or developing outside the time frame of the trial.
But if RCTs can make effects of a drug or vaccine that are as common or even more common than the primary endpoint vanish, or if other effects are rare or idiosyncratic but serious, then the basis for claiming a favourable Risk-Benefit ratio for drugs or vaccines vanishes.
This clearly is likely to be even more the case for novel mRNA technologies.
There is furthermore no metric, or algorithm, for making Risk-Benefit judgements on a population level. These are a matter of individual decision.
The development of modern therapeutics has put pharmacovigilance at the heart of medicine. While there are epidemiological and other processes that pharmacovigilance can turn to, the central discipline involves a doctor and patient deciding on the basis of an examination of the patient what is happening when the patient reports some change after taking a medicine. The event being considered could be a benefit in which case doctor and patient will opt to continue treatment. Unless such decisions are ordinarily correct, medical practice could not continue.
Or the event could be that the treatment is not working or is causing a harm.
This assessment by doctor and patient is or should be a judicial process. Judicial processes are commonly viewed as something distinct from science. A judicial process, however, necessarily adheres to available data, and only available data, just as closely as science does. Judicial processes have rules of evidence as strict as science. Speculation has no part in a judicial process. Nor since the execution of Walter Raleigh in 1618 does hearsay. If a witness cannot be brought into the examination room and cross-examined, their evidence is discounted.
The explanation, the verdict, must match the facts presented. It must also achieve a consensus that overcomes the biases of 12 different jurors.
In a clinical encounter, just as in a legal process, the answer is provisional, as are all answers in science. Further facts may come to light that challenge a provisionally accepted view.
As outlined here, this is a process that is as scientific as any of the demonstrations of physical or chemical phenomena undertaken in the Royal Society from 1660 onwards that we view as establishing the scientific paradigm, namely that science seeks to explain observable data, that it challenges bias by sticking to this rule and that all its verdicts are intentionally provisional with the process encouraging others to experiment further.
Good clinical practice, when undertaken as outlined here, is inherently scientific. More scientific than any practice shaped by unavailable RCT data, misleadingly represented in ghostwritten publications that hype the benefits of a treatment and hide the hazards.
Clinicians, who attempt to be scientific in this manner, however, are increasingly called on to account for the mismatch between their judgements based on what they see in and hear from patients, for instance that this SSRI made this person suicidal, and the denial of this possibility in an apparently scientific literature, along with the NICE or other guidelines that are based on this literature.
The goal of medicines regulation, as in traffic, financial, or airline regulation, is to enhance safety. It is not to enhance efficacy.
A set of Amendments to the US Food and Drugs Act put in place in 1962 introduced RCTs to the regulation of medicines for the contribution they might make to assessing therapeutic efficacy, in the hope that eliminating ineffective treatments might contribute to safety.
Safety, however, has not been enhanced by these regulations, at least not since the 1980s. Prior to 1962, the benefits of treatment had to be evident to clinicians and patients, whereas now, if a sufficiently large number of patients are recruited to a trial, statistically significant effects can be established on surrogate outcomes with these leading to approvals without any doctors or patients seeing an evident benefit, and with more deaths on treatment than if the patient were left untreated.
It was thought in 1962 that clinicians would remain, as they had been before, the principal agents in medicines’ regulation, with the bureaucrats we now call regulators continuing to occupy a relatively minor role.
In contrast to clinicians, the bureaucrats working in the United States Food and Drug Administration (FDA) and Britain’s Medicines and Healthcare Regulatory Agency (MHRA) do not engage with the judicial processes outlined above for four reasons.
1). When doctors or patients report adverse events to MHRA, the first step, if the reporting process has not already achieved this, is for the bureaucrat to strip the names of patients from any communication – ostensibly to comply with clinical confidentiality requirements. This transforms reports from doctors (or patients) into Hearsay.
Anonymization makes reports from injured patients and their doctors inadmissible as evidence in court in a way that case reports with a doctor and patient’s name on them remain admissible. A patient who has been injured by treatment, and further injured by anonymization, cannot be examined and cross-examined and it is not possible to establish cause and effect.
In the case of the Covid vaccines, MHRA, despite huffing and puffing about searching night and day for the needles of causation in a haystack of reports, are faced with a needlestack of reports but are unable to determine any cause-and-effect relationships. While regulators have conceded a link between current vaccines and both thromboses and myocarditis, this was only after clinicians established that these are happening based on their assessment of patients in front of them. To save face regulators have had to agree.
In the event of reports to them, pharmaceutical companies, in contrast to regulators, are legally obliged to follow up patients and their doctors over time to establish whether the company’s drug might be causing a problem. Companies do this and decide their drug has caused a harm and add this harm to their drugs label, while regulators stack up thousands of reports and claim they have never made a causal link in a single case.
2). A further factor inhibits regulators from making causal determinations in respect of deaths on vaccines and drugs. In the case of deaths, regulators like MHRA in the UK ordinarily wait for inquests, before coming to a view. These inquests have inputs from the patient’s doctor. There is also a coroner in place who comes to a view as to what the cause of death has been.
While coroners can indicate a street drug has been a cause of death, they do not have an option (a box to tick) to indicate that a prescription drug or vaccine has caused a death.
Physicians, meanwhile, if asked to attend an inquest or prepare a report on a death, will be advised by their medical insurer to deny a link to any treatment to which they were party. Doctors at inquests are routinely advised in this manner, by the representatives of a business, whose interests lie in not incurring further costs. Containing a medical insurer’s costs is done by deflecting attention from a treatment to an illness.
Coroners, who are normally not medically qualified, are not in a position to gainsay a treating physician who denies a link to treatment. If a coroner is very concerned about a problem, s/he can make a regulation 28 report to MHRA, who will ordinarily opt not to gainsay the view of the treating doctor.
Even the press reporting on inquests are caught in this web, as advice to journalists in the UK, for instance, from the Independent Press Standards Organization (IPSO) on reporting suicides makes clear. IPSO explicitly tells journalists not to go against the conclusion of the coroner, which for the reasons outlined will never implicate a treatment.
3). Regulators support the mantra that RCTs provide gold standard evidence on drugs, partly because RCTs make the bureaucratic job easier. This faith in RCTs extends to claiming that if an effect has not been shown to happen to a statistically significant extent in an RCT, then no matter how plausible it might be, we simply do not know that it happens.
Ian Hudson, the recent CEO of MHRA, said exactly this under oath in the 2001 Tobin v SmithKline trial, when he was working for GSK. The jury rejected the argument.
4). In line with Hudson’s approach, when considering adverse events, FDA and other regulators claim to have dedicated epidemiologists for this task. Epidemiologists have very little role in pharmacovigilance other than in monitoring the risk of birth defects on treatment.
Neither FDA, nor MHRA, have any clinicians trained in assessing adverse event causality. They lack procedures to assess causality. If were they to establish a link between treatment and an event, pharmaceutical companies effectively maintain a drugs label and have to agree before anything happens. Finally, there is reluctance to acknowledge a Suspected Unexpected Serious Adverse Reaction (SUSAR) because of all the work this leads to.
The anonymized reports of deaths and injuries on vaccines are worthless from an evidential point of view in legal and scientific settings, although some value can be extracted from proportional reporting rates.
In contrast, for sixty years the results of RCTs have been taken in legal settings as meeting a Hearsay exemption.
For the first thirty of those sixty years, however, the courts could request the presence of physician authors on the papers reporting the result of company trials, where authorship meant physicians who treated patients who existed with the physician in a position to evaluate the full effect of these drugs in patients.
For the thirty years from 1990 to now, few of the authors on papers reporting company RCTs will have met any of the patients in trials and several authors have ended up in jail for recruiting non-existent patients to trials. Between non-existent patients, and lack of access to the patients who do exist, the results of RCTs arguably should no longer qualify for a Hearsay Exemption.
As per scientific norms, there is an assumption that regulators, investigators, and the authors whose names appear on publications of RCT results, see the clinical trial data. None do.
Instead, the regulation of pharmaceutical company products is a business process where commercial confidentiality counts for more than the norms of science. Unlike drugs, company involvement in the case of vaccines is relatively recent. Vaccines were made by countries, who were not constrained by profit considerations.
Companies hold the trial data and submit Clinical Study Reports (CSRs) to the regulator. These are the company representation of what its clinical trial has shown. The CSR will often contain large amounts of largely irrelevant figures set out in Tables. The Tables are in principle transcribed from Clinical Report Forms (CRFs), which, until a recent turn to electronic capture of figures, might have up to 100,000 pages for a 300–400-person trial.
FDA insist they get everything from companies, but if they do get CRFs FDA do not read them other than for audit purposes – checking to see if there are any hints that every twelfth patient might not exist.
As I can attest, based on a more detailed examination of CRFs than FDA undertake, if a significant question mark arises about some harms, FDA will invite the company to revisit its records and tell the regulator what the score is in the light of some concern that has arisen. When matters get this serious, companies are still liable to mislead regulators.
(FDA is mentioned here rather than other regulators because of their claims to thoroughness rather than because they offer a gold-standard in regulation).
More recently FDA led the way in getting what they refer to as individual patient level data (IPD). IPD gives the impression a regulator is getting to grips with something like the raw data but in fact IPD is essentially a spreadsheet with figures primarily linked to efficacy that allow FDA to check on some of the claims for benefits, based on selected data provided. It does not allow FDA or any other regulator to explore hazards or interrogate subjects from trials.
The CSR does include some individual patient level data in the form of narrative reports on serious adverse events (SAE) – events that result in hospitalization or death.
From these narratives, it might be possible to guess that the man whose death was coded as death by burns died because he poured petrol over himself intending to kill himself but only died 5 days later and question the company as to whether this death occurring on a new psychotropic drug rather than placebo might have been better coded as suicide.
But companies have found ways around the obligation to write narratives for SAEs. Patients who drop out of trials because of an adverse event can be designated as having intercurrent illness or some related term. There is no obligation to write a narrative on a patient like this, which leaves anyone who might get to see the records none the wiser as to what has gone on.
In Study 329, a 15-year-old boy was picked up by the police on the street waving a gun and threatening to kill people. He was brought to hospital. This was almost certainly a paroxetine related adverse event. It vanished under an intercurrent illness coding – as did events that befell three other children in this trial, all of whom were on paroxetine and none on placebo.
Ultimately people are the data in clinical trials. Without access to their names and contact details, which should be possible in a vaccine trial where volunteers are not being treated for an illness, no-one can establish what has happened in a clinical trial.
It takes a clinical interview to establish if the person in the trial likely became suicidal as a result of their illness or their drug. The clinical interview is the only place where all the data is present – the later trial database loses data as a result of a semi-automatic allocation of features that an illness and a treatment might share in common, such as suicidality, to the illness.
Up to half of the symptoms that people have in a trial may be features shared in common between a treatment and an illness. This can be teased out in healthy volunteer trials of drugs which remove the illness confounder. In vaccine trials, everyone is a healthy volunteer but companies have found ways to discount the injuries that have happened.
Before Covid, there was growing evidence that life expectancies were falling, or improvements in life expectancy had stalled, in many Western countries, including the UK and US. While poverty and inequality kill, as outlined in Shipwreck, or El Naufragio de lo Singular, we also know that a polypharmacy, made possible by the business and regulatory processes outlined above, also kills.
Few of us were on more than a brief course of one medicine per day in the 1980s. Now heading toward 50% of us over the age of 45 are on 3 medicines per day and approaching 50% of us over 65 are on 5 medicines or more per day. The evidence that reducing medication burdens can increase life expectancy, reduce hospitalizations and improve quality of life is strong enough for Health Departments to support efforts in this area.
An unwillingness to tackle the issues raised here, however, semi-mandates a continuing increase in medication burdens, regardless of efforts to reduce meds even in the hopes of saving money. These points are ones that those in favour of vaccine mandates might consider more carefully.
Points 1-10 above have been made to regulators, Ministers of Health, Chief Medical Officers, guideline makers, the editors of medical journals and others. Much of the correspondence is available Here
They have been presented at academic meetings on all continents, featured in articles in major journals and in University Press books without being contested or leading to legal action.
But Health Systems have a bias toward pulling the Horse inside the Citadel.