In Medicine there’s a saying “if you hear hoofbeats, don’t look for zebras.”
It’s a reference to the wisdom of looking for the most likely explanations when making a diagnosis rather than looking for those that are rare and unusual. Hoofbeats are of course more likely to be the common horse than the rare zebra.
My encounters with New Zealand’s pharmacovigilance system over the past four years have been akin to a safari, where I have witnessed scientists involved in pharmacovigilance and medicines regulators wildly hunting zebras while a rather large and obvious horse was standing on their toes.
Pharmacovigilance is “the science and activities relating to the detection, assessment, understanding and prevention of adverse effects or any other drug-related problem.”
International pharmacovigilance programmes were established following the Thalidomide disaster in which the fraud, dishonesty and greed of a pharmaceutical company and negligence of regulators and doctors, led to over 10,000 babies being born with severe deformities.
Pharmacovigilance programmes address the fact that, as the journal of Drug Safety explains
Clinical trials are inherently limited in their ability to produce data regarding adverse effects, especially when these are rare and unexpected. Therefore, at the time of marketing of a drug, the knowledge of its tolerabiltiy is inevitably incomplete. During the years following the launch of a drug, this knowledge of its tolerability is inevitably incomplete.
Pharmacovigilance programmes are designed to gather information on what effects drugs have in the real world rather than in groups of carefully selected clinical trial populations.
Most people believe that new drugs must be proven to be safe and effective prior to their release on the market, that clinical trials will have identified all possible risks and adverse reactions and that regulators would not approve drugs unless that was the case. They have no idea that the real testing of the drug occurs when it begins to be prescribed to the general public.
In New Zealand we feel pretty well protected from the harmful effects of prescription drugs. Both the World Health Organisation and the New Zealand Government tell us that New Zealand has one of the best pharmacovigilance programmes in the world and consequently, the highest adverse reporting rate internationally.
In New Zealand therefore, we can expect that should a drug have adverse effects not identified in clinical trials, our regulators will be world leaders in identifying and responding to those effects.
According to the WHO the aims of Pharmacovigilance are “to enhance patient care and patient safety in relation to the use of medicines; and to support public health programmes by providing reliable, balanced information for the effective assessment of the risk-benefit profile of medicines.”
So does New Zealand have a superior record in patient care and safety in relation to medicine use? Does it inform its public health programmes with reliable, balanced information based on effective risk-benefit assessment of medicines?
In a word… No. In several words, the evidence is that we actually do worse on these measures than other countries. We for example, have no warnings on the packaging of drugs and do not have patient information leaflets advising of risks associated with the drug, inside the packets.
According to Medsafe, New Zealand reports approximately 5% of all adverse reactions. When I asked for the evidence of this recently, I was told (somewhat patronisingly) that our adverse reaction reports are higher than the reports of other countries per head of population. Not trying to be annoying (but failing I think), I then asked how they knew this meant we had a higher reporting rate and not a higher actual rate of adverse reactions than other countries. Trying to be helpful (but not quite succeeding I think), I suggested this might be as a result of our high prescribing rates, significant population of people from ethnic groups with higher prevalence of genetic variations which influence the metabolism of drugs (and therefore the adverse reaction rates) and a host of other factors.
I was then told (somewhat sternly ) that it’s impossible to know exactly what percentage of adverse reactions are reported because we do not know what the actual number of adverse reactions are and therefore cannot calculate what proportion of those reactions are reported.
Really? I would have thought that this was relatively easily done. Maybe all that zebra hunting has exhausted them, or given them sunstroke or something.
You see, if CARM went and asked our drug buying agency Pharmac, how many of a particular class of drugs were prescribed in a particular year, Pharmac would be able to produce those figures. I know, I ask Pharmac for antidepressant prescribing figures all the time and they are very obliging.
Medsafe, our regulator, provides figures in its data sheets for health professionals on the expected frequency of specific adverse events based on clinical trial and post marketing data.
CARM knows how many adverse reactions were reported.
Easy as 1,2,3. With these three pieces of information, one can make a reasonably intelligent estimate of the adverse reaction reporting (adr) rate.
The formula is pretty simple:
Number of ADRs reported ⁄ actual (or estimated) ADRs = rate of ADR reporting
I.E.; the number of adverse reactions reported divided by the number of actual (or estimated) adverse reactions in the population, calculated as a percentage. For example, if 10 reactions were reported and you know that 100 people suffered adverse reactions then calculating 10 as a percentage of 100 gives an adverse reaction reporting rate of 10%.
If actual numbers of adverse events in the population is not available, then calculating estimated ADRs would involve the following calculation:
Number of people prescribed the drug X prevalance of ADRs in clinical trials = Estimated users experiencing ADRs.
I.E.; the percentage of people who suffered adverse reactions in clinical trials applied to the number of people on the drug. So, for example if 15% of people in the clinical trial got headaches, then it’s likely 15% of people on the drug in the real world would also get headaches so we just find 15% of the total number of people prescribed the drug.
For completed suicide as an adverse drug reaction, we have actual numbers of suicides where the victim was using SSRIs. Our Chief Coroner advises that “Of the 497 deaths ruled as suicide in the 2008 calendar year, 162 had been prescribed some sort of anti-depressant, anti-psychotic or anti-anxiety medication at the time of their death.”
So in what proportion of those cases did a medical professional report this most serious of suspected adverse reactions to our pharmacovigilance agency CARM? Surely if a person died by suicide, a doctor would send off a report to CARM that they were taking a drug known to be causally associated with suicide at the time they died. You’d expect a good proportion to be reported, right?
Using our formula we can calculate the rate at which the adverse reaction ‘completed suicide’ was reported to CARM:
Number ADRs ⁄ actual (or estimated) adrs = rate of ADRs
162 / 0 = 0
That right. Of the 162 suicides, not one of the victims’ doctors reported to CARM that their patient had died in the days or weeks since they had been prescribed a drug. Ok I know I’m repeating myself but zero percent of the suicides that year, of whom 33% of the dead had current prescriptions for psychiatric drugs with known links to suicide at the time of their death, were reported to CARM.
It is hard to find a plausible explanation for this that does not cast aspersions on the ethics of the medical profession but perhaps suicide is a special case. Maybe there’s a causal relationship between the suicide of a patient and Physician Typing Finger Paralysis Disorder or something.
Perhaps the situation is different in relation to other other serious adverse reactions. Surely doctors take the responsibility to prevent medication induced harm seriously and report other adverse reactions to CARM. They’re healers. They wouldn’t put the risk of their prescribing practice coming under scrutiny above their commitment to patient safety.
Lets look at another example.
The information Medsafe provides to health professionals on Fluoxetine advises that in pooled SSRI trials, 15% of those on the drug withdrew as a result of adverse reactions. Taking this figure and applying it to the number of patients prescribed SSRIs in New Zealand in 2008 we can use our formula to generate the expected number of adverse drug reactions in New Zealanders using SSRIs.
Number of people prescribed the drug X prevalance of ADRs in clinical trials = estimated users experiencing ADRs
249,048 X 0.15 = 37,357
Pharmac tells us 249,048 patients were prescribed SSRIs in 2008. Medsafe tells us that 15% of those in clinical trials of SSRIs dropped out as a result of adverse reactions. So, we can expect 15% of 249,048 people, or 37,357 people on SSRIs in 2008, to have suffered adverse reactions of sufficient severity that they discontinued use of the drug.
How many of these expected ADRs were reported to CARM? According to their database, the number of adverse reactions reported to Medsafe for SSRIs in 2008 was 75. So using the formula again, we can calculate the percentage of adverse reactions reported to CARM
75 / 37,357 X 100 = 0.2%
Hmm…0.2%. So much for Medsafe’s claims that their ADR reporting rate is 5%. Perhaps this is another aberration though. It may be that when a patient discontinues their drugs, doctors suffer Potential Loss of Revenue & Pharmaceutical Company Incentives Typing Finger Paralysis. Its important to examine factors such as this before rushing to condemn doctors for what may be a medical condition rather than negligence.
Putting suicide and reactions resulting in drug discontinuation aside, lets look at another example. Medsafe’s datasheet advises that 4% of children and young people on SSRIs in clinical trials developed suicidal thinking or behaviour. In 2008, a total of 6668 children under 18 were prescribed SSRIs in New Zealand. Using the Medsafe data, 266 of these children could be expected to have developed suicidal thinking or made suicide attempts on the drugs. How many reports of suicidality in paediatric patients were reported to CARM that year?
That’s right. No reports were made. That’s a reporting rate of 0% of suicidality reactions in a country that has the highest rate of youth suicide in the OECD and in which doctors express their dedication to suicide prevention.
In the interests of fairness, its worth considering that this lack of reporting may stem from a form of finger paralysis preventing the typing of an adverse reaction report which arises from naughty, attention seeking children threatening to kill themselves, rather than any form of disinterest in child safety on behalf of the doctor involved. Naughty, Attention Seeking Child Typing Finger Paralysis. Possible.
If doctors are not reporting severe adverse reactions like suicidal thinking, suicide attempts and completed suicide, what about more common adverse reactions? I mean, a doctor isn’t going to get into trouble for prescribing a drug that causes their patient to get a headache or feel nauseous, so surely (aside from Typing Finger Paralysis Disorder) there are surely no barriers to reporting those adverse effects.
The Medsafe data sheets list fatigue and nausea as adverse reactions occurring in at least 10% of all users. On that figure, 24,904 New Zealanders could be expected to have experienced these as adverse reactions to the SSRI they were taking. In 2008 reports of these common adverse reactions numbered 5 (nausea) and 2 (fatigue), representing an adverse reaction reporting rate of 0.002 and 0.0008 respectively.
% ADRs Reported
It seems that New Zealand’s claims to have an adverse reaction reporting rate of 5% may be somewhat exaggerated at least in the case of SSRIs.
But let’s not labour the point. Size isn’t everything and CARM’s database might be small but even the underendowed can do the job if they how to use what they’ve got, right? So lets move on and look at how our regulator, Medsafe, uses the adverse reaction reports provided to them.
Champix (Varenicline), the smoking cessation drug is a good example of a drug for which Medsafe has been provided with a significant body of evidence that it induces suicide.
Recently, Champix was subject to an intensive monitoring campaign by CARM who released a report which concluded
This intensive postmarketing study of 3415 New Zealand patients demonstrates that psychiatric adverse events are commonly reported in patients taking varenicline. Approximately 3% of patients experienced symptoms of depression and the majority of these cases appeared to have a causal association with varenicline. Serious psychiatric reactions including suicide, suicidal ideation and psychotic reactions were also identified, but these were less frequently reported.
People died from suicide and became suicidal on the drug during the period it was being monitored. People became depressed and psychotic. So what action did Medsafe take on this information? They refused to accept that Champix was causally assosicated with suicide. They continued to claim in their Medicine Data Sheets that no causal association had been established and that people who suicided on Champix did so because of depression caused by giving up smoking – despite the fact that many who died had not actually stopped smoking and that the depression many reported ceased when they stopped using Champix.
I know its stating the obvious but even if we did have the highest rate of ADR reporting in New Zealand, it would be kind of irrelevant if Medsafe fails to take any action on the pharmacovigilance data collected.
The adverse reaction report I filed after my son died from suicide while on Fluoxetine (Prozac) prompted a similar lack of action to the Champix report.
CARM assessed the relationship between the drug and my son’s death as ‘possible.’ I challenged this (gaining yet another gold star on the annoying rather than helpful chart I suspect) and after a brief conversation, the Director of CARM agreed the correct assessment was that the drugs relationship with my son’s death was ‘probable.’ That means, according to WHO definitions, that the most likely cause of my son’s death was the drug he was taking and that no underlying psychiatric disorder caused him to take his life.
A couple of weeks ago I spoke to Medsafe and asked them what action they had taken as a result of this finding.
Now you might think that a finding that a child had died a horrible, painful death as a result of a drug prescribed by a doctor, might prompt the regulator to take some action to ensure this doesn’t happen to other children. You might think that it would result in a reversal of the current position where these drugs have no warnings on the packaging and no patient information leaflet in the packet.
You would be wrong.
Medsafe’s response to my enquiry about what they had done with the information CARM had provided them around my son’s death, a finding CARM tell me is very significant and should result in some action, was to say that they thought they had issued a Prescriber Update. When they checked the database however they discovered that oops…no…in fact they had done nothing.
I’m trying to be fair and objective here but just can’t accept that the staff of a whole regulatory agency would concurrently suffer Sudden Onset Every Freaking Part of the Body and Brain Paralysis of sufficient severity to prevent them taking any action for three years.
Perhaps they are influenced by the Zebra that the government has obsessively fed and watered since my son died – that his mother’s belief that the drug he was prescribed killed him is nothing more than a grief reaction. Once again, the poor old common horse – the CARM causality assessment has been ignored. I wonder if they think that the Director of CARM is suffering a grief reaction in relation to my son’s death? My view, for what its worth, is that the only adverse event he has suffered, is a kick in the head from a rather large horse…and incidentally that as a result, he may have stumbled across the cure for Typing Finger Paralysis Disorder.
Anyway, there are probably many who think it would be unreasonable to expect a product to be labelled with a safety warning based on the death of one child. Here’s the thing though. In New Zealand this happens all the time, not just in relation to the death of one child but in response to the possibility of harm to a child.
On September 15 last year, a supermarket chain removed baby wipes from its shelves as they contained a chemical that may be harmful to babies if inhaled. While no babies had died or been harmed as a result of the wipes, the Ministry of Health praised the company for withdrawing the product.
Five days later, on 20 September, the Ministry of Health issued a warning to parents over the use of baby slings after the death of an infant in Australia. Again, no New Zealand babies had been harmed by use of the product but the Ministry considered it should warn parents of its risks.
Twenty seven food products were recalled from supermarket shelves in NZ in 2011 and to date in 2012 the tally is 17. No one died from using these products. There was no massive hunt for Zebras in relation to these products. No blaming of consumers, no denying the products benefits outweighed the risks. No frantic search for increasingly bizarre alternative explanations for the harm the product caused or could cause to those who used it. No hesitation in ensuring the safety of consumers was put before the interests of manufacturers, even where no serious harm had been caused.
Its also important to consider that according to CARM, my son’s death is one of a total of 3 deaths reported of patients on Fluoxetine (Prozac) since 1989. Remembering that the adverse reactions reported to CARM are by their own estimates, only 5% of actual adverse reactions, lets look at how many deaths like these three deaths could be expected to have actually occurred in that period.
|Total Deaths if ADR reporting rate = 5%||
Yep, sixty suicides. Surely that is sufficient to at least put a warning on the drug, if not withdraw it from the market.
Why are SSRIs not subject to the same regulatory action as other potentially harmful products. Why is CARM not putting energy into collecting better data? Why is Medsafe not using the data provided to it by CARM? Why do patients and their families in New Zealand not receive the same safety warnings that those in other countries do?
The answer is simple. Because fraud, greed and negligence characterise pharmaceutical companies, doctors and regulators to no less extent now, than they did during the Thalidomide disaster. Because now, like then, we focus on Zebra-spotting while a bloody great horse tramples our children to death underfoot.
The adverse reaction data I have used in this blog to evaluate the performance of our regulator in collecting and responding to Adverse Drug Reactions is available in other jurisdictions. I urge people to access it, conduct a similar evaluation on their regulators and challenge them on their approach and the outcomes they achieve. What proportion of adverse reactions are being reported in your country? How, if at all, is the data collected being used by your regulator to improve patient safety?
It is clear to me that neither the WHO, doctors, nor regulators have sufficient commitment to monitoring and addressing the harm caused by psychiatric drugs. That means that we, the people at risk of being harmed as consumers of prescription drugs, need to take control of adverse reaction reporting and of the regulation of prescription drugs. We need to support Professor Healy’s independent adverse reaction reporting site Risk.org, report adverse reactions there and to our pharmacovigilance agencies rather than rely on doctors to do so and we need to do what we can to ensure that when doctors and scientists hear hooves, they stop wasting time and lives by looking for zebras and stop closing the stable door after the horse has bolted.
 Adverse Drug Reaction
 This figure excludes suicides where the victim was withdrawing from a psychiatric drug or where data was not provided to the court on the victims use of prescription drugs.