Risks and benefits: Interview with John Campbell

We discuss the different measures of risk (absolute risk reduction, relative risk reduction, number needed to treat etc) and the importance of understanding the difference between them in order to understand whether a medical treatment is safe and effective.

Slides (pdf) available here:

This video is also on John’s youtube channel:

Written by Norman Fenton


Leave a Reply
  1. Hey Norman Fenton, have you seen that Awful, Back To The Science, channel on ytube? I am not up on the maths but something about it just does not add up. She loves dismissing you and John Campbell.

  2. I watched the video on this channel , but when we had 500 views here , there were 20 k on John's channel . But you made a huge work to make this topic understandable , and thanks really a lot . From France .

  3. Great interview

    Watching the body language and comments from 8:55. How the risk is presented vs the risk / benefit.

    Dr John C. stroking chin, then takes fountain pen to make a 'famous tick or note' 🙂

    Very diplomatic. 😉

    Love it…

  4. I am recently living through the biggest corporate fraud in human history, and they didn't even give me a t-shirt to prove that I was there! When all is said and done, this will make Enron or the 2008 financial meltdown seem about as big as PeeWee Herman's missing bicycle!

  5. Another important thought to layer on top of all this is that when we are looking at mortality rates from the disease vs the intervention, the numbers are presented as binary lives, but if we look at it in terms of life lost, those losing life to the intervention are losing more than an order of magnitude more life-years, especially at this point in time, where almost everyone dying with the disease is someone who would probably have died within weeks, from one thing or another.

  6. Another angle to look at the big picture would have been to randomly test people for the virus with low-cycle (25 cycles, perhaps) PCR in a population and create graphs where the X axis is not calendar time, but cumulative infections, and in the Y axis, graph cumulative all-cause excess deaths, deaths attributed to the virus, and deaths from specific disease categories like circulatory and respiratory. That way, we can see how immunity developed over the course of natural infection, rather than the course of time, because the "waves" when charted against time are very distracting and obfuscating. I suspect that all of these negative outcomes would be seen as diminishing gradually throughout the pandemic, and we'd see no special unique immunity event around the rollout of the intervention to the elderly, except perhaps in remote, rural areas where the virus had not penetrated much before the rollout.

  7. I'd just like to add, that in order to really get a good idea on efficacy, they would need to continue tracking the same people throughout the entire time period, right up until today. The "efficacy" changes over time, each new data set would create a new table, added to the last and so on. This would show you the effects of natural immunity vs the jab. For instance, the original efficacy data doesn't even include the possibility of repeated infections in one group or the other. It's possible that over time the Jabbed, are actually re-infected more often. If that were true you would add this new data to the old data. The new data set created by the two sets would give you your new efficacy numbers. One data set would show original efficacy the next would show current efficacy the third would show the total of the two. I'm dumbing this down, but I hope I made my point. A snapshot in time, doesn't tell you much about efficacy OVER TIME.

  8. Thanks so much Norman. It would make more sense if they removed anyone who contracted the disease during the not effective stage (both the placebo and vaccine group up to 2 weeks after the last person getting the second shot). This would even out the playing field and I feel from my limited analytical knowledge this would make more sense than categorising them all as unvaccinated in the trial.

  9. Regarding Relative vs Absolute Risk Reduction I would like to play devil's advocate here: As said in the video, the absolute risk reduction is dependent on the prevalance of the desease being treated, and maybe back in April (or whenever it was) 2020 at the beginning of the study there was still a low prevalance because of the covid measures, which causes a low abolut risk reduction. But back then it was a reasonable assumption to make, that the prevalance will rise dramatically over the course of year, and therefore the relative risk reduction (which doesn't depend on prevalance) was the more sensible measure to use during the trial.
    But I do agree, that ommitting the abolute risk reduction without explaining the reasoning and the pushing of the 95% effectiveness was purely to sell the product.

  10. Statin drug manufacturers typically have stated the benefit of the drug in preventing cardiovascular events (heart attacks, strokes, etc.) as relative risk reduction but stating the adverse effect risks as absolute risk, which gives the impression that the drugs are much more effective than they really are in absolute terms and that the adverse effects are very rare when they're not. Even worse, the statin drug trials sometimes gave the patients in the treatment group a 6-week trial period (a "run-in period") of the drug and excluded any of them who were intolerant (had adverse reactions) before running the longer trial of the drug and computing the risk of adverse reactions. That's pure deception.

  11. Your use of NNT seem naive at best. The majority of the population had no immunity at the time of the trial. One prevented case in the trial participants, may have prevented many cases in the non trial population. Also it only takes into account the period of the trial.

Leave a Reply

Your email address will not be published. Required fields are marked *