The single most useful question about any health claim is when the decline actually started. It does more work than any amount of expertise, and most accounts of a medical triumph quietly skip it.
Whooping cough makes the point quickly. English child deaths from it stood near 1,400 in every million back in 1850, and by 1952 they were barely detectable. Scientists pinned down the bacterium responsible somewhere around 1906. By then the figure had already dropped to about 900. Most of the fall happened before anyone could name the culprit.
That is not an argument about any particular treatment. It is a habit of reading, and it is the one this article hands you. If you want to build the steadiness to use it, the tryit.tv course opens by teaching you to watch your own reaction before you decide what you think.
Three checks you can run on almost any health claim you meet:
- When did the decline actually begin, and did anything change at the moment the treatment arrived?
- What else were these people breathing, eating, or being given at the same time?
- What did the trial compare the treatment against, and for how long did it watch?
None of them needs a laboratory. All three need only the dates.
The curve was already falling
You can watch this play out in the tuberculosis record, and a Harvard physician named Edward Kass laid it out for a room of specialists in 1970. He had led that field himself. What he showed them was not what they expected.
Deaths had been sliding since the mid-1800s, and they kept sliding in an almost straight line for a century. Then he listed everything that failed to bend that line. Finding the bacterium did not. Nor did the tuberculin test, nor BCG, nor screening programmes, nor even streptomycin.
Polio traces a similar arc. Deaths dropped by 47 percent across America and 55 percent across Britain in the thirty years to 1953, all of it ahead of the vaccination campaigns. What genuinely shifted over those decades was what people ate, whether their water was clean, and how many bodies shared a bedroom.
So the check is simple. Find the year the treatment arrived, mark it on the curve, and ask whether the line bends there or carries on as it already was.
What else was in the air
The second check asks what shared the time and place with the illness, and it changes what polio looks like.
Sweden saw the earliest outbreak in 1887. Line that up against two other dates. A chemist had synthesised DDT thirteen years before, and the mechanical crop sprayer had arrived fourteen years before. Then Massachusetts switched from one arsenic pesticide to a nastier one, lead arsenate, in 1892. Its own outbreak followed within two years.
Bench work had already hinted at this. Back in 1878 a neurologist, Alfred Vulpian, poisoned dogs with lead and watched them develop what looked like the human illness. A Russian researcher called Miezeyeski Popow reproduced that paralysis with arsenic within five years.
Then the timing again. American polio incidence peaked in 1952 and fell away quickly, three years before the Salk vaccine was introduced.
None of that settles the question on its own. It just means a second explanation was available and went untested, and holding one open long enough to check it is harder than it sounds. I have watched how much can rest on that willingness in my own work.
Cattle give an even tidier illustration. From 1985 British farmers were legally obliged to pour phosmet, a fiercely toxic insecticide, onto their animals' necks. The strongest doses went on in Britain, in Northern Ireland, and in Switzerland. Those three are where nearly every case of mad cow disease turned up. Meanwhile an organic farmer who fed the suspect meal but skipped the chemical watched his herd stay healthy.
Ask what the trial compared
A trial's meaning lives in its comparison group, which is the part press coverage almost never mentions.
Consider the 1987 trial that put the first AIDS drug on the market. It lasted four months. The company selling the drug paid for it. Patients quietly had their tablets chemically analysed so they could work out which arm they were in, which destroyed the blinding. And one arm got thirty rounds of blood transfusion where the other got five, so the groups were never really comparable.
Vaccine testing poses the same puzzle wearing different clothes. Rather than an inert dummy shot, the Gardasil trials measured against an aluminium adjuvant. That comparison substance caused adverse reactions in 85 out of every 100 people who received it. The vaccine caused them in 90. Five points is not much of a gap to hang a safety verdict on when your yardstick provokes reactions of its own.
Duration counts just as much. Other medical products get watched for five years or more, while a vaccine trial can wrap up in a matter of days. When the Institute of Medicine went through 231 reported adverse events, it concluded that for 184 of them the evidence simply could not settle the question either way. Nobody had shown harm. Nobody had shown safety either. Those are two different findings, and coverage tends to collapse them into one.
A positive result is not a diagnosis
Base rates change what a test result means, and cervical screening shows it plainly.
As many as four women in five will carry human papillomavirus at some stage of life. In four out of five of those, the body clears it unaided. Of everyone who tests positive, well under one in a hundred ever develops cervical cancer.
Now carry those proportions into a screening programme. Watch 1,000 women for 35 years and you avert a single death. Along the way 150 of them get a frightening result and 50 undergo treatment for cancer. American surgeons remove roughly 200,000 wombs annually, in a country where 14,000 women develop the disease.
How much testing you do shapes the count as well. Germany logged 7,582 positives in one week of March 2020 and 23,820 the week after. Over those same days, the number of tests being run had roughly tripled. Read the case figure without the testing figure beside it and you mistake arithmetic for news.
Follow the money before you weigh the claim
Funding relationships are usually printed alongside the research, which makes this the easiest check of the three. Where a figure is buried in a long report, you can put the question to the source and get the number back rather than hunting for it yourself.
One study in a major journal went through the advisory panels convened by America's medicines regulator. Nearly three-quarters of those meetings, 73 percent, included someone with a financial conflict. Payments topping $100,000 featured in almost a quarter of the contracts. Of the conflicted members, a single percent were kept out of the vote.
British MPs reported in 2005 that drug firms bankroll three out of every four clinical studies appearing in the top journals. One of those very journals had loosened its rules for contributors three years before, on the stated grounds that untied experts had become too scarce to find.
Marcia Angell edited that journal once, and she leaves you with a question requiring no expertise whatsoever. If these medicines are as good as claimed, why do they need such hard selling? Treat a heavy marketing push as your cue to examine the evidence more closely. Something that genuinely works does not rely on persuasion to find the people it helps.
What openness actually costs
Here is the part that catches people, and it is not really about medicine.
Running these checks is easy. Sitting with what they turn up is harder, because the answer sometimes disagrees with something you already decided, or with what someone you trust told you.
Where I see that resistance bite hardest, the block is never the method itself. It is the unwillingness to let a rival explanation stay possible long enough to be tested. Openness is not credulity. It is holding two accounts side by side and letting the evidence choose between them.
That posture is the whole of what this book asks for. Its four authors, led by the investigative journalist Torsten Engelbrecht, did not argue from theory. They wrote to the institutions named in each chapter, asked for the specific studies, and published what came back and what did not.
Where the checks leave you
What you gain here is better than a position. You gain a method you can use again.
When the next claim arrives, and it will, you have three questions ready. When did the decline begin. What else were these people exposed to. What did the trial compare against, and for how long.
Asking them calmly, especially about a decision that frightens you, is its own skill. None of this is medical advice, and none of it argues against ordinary care. Your own doctor stays the place to take any decision about your health, and these checks work alongside that conversation rather than instead of it.
What you are left with is not a verdict on any particular illness. It is a way of reading that travels, and it costs you nothing but the patience to check three dates before you decide. The question that started this is the one worth keeping. When did it actually start falling?
If a health decision is weighing on you right now and you would rather not reason it out alone, you can talk me through what you are weighing up. And if it is the evidence itself you want to press on, the course builds the steadiness that clear reading needs.