Three Men in a Boat syndrome: the use and abuse of statistics

I remember going to the British Museum one day to read up the treatment for some slight ailment of which I had a touch—hay fever, I fancy it was. I got down the book, and read all I came to read; and then, in an unthinking moment, I idly turned the leaves, and began to indolently study diseases, generally. I forget which was the first distemper I plunged into—some fearful, devastating scourge, I know—and, before I had glanced half down the list of “premonitory symptoms,” it was borne in upon me that I had fairly got it.

I sat for awhile, frozen with horror; and then, in the listlessness of despair, I again turned over the pages. I came to typhoid fever—read the symptoms—discovered that I had typhoid fever, must have had it for months without knowing it—wondered what else I had got; turned up St. Vitus’s Dance—found, as I expected, that I had that too,—began to get interested in my case, and determined to sift it to the bottom, and so started alphabetically—read up ague, and learnt that I was sickening for it, and that the acute stage would commence in about another fortnight. Bright’s disease, I was relieved to find, I had only in a modified form, and, so far as that was concerned, I might live for years. Cholera I had, with severe complications; and diphtheria I seemed to have been born with. I plodded conscientiously through the twenty-six letters, and the only malady I could conclude I had not got was housemaid’s knee.*

Millions of people in Britain use medical drugs regularly. Each packet has an inserted document, which amongst other things, indicates possible side effects. Presumably the regulator has insisted that this information is inserted to alert users to possible dangers from correctly using the drugs in question. All very laudable. But is the information helpful, or essentially meaningless?

Atorvastatin are pervasively used in Britain one of which has the brand name Lipitor. The leaflet begins with a pink box designed to attract your attention. It says “Read all of this leaflet carefully before you start taking this medicine because it contains important information for you.” Section 4 (on page 4) lists possible side effects. The first two sections are ‘Rare’ (1/1000 people) and ‘Very rare’ (1/10000). There are two further sections ‘common side-effects,’ where the range is expressed as ‘up to 1/10’ and ‘uncommon side effects (1/100)’. So for common side-effects there is an escalation from 0.01% to 10% in anticipated adverse outcomes. Let us focus on ‘common side effects’. (It’s only fair to quote the actual document itself, see addendum).

Statins are heavily promoted by the NHS as a preventative drug and have very large numbers of users. Patients use them for extended periods of time. About 14 million British adults are eligible to use them- but not all of them do so- (Lipitor is only one statin brand). At 1/10000 very rare events could impact on about 14,000 people per year**. For ‘rare’ events (1/1000 people) the figure is 140,000. These figures are very much larger than people might expect reading of ‘very rare’ and ‘rare’ adverse events. The Lipitor document says that one rare event “can be life threatening and lead to kidney problems.”

For anticipated adverse reactions the population is very much larger. There are seven principal categories one of which contains five sub-categories but there is no indication which are more likely to occur (see addendum). Remember that in this section key statistics are prefaced by ‘up to 1 in 10 people’. So 10% is the maximum but how many hit this dangerous level? Is the list organised in priority order with the most anticipated at the top (or bottom); is the list randomly organised? All of this matters because ‘up to’ 10% could cluster at 0.1%. This would be a negligible risk whereas a mean of 9% is statistically significant. This drug is being actively promoted to the general population and if carrying significant negative risk some might fear that the cure was worse than the disease.

The Lipitor warnings are more-or-less meaningless but what happens if more than one drug is taken, which adverse effect is caused by which drug? If a Lipitor user also uses Ramipril the careful reader of the information inserts would read that headache is a common side effect (category 4, Lipitor). Is this adverse impact compounded, or multiplied, or neither? Ramipril also has, as a common side effect muscle pain (category 6, Lipitor). Diffundox XL shares with Ramipril an adverse effect of dizziness. Even though dizziness is the sole common side effect of Diffundox XL they still follow the same form of words, that is ‘may effect up to 1 in 10 people’. A diligent drug company should know what percentage of users suffer this common side effect as there is only one category identified. The commonplace Dispersible Aspirin Tablet have an even less precise warning. Their information insert says “Most people will not have problems, but some may get some.” A masterclass in vacuity.

Careful reading of the information inserts could well induce ‘Three men in a boat syndrome’. Therefore to avoid the feared but non-lethal Three men a in boat syndrome don’t read the inserts in boxes of medicines- unless you like being frightened.

ADDENDUM
Common side effects [of Lipitor] (may affect up to 1 in 10 people) include:
1 inflammation of the nasal passages, pain in the throat, nose bleed
2 allergic reactions
3 increases in blood sugar levels (if you have diabetes continue careful monitoring of your blood sugar levels), increase in blood creatine kinase
4 headache
5 nausea, constipation, wind, indigestion, diarrhoea
6 joint pain, muscle pain and back pain
7 blood test results that show your liver function can become abnormal.

*Jerome K Jerome Three men in a boat (to say nothing of the dog) pp1-2
**Assuming that Lipitor is characteristic of all statins. Pfizer has sold c. $112.1bn of Lipitor tablets since it was launched in 2003 but its sales are petering out due to competition.

Chris

Advertisements
This entry was posted in Health, Humour, statistics and tagged , , , , , , , , . Bookmark the permalink.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s