Skip navigation

Risk: aspirin or car?

Clinical bottom line

Risks of bad things happening with medical interventions are of the same order as those with the rest of life - often less. Thinking about how we deal with risks and benefits of treatment is something we have to get to grips with.


A paper [1] on different sorts of risk adds to a policy debate on risks and benefits of drugs, and the role of the FDA, currently ongoing in the USA. It examines the mortality of several different drug therapies, and compares them with mortality estimates of different occupations and modes of travel. Its sources are detailed, mainly US, and selected to provide a range of estimates form different sorts of risk. Table 1 captures the main risks they have examined.

Risks of drug treatment, occupation, and transport provided in a variety of ways

Treatment/Occupation/Transport
Cause of death
Risk of death
per 100,000 person years
Risk of death
1 in X
Medical treatment
Smallpox vaccine Smallpox
0.07
1,430,000
First generation antihistamines for allergies for four months a year Increased risk of non-motor vehicle related injury
2.8
36,000
Aspirin for CV disease in 50 year old men Stroke
10.4
9,600
Clozapine for schizophrenia Agranulocytosis
35
2,800
Natalizumab for multiple sclerosis Progressive multifocal leukoencephalopathy
65
1,500
Rofecoxib for cancer prevention Cardiovascular events
76
1,300
Type of occupation
Office and administrative  
0.4
250,000
Fire-fighter  
10.6
9,400
Construction labourer  
29.3
3,400
Taxi driver/chauffeur  
36.1
2,800
Truck driver  
44.8
2,200
Logger  
114
877
Tree maintenance  
358
279
Transportation
Train  
0.11
910,000
Air  
0.15
670,000
Bus  
0.19
530,000
car  
11
9,100
Motorcycle  
450
222

For readers who are already questioning some of the medical risks, it needs to be emphasised that they are based on much less solid information than those for occupation and transport, which are actuarial and usually based on relatively large numbers of events and sensible assumptions. The medical risk estimates are more shaky. For instance, the rofecoxib information comes from a single trial in colorectal cancer prevention, where any risks are far higher than for arthritis, where rofecoxib is licensed, and based on only 10 events in total, a difference of 6 in 3,059 patient years for rofecoxib and 4 in 3,359 person years for placebo.

Extrapolation to a risk of 1 in 1,300 per year of cardiovascular death is a bit of a stretch. But stick with the numbers as they are, because it is some of the thinking behind the table is interesting.

Cohen and Neumann make the point that just comparing risk rates, however calculated, is not as simple as it seems. he risks are not necessarily equivalent, or may not be to everyone. They make these points about how we may judge the risk:

Comment

There is no simple answer, so don't go looking for one. What we do have in a useful and thoughtful discussion of the issues, including the benefit side of the equation. Despite known risks, some patients (not all) will choose to run the risk because of benefits of treatment. They quote a paper saying that half MS patients would accept a risk of 1 in 1,000 of something bad happening, and about one in six would accept a risk of 1 in 100.

This discussion about risk and benefits of new therapies (not just drugs) compared with other risks in life is something we need to get thinking about. We risk abrogating decision-making to some remote body, rather than letting individuals decide for themselves. There are tough calls to be made.

Reference

  1. JT Cohen, PJ Neumann. What's more dangerous, your aspirin or your car? Thinking rationally about drug risks (and benefits). Health Affairs 2007 26: 636-646.