Saturday, 8 June 2013

Unintended consequences and the homeostasis of risk

I recently saw a patient who broke her ankle slipping on the shiny yellow paint they use to highlight the edges of steps. Is this an example of the law of unintended consequences, or the theory of risk homeostasis (risk compensation)? Either way, a well-intended intervention backfired (what the CIA colourfully label ‘blowback’), a phenomenon more common than we think.

Explanation
1. The law of unintended consequences states that intervening in a complex system tends to create unanticipated and often undesirable outcomes (in attempting to reduce the risk of falls by making the steps easier to see, they made them slippery, thereby increasing the risk of falls).
2. The theory of risk compensation (homeostasis) states that if the perceived risk is lowered, a person's behaviour is adjusted so that the risk level rises towards the pre-set ‘acceptable’ level (if steps are easier to see, people will not take as much care negotiating them).

As you will see, there are many examples of both, and the two are related (strictly speaking, risk compensation is just one of many causes of unintended consequences), but they both lead to the same place: the effectiveness of an intervention being less than expected.

Economics
Similar phenomena occur in economics.
  • The introduction of a first homebuyers grant in my home state simply increased the price of property.
  • If we put aside ‘X’ billion dollars to provide health care for all the disadvantaged who need it, the number of disadvantaged that “need” it will probably increase, and the cost of healthcare will also rise with the demand.
  • Those who take out insurance are more likely to behave in a way that will result in claiming that insurance. See the economic theory of moral hazard.


Risk homeostasis
The theory of risk homeostasis has been mainly used in road traffic injury research (overview here). This health-related overview from The Lancet (on condoms and seat belts) is very well written and is a great introduction to the topic.

I would argue that a strict theory of risk homeostasis (that the risk always returns to the same level) cannot be supported. Many examples of risk homeostasis result in a net benefit or net harm, rather than a neutral effect. For example, the road toll is gradually falling (possibly due to a gradual change in the level of ‘acceptable risk’?). Few would argue, though, with the idea that some degree of risk compensation occurs with many interventions.

Unintended consequences
It should be pointed out that unintended consequences are not always bad. They can be positive, they can result in a reduction in the benefit from an intervention, or they can result in net harm. Here, we are concentrating on the latter two.

Examples
  • The Streisand Effect states that attempts to censor information result in an increase in circulation of that material.
  • Helmets on bicycle riders decrease the effect of a direct blow to the head, but laws making their use mandatory result in a decrease in bicycle use, and consequent negative health effects.
  • Providing condoms to decrease HIV transmission resulted in an increase in sexual activity, which acted to increase the transmission rate back to previous levels.
  • Providing cars with anti-lock brakes or air bags results in more aggressive driving. So does increasing street lighting.
  • Patients with severe osteoarthritis of the knee often complain that they cannot improve their health by losing weight pre-operatively due to the arthritis. After a total knee replacement, their ability to walk is improved, but this study showed that patients tend to increase weight after a knee replacement. The surgery was meant to improve their health, but it led to other consequences that oppose that improvement.
  • In Sweden in 1967, they switched from left-hand drive to right-hand drive cars. They thought it would be a disaster, but the road fatality rate actually fell by 40% because everyone drove more cautiously. Because the perceived risk was very high, drivers adjusted it downwards). Within 18 months, it was back to previous levels.
  • Parents are less careful in handling childproof medicine bottles, causing no change in the number of accidental poisonings.
  • Booth’s rule #2: “The safer skydiving gear becomes, the more chances skydivers will take, in order to keep the fatality rate constant”.


The bottom line
My point is not to prove the theory of risk homeostasis, but to raise awareness of the phenomena of risk compensation and unintended consequences. People should understand that an intervention that sounds (superficially) like it would have a clear benefit, probably has unintended consequences and may lead to other changes that diminish its effectiveness.

I steered clear of a many obvious medical examples, but you only need to browse this blog to see them. 

2 comments:

  1. I think you make a good point and I think a common thread appears to be the person's behavior/thinking.

    Your point of raising awareness is key, just as we need to raise awareness of our bias as well.

    Perhaps we can become a better species if we are able to actually become of aware of our behaviors, thinking, and bias. Science would flourish and all the useless stuff would be thrown out. Wishful thinking...

    ReplyDelete
    Replies
    1. Thanks, and I agree: we need to be aware of bias (out own and that of others). Bias is pervasive and easily overlooked. Perhaps excusable in some, but not in those who purport to be scientists.

      Delete