In recent days there have been reports of a jump in the number of cases of H1N1 influenza (swine flu) in the UK. There have been 29 deaths associated with pandemic influenza in the UK, and there are 652 people in hospital in England with the flu. Faced with the prospect of primary health care services becoming overwhelmed, the government has set up a telephone hotline to allow those affected by the flu to access antiviral drugs (for example oseltamivir or Tamiflu) without needing to see a doctor. But there are also suggestions that not all patients with flu-like symptoms should be treated. Patients with mild or vague symptoms of the flu, without other medical conditions that put them at particular risk, may not be given medication.
This sets up a problem for patients who develop mild flu-like symptoms. Although there is only a small chance of them becoming seriously ill or dying from the flu it is possible that early treatment with anti-virals would reduce that risk. (Antivirals were only effective in trials if given in the first 48 hours of illness) Should they demand treatment from their doctor in the hope of avoiding a serious complication of influenza? Should they exaggerate their symptoms? If the doctor refuses, should the patient self-treat with medications that they have obtained privately (for example over the internet)? There is a form of the classic prisoner’s dilemma involved in such questions.
In the prisoner’s dilemma interrogators question two suspects separately. They are told that if they confess and implicate the other prisoner they will be treated leniently and serve a reduced sentence. If they deny the crime but are implicated by the other prisoner they will serve a full sentence. If both deny the authorities will be forced to release them after short imprisonment. The dilemma is that the prisoners don’t know what the other prisoner is going to do. The best strategy would be for all prisoners to refuse to confess, but the risk is of being implicated and suffering the worst outcome.
The antiviral decision has some similarities with the prisoner’s dilemma. It would be better for everyone if patients with mild flu-like symptoms were not treated with Tamiflu. Widespread use of antiviral drugs would potentially accelerate the development of viral drug resistance. It would place more pressure on the limited stocks of antivirals. Both factors would increase the risk that a later serious infection with influenza would not be able to be treated. But an individual now with symptoms of early mild flu may reduce their (small) chance of dying by taking Tamiflu. What is prudentially best for the patient may be worse for the community if everyone behaved similarly.
One difference between this dilemma and the prisoner’s dilemma is that there is a separation in time between the risk reduced by taking antivirals now, and the risk created (in the future) of a drug-resistant strain of the virus. It is more like an iterated prisoner’s dilemma, where the individual faces multiple choices. The second difference is that uncertainty makes the outcomes and risks less tangible. Individuals with mild-flu-like symptoms do not usually know whether they actually have influenza. They do not know whether their mild symptoms will stay mild, or will develop into something much more serious. They don’t know whether or not Tamiflu would reduce that risk. It is also difficult to know how much (if at all) an individual decision to take Tamiflu would affect the development of oseltamivir resistance. The other difference is that in the flu case the patient may actually secure the best outcome for himself by taking the drug now. If they do have H1N1 influenza taking antivirals may reduce their chance of serious illness now while rendering them immune to a later drug-resistant strain.
So what should we do? The best strategy as a community is probably to avoid giving antivirals to patients at very low risk of developing complications. Guidelines that discourage prescription to the mildly unwell would probably be appropriate. It would also be better as a rule if we all refrained from obtaining Tamiflu over the internet and self-prescribing when mildly symptomatic. But what about those who are symptomatic (or whose children are symptomatic) and strongly desire treatment? One option, rather than demanding a script for tamiflu, would be to request (and potentially pay for) a rapid diagnostic test for H1N1 influenza. Results can be available within a few hours. If the test were positive individuals could be treated with antivirals. This would reduce the risk of completely unnecessary treatment.
Ultimately there is no easy way out of the prisoner's dilemma/patient's dilemma. If we each do what is rationally best for us as individuals we may all end up worse off.