As the temperature reaches the low 30°Cs it may seem difficult to imagine, but influenza season will soon be upon us. One of the many dilemmas of flu season is at what point one should go sick or keep a child at home. A recent study from the United States has looked at this very subject, but from a public health perspective, what is the effect of someone with flu going to work on the wider community? (Kumar et al (2013) Am J Pub Health 103 1406-1411).
This was a study done using agent based modelling, in other words it was a computer simulation. Probably the most important thing to do when you look at a model such as this is to check the assumptions made in the model, so for example here they had different scenarios: one where not everyone had access to paid sick leave; one where everyone did; and some where employees were able to take either 1 or 2 ‘flu days’ regardless of access to paid sick leave. In all models it was assumed that 28% stoically went to work even if they had the flu; and that the R0 (the average number of infections caused by each infected person) was 1.4 – pretty typical of seasonal influenza. However, in some circumstances, such as schools and nurseries it may be much, much higher.
The results of the modelling showed that the attack rate (the proportion of those without immunity who were exposed and caught the flu) was 11.54% in the mixed sick-leave scenario; 10.86% where paid sick leave was universal; only 8.62% where 1 flu day was allowed; and 7.01% where employees were able to take 2 flu days. In a simulated population of 575 866, this equates to 66 444; 62 538, 49 611; and 40 386 infections respectively. That is a lot of numbers, but what it is saying is that the 2 flu day scenario reduced the number of infections at work by 26 059 or 39% from the baseline mixed sick-leave scenario.
So the lesson from this is that it may be better for the economy for people to go off sick rather than go to work when they are ill. Remember that once one has the flu, you can then infect other people who may be in one of the high risk categories for severe disease, so the impact is almost certainly even wider than this study suggests. This can of course be mitigated by getting vaccinated - and the flu vaccine is gradually being rolled out to all children aged over 2-years (not this is not an injection!); and others in the high risk categories should already be getting it. More details here (it is a bit wordy - but keep going - it is on page 10!).
Now major warnings in interpreting these and similar data:
- All models are wrong – the trick is to get the model that is least wrong
- Look at the assumptions – do they seem sensible?
- Does the model allow for ‘random behaviour’? These are called stochastic models and while they are more lifelike, they are also much more complicated. The opposite, where people essentially do as they are told or behave the same are called deterministic models.
- In epidemiological studies always be clear about the case-definition; how do you know someone has the flu? This is crucial, because we need to know that we are talking about influenza and not just a bad cold. The UK case definition and other epidemiological data are here.
Finally reading this paper I learned a new term, which is ‘presenteeism’ – which is going to work or school when ill, something I may have been guilty of in the past... More about flu to come, in the meantime enjoy the sun!