Measuring the tip of the iceberg
The Health Protection Agency estimates that there were 110,000 new cases of swine flu in England last week,10,000 more than the week before. But how did they work this out and how accurate is it?
They start out on fairly safe ground with the GP consultation rate. They have a pretty good idea of this from QSuveillance (which tracks computer codes entered by GPs and covers 40% of the UK) and RCGP data. GP consultation rate, however, is only the tip of the iceberg: many more will have symptoms but not see their GP. The HPA have tried to estimate what percentage of people with flu do not seek medical attention.
There is currently no reliable information about the proportion of pandemic (H1N1) 2009
cases who consult their GP. It is thought, however, that this proportion is likely to lie within the range 0.2 (20%) to 0.5 (50%). This proportion is based on research demonstrating that approximately 10% of those with symptoms visit their GPs during normal flu seasons; and an assumption that this percentage is likely to be higher at present due to heightened awareness (and anxiety) about pandemic H1N1.
In other words, for every person with flu like symptoms who presents to their GP with flu, between one and four do not. The HPA took the middle point of this and scaled up the number of GP consultations.
But what about the National Pandemic Flu Service (NPFS)? How has that affected the numbers who consult their GPs? In the age groups where the service was available it is thought to have reduced consultation rates by about 10%. To reach this figure the change in GP consultation rates in the 15-65 year old age group who can consult the service was compared with consultation rates in children under 1, who are not managed by NFPS. There was a 10% relative reduction in consultations in the 15-65 age group compared with children under 1. Quite a modest reduction given all the hype in the press and anecdotal reports from GPs.
Thus, the scaling factors of 2 (50% of cases visit their GP) and 5 (20% of cases visit their GP) were increased to 2.22 and 5.56, respectively, for those aged 1 or older and under 65 in the week 21 July to the 27 July.
All of these statistics may leave many people cold, but they do raise an interesting issue around margin of error. The level of uncertainty in the figures – particularly the range of possible proportions of people who do not contact their GPs – could be thought to make a 10% difference from one week to the next seem irrelevant. Should the HPA include confidence intervals in their figures or quite them as a range?
What’s in a name?
The media call it swine flu. The government (and the BMJ) call it pandemic influenza . The WHO goes one better by calling it pandemic (H1N1) while the HPA probably goes too far when it calls it the “influenza A/H1N1v (swine flu) pandemic”. Catchy. To the Burmese Junta it’s human flu, whereas a reader of Dr Ellie Cannon’s column in the Daily Mail calls it Tamil flu. Where will it end? What will the history books call it? Flu?
Tom Nolan is a GP trainee in London.