[Skip to content]

Sign up for our daily newsletter
The Actuary The magazine of the actuarial profession

Modelling/Life: H1N1 — The swine who?

In January 2010 the Health Protection Agency published in its weekly update that the number of deaths related to swine flu in England had reached 251. It took care to point out though that the number “does not represent the number of deaths that can be attributed to swine flu” but rather “the number of deaths in individuals with swine flu”.

As aggregated national mortality statistics for 2009 become available, it is possible to get a clearer picture of the impact of the 2009 H1N1 swine flu.

Late last year the UK Office for National Statistics (ONS) released publication DR_09, which sets out statistics of the deaths recorded in England and Wales during the calendar year 2009. The ONS is among the quickest of the national statistical agencies around the world to report population mortality statistics and its data allows the additional deaths in the population caused by the 2009 swine flu outbreak to be assessed.

The publication counts total influenza deaths in 2009 as 227, of which 149 were from avian or swine flu. To put that into perspective, the combined influenza death count in the three years prior to 2009 was just 87. In 2009 not only did the total number of influenza deaths dramatically rise but also the percentage of influenza deaths that came from the working age population soared. In 2009 around half of influenza deaths were in the working age population with infants and seniors accounting for just over one-third. In the three years prior to 2009 over 80% of reported influenza deaths were in seniors and infants, with the working population making up just over 5%.

Yet this published data on deaths by cause may not reveal the whole picture. In a pandemic year we could expect the deaths directly attributed to the pandemic to be under-reported, with there being additional deaths for which influenza was not the primary cause but where its presence is likely to have been a significant contributing factor in the death. One way to assess the impact of the influenza on death rates is to make a comparison of deaths observed in the pandemic year to those observed in recent years.

Figure 1 uses 2006 and 2007 data as a base expectation from which to perform an actual to expected analysis of the male deaths in 2008 and 2009. The red lines indicate a 95% confidence interval for the actual to expected ratio after allowing for general mortality improvements. There is no statistically significant increase in the number of deaths in 2009 with the exception of people in their late 30s and early 40s. Higher actual to expected deaths at these ages are not exclusive to 2009 but are also present in 2008 suggesting that they should not be wholly attributable to the presence of the swine flu. In fact, the ONS population estimates that the middle ages are more heavily influenced by net migration estimates than other age groups so it could be that the increase in deaths is a result of an understated population rather than an increase in mortality rates.

To estimate net migration in any year the ONS uses a stratified sample survey conducted at UK points of entry and exit. A publication from November 2010 shows that estimated immigration to the UK within the 25-44 age group during 2008 and 2009 was broadly flat compared to significant annual increases over the prior decade. Estimated emigration from the UK for the same age group during the same two-year period was significantly increased compared to prior years. An underestimation of those entering the country and an overestimation of those leaving would make population estimates lower than the actual population from which deaths occur. It is plausible that the compounded effect of any errors in net migration estimates since the last population count at the 2001 census could lead to a fairly sizeable mis-estimation of the true population.

There is also a statistically significant decrease in the deaths during 2009 at senior ages. A level of immunity to the H1N1 strain from the last time it was in circulation could partially explain this decrease in deaths. In particular, the H1N1 strain may have crowded out the seasonal flu strains and caused their presence as a secondary cause of death to be lower than in a normal year. However, accelerating improvements to old age mortality could equally explain such a movement.

The analysis of the data supports the assertion that swine flu was responsible for surprisingly few deaths as either a primary or secondary cause despite its rapid spread through the population. However, the swine flu pandemic serves to remind all of us in the life insurance industry that a global viral outbreak is a very plausible threat to capital and solvency.


Dan Knipe is the life portfolio manager at Leadenhall Capital Partners