In today’s Guardian, Professor David Salisbury, Director of Immunisation at the Department of Health, is quoted as saying that H1N1 vaccine is completely safe for pregnant women.

No sensible scientist goes to the stake avowing "complete safety". More cogently, Sir Liam Donaldson, the Chief Medical Officer, argued that H1N1 risks for/to pregnant women are real, not theoretical, but even these empirical risks, he regretted, had not been properly quantified.

Let us not only sympathise with the CMO’s dilemma but try to resolve it; or at least put forward a basis for resolution.
 
First, UK has about 12.2 million females aged 16-44 years. Fifteen years ago, the original National Survey of Sexual Attitudes and Lifestyles reported that the ‘past year’ miscarriage/stillbirth/abortion rate was about 4 per cent in women aged 16-44 years, and so approximately 500,000 such events (0.5m). In addition, the UK has about 700,000 maternities a year (that is: live-births together with still-births that have reached at least 24 weeks gestation).
 
Now, as a first approximation, we shall ignore both multiple births within maternities and the small overlap between a) miscarriage/stillbirth/abortion and b) maternities. Moreover, we shall count each a)-event as contributing 3-pregnant-months and each b)-event as contributing 9-pregnant-months so that pregnancy-related events in a given year contribute an estimated 3 × 0.5m plus 9 × 0.7m = 7.8m pregnant-months. Hence, to a first approximation, there may be 650,000 women pregnant in the UK in any month.
 
And, in the six months from mid-April to mid-October 2009, we may reckon on there having been about 3.9m pregnant-months (nearly 4 million pregnant-months). How many UK hospitalisations, Hp, of pregnant women aged 16-44 years have there been for confirmed H1N1? The severe-H1N1-risk to pregnant women in UK’s initial wave of H1N1 could then be expressed as Hp/3.9 per 1 million pregnant-months.
 
Scotland has about 60,000 births a year and 1.05 million women aged 15-44 years. Scotland knows its number of confirmed H1N1 hospitalisations in women aged 16-44 years to have been 55, of whom eight (ScotHp) were pregnant. Let us assume (on the basis of the ratio between the UK and the Scottish population) that Hp = 12 × ScotHp and so we posit that the severe-H1N1-risk to UK’s pregnant women has been: 12 × ScotHp/3.9 per 1 million pregnant-months, or about 3.1 × 8 = 25 per 1 million pregnant-months (with Poisson uncertainty: 11 to 49).
 
In Scotland, the severe-H1N1-hospitalisation risk to 16-44 year old women who were not pregnant at the time was roughly 47 per (6,090,000 – 325,000 = 5.765 million) female-months, or 8 per 1 million non-pregnant-months (with Poisson uncertainty: 5.8 to 10.5)
 
Corresponding calculations for fatal-H1N1-risks need access to UK-wide data, which has been called for in past Straight Statistics articles. UK-wide data would also allow the severe-H1N1 risk in pregnant women to be estimated with much greater precision.
 
Now, the number of women who are pregnant in any month is unlikely to differ very much by season. But the winter-related prevalence of H1N1 is likely to be greater in the six months from mid-October to mid-April than during its initial phase. How much greater remains to be determined. Therefore, the above currently assessed severe-H1N1-risk to pregnant women may need to be multiplied up several-fold depending upon the prevalence of H1N1 during the coming six months.
 
The relative risk per pregnant/non-pregnant month for women aged 16-44 years may be more stable.
 
Sheila M. Bird is at the MRC Biostatistics Unit, Cambridge, and Chris Robertson is at Health Protection Scotland and Department of Mathematics and Statistics, University of Strathclyde.