RSS demands better flu monitoring

The Royal Statistical Society has fired a shot across the bows of  Sir Liam Donaldson, the Chief Medical Officer, by demanding changes in the way information on the H1N1 flu pandemic is collected and published. 

Its letter follows many of the themes that have been rehearsed by Sheila Bird on this website (see related articles and related news). So far, the information available on H1N1 in England has fallen far short of that available in Scotland, and has not been sufficient for experts to draw conclusions on who is at greatest risk, or on the occurrence, frequency and outcome of more extreme cases of flu, including patients who need intensive care.

The letter, from Professor David Hand, President of the RSS, says that he understands the switch from daily to weekly reporting on the epidemic, which took place on July 2, was motivated by the need to focus on data quality. He welcomes this focus, but goes on to point out areas of reporting standards and epidemic themes that the weekly data needs to include if it is to be achieved.

The RSS statement lists four areas in need of attention:

Statistical Reporting standards

Users of the data need numerators and denominators, not just percentages, and an explanation of how the HPA is calculating how many cases of H1N1 flu there have actually been.

Demography of those in hospital

They also need to know the numbers in hospital, and in intensive care, and demogaphic details of these cases, as well as the presence or absence of other risk factors. This will make it possible to work out how age-group and sex determine the relative risk of needing hospital admission.

Epidemic monitoring of incident severe cases

Admission to hospital and intensive care need to be reported by date of admission, and the cumulative total reported. Daily reporting-in by hospitals is needed - a task that is "tricky but crucial", the statement says.

Epidemic monitoring of extreme events

Deaths and cases of H1N1 in pregnant women need to be recorded by date, and the cumulative totals of such extreme incidents to date should be reported, so that the doubling time of such extreme events can be tracked. Both hospitals and GPs need to be involved, as not all deaths will occur in hospital.

The statement, sent to Sir Liam yesterday and made public this morning, concludes with the key analytical goals, which are:

Properly to estimate case fatality rates for hospitalized patients with suspected or confirmed H1N1, and to do so demographically and by taking account of pre-existing co-morbidity; to monitor pressure on NHS hospital facilities; and to track, and report publicly, the epidemic course of severe or extreme H1N1v incidents.

UPDATE:

In a reply released this afternoon, Sir Liam has promised to look in detail at the RSS suggestions and see what can be done to address them.

But he warns that the effort of putting a common and consistent set of analyses into the public domain regularly is "huge" and requires staff time. "Busy NHS staff, both managers and clinicians, are notoriously reluctant to fulfil extra reporting requests given the heavy burden of gathering information already in the system. Understandably so." And anything that involves additional reporting on a cohort or the linkng of information is very difficult to construct in continuous data gathering.

That sounds a bit like clearing the throat preparatory to saying no. But we mustn't hasten to judgement. Sir Liam has responded swiftly and courteously, and said that he is an enormous enthusiast for statistical data. The question now is how much he can provide.