Missing swine flu data for patients admitted to hospital

Cases of H1N1 have been creeping up again, with Scotland - where children went back to school several weeks earlier - leading the way. 

Health Protection Scotland has been reporting a gradual increase in its weekly estimated H1N1 cases, particularly in the younger age-groups. This anticipated increase has been without unduly alarming press reportage. A similar trend is likely to follow in England and Wales, and should be evident in this week’s estimates.

This will enable comparison to be made between countries within the UK of the rate of increase in: estimated H1N1 cases (overall, and in the younger age-groups); H1N1 positive rate in persons with suspected swine-flu, sampled from GP consultations, from National Pandemic Flu Service authorizations (overall, and in the younger age-groups); authorizations by NPFS (overall, and by age-group but excluding Scotland); and incident hospitalizations for H1N1 (overall and by age-group).

Andrew Pelling, MP for Croydon, has persisted in his quest for key data by which to work out the H1N1-related hospitalization rate by age-group for (i) the 1,052 cases of swine-flu confirmed in England, Wales and Northern Ireland by 16 June 2009 and for (ii) the 4,410 cases of swine-flu confirmed in England, Wales and Northern Ireland between 17 and 30 June 2009.
 
Either the Health Protection Agency doubted that a Member of Parliament was capable of posing an epidemiologically-adept PQ and hence misread it, or HPA chose helpfully to answer a different question which they could answer. Whatever the reason, HPA provided information on swine-flu hospitalizations up to 16 June (despite the fact that some of cohort i might yet be hospitalized during 17-30 June); and on hospitalizations during 17-30 June 2009 (despite the fact that some of cohort ii might yet experience swine-flu-related hospitalization after 30 June 2009).
 
The data provided to Mr. Pelling show that imported confirmed H1N1 cases were a decreasing minority: 19 per cent of cohort i, but only 3 per cent of cohort ii because, by then, transmission within the UK was dominant. The modal age-group for imported cases was 20-29 years, higher than for ALL confirmed swine-flu cases to the end of June 2009, for whom modal age-group was 10-19 years.
 
        
The data for the two cohorts clearly indicate - as BBC correspondent Eleanor Bradford had deduced – that hospitalization data for England were initially incomplete. The early data on swine-flu hospitalizations for cohort i are not credible: too few were reported by 16 June 2009 relative to reports by 30 June 2009 for cohort ii. Can these deficiencies be remedied in retrospect? The answer is yes, see later.

Can even the data on H1N1-hospitalizations during 17-30 June 2009 can be relied upon? They make sense insofar as the hospitalization rate is high in the very young and increases in older age.

Correct cohort-specific hospitalization rates by age are needed. Because virological confirmations ceased in early July, Andrew Pelling tried to obtain these hospitalization rates for the above two crucial early cohorts of H1N1-confirmed cases. The tragedy is that he could not . . .
 
The Health Protection Agency should now urgently use record linkage to establish crucially missing data on hospitalizations for the above two cohorts of over 5,000 confirmed early cases of H1N1.
 
The same record-linkage methodology can them be used to follow-up on the hospitalizations of representatively sampled suspect swine-flu cases in the coming weeks from whom nasal or other swab is requested for virological confirmation of swine-flu. But, we need swab rates to be much higher . . .
 
Meantime, HPA is to be congratulated for now including in its weekly update a monitoring scheme which compares observed versus expected deaths in England and Wales. However, age matters! And so, I therefore urge that this monitoring be reported by age-group because H1N1-exposed age-groups in which the death-rate is otherwise low are likely to signal soonest, precisely because low numbers of deaths are otherwise expected.