Prevalence of non-pneumonic infections with SARS-correlated virus

Sir

Patrick Woo and colleagues' findings¹ are certainly pertinent to health-care-system planning for potential SARS outbreaks. However, I wish to express my concern over the comparison of a figure reported as a rate for the study population with the cumulative incidence of pneumonic SARS for the Hong Kong population.

The rate for the study population consists of one non-pneumonic paediatric patient and three blood donors with positive results divided by the overall number tested in the study. I would argue that the one paediatric patient with a compatible illness divided by the total paediatric study population represents the cumulative incidence for non-pneumonic SARS in that population alone (which might or might not be representative of the total paediatric inpatient population over the study period or the paediatric population of Hong Kong).

With respect to the blood donors, unless these people were identified, I believe it would be impossible to know whether or not they had pneumonic or non-pneumonic SARS, or whether their illness was in fact within the study period. Assuming these three people were not among the 1728 who were known to have developed SARS pneumonia, and assuming no one in Hong Kong was exposed to SARS coronavirus before March, 2003, these three confirmed cases divided by the total number of blood donors in the study population would represent the cumulative incidence of non-pneumonic SARS in this population (which again might or might not be representative of all blood donors, depending on whether they were selected at random and might or might not be representative of the wider Hong Kong adult population).

If we accept that the blood donors are likely to be representative of the adult population of Hong Kong, comparison should be drawn between this cumulative incidence figure and the pneumonic SARS cumulative incidence for the adult population of Hong Kong. Similarly, if it is fair to assume that the other study groups are representative of some broader community group, these cumulative incidence figures could be compared, although care in interpretation would need to be taken given the low numbers in the study population.

The overall study population, however, is unlikely to be representative of the entire Hong Kong population and therefore conclusions drawn by comparing pneumonic SARS illness in Hong Kong with the cumulative incidence of antibodies to SARS coronavirus in the study population is fraught with danger. This situation arises particularly given the known effect of age and chronic medical conditions on the epidemiology of SARS, but could well be influenced by other (known and unknown) confounders too.