about Epidemiology & the department

Epidemiology academic information

Epidemiology faculty

Epidemilogy resources

sites of interest to Epidemiology professionals

Last Updated

07 May 2003

Source: New York Times, May 7, 2003

Study in Hong Kong Suggests a Higher Rate of SARS Death


The death rate from SARS may be significantly higher than health officials had thought, up to 55 percent in people 60 and older, and up to 13.2 percent in younger people, the first major epidemiological study of the disease suggests.

Mortality rates are bound to change somewhat as an epidemic continues. But unless the numbers fall drastically, SARS would be among infectious diseases with the highest death rates. Until now, fatality rates reported by the World Health Organization had ranged from 2 percent, when the epidemic was first detected in March, to 7.2 percent.

The new findings come from a statistical analysis of 1,425 patients suspected of having SARS who were admitted to Hong Kong hospitals from Feb. 20 to April 15. Over all, their mortality rate was estimated to be as high as 19.9 percent. By contrast, the influenza pandemic of 1918, which killed tens of millions of people worldwide, had an estimated mortality rate, over all, of 1 percent or less.

But calculating mortality rates for newly emerging diseases is a notoriously difficult challenge for epidemiologists, especially if there is no definitive diagnostic test, as is the case with severe acute respiratory syndrome.

It is possible, for example, that some people infected with the virus believed to cause the disease never fall ill, or develop symptoms so mild that they do not seek medical assistance. If that is the case, mortality rates could be much lower.

The research, led by Dr. Roy M. Anderson and Dr. Christl A. Donnelly, both of Imperial College in London, is to be reported in Saturday's issue of The Lancet, the British-based medical journal. Its editors posted the report on the journal's Web site,, yesterday, citing its public health importance.

The principal authors of the study — from Imperial College, the University of Hong Kong, the Chinese University of Hong Kong and the Hong Kong health department — said their findings underscored that SARS was a serious threat to the public and that health officials and workers must act to contain it.

The authors credited Hong Kong's efforts to reduce the time from the onset of symptoms to the isolating of patients in hospitals as an important step in controlling the disease.

Reducing the time did not affect the course of the illness in individual patients, they said, but expediting the isolation process reduced the period when patients could transmit the virus to others.

Yesterday, the W.H.O. said the number of new SARS cases in Hong Kong had steadily declined, which they said suggested that the outbreak there had reached a peak. So far, Hong Kong has reported 1,646 probable cases and 193 deaths, which would mean that the death rate is 11.7 percent.

But death rates calculated from this kind of data — what epidemiologists call case-fatality rates — can underestimate actual mortality rates, in part because they do not take into account patients who remain ill, some of whom may die.

By contrast, the authors of the new study arrived at their figures by studying people admitted to Hong Kong hospitals each week. They began their analysis with patients admitted in the week after Feb. 26; until then there were too few to make statistically significant calculations, they said. Their analysis ends with patients admitted in the week ended April 15. For those admitted later, they said, "too little time has elapsed" after the onset to allow analysis of mortality rates.

In an interview, Dr. Michael T. Osterholm, an expert on infectious disease now at the University of Minnesota, said that the Lancet study "looks solid and provides an important source of information about the evolving epidemic."

"Anyone who tries to make a point of precision right now about case-fatality rates does not understand them," he said.

Dr. Klaus Stöhr, the scientific director of the SARS investigation for the W.H.O., said in an interview that the agency was reviewing the new data and the statistical models used to obtain it.

"The case-fatality rate can be calculated in many ways and can depend on various scenarios," Dr. Stöhr said. Data that the W.H.O. has received from Hong Kong officials suggests a death rate closer to 15 percent than 20 percent.

Dr. Stöhr said he wished the W.H.O. had received the new data sooner. He said it had been available for more than a week by the time it was posted.

Dr. Stöhr said that earlier receipt of the data could have helped in the evaluation of another finding in the study — the time between exposure to the virus and the onset of symptoms, or the incubation period. The study estimated it to be 6.4 days. But other statistical calculations based on a small number of patients in the study showed that the incubation period could be as long as 14.22 days.

The figure is important because health officials use it to determine how long people exposed to the virus must remain quarantined. Also, as a rule of thumb, health officials double the estimated incubation period to determine when they can declare an outbreak over in a certain area.

Health officials have been using 10 days as the longest incubation period and so have used a 20-day period as the criterion for declaring an area to be no longer affected by SARS. But a 14.22-day figure could mean lengthening the quarantine period and the criterion for declaring an outbreak over.

Dr. Stöhr said that if the data concerning the longer incubation period held up in further analysis, the W.H.O. might change its recommendations for quarantine.

The authors said it might turn out that the incubation period depended on where the SARS virus entered the body. Health officials say that most cases result from the spread of droplets from coughs and sneezes. The SARS virus can survive on surfaces for up to four days, and although scientists have not proved that transmission has occurred through such contamination, Hong Kong officials suspect that a big outbreak at an apartment complex there was spread through sewage.

In conducting the Lancet study, the researchers used different statistical methods to calculate the case-fatality rate over the first nine weeks of the epidemic in Hong Kong.

By one method, they found that the case-fatality rate, among those admitted to hospitals, was 43.3 percent among patients 60 and older and 13.2 percent among those less than 60.

Using another method, they found the case-fatality rates to be 55 percent among those 60 and older and 6.8 percent in those younger than 60.

Dr. Anderson and his team said it would be important to learn whether public health interventions, including news reports, encouraged infected individuals to report their symptoms quickly.

"The epidemic has demonstrated the need," he said, "for communication of risk which will inform and warn the public, in a way which will improve personal protection, without inducing high levels of anxiety and fear, as an essential part of epidemic control. A change in risk perception would potentially lead to an increase in early reporting of symptoms as well as improvements in hygiene and prevention of transmission."