CDC's new seasonal flu death estimates reflect disease variability

Aug 26, 2010 (CIDRAP News) – The US Centers for Disease Control and Prevention (CDC) released new death estimates for seasonal flu today that are designed to move away from a single number and instead take into account the disease's unpredictability and the extra toll inflicted when influenza A H3N2 is the dominant strain.

A 2003 CDC estimate of 36,000 flu deaths a year has been used by the media and some researchers as a general benchmark, but some experts have criticized that use of the number, because it is likely to overestimate deaths, given that the estimate is based on flu seasons during the 1990s when H3N2 was the main circulating strain during eight of nine flu seasons.

The agency published a report containing the new estimates today in its latest issue of Morbidity and Mortality Weekly Report (MMWR). The report uses statistical models with data from death certificate reports and covers 1976 through 2007, the most recent year for which the certificates are available. It adds four influenza seasons from the CDC's last such report in 2003.

David Shay, MD, MPH, a medical offer with the CDC's National Center for Immunization and Respiratory Diseases and lead author of the report, told reporters at a media briefing today that the new way of reporting flu deaths is designed to give the public a more accurate context of the flu's burden and show how fatalities can vary from year to year. "A simple average fails to give the impact of flu in an understandable fashion," he said. "We want to focus more on a range than a simple average."

"Flu really is unpredictable. At the beginning of the season we don't know what the impact will be," he said. Aside from the circulating strains, other factors can influence flu illnesses, hospitalizations, and deaths, such as the length of the season, how many people get sick, and who gets sick, Shay said.

The CDC's new flu death estimates use a two-tiered approach. The bottom estimate is based on deaths from pneumonia and influenza causes, and the upper estimate, designed to reflect flu deaths that occur from secondary infections and underlying chronic conditions, is based on underlying respiratory and circulatory mortality data, which includes pneumonia and flu causes.

The upper bound number is likely to provide the most accurate estimates, which for the overall population over the last three decades ranged from a low of 3,349 to a high of 48,614 deaths. The lower estimate, which included only pneumonia and flu deaths, ranged from 961 to 14,715.

For both levels, the mortality rates for the 22 seasons during which H3N2 was the dominant strain was 2.7 times higher than for the nine seasons when the subtype wasn't dominant.

Shay told reporters that although researchers don't understand all of the reasons seasonal H3N2 flu appears more lethal, he said that particular virus tends to go through genetic changes more quickly than do other seasonal flu viruses, meaning that people with past H3N2 infections may be more likely to become infected again.

Also, the CDC broke the estimates into three age-groups: under 19, 19 to 65, and 65 and over. Shay said the researchers saw no surprises when they looked at flu death age breakdowns. As reported in past flu death estimates, about 90% of flu-related deaths occur in adults ages 65 and older.

The CDC said the new estimates were subject to four limitations: The model doesn't account for respiratory syncytial virus and other circulating pathogens, changes in influenza testing practices, increasing number of seniors in the population, and a lag in the availability of death certificate data.

To better portray the mortality burden on younger populations, future research should consider years of life lost, Shay said.

At today's press conference, journalists said it will be a challenge to explain the nuances of the new estimate to the public, especially in short radio and TV broadcasts. Shay suggested that a brief way to portray the more complex message is that in many years, flu causes tens of thousands of deaths, though the number can vary depending on the predominant strain.

Michael T. Osterholm, PhD, MPH, director of the University of Minnesota's Center for Infectious Disease Research and Policy, publisher of CIDRAP News, has raised concerns about using the 36,000 annual death figure for flu, and during the 2009 H1N1 pandemic he coauthored a report on years of life lost during the recent outbreak. He said he congratulates the CDC on publication of the new flu death estimates. "It's a much more realistic way to show the impact of influenza mortality," he said.

CDC. Estimates of deaths associated with seasonal influenza, United States, 1976-2007. MMWR 2010 Aug 27;59(33):1057-62 [Full text]

This week's top reads