Study: In life-years lost, H1N1 pandemic had sizable impact

Mar 23, 2010 (CIDRAP News) – A new study argues that because the H1N1 influenza pandemic cut many young lives short, its real public health impact has been substantially greater than is generally perceived.

In the study, a team of government and academic researchers came up with new estimates of deaths in the pandemic. By combining those with data on the age distribution of deaths, they estimated the number of "years of life lost" because of the pandemic. By that measure, its impact was at least as severe as a tough seasonal flu epidemic and possibly greater than the pandemic of 1968-69, they contend.

"We conclude that the 2009 A/H1N1 pandemic virus had a substantial health burden in the US over the first few months of circulation in terms of years of life lost, justifying the effort to protect the population with vaccination programs," says the report, published last night by PLoS Currents: Influenza. In the interest of rapid dissemination, the online journal publishes studies that have been screened by experts but have not undergone formal peer review.

The lead author of the study is Cecile Viboud, a research scientist at the National Institutes of Health's Fogarty International Center, and the senior author is Lone Simonsen, research professor at the Department of Global Health, George Washington University, School of Public Health and Health Services, Washington, D.C., and Bethesda, Md. Another co-author is Michael T. Osterholm, PhD, MPH, director of the University of Minnesota Center for Infectious Disease Research and Policy, publisher of CIDRAP News.

Their report seeks to counter a popular view that the pandemic, at least to date, has been much "milder" than expected. This perception has prompted criticism that the threat was exaggerated by public health officials and by pharmaceutical companies hoping to profit from it.

In particular, critics point to the Centers for Disease Control and Prevention (CDC) estimate that the pandemic has caused 12,000 deaths and note the CDC's own oft-quoted estimate that seasonal flu kills about 36,000 people annually.

The authors of the PLoS study contend that such comparisons are misleading, because the estimates were derived by different methods and because seasonal flu kills almost exclusively elderly people, whereas the H1N1 pandemic has primarily preyed on children and younger adults.

They sought to devise an "apples to apples" way to compare pandemic and seasonal flu mortality. To estimate the age distribution of pandemic deaths, they relied mainly on a European study of 468 laboratory-confirmed pandemic deaths, published in August 2009. It showed that more than 85% of the deaths were in people younger than 60, with an overall mean age of 37.4, as compared with an estimated mean age of 76 in those who die of seasonal flu.

The team then developed an estimate of pandemic deaths, given that only a fraction of cases and deaths are laboratory tested and that final statistics will not be available for another 2 to 3 years. This task involved comparing preliminary mortality data from the CDC's 122 Cities mortality surveillance system with final mortality data from the National Center for Health Statistics for 1999 through 2006. The team also used data on deaths in non-flu-season months to estimate the number of monthly deaths that would occur in the absence of flu.

Using these variables and the 122 Cities data for April through December 2009, the researchers estimated the pandemic death toll to be between 7,500 and 44,100. The lower number is based on deaths classified as due to pneumonia and influenza (P & I). The higher number, Viboud told CIDRAP News by e-mail, is an estimate of "excess all-cause mortality, which is more inclusive as it also takes into account excess deaths from all respiratory and cardiovascular diseases that are associated with influenza infection, but may not be reported as such."

She added that the "excess all-cause mortality" approach has traditionally been used to measure the impact of seasonal epidemics and for the 1957 and 1968 pandemics, while the P&I approach has been used in assessing the 1918 pandemic. The team used both approaches to allow fair comparisons with past flu seasons, she said.

The team's range of estimates is considerably wider than the CDC estimate of between 8,500 and 17,600 H1N1 deaths in 2009. Viboud said the CDC estimate relies primarily on hospitalizations for lab-confirmed flu and related hospital deaths, with a correction for underreporting.

By applying the age-group mortality data from the European study to the estimated deaths, the researchers calculate that the pandemic took a toll of between 334,000 and 1,973,000 years of life lost (YLL).

The team used the same method of estimating flu-related deaths to come up with YLL estimates for previous pandemics and for seasonal flu. The results:

  • The 1968 pandemic, with 86,000 deaths and victims averaging 62.2 years old, caused 1,693,000 YLL.
  • The 1957 pandemic, with 150,600 deaths and a mean age of 64.6, caused 2,698,000 YLL.
  • The 1918 pandemic, with an estimated 1,272,300 deaths and a mean age of only 27.2, exacted a toll of 63,718,000 YLL.
  • An average flu season dominated by influenza A/H3N2—which generally causes more severe epidemics than other strains—causes 47,800 deaths and 594,000 YLL, with a mean age of 75.7.

Thus, the authors say, the lower end of their YLL estimate for the H1N1 pandemic is comparable to the estimate for an H3N2-dominated flu season, while the upper end is greater than that for the 1968 pandemic. Those impacts, of course, are dwarfed by that of the catastrophic 1918 pandemic.

"Based on US mortality surveillance data, we conclude that the YLL burden of the 2009 pandemic may in fact be as high as for the 1968 pandemic—but that at this time the assessment is still tentative," the report states. More waves of H1N1 cases are likely to come over the next few flu seasons, and later waves could be worse, it says.

The 1968 pandemic is regarded as the least severe of the three 20th century pandemic and has been described as mild. But Viboud objected to the latter characterization.

"We do not like to use qualifiers such as 'mild,' 'moderate,' or 'severe' for influenza pandemics," she commented. "While the 1968 pandemic was associated with fewer deaths than the 1957 or 1918 pandemics, those deaths occurred in younger than usual age groups, as compared with typical seasonal influenza epidemics. By using the years of life lost metric, which takes into account both the number of deaths and their age distribution, the burden of the 1968 pandemic is estimated at about three times that of a typical influenza season in recent years, after adjusting for time trends in size and age structure of the US population."

In their report, the authors write that using terms such as mild, moderate, and severe to describe the health impact of seasonal and pandemic flu "is insufficient, and possibly inappropriate," because they are based on just one outcome measure: total deaths.

"We recommend not using those terms anymore, because they really don't portray what happened," said Osterholm in an interview. "If there are 300,000 years of life lost, is that something mild?"

Terms like mild and moderate not only don't reflect YLL, but also fail to count lost work productivity when young people die and the pandemic's impact on hospital intensive care units, he added.

Concerning the CDC estimate of 36,000 deaths in an average flu season, Osterholm commented, "Those deaths in a traditional flu season are primarily in the elderly, where there's a question of how many cases you can prevent with vaccine anyway, whereas these [H1N1 pandemic deaths] are in a much younger population. From a public health perspective, it's clear that a death is not just a death."

"I think we can all agree that we'd all like to live long, healthy, prosperous lives, and anything that has a median age of death of 37 years is something that cuts short many healthy prosperous lives," he said.

The researchers don't claim to have found a single outcome measure that best describes the impact of a pandemic, but say the YLL approach is the best for now.

"We're wide open to suggestions on how to better measure pandemic flu in relation to seasonal flu. How do you do it?" said Osterholm.

Viboud C, Miller M, Olson D, et al. Preliminary estimates of mortality and years of life lost associated with the 2009 A/H1N1 pandemic in the US and comparison with past influenza seasons. PLoS Curr Influenza 2010 Mar 20;2:RRN1153 [Full text]

This week's top reads