Media and Government Reports May Have Skewed Statistics, Says Georgetown Professor
Michael Stoto, professor of health systems administration and population at Georgetown University, has concluded that—while incidence and hospitalization rates caused by H1N1 were higher among children—death rates were substantially higher in seniors.
This is a trend that that would not be considered anything out of the ordinary for seasonal influenza. But H1N1 was generally seen as having more severe rates of infection and death among younger people. Stoto presented his findings on H1N1 at the Public Health Preparedness Summit in Atlanta this past February.
"[A]lthough no one knows the extent of this bias, it looks as if those 65 and over are at a higher risk of dying, and perhaps being hospitalized, than those under 18,” Stoto said in an e-mail interview. “This is true both among those infected and on a population basis.”
Stoto points out that more H1N1 cases were counted and reported among children over the course of the pandemic. In press briefings in October, CDC officials were reporting their observations of a very different pattern from the usual impact of seasonal influenza—with higher rates of mortality among young people.
With seasonal influenza, 60 percent of those hospitalized, and 90 percent of those affected could be seniors. Between Sept. 1, 2009 and Oct. 10, 2009, the CDC had recorded that 90 percent of hospitalizations and deaths caused by H1N1 were in those 64 and under.
Stoto argues that more cases of H1N1 may have been confirmed for those under 18 years of age, possibly boosting recorded rates of hospitalizations and deaths among those age groups. This is because parents, caregivers, and physicians were aware of cautionary reports in the news about the impact of H1N1 on younger people. More children than adults may have presented themselves for medical care and been tested for the virus.
“Physicians, aware of the…news about children being at higher risk and also having been told that people born before 1957 were probably not ‘at risk,’ followed public health recommendations issued in early May that emphasized testing only those with the most severe disease,” said Stoto. “[It] may…have been more likely to have a child tested than an adult with the same symptoms.”
He describes the early period of an influenza pandemic as a “fog of war” because of the uncertainty about how it will spread and how severe it will be. But he said that population-based statistical approaches to surveying may lend more clarity on the disease.
Population-Based Method for Measuring Impact of Pandemics
According to Stoto, biases in reporting arise from syndromic surveillance because the method measures reactions or behaviors to disease symptoms, such as seeking medical care, staying home from school or work, or buying medications.
“Since people’s actions are influenced—at least to some extent—by what they hear in the news, syndromic surveillance data runs the risk of circularity, reflecting what people hear rather than their illness,” said Stoto.
Stoto’s proposed population-based statistical approach would be based on surveys such as the National Health Interview Survey (NHIS) and the Behavioral Risk Factors Surveillance System (BRFSS). This would require adding specific questions about influenza to the surveys. He hopes this method will lead to more clarity and accuracy in the reporting of flu pandemics.
Now that H1N1 is on the wane, and public health professionals are emerging from a period of intense activity, hindsight may continue to bring insight on how response efforts for pandemics can be improved.
The H1N1 pandemic taught public health officials and others working in response to the pandemic some very valuable lessons. “We need to assess the public health system’s ability—from the global to the local level—not only to detect outbreaks but also to characterize the pathogen and its epidemiologic characteristics,” said Stoto. “And if 2009 H1N1 taught us anything, it is to expect the unexpected.”