Sunday, January 28, 2018

Why Flu Fatality Numbers Are So Hard To Determine


Now that the effects of this winter's severe flu season are being felt across the country - in the form of school closures, employee absenteeism, long ER waits & Tamiflu shortages - I'm beginning to see social media posts suggesting the CDC is deliberately `downplaying' the true severity of this year's outbreak.
While it is true that the numbers provided by the CDC each week in their FluView reports under represent the true burden of flu in the country, the CDC  has always been up front about the limitations of influenza surveillance and reporting.
Unfortunately, most reporters only see the weekly numbers - and report them as if they are  `facts' - rather than explain why these numbers are only representative of the relative severity of the flu season, not a true accounting of the impact.

In Friday's CDC press conference, Dr. Dan Jernigan once again addressed the problem of getting a handle on the number of flu fatalities (underlining mine).

So with pediatric deaths, these that are in the graphs and the ones we report again, probably also are an underestimation of the actual deaths that are out there. There may even be as much as twice than the number we have.
For the adults, curves that we have in the graphs from the pneumonia and influenza morality reporting system, that data comes from death certificates. You can look and see that the rapid rise that we have in this past couple of weeks, really looks very similar to what happened in 2014-15. With that, it’s possible it may go a little above it. It may peak at the same time. We don’t know right now, but again that information has some of the same lags because of information getting into the death certificates as well.

ROB STEIN: Do you have a total number of deaths?

DAN JERNIGAN: We don’t, partly because the systems that we collect those numbers from only are representative of parts of the U.S. So what we do then is look at the rates, so we can have trends over time.
In years where there is an H3N2 that is at that higher level of severity, we have estimated that at the end of the year there were 56,000 deaths. So the H3N2 years are the ones that have the higher number of deaths and that upper level is about 56,000 for the total year.
Unlike adult fatalities, pediatric flu deaths are required to be reported to the CDC. Even with this reporting requirement, many pediatric deaths are missed. Often because not all child deaths at home are tested for influenza, particularly when there are other comorbidities.

In October 2010's Lancet: Pediatric Mortality Related To Pandemic H1N1, we looked at the retrospective accounting by the CDC of adult and pediatric deaths during the pandemic.  Officially, there were roughly 280 pediatric fatalities reported across the country.

As bad as these numbers were, the CDC actually estimated that only about 25% of pediatric influenza deaths were identified – and put the `real’ number at closer to 1300.

CDC Pandemic Est

With adults - particularly older adults - influenza deaths are even harder to quantify. 

Take it from a former paramedic, thousands of people die at home every day, and unless there is something suspicious or unexpected about a death, the patient's physician will generally sign the death certificate and no tests or autopsy are performed.  
If the patient had cancer, or heart problems, was of advanced age, or had any other significant comorbidity - that, not influenza or pneumonia - is usually listed as the primary cause of death.
Just four days ago we looked at another study that found the risks of a heart attack go up significantly in the first week after an influenza infection (see NEJM: Acute Myocardial Infarction After Laboratory-Confirmed Influenza Infection).

Only rarely would one of these heart attack victims be charted as a `flu fatality'.  And without that on the death certificate, it never makes it into the flu statistics.

Globally, it has been estimated that seasonal flu kills between 291,000 and 646,000 people each year (see The Lancet: Estimates Of Global Seasonal Flu Respiratory Mortality).

But even these recently revised numbers likely understate the burden of influenza, since they are based strictly on respiratory-mortality, and ignores other influenza-associated causes of death (heart attack, stroke, etc.).

Closer to home, the CDC explains how they derive their flu numbers:
Disease Burden of Influenza

The severity of influenza disease in the United States can vary widely and is determined by a number of things including the characteristics of circulating viruses, the timing of the season, how well the vaccine is working to protect against illness, and how many people got vaccinated. 
CDC tracks severity principally through its national Influenza Surveillance System that monitors key indicators like the percentage of deaths resulting from pneumonia or influenza, rates of influenza-associated hospitalizations, pediatric deaths and the percentage of visits to outpatient clinics for influenza-like illness.

In addition to using surveillance data, CDC uses mathematical models to fill in the picture of the disease burden and the impact of influenza immunization programs. Models are used to augment surveillance because most of the surveillance systems only look at portions of the U.S. population and in some cases there can be significant under-reporting of influenza deaths and hospitalizations. This page includes links to key resources on the burden of influenza.
CDC Burden of Influenza

While it would be great if we had a way to definitively say how many flu deaths occur each year, there is simply no practical way to do so.  The same holds true in the United States, and around the world, for just about every cause of death.
But what we can say is this influenza season is particularly severe, and that tens of thousands of lives will be lost across the country this winter either due directly or indirectly to influenza infection.
The severity of this year's flu season isn't exactly a surprise. In late December the  CDC issued a HAN Advisory warning of an aggressive H3N2 season ahead, reduced vaccine effectiveness, and stressed the importance of early and aggressive treatment of severe influenza - particularly for `high-risk' patients - with antivirals. 
A month ago in Yes, We Have No Pandemic . . . But Line Up A Flu Buddy Anyway, I urged my readers to have, and to become, flu buddies - particularly to those who live alone. 
Hopefully we'll see a peak in the flu season soon, but even so, we've probably got another two months of flu ahead.  We might even see a second `surge' of H1N1 or Influenza B as H3N2 begins to wane in the spring.

As rough as this flu season turns out to be, it will pale compared to the next influenza pandemic.  Hopefully the lessons we learn from this winter will be taken to heart, and acted on, rather than simply fading away with the arrival of summer.

After all, we can only expect to get so many warnings.

No comments: