How CDC Estimates the Burden of Seasonal Influenza in the U.S.

The burden of influenza on the United States can vary widely from season to season and is affected by a number of factors including the characteristics of circulating viruses, the timing of the season, population immunity to circulating viruses, how well influenza vaccines are working, and how many people have gotten vaccinated. While the impact of influenza varies from season to season, it places a substantial burden on the health of people in the United States each year.

CDC uses a mathematical model to estimate the numbers of influenza illnesses, medical visits, hospitalizations, and deaths in the United States, (1-4) as well as, the impact of influenza vaccination on these numbers. The methods used to calculate the burden of influenza have been described previously (1-2). More recently, the same model was adopted to estimate influenza-associated deaths in the United States. This methodology has been used to retroactively calculate influenza burden, including deaths, going back to 2010.

This page includes information on current and past methods for estimating the number of influenza illnesses, medical visits, influenza-associated hospitalizations, and influenza-associated deaths that occur in the U.S. during a given season.

How CDC estimates Influenza-Associated Hospitalizations in the U.S.

What methods are used to estimate the number of influenza-associated hospitalizations in the U.S.?

Laboratory-confirmed influenza-associated hospitalization rates are obtained from the Influenza Hospitalization Surveillance Network (FluSurv-NET), a collaboration between CDC, the Emerging Infections Program Network, and selected state and local health departments in 13 geographically distributed areas in the United States that conduct population-based surveillance. The network includes hospitals that serve roughly 9% of the U.S. population. The reported numbers of hospitalizations are used to calculate hospitalization rates and the rates are adjusted to correct for under-detection of influenza.  This adjustment is done by using the percent of persons hospitalized with respiratory illness who were tested for influenza and the average sensitivity of influenza tests used in the participating FluSurv-NET hospitals. The data on influenza testing can lag up to two years after the end of the season, so for more recent seasons, testing data from prior seasons is used (1).

Adjusted rates are applied to the U.S. population by age group to estimate the total number of influenza-associated hospitalizations.

How CDC estimates Influenza Illness in the U.S.

What methods are used to estimate the number of influenza illnesses and medical visits in the U.S.?

The numbers of influenza illnesses were estimated from hospitalizations based on how many illnesses there are for every hospitalization, which was measured previously (5).

Some people with influenza will seek medical care, while others will not. CDC estimates the number of people who sought medical care for influenza using data from the 2010 Behavioral Risk Factor Surveillance Survey, which asked people whether they did or did not seek medical care for an influenza-like illness in the prior influenza season (6).

How CDC estimates Influenza-Associated Deaths in the U.S.

What methods are used to estimate the number of influenza-associated deaths in the U.S.?

The methods to estimate the annual number of influenza-associated deaths  have been described in detail elsewhere (1-2). The model uses a ratio of deaths-to-hospitalizations in order to estimate the total influenza-associated deaths from the estimated number of influenza-associated hospitalizations.

We first look at how many in-hospital deaths were observed in FluSurv-NET. The in-hospital deaths are adjusted for under-detection of influenza using methods similar to those described above for hospitalizations using data on the frequency and sensitivity of influenza testing. Second, because not all deaths related to influenza occur in the hospital, we use death certificate data to estimate how likely  deaths are to occur outside the hospital.  We look at death certificates that have pneumonia or influenza causes (P&I), other respiratory and circulatory causes (R&C), or other non-respiratory, non-circulatory causes of death, because deaths related to influenza may not have influenza listed as a cause of death. We use information on the causes of death from FluSurv-NET to determine the mixture of P&I, R&C, and other coded deaths to include in our investigation of death certificate data. Finally, once we estimate the proportion of influenza-associated deaths that occurred outside of the hospital, we can estimate the deaths-to-hospitalization ratio.

Data needed to estimate influenza-associated deaths may lag for up to two years after the season ends. When this is not yet available for the season being estimated, we adjust based on values observed in prior seasons (e.g., the 2010-2011 season through the 2016-2017 season) and update the estimates when more current data become available.

Why doesn’t CDC base its seasonal flu mortality estimates only on death certificates that specifically list influenza?

Seasonal influenza may lead to death from other causes, such as pneumonia, congestive heart failure, or chronic obstructive pulmonary disease. It has been recognized for many years that influenza is underreported on death certificates. There may be several reasons for underreporting, including that patients aren’t always tested for seasonal influenza virus infection, particularly older adults who are at greatest risk of seasonal influenza complications and death. Even if a patient is tested for influenza, influenza virus infection may not be identified because the influenza virus is only detectable for a limited number of days  after infection and many people don’t seek medical care in this interval. Additionally, some deaths – particularly among those 65 years and older – are associated with secondary complications of influenza (including bacterial pneumonias). For these and other reasons, modeling strategies are commonly used to estimate flu-associated deaths. Only counting deaths where influenza was recorded on a death certificate would be a gross underestimation of influenza’s true impact.

Limitations of Influenza Burden Estimates

These estimates are subject to several limitations.

First, rates of influenza-associated hospitalizations are based on data reported to the Influenza Hospitalization Surveillance Network (FluSurv–NET) that are current as of the time estimates are made. Final case counts may differ slightly as further data cleaning is conducted by FluSurv–NET sites. The most updated crude rates of hospitalization for FluSurv-NET sites are available on FluView Interactive (8).

Second, national rates of influenza-associated hospitalizations and in-hospital death are adjusted for the frequency of influenza testing and the sensitivity of influenza diagnostic assays, using a multiplier approach (3). However, data on testing practices during the current season may not be available at the time of estimation. In such cases, we adjust rates using data from prior seasons. Burden estimates from a given season will be updated at a later date when data on contemporary testing practices become available.

Third, estimates of influenza-associated illness are made by multiplying the number of hospitalizations by the ratio of illnesses to hospitalizations; estimates of medical visits are made by a similar process.  These multipliers are based on data from prior seasons, which may not be accurate if patterns of care-seeking have changed.

Fourth, our estimate of influenza-associated deaths relies on information about location of death from death certificates.  However, death certificate data for a given season may not be available at the time of estimation. When this occurs, we use death certification data from prior influenza seasons where these data are available from the National Center for Health Statistics. Specifically, our model uses the frequency of influenza-associated deaths that have cause of death related to pneumonia or influenza (P&I), other respiratory or cardiovascular (other R&C), or other non-respiratory, non-cardiovascular (non-R&C) to account for deaths occurring outside of a hospital by cause of death. If these frequencies are not available from a given season at the time of estimation, we use the average frequencies of each cause from previous seasons.

Fifth, estimates of burden were derived from rates of influenza-associated hospitalization, which is a different approach than the statistical models used in older published reports. This makes it difficult to directly compare our estimates for seasons since 2009 to those older reports, though the estimates from our current method are largely consistent with estimates produced with statistical models for similar years (9-10).  However, it is useful to keep in mind that direct comparisons to influenza disease burden decades ago are complicated by large differences in the age of the US population and the increasing number of adults aged ≥65 years.

More information about how these estimates compare to other burden estimates and past studies can be found on CDC’s Frequently Asked Questions about Estimated Flu Burden.

Figure 1: Illustration of Influenza Burden Estimates Model

508 Text Description of Image

Starting on the left, there are two sequential boxes with arrows that lead to a pyramid. The left-most box is labelled “Reported rate of hospitalization”. There is an arrow leading right to the next box, which is labelled “Adjusted rate of hospitalization”, and over the arrow is a label “1. Correct for under-detection”. From the box, labelled “Adjusted rate of hospitalization”, is an arrow leading right to the middle of the pyramid, which is labelled “Hospitalized” and over the arrow is a label “2. Extrapolate to U.S. population”. In the pyramid, the base is labelled “Symptomatic Illness”, the middle section is labelled “Hospitalized”, and the top is labelled “Died”. On the right-side of the pyramid are arrows coming from the middle section, with one leading up to the top section, labelled “Died”, and another arrow leading down to the bottom section, labelled “Symptomatic Illness”. Next to the arrow leading up to “Died” is a label stating “3. Calculate deaths equals multiply by ratio deaths to hospitalizations”. On the arrow leading from the middle of the pyramid down to the base, which is labelled “Symptomatic Illnesses”, there is a label stating “4. Calculate illnesses equals multiply by ratio of cases to hospitalizations”.