By Keith Passwater, Dave Nelson, and Thomas Friedrich
Influenza (flu) is an infectious respiratory illness caused by a family of influenza viruses. Common symptoms of the flu include fever, cough, sore throat, runny nose, muscle aches, headaches, and fatigue. The United States Centers for Disease Control and Prevention (CDC) estimates that each year since 2010 in the United States, the flu has caused between 9 million and 36 million illnesses, 140,000 to 710,000 hospitalizations, and has resulted in 12,000 to 56,000 deaths. The annual health care cost for this recurring illness averages greater than $10 billion in the U.S. alone. And lost productivity due to employee absences contributes billions more in cost each year.
This article will offer background on the flu, observations on the actuarial role, and suggested sources of data.
Although this article focuses on flu impacts to humans, the flu infects several types of animals as well, including swine, dogs, cats, some sea mammals, bats, and many types of birds. It’s thought that most influenza viruses start in birds and that new influenza viruses originate from a wide pool of different influenza subtypes circulating, particularly in wild birds. Birds are the natural hosts for avian influenza viruses, and a significant risk to humans is the emergence of new flu variations from birds.
Flu Types and Variants
There are four types of influenza viruses: A, B, C, and D. Types A, B, and C infect humans, whereas D is only known to infect cattle. Type C infects humans but tends to be mild. Types A and B both vary in severity depending on the specific flu virus. Type A is further differentiated into strains based on the particular variant of two distinctive proteins on the surface of the virus, the hemagglutinin (H) and the neuraminidase (N). The H protein is known to have 18 different variations, while the N protein is known to have 11 variations. These protein variant combinations are then notated with the variant number following the protein letter, such as H3N2, which was the prevalent strain in the U.S. for the 2017–2018 flu season. Another example, H1N1, was the prevalent strain during the 2009–2010 flu season.
As mentioned earlier, birds serve as a host for flu, and new variants do emerge and transmit from birds to humans. For instance, in 1997, H5N1 was transmitted to humans from poultry in Hong Kong, and since 2003, this avian flu virus has spread from Asia to Europe and Africa. Of grave concern, the mortality rate of H5N1 is 60 percent.
An important distinction is the difference between what is called seasonal influenza and pandemic influenza.
Seasonal influenza viruses cause the familiar annual flu seasons. Seasonal flu is caused by influenza A and B viruses that are basically completely adapted to humans. The severity of flu seasons depends on several factors, including the inherent virulence of the strains circulating that year and the degree to which the circulating viruses match the ones chosen for the vaccine.
Because seasonal influenza viruses continuously evolve, public health leaders frequently update the vaccine formulations to keep pace with the emerging seasonal virus(es). Sometimes the vaccine formulation is chosen well and there is a good antigenic match—and, therefore, vaccine effectiveness (VE) is high. On the other hand, some years the choice is poorly matched to circulating flu strains and VE is low. For example, the 2017–2018 flu infection rate was high due to low VE. Health actuaries are wise to pay careful attention to emerging evidence of VE each flu season to understand risk and disease burden.
To understand the historical impact of flu, let’s consider influenza A viruses more broadly. Influenza A viruses cause seasonal epidemics, but they also cause occasional pandemics. Pandemics occur when antigenically novel influenza A viruses emerge that are transmissible in humans. Antigenically novel means that humans do not have immunity to these viruses, which is why the flu strains can cause infections in a huge swath of the global population. These pandemic viruses ultimately come from birds, though in most pandemics it’s thought that the bird viruses first “mix” with mammal flu viruses in an intermediate host, usually pigs. Pigs are more susceptible to bird viruses than humans are, so by circulating first in pigs, some bird viruses can acquire the ability to infect and be transmitted in humans.
In 1918 an H1N1 virus made the jump to people, maybe directly from birds, or maybe via some other animal. Of critical importance, in the years prior to 1918, H1N1 viruses had not circulated in humans, so few or possibly no humans had prior immunity to the H1N1 subtype. The flu generally presents a greater threat of mortality to those under age 5 and those over 65 due to their relatively weaker immune systems. However, the 1918 strain also took the lives of those between the ages of 20 and 40, which may have been caused by a cytokine storm (an overreaction of the body’s immune system).
It has been postulated that strong immune reactions of young adults ravaged the body, whereas the weaker immune systems of children and middle-aged adults resulted in fewer deaths among those groups. However, more recent investigations, mainly based on original medical reports from the period of the pandemic, postulated that the viral infection itself was not more aggressive than any previous influenza but that the special circumstances (malnourishment, overcrowded medical camps and hospitals, poor hygiene) promoted bacterial super-infection that killed most of the victims, typically after a somewhat prolonged deathbed. Whatever the mode of impact, the 1918 pandemic was devastating. The prevalent strain that year was H1N1, which infected an estimated 500 million people—one-third of the world’s population—and took the lives of 50 million.
Once established in humans, H1N1 viruses came back to cause seasonal flu from 1919 to 1957. In 1957, a new virus emerged of subtype H2N2, which completely replaced H1N1 viruses. So, all seasonal influenza A after that has been caused by H2N2 viruses. Then in 1968, H3N2 viruses emerged and displaced H2N2. In each case a large proportion of people had no immunity to these new viruses, which is why infection spread rapidly worldwide. H1N1 viruses re-emerged in 1977, and since then seasonal flu A has been caused by a mix of H1N1 and H3N2 viruses.
Usually one subtype or the other predominates in a given flu season, for reasons we don’t understand. In 2009, yet another new H1N1 virus emerged, this time from pigs. Interestingly, it resembled the 1918 virus more than H1N1 viruses that had been circulating recently. So again, most people lacked immunity to that virus, and it caused a pandemic. That pandemic was not as severe as the one in 1918, but nonetheless the 2009 pandemic H1N1 virus completely replaced H1N1 strains that had been circulating previously.
Most infectious disease scientists and public health officials agree that a future influenza pandemic is one of our top public health concerns. It’s clear that there have been pandemics in the past, and we were probably fortunate that the 2009 pandemic was caused by a virus subtype that had infected humans before (H1N1) and that the strains that circulated, although they were very transmissible, generally did not cause very severe disease.
Health leaders worry, in particular, about pandemics that might be caused by avian viruses like H5N1 or H7N9, which have infected many people but so far have not acquired the ability to be easily transmitted between people. An additional risk is that canine viruses (type A, H3N8) may spread to humans given the close proximity of pet owners and dogs in North America and Europe. Of grave concern, most people in the world have no immunity to these animal strains, so a pandemic caused by these viruses could be quite devastating. It is, however, very difficult to quantify the risk of an animal influenza pandemic because we do not know enough to predict such an emergence.
Health insurance companies rely on actuaries to understand flu costs for the purpose of explaining claims trends to management (and investors), updating company reserves, adjusting prices, and making sure company assets are adequate in extreme circumstance.
During the flu season, actuaries on the companies’ claim trend team prepare regular reports—monthly or even weekly—summarizing flu indicators. These indicators include CDC flu incidence reports, Tamiflu scripts per member, and flu-related company-specific claims per member per month (PMPM). Like most actuarial work, attention to detail is an important part of this trend identification work. For example:
- To estimate the PMPM impact of the flu, actuaries review distinctive CPT (Current Procedural Terminology) codes and ICD-10 (International Classification of Diseases, Tenth Revision) codes to identify which claims are associated with the flu.
- To determine likely flu incidence rates, actuaries rely on regional flu data provided by the CDC.
Actuaries use all of this information to make adjustments to claim reserves for weeks 42 through 18 where flu is a factor.
Senior management, board members, and stock analysts are often inundated with news reports about the flu. For that reason, actuaries often share their regular flu updates with senior management ahead of probable questions. In addition, it is common for actuaries working for public companies to prepare a complete flu summary in advance of the companies’ quarterly earning calls so that claims trend related to the flu can be explained to investors.
Health insurance rates typically include a risk factor to maintain adequate assets when unexpected costs develop, such as spikes in the incidence of the flu. Predicted season-specific flu spikes do not usually impact pricing, because health insurance rates are usually set well before spikes can be confidently predicted.
Flu incident rates consistent with a pandemic are often one of the stressors tested in the companies’ enterprise risk management (ERM) effort. Actuaries often prepare these scenarios or provide input to the development of the scenario.
Flu Data Sources
There are a number of data sources, including insurance companies’ claim files. When extracting flu data, it’s important to recognize that the designated ICD-10 codes used by physicians are quite often not assigned to patients suffering from flu because the flu symptoms can be atypical, particularly in older patients. Therefore, it’s important for the conscientious actuary to scour ICD-10 codes for a broader range of respiratory conditions when watching for the emergence of flu in the fall/winter weeks.
As mentioned above, actuaries can monitor reports of vaccine effectiveness and antigenic match. Actuaries can also look for evidence of outbreaks, particularly in China, which is a hotspot for influenza evolution due in part to the frequent close proximity between humans and animals in parts of China. Other geographies to watch include places where avian viruses sometimes circulate in domestic poultry (e.g., Egypt), although visibility in these areas is probably less good.
Health insurance companies rely on actuaries to understand flu costs for the purpose of explaining claims trends to management (and investors), updating company reserves, adjusting prices, and making sure company assets are adequate in extreme circumstances.
Additionally, actuaries can monitor the opposite hemisphere, where seasons are the inverse of the actuary’s focal geography. So, for instance, actuaries practicing in the Northern Hemisphere should monitor the Southern Hemisphere; Australia offers excellent reporting. Another source of flu data, though anecdotal and lagging, are public company earnings reports. For instance, one large health insurer reported on its first-quarter 2018 earnings call that in the quarter, the impact of influenza on medical cost ratio would be about 40 basis points.
Additionally, the following public health data sources are available.
Centers for Disease Control and Prevention Influenza Division: www.cdc.gov/flu/index.htm
World Health Organization: www.who.int/influenza/en/
European Centre for Disease Prevention and Control: www.flunewseurope.org
Public Health Agency of Canada: www.phac-aspc.gc.ca/fluwatch/
Public Health England: www.gov.uk/government/statistics/weekly-national-flu-reports
The University of Minnesota: www.cidrap.umn.edu
Australia Government Department of Health: www.health.gov.au/flureport
Flu is a very significant recurring seasonal illness that unfortunately affects many lives each year and drives significant cost. Additionally, influenza viruses present some of the greatest risks of a worldwide health catastrophe, which would be devastating to life and world finances. Therefore, it is an important consideration in health claim trend analysis, reserving and pricing, as well as enterprise risk management and ruin analysis. There are many public resources available to the actuary—and yet many prediction problems remain to be solved.
So as the annual flu season gets underway and the alarming headlines begin to swirl, we hope this article has given you some additional information to help you sift through the noise—and hopefully you’ll stay healthy through the months ahead as well.
KEITH PASSWATER, MAAA, FCA, FSA is managing director of KTPassCo, providing leadership, strategy, finance, and actuarial consulting; he can be reached at KPasswater@ktpassco.com.
DAVE NELSON, MAAA, FSA, is a retired actuary, having served as vice president and chief actuarial officer at Blue Cross Blue Shield of Michigan.
THOMAS FRIEDRICH, PhD., is associate professor of pathobiological sciences at the University of Wisconsin-Madison.
Endnotes Centers for Disease Control and Prevention (CDC), Disease Burden of Influenza.  CDC, Make it Your Business to Fight the Flu.  CDC, Types of Influenza Viruses.  World Health Organization, FAQs: H5N1 influenza.  Brundage & Shanks, “What Really Happened during the 1918 Influenza Pandemic? The Importance of Bacterial Secondary Infections,” Journal of Infectious Diseases, 196 (11): 1717–1718.