The image of an aggressive shark in the deep ocean is graphic and terrifying, but the risk of mundane threats far outweighs the risk from shark attack. The dead man’s brother should worry much more about heart disease, which felled his brother, and devote his attention to lowering that and similar risks. There is only so much time and energy; each unit of energy spent on lowering the risk from sharks is one less unit that can be spent on hearts.

What should we fear? What threats are most likely to kill us? Setting aside cataclysmic events such as nuclear wars and planet-altering meteorites, there are some risks that generate a lot of fear but few deaths, such as shark attacks, terrorism, and killings by police. On the other end of the spectrum are everyday risks that kill a large number, such as heart disease and cancer. In between are risks from motor vehicle collisions and the seasonal flu. And this year there is a new risk: COVID-19.

Putting small risks in perspective: Let’s start with the risk from our introduction. In 2019, there were two deaths from unprovoked shark attacks in the whole world, though the longer-term average is about four per year. Based on the latter number, the risk to the average person of dying from a shark attack works out to 1 in 2 billion per year. For someone who is in the ocean frequently, that risk is higher. For someone who eschews open waters, the risk is lower.

To put that risk in perspective, there were just over 40,000 motor vehicle deaths in the United States in 2017, a typical year. That works out to 1 death in 8,078 people, or one death per every 80 million miles traveled. So, the annual risk of dying from an unprovoked shark attack equals the risk of driving 0.04 miles — or 217 feet.

What about the risk of terrorism, which has received a lot of attention (and government spending) over the last two decades? During the 22-year period from 1995 to 2016, which includes the 9/11 attacks in 2001, there were a total of 3,277 fatalities from terrorism in the United States, according to data from the University of Maryland’s National Consortium for the Study of Terrorism and Response to Terrorism. Using those numbers, the risk of death from terrorism for the average American is about 1 in 2.2 million per year. That equals the risk of driving 37 miles. Put another way, in a given year we should be as fearful of driving 37 miles as we are of dying from a terrorist attack.

The troubling deaths of George Floyd and some other African Americans at the hands of police officers in recent years have raised questions about the risk posed by police, especially to African-Americans. According to a database assembled by the Washington Post, an average of 62 unarmed people die each year at the hands of the police. That is about the same number of deaths caused by “contact with hot tap water” and “contact with hornets, wasps and bees,” according to data from the National Safety Council. That works out to a risk of 1 in 5.2 million. For males, the risk increases to 1 in 2.8 million per year (57 out of 159 million men) and 1 in 1 million for black men (21 out of 21 million black men). That is roughly the same risk as death from “exposure to excessive natural heat” or from drowning in a bathtub, according to the National Safety Council. Compared to the risk from motor vehicles, the chance of death for the average unarmed American is about the same as the chance of death from traveling 15 miles by car; for an unarmed male it is about the same as the risk of traveling 29 miles, and for an unarmed African-American male it is about the same as the risk from traveling 80 miles. While worrisome as a justice issue, those deaths of black men are not much of a risk concern.

Larger risks: The typical American faces much greater risk of death from comparatively mundane causes. Heart disease kills about 1 in 502 Americans each year, while cancer kills 1 in 542.

The number of deaths from seasonal flu varies significantly from year to year, but it has averaged about 40,000 in the United States in recent years, which works out to 1 death in 8,125 Americans. The good news is that rate has fallen significantly over the decades; if the death rate from flu in the 1950s and 1960s were applied to today’s population, we would see over 160,000 deaths per year.

If the death rates from these diseases seem high, it is because they are. Heart disease alone kills as many Americans each year as the combined U.S. combat casualties from all American wars.

Where does COVID-19 fit? As this was written in mid-October, COVID-19 had killed more than 220,000 Americans. How many Americans will it kill in 2020? No one knows, but a reasonable guess is around 300,000, which would be about 11% of total U.S. deaths in 2017. That would mean that 1 in every 1,100 Americans will have died from COVID-19 in 2020. That would be higher than the death rate from the flu in the 1950s and 1960s, but it would be substantially below the tolls for both heart disease and cancer. Put another way, COVID-19 subjects us to a risk equivalent to that of driving about 73,000 miles. Table 1 presents statistics on various risks.

Putting all these numbers in perspective, the average American is 4 million times more likely to die from heart disease or cancer than from a shark attack and 20,000 times more likely to die from heart disease or cancer than from the police (assuming the person is unarmed). The average American is 4.2 times as likely to die from heart disease or cancer as from COVID-19 and 9.4 times as likely to die from anything other than COVID-19 as from COVID-19.

FDA reform: What policy responses could lower Americans’ risk of dying from the medical conditions discussed above? We have one recommendation: reform the U.S. Food and Drug Administration.

In 1962, Congress passed the Kefauver–Harris Amendments, which substantially increased the threshold for FDA approval of new drugs by requiring drug companies to prove efficacy in addition to safety. This may seem like a good idea, but it significantly slowed the approval of new medicines and reduced the number of new drug launches, while apparently not improving the efficacy of those medicines.

In 1973, economist Sam Peltzman analyzed the effects of Kefauver–Harris by comparing the number of new chemical entities (not just reformulations) approved by the FDA before the law changed with the number approved in the decade after the change, as well as econometrically estimating values for that same period if there had been no policy change. He found that the number of new drugs approved dropped by 60%, a number that should disturb Americans. According to his model, there should have been about 40 new approvals each year after the new law, but instead there were just 16.

Subsequent studies have concluded that this change in the law did not result in weeding out inferior drugs. According to Henry Grabowski and John Vernon, “In sum, the hypothesis that the observed decline in new product introductions has largely been concentrated in marginal or ineffective drugs is not generally supported by empirical analyses.” Peltzman himself came to this same conclusion, seeing the culling “as … if an arbitrary marketing quota … had been placed on new drugs after 1962.”

How could the FDA’s new rules not have affected the ratio of good to bad drugs? Peltzman surmised, “The penalties imposed by the marketplace on sellers of ineffective drugs before 1962 seem to have been sufficient to have left little room for improvement by a regulatory agency.” In other words, the market did as good a job of weeding out ineffective drugs before 1962 as Kefauver–Harris did afterward, without the harmful side effect of dramatically lowering the release of new drugs.

Why should we care that there have been fewer new drug approvals? Columbia University’s Frank Lichtenberg studied the effects of pharmaceutical innovation by comparing the vintage (the launch year) of drugs used in a country with the increase in national life expectancy at birth. This research was motivated, interestingly enough, by research that showed a positive correlation with higher productivity in manufacturing firms that used more modern, or later-vintage, equipment. Could a similar pattern be seen with drug usage?

Here is what he found:

As with all retrospective studies, Lichtenberg’s work shows only correlations and cannot prove cause-and-effect. Yet if this relationship is causal, three-quarters of the increase in life expectancy we have enjoyed in recent times is due solely to our adoption of newer drugs.

New pharmaceuticals are friends, not foes. Yet, the FDA treats all new drugs as guilty until proven innocent. The Kefauver–Harris Amendments have augmented that attitude. That attitude should be reversed.

If we want to lower our risk from COVID-19, one of the most fruitful approaches would be to speed up research and approvals for drugs that could treat the disease. An easy way to do that during the coronavirus pandemic would be to roll back the FDA’s efficacy requirement for COVID-related drugs. The same goes for cancer and heart disease drugs. If drug companies can demonstrate safety, let the drug makers market or otherwise distribute their new medicines and allow researchers, doctors, patients, and hospitals to evaluate efficacy. This approach has the potential to save lives today.

• “Pharmaceutical Innovation and Longevity Growth in 30 Developing and High-Income Countries, 2000–2009,” by Frank R. Lichtenberg. Health Policy and Technology 3(1): 36–58 (2014).
• Regulation of Pharmaceutical Innovation, by Sam Peltzman. University of Chicago Press, 1974.
• The Regulation of Pharmaceuticals: Balancing the Benefits and Risks, by Henry G. Grabowski and John M. Vernon. American Enterprise Institute Press, 1983.