By Bernard L. Cohen
Posted: Wednesday, February 27, 2002

ARTICLES
Publication Date: February 27, 2002

My favorite way of putting risks into perspective is to consider the average loss of life expectancy they cause, LLE (indicated in parentheses throughout this article). I present here a brief catalog of these, taken from my paper published in the September 1991 issue of Health Physics Journal.

Historically, diseases were one of the most important causes of life shortening, but now only heart disease (4.4 years), cancer (3.4 years), and stroke (250 days) cause LLE of more than six months. Aside from diseases, the principal direct causes of death are accidents (366 days), suicide (115 days), and homicide (93 days). Over half of all accident deaths are due to motor vehicles, and half of these are alcohol related. The most important other type of accidents are falls (28 days), suffocation (28 days), drowning (24 days), poison (20 days), and fires (20 days). From the standpoint of fatal occupational accidents, the most dangerous industries are construction (227 days) and mining (167 days); much safer are services (27 days), trade (27 days), and manufacturing (40 days).

Perhaps the best known risky behavior is smoking cigarettes (6.6 years for men, 3.9 years for women). Even more dangerous is being an alcoholic (12 years). Over-eating gives an LLE of about 36 days per pound, or one year for each 10 pounds overweight; being 20% overweight increases the fatality risk of heart disease by 29%, of cancer by 10%, of stroke by 15%, and of diabetes by 130%.

Having very poor, vs. very good, social connections correlates with LLE of 9 years. As one manifestation of this, remaining single rather than married has LLE of 5 years.

One of the greatest risks to an individual is living in poverty, LLE = 9 years for 19 large U.S. cities and for Montreal. In Britain, the difference in life expectancy between professional people and unskilled laborers is 7.2 years, and in Finland it is also 7.2 years. When Canadian men are ranked by income, those in the 90th percentile live 6 years longer than those in the 10th percentile. The latter have a higher mortality rate by 32% for heart disease and stroke, by 34% for cancer, and by 88% for accidents, poison, and violence. On an international scale, poverty plays a much bigger role — life expectancy is typically 30 years longer in well-to-do countries than in poor countries.

Life expectancy varies substantially with occupation. Post office employees, university professors, and workers in clothing manufacturing and in communications industries live 1-2 years longer than average, and miners, policemen, firemen, truck drivers, and fishermen die 2-3 years younger than average. But the most dangerous job is no job at all — unemployment. A 1% increase in national unemployment results in 37,000 deaths per year (plus 4200 admissions to mental hospitals and 3300 admissions to prisons).

If you believe the most dire warnings of environmental activists, you might add several other threats, though they would still be small compared to many of the problems listed above: air pollution (40 days), drinking water pollution (20 days), chemical residues in food (20 days), and chemicals released from consumer products (20 days). Media give wide publicity to cancer causing pollutants. Some of these are pesticide residues in food (12 days), tobacco smoke (8 days), other indoor pollutants (2 days), industrial air pollution (4 days), hazardous waste sites (2.5 days), drinking water contaminants (1.3 days), and all radioactivity releases from nuclear power including accidents (0.04 days). Note that nuclear power is an insignificant contributor to radiation exposure compared to radon in homes (25 days), other natural sources of radiation (10 days), and medical exposures (10 days).

Broiling meat produces carcinogens (0.1 day) and we produce similar carcinogens in bread crusts, toast, and fried potatoes. But everything man does, purposely or through pollution, is trivial in comparison with nature's contribution. All plants contain toxic chemicals to protect them from their natural enemies. Many of these chemicals can cause cancer, like nitrosamines in beets, celery, and lettuce; aflatoxin in peanuts, corn, and milk; sterigmatocystin in salami, ham, and wheat; hydrazines in mushrooms; allyl isothiocyanate in mustard, broccoli, and cabbage; safrole in pepper; tannins in coffee, tea, and wines; psoralens in celery and parsley; ethyl carbamate in bread, yogurt, beer, and wine; formaldehyde in fruits; benzene in eggs; methylene chloride in fats; coumarin in candy; diacetyl in coffee and butter; and flavonoids in fruits and vegetables. These are nature's pesticides, and per quantity ingested, they are typically as carcinogenic as man-made pesticides. But we eat 10,000 times as much of nature's pesticides as of man-made ones.

Natural catastrophes in the U.S. give the following relatively small LLE: hurricanes and tornadoes — 1.1 days; lightning — 0.7 day; storms and floods — 0.9 day, earthquakes and volcanoes — 0.2 day, heat waves — 0.7 day, cold waves — 2.1 days. Some similarly low risks are venomous plants and animals — 0.5 day (half from bee stings, and only 15% from snakes, lizards, and spiders) and dog bites — 0.12 day.

Historically, the great killers have been pestilence, war, and famine, with war often causing the other two. The best known pestilence epidemics have been the "Plague of Justinian" in AD 500-650 which killed 100 million, the "Black Death" in 1347-1351 which killed 75 million in Europe plus perhaps more than that in Asia, various diseases among American Indians due to contact with Europeans after 1492 that killed untold millions ( a large fraction of the Indian population), and the influenza epidemic of 1918-1919 which killed 20-50 million including a half million in U.S. AIDS is killing 12,000 Americans per year (LLE — 55 days for the average American). Fortunately, it does not spread through such efficient channels as coughing or food, but we have no guarantee against development of a new, equally powerful virus that does spread efficiently. And the best-understood natural disaster that could wipe out nearly all of mankind is the impact on the Earth of a large asteroid, expected once in a million years.

To help put some of the risks we have described (plus others) into perspective, a bar graph is attached in which the length of the bars gives the LLE. Asterisks (*) refer to effects averaged over the entire population, while those without asterisks refer just to people involved in the activity. The largest risks are shown at the top: alcohol, poverty, smoking, poor social connections, heart disease, and cancer each take years off a person's life expectancy. The smallest risk in that left section (the smallest of the large risks), motor vehicle accidents, is also shown as the largest risk in the middle group for which bar lengths have been multiplied by 20. This middle group consists mostly of risks widely recognized but not greatly feared. The smallest risk in this middle group, bicycles (the most dangerous transport per mile traveled), is also shown as the largest bar in the bottom group, for which bar heights have been multiplied by another factor of 50, a total factor of (20 x 50 =) 1000 over the high risk group.

In a rational society, the low risks shown in the bottom group should receive little consideration, but the public's attention is determined more by media coverage than by results of scientific risk analysis. The most glaring example of this is nuclear power, which is widely perceived by the public as being dangerous. We see from the bar diagrams that its perceived risk is a thousand times inflated. Few people take the time to rank risks rationally.

Bernard L. Cohen is a Professor Emeritus of Physics, University of Pittsburgh

Responses: March 1, 2002 Absolutely wonderful! This is one of the most purely intelligent things I've ever seen. People might remember it better if you called it "LOLE" rather than "LLE," but by any name, it is a profound analysis. All the best, —Bill Adams March 4, 2002 To: Elizabeth Whelan, ACSH President Congratulations on the excellent articles your organization provides for the general public. I just finished reading "How to Rank Risks." Unfortunately, it seems that the horror stories often provided by the various media have a much greater and more far reaching effect because they are so pervasive. I was a professor of Environmental Geology for more than thirty years and was constantly fighting the battle for responsible scientific evaluation of environmental issues. Most students, however, have been exposed to the typical doomsday rhetoric since their grade school days, and, therefore, have difficulty moving from their established emotional views to a more rational scientific view. When your book Toxic Terror was published, many of the sections were required reading for my classes, especially the section on Love Canal. Students were always amazed at the differences between the presentation in your book vs. the ones they typically read in environmental textbooks. I sincerely hope that your work is making inroads in the proper edification of the public on environmental and health issues. The general public needs people like you and your organization to save them from themselves. Keep up the good work. Sincerely, —Doug Sherman March 12, 2002 The risk assessment reasoning used by Bernie Cohen has a further dimension of importance after 9/11. As a risk analyst myself and a member of a current anti-terrorism panel (NAE), I know that panic reactions (public ones, corporate ones, and ones at multiple levels of government) are some of the most damaging consequences of possible future terrorist attacks. The misallocation of the national resources used to remedy life-reducing exposures and behaviors is implicit in the bar graph of Bernie Cohen's article. This misallocation has been called "committing statistical murders" by John Graham (on leave from Harvard Risk Center to OMB). If similar misallocation of resources occurs in the campaign to defend against terrorism, the effect of misinformation will be to multiply the damage and effectiveness of terrorist attacks. Misinformation about risks plays into the hands of terrorists. Education of the public is vital. Education of our teachers is perhaps equally vital. It is discouraging to see people with high academic standing repeat misinformed "conventional wisdom" on hazards and risks when testifying before Congressional committees. —Ed Zebroski March 15, 2002 Dear Professor Cohen: Your article on the ranking of risks, published by the American Council on Science and Health, provided a welcome antidote to the hysteria surrounding much relatively safe behavior that the general public considers to be very risky. I was surprised, however, to find that you ranked radon in homes as having an average loss of life expectancy (LLE) of 25 days compared to other sources of radiation and medical exposures (10 days each). That seemed relatively high. In writing Haunted Housing (Washington, D.C.: Cato Institute, 1997), I relied heavily on your finding that lung cancer rates decreased with increasing radon levels. Your article in Health Physics (vol. 57, no. 6) supported the theory of hormesis by showing that the average radon level in the "highest" lung cancer counties was only about one-half that in the counties with the lowest lung cancer. Since exposure to indoor radon may be beneficial, why should radon rank above pesticides in your chart? Have I missed a causative link? —Cassandra Chrones Moore Adjunct Scholar, Cato Institute Cohen replies: In my paper, I used the linear-no threshold theory of radiation-induced cancer, even though I personally am convinced that it greatly exaggerates the risks from low level radiation. I did this because linear-no threshold theory is still officially accepted — and even believed in many quarters — and I wanted to avoid side controversy in the paper on risks. It would not materially strengthen that paper. —Bernard L. Cohen Physics Dept. University of Pittsburgh May 24, 2002 Dr. Cohen, I found your article very thought provoking. While I think smoking is likely the riskiest environmental hazard faced by most Americans, I disagree with your view on radon. Very knowledgeable federal and state health officials have rated radon as one of the top environmental hazards. Surely, they are not exaggerating. A review of health-based comparative risk assessments in the United States by B.L. Johnson of the Department of Environmental and Occupational Health, Rollins School of Public Health, Emory University, Atlanta, Georgia (Reviews on Environmental Health 2000 Jul-Sep;15(3):273-87) found that comparing the risks posed by specific environmental hazards has become attractive to policy makers and legislative bodies as an aid to budgeting and other policy decisions. A consolidation of the comparative risk assessments of fifteen states revealed good agreement with federal health-based environmental hazard priorities and partial agreement with local-government health departments. In descending order of priority, the highest-ranked environmental hazards to human health were indoor air pollutants (excluding radon), criteria air pollutants, hazardous air pollutants, indoor radon, lead contamination, inactive hazardous waste sites, and drinking water at the tap. Other very credible studies have also suggested radon is indeed a serious health threat. See: http://www.epa.gov/iaq/radon/iowastudy.html Sincerely, John Boll January 31, 2003 Bernard Cohen's article "How To Rank Risks" is useful and thought-provoking. However, I have to take issue with one comment. He states: "the most dangerous job is no job at all — unemployment. A 1% increase in national unemployment results in 37,000 deaths per year (plus 4,200 admissions to mental hospitals and 3,300 admissions to prisons)." Presumably, these figures were assessed based on the relative risks of death and admission to mental hospitals or prisons among unemployed versus employed people. It is not at all clear that unemployment caused those problems; in fact, it seems equally plausible that one risk factor for unemployment is having serious psychological or other problems that pre-date the unemployment. At times when the unemployment level is extremely low, the majority of unemployed will be people with serious barriers to employment, such as psychological problems, alcohol or drug addition, or low intelligence. At times when the unemployment level is higher, then a larger fraction of the unemployed will be people who are otherwise normal and healthy, but just unlucky. Therefore, the figures cited by Cohen are probably upper bounds on the deaths and admissions that would be caused by a 1% increase in unemployment, and quite possibly very loose upper bounds. Vicki Bier University of Wisconsin