Avoidable Mortality

V. Key Health Policy Issues >> A. Burden of Illness >> Avoidable Mortality (last update 12.4.17)

Total Deaths, by Cause

Major Causes of Death

Methods of Recording and Classifying Deaths

Avoidable Causes of Death


These items all provide detailed syntheses of evidence regarding the leading causes of death in the U.S. (listed in reverse chronological order).

  • Wikipedia. Annual number of deaths and causes
  • J. Michael McGinnis. Table 3-1. Rounded Estimates of Preventable Factors Causing U.S. Premature Mortality in 1990, 2000, and 2010. Updates McGinnis/Mokdad estimates to 2010.
  • Ali H. Mokdad; James S. Marks; Donna F. Stroup; Julie L. Gerberding. Actual Causes of Death in the United States, 2000 Journal of the American Medical Association. 2004;291(10):1238-1245. This updates to 1993 McGinnis/Foege estimates to 2000. The leading causes of death in 2000 were tobacco (435 000 deaths; 18.1% of total US deaths), poor diet and physical inactivity (400 000 deaths; 16.6%), and alcohol consumption (85 000 deaths; 3.5%). Other actual causes of death were microbial agents (75 000), toxic agents (55 000),motor vehicle crashes (43 000), incidents involving firearms(29 000), sexual behaviors (20 000), and illicit use of drugs (17 000).
  • C. Conover. Leading Cause of Preventable Deaths, 1993. Duke University, Center for Health Policy. Adds additional causes of preventable death such as hospital deaths due to negligence and updates figures to 1993.
  • J. Michael McGinnis and William H. Foege. Actual Causes of Death in the United States. Journal of the American Medical Association November 10, 1993; 270 (18): 2207-2212. Provides 1990 estimates of leading causes of death based on underlying determinants (e.g., tobacco, diet/inactivity) rather than disease classification.

Major Causes of Avoidable Deaths

Interventions that Significantly Reduce Death Risk

Listed in order of percent reduction in lifetime risk of death.

Running. A 2017 study shows running adds 7 hours of life expectancy for each hour spent running. This may or may not be a sensible investment depending on how much an individual discounts future gains in life expectancy.

Avoiding Obesity. A 2016 meta-analysis of 239 worldwide studies on obesity concluded that relative to those with BMI of 22.5-25.0 (normal weight) the increase in death risk for those who are slightly overweight (25.0-27.5) is 7%; for those who are more overweight (27.5-<30.0) is 20.0%; for those who are moderately obese (30-<35.0) is 45%, those who are very obese (35.0-<40.0) is 94% and those who are morbidly obese (40.0+) is 176% (Table 2).

Church-Going. Researchers used data from a long-term study of 75,534 women that tracked their health and lifestyle, including their attendance at religious services, over 16 years through 2012. After controlling for more than two dozen factors, they found that compared with those who never went to church, going more than once a week was associated with a 33% lower risk for death from any cause, attending once a week with a 26% lower risk, and going less than once a week a 13% lowered risk (JAMA Internal Medicine, June 2016).

Owning a Dog. A Swedish study on 3.4 million Swedes ages 40 to 80 found that owning a dog was associated with a 20% reduction in all-cause mortality. This study followed adults over a 12 year period and controlled for a variety of socioeconomic, demographic and lifestyle characteristics. It was 100 times larger than any previous study of this relationship (Scientific Reports. 11.17.17).

Eating Whole Grains. Using data from 45 studies, researchers calculated that compared with eating none, 90 grams of whole grains a day reduced the risk for all-cause mortality by 17% (BMJ, 6.14.16). A different analysis of 14 prospective studies found that compared to those who at the least whole grain foods, those who ate the most had a 16% reduction in all-cause mortality (Circulation, 6.14.16).

Mortality Risk

Annual and Lifetime Risk of Death

  • Insurance Information Institute. Odds Of Death In The United States By Selected Cause Of Injury, 2014. Annual and lifetime risk of death in U.S. for 17 different types of accident or injury.
  • Political CalculationsThe Odds of Dying in the U.S. Annual and lifetime risk of death in U.S. for 2 dozen different types of accident or injury.
  • Political CalculationsThe Odds of Dying, Again!  Lifetime risk of death in U.S. for 3 dozen different causes.
  • Political Calculations. Probability of living to 100 calculator.
  • DeathRiskRankings.org (Carnegie Mellon University) calculated your risk of dying in the next year and allows you to compare that risk to others in the world. Not available since 2011. Wayback Machine archived version.
  • American College of SurgeonsACS Surgical Risk Calculator. The calculator estimates the chance of an unfavorable outcome (such as a complication or death) after surgery. The risk is estimated based upon information the patient gives to the healthcare provider about prior health history. The estimates are calculated using data from a large number of patients who had a surgical procedure similar to the one the patient may have.

Life Expectancy Gains from Elimination of Causes of Death

Life Table Analysis. The most accurate way of calculating LE gains from elimination of a single cause of death is to use a life table. This entails setting the age-specific-mortality-rate (ASMR) for any given risk equal to zero while keeping ASMRs for all other risks at their current level and then simulating the new (higher) life expectancy from the old life expectancy to determine the increase (p. 99).

Note this only works if mortality risks by age are readily available for a given cause of death and if the cause of death eliminated is uncorrelated with other causes of death (which may be true for “shocks” such as accidents but often is not true in the case of chronic illnesses: in the latter cases, this method will tend to overstate the gains in life expectancy because even if a given cause is eliminated, the individual will die of a competing cause at a very similar age etc.).

GM Longevity Gain Correlation. However, a shortcut produces reasonably good approximations of mortality gains as using the more complex life table analysis (p. 99):         Longevity Gain= Crude Mortality Rate x .02, where CMR=deaths per 100,000.

Discounted Loss of Life Expectancy (LLEd)

Definition. This metric, LLEd, was first proposed by health economist Kip Viscusi and colleagues. LLE is defined as the mean average number of years of life expectancy lost per victim of a condition. That is, for each victim, remaining years of life expectancy can be obtained from a life table based on age of death and these are averaged across all victims of that condition to obtain the mean value across all victims in a given year. LLEd simply discounts future years based on a standard discount rate, e.g., 3%.

Computation of LLEd

Conditions Causing Immediate Death. E(YLL) is equal to LLE among decedents of a given condition times the lifetime probability of death from that condition:

p x ([(1+r)^(LLE)]-1)/r(1+r)^(LLE-1)

So if lifetime probability of drowning is 13 in 10,000 and LLE for drowning victims is 42.89 and a 3% discount rate is used (Table 3),

.0013 x ([(1+.03)^42.89]-1)/(.03(1+.03)^41.89) = .03 years

Conditions Causing Delayed Death. In the case of risky behaviors (e.g., many cancers) whose effects are not immediate, the formula is identical except that it is multiplied by (1+r)^-s, where s is the length of the lag time before that condition increases mortality risk. If s=10 years, then (1+r)^-10 = 0.744 meaning discounted E(YLL) is reduced by an additional 26.6%.

Estimates of LLEc for Common Activities

W. Kip Viscusi, Jahn Hakes and Alan Carlin. Measures of Mortality Risks. Report prepared for Office of Policy, Planning and Evaluation, EPA, October 1995.

  • Table 1 provides probability of death, undiscounted LLE and E(YPLL) and discounted LLE and E(YPLL) for 12 major conditions ranging from heart disease to congenital anomalies.
  • Table 3 provides probability of death, undiscounted LLE and E(YPLL) and discounted LLE and E(YPLL) for 29 various conditions ranging from heart disease to measles.

Undiscounted Loss of Life Expectancy (LLEc)

Definition. This metric, LLEc, first proposed by physicist Bernard Cohen in 1979, converts death risks into an average (undiscounted) loss of life expectancy for an entire population. Cohen uses the term “LLE” but this value is denoted LLEc here to differentiate it from the undiscounted LLE metric used by Viscusi (above). What Cohen terms LLE is equivalent to the undiscounted E(YLL) figures estimated by Viscusi. Among health economists, the discounted E(YLL) is theoretically superior since it takes into account that most people value future years of life less than the current year. However, the LLEc estimates are still useful insofar as they can be readily converted to discounted E(YLL) estimates simply by taking the present value of the LLEc: (((1+r)^LLEc)-1)/(r*(1+r)^(LLEc)).

Computation of LLEc

In his Updated Catalog of Risks, Cohen offers a variety of ways of calculating LLEc depending on what information is available.

  • Life Table Analysis. If information is available, the life table analysis described above is the most accurate way of calculating LLEcs.
  • Fixed Population-wide Increases in Risk. If the entire population is subject to an increase in a risk less than 1/1000, then LLEc can be approximated with the formula 1.1 x 10^6 d x r, where 10^6=10 to the 6th power and r=increase in risk. Thus if the entire population faces an increased risk of 1 in 10,000 (e.g., due to asteroid strike), then 1.1 x 10^6 x 10^-4 = 110 days (equation 4).
  • Proportionate Population-wide Increases in Risk. If the entire population faces a risk that ranging from an 80% reduction in the chances of death to a quintupling of the chance of death, then LLEc can be approximated with the formula 13 yr. x ln f, where f is a multiple of the current risk of death. Thus if the entire population faces a mortality increase of 50% across all ages (i.e., f=1.5), then 13 x ln 1.5 = 5.3 years (equation 5).
  • Lifetime Risks. The lifetime probability of death associated with a condition closely matches the proportions of fatalities in a given year attributable to each cause of death after allowing for demographic changes in age distribution. Consequently, LLEc can be estimated by multiplying the fraction of annual deaths due to a given cause times the average LLE among decedents from that cause: p x N/(Total Deaths), where p=average remaining life expectancy among those who die and N=total annual deaths from that cause. There were 2,712,630 deaths in 2015. There are 1,269 deaths due to choking and the average age at death is 65.1 meaning remaining life expectancy for victims is 20.2 years (note 5). Thus, the lifetime risk of death from choking is (1,269/2,712,630) x 20.2 x 8760 hrs./year = 82.8 hours of lost life expectancy (equation 8).
  • Annual Loss of Life Expectancy. If the risk of death from an activity is 1 in 1 million and the average age of death of participants in that activity is 40, then one can use a conventional life table to find the remaining years of life expectancy at age 40 (e.g., 41.7 years from Table 1) and then compute annual LLEc in minutes as (1/1,000,000) x (41.7) x 365 x 24 x 60 = 21.9 minutes.
    • This means that each person engaged in that activity implicitly faces an expected loss of 21.9 minutes of life expectancy (just as someone who bets $6 to roll a 1 on a die loses an expected $5 per roll: sometimes that person wins $6, but 5 out of 6 times, they lose $6, so the average loss per roll is $5 etc.).
    • Put another way, if 1 million participate, 1 unlucky person will lose the remaining 41.7 years of their life, but averaged over everyone (since there is no way of telling ex ante who this unlucky person will be), the loss per person is 1 millionth of this amount.
  • Note that these calculations all provide a very rough approximation of undiscounted LLE and is quite different than how gains in life expectancy from elimination of risks would be calculated by demographers. That method, however, using life tables, is considerably more complicated.

Estimates of LLEc for Common Activities

  • Cohen, Bernard. Chapter 8: Understanding Risk, in The Nuclear Energy Option. Extensive listing of LLEc for nearly 50 everyday risks, including explanation of how calculations derived and extensive list of sources. Fig. 1 lists the risks and their associated LLEc in rank order.
  • Cohen, Bernard. Catalog of Risks Extended and Updated. Health Physics, September 1991. Extensive listing of LLEc for a variety of everyday risks, including explanation of how calculations derived and extensive list of sources.
  • Cohen, Bernard. Risks in PerspectiveJournal of American Physicians and Surgeons Volume 8 Number 2 Summer 2003. Updated listing of loss of LLEc for nearly 50 everyday risks; sources listed are prior Cohen compilations. Fig. 1 shows the list in rank order. He includes:
    • Disease-Caused Mortality
    • Accident-Caused Mortality
    • Occupation-Related Mortality
    • Lifestyle-Related Mortality
    • Economic Status
    • Environmental Risk Factors
    • Natural Hazards


First proposed by Stanford professor Ronald Howard,  a micromort equals a risk of 1 death per million and is the easiest way of comparing everyday risks that are relatively small in magnitude.

  • Understanding Uncertainty. Micromorts. Calculator contains a number of mortality risks, expressed in micromorts (chance of death per million), including various types of transportation (miles traveled per micromort; micromorts per 100 miles); selected activities (e.g., skiing); risk of getting out of bed at various ages; and various health activities (e.g., giving birth, having anesthesia).
  • Hassan Vally. What’s most likely to kill you? The more micromorts, the more chance of dying. Includes micromort estimates for about 10 causes of death, including running a marathon, climbing Everest, swimming, shark attacks etc.

Life Expectancy

Life Expectancy and Economic Development

  • Haacker, Markus.  Contribution of Increased Life Expectancy to Living StandardsThe paper provides an analysis of the contribution of increasing life expectancy to living standards across countries. Building on an intertemporal utility framework and the literature on the value of statistical life, it analyzes contributions of economic growth and increasing life expectancy to welfare for 20 countries from 1870, offers an analysis with a near-global scope from 1950, and covers the adverse implications of a negative health shock (HIV/AIDS). Various measures of life expectancy are explored, and the implications of different methods of discounting are discussed.

One Response to Avoidable Mortality

  1. Pingback: ルイヴィトン サイフ

Leave a Reply

Your email address will not be published. Required fields are marked *