Posts Tagged ‘baseball’

Baseball is a sport that is enjoyed in many nations around the world. Beyond the United States and Canada, baseball is also played professionally in several countries in Asia, Latin America and the Caribbean, and players from more than 50 nations have played in the Majors.

Life expectancy at birth (a composite measure of mortality experience of a population) has been shown to vary greatly by country. While it seems clear that economic factors play a major part in these differences, there is the obvious potential for independent genetic and cultural causes as well.

This led me and my colleague Steven Day to wonder if the relative mortality experience of baseball players from other countries would mirror that of the general populations in question. That is, would players from somewhere such as the Dominican Republic — which has a lower life expectancy at birth than does the U.S. — have higher mortality than U.S. players, or would it even matter? What about players from Canada, where life expectancy is actually higher than the U.S.? Would those players continue to experience lower mortality or would their participation in MLB alter their experience somehow? We hypothesized that players from first-world countries with high standards of living would display no differences in mortality compared to U.S. players, while countries from Latin America and the Caribbean (with their high crime rates and extreme poverty) would demonstrate higher mortality rates than U.S. players. This research is the basis for a poster we are presenting at the 3rd North American Congress of Epidemiology next week, on June 23, 2011.

To test these hypotheses we used data on all baseball players who debuted in MLB between 1900 and 1999. From these, we selected all the players from the 6 best-represented nations: Canada, Cuba, Dominican Republic, Mexico, Puerto Rico, and Venezuela. Since many of these nations did not have any players in MLB until the 1950s, the study period was limited to 1950-1999. We calculated age and decade-specific Standardized Mortality Ratios (SMR) using mortality rates from U.S. baseball players as the population comparison rate.

Results show that for the most part MLB seems to homogenize mortality among the players. SMR for all the nations except D.R. and Venezula were very close to 1.00 with confidence intervals that included 1.00. The SMR for D.R. was fairly high at 2.38, but the 95% confidence interval of 0.90 to 5.06 includes 1.00. Venezuela had significantly elevated mortality risk, with an SMR of 3.14 (95% CI = 1.10, 6.95).

Though we cannot be sure, we speculate that Venezuela’s increased mortality is the result of Third-World violence. The CIA Factbook warns would-be travelers against Veneuzuela’s high crime rate, particularly murders, for which Venezula is a world leader. That D.R.’s mortality rates were elevated and approached significance supports this idea, as they too have widespread poverty and a high crime rate, though to a lesser extent than Venezuela. Though it may be tempting to wonder why Mexico did not show an increase in mortality given its recent rise in violence, recall that this study was through 1999 only.

It seems then that the lifestyle of MLB — physical fitness, travel, money, fame, access to health care, etc. — pulls the mortality of players from most nations up or down to unity with that of players from the United States, an idea we intend to test further in the near future. If true, it is an important piece of the puzzle concerning mortality rates, athleticism, and economics.


Read Full Post »

In my last post, I talked about how Major League Baseball makes for an excellent longitudinal occupational research cohort. In this post I would like to quickly discuss some of the published research that has come from MLB data.

By far the most popular epidemiological outcome using the MLB data is mortality. The first analysis of baseball player mortality was published in 1975 in the Statistical Bulletin of the Metropolitan Life Insurance Company (7). In the days before personal computers this study was conducted by abstracting the vital status of over 10,000 baseball players from the 1974 Baseball Encyclopedia and calculating standardized mortality ratios (SMRs). In comparison to the US general population, players whose careers began between 1876 and 1900 experienced only 97% of the expected mortality, those who debuted between 1901 and 1930 experienced only 64% of the expected, and those who debuted between 1931 and 1973 experienced a mere 55% of the expected deaths. Clearly even by the 1930s Major League baseball players were exhibiting the healthy worker effect and/or the health benefits of being a professional athlete.

A much more detailed analysis was conducted in 1988 by Waterbor et al. and published in the New England Journal of Medicine (9). This study followed 958 players who debuted between January 1, 1925 and December 31, 1984. The authors obtained death certificates on the decedents this time, and coded the deaths using the ICD-9 system. All-cause mortality was significantly reduced for the cohort as a whole (SMR=94%), several causes of chronic disease were insignificantly reduced, and the risk of death by arteriosclerotic heart disease and cancer were insignificantly elevated. Stratifying the analysis by player position, infielders were found to have the lowest SMR, and catchers had the highest; all other positions were not statistically different from 100. Longevity was also related to in-game achievements and weight-to-height ratio.

A group of researchers based at the University of Houston and the University of Texas Health Science Center at Houston published a paper on life expectancy in MLB in 2008 (8). They found that MLB players had a life expectancy anywhere from 4 to 6 years greater than the US GP at ages up to 50, and 1 or 2 years greater life expectancy at ages over 50.

Subsequent mortality analyses using baseball data focused on the effect of sinistrality (aka “lefthandness”) (1)(3)(4)(5), the effect of education (6), and the Healthy Worker Effect (2).

Baseball data are naturally useful for other purposes as well. MLB data have been used extensively in economics research, sports medicine, political science and of course, in analyses about the best way to play and win the game of baseball itself. As you can see, baseball offers a rich dataset that is useful in answering questions in a number of scientific disciplines.


1. Abel, E. L. and M. L. Kruger (2004). “Left-handed major-league baseball players and longevity re-examined.” Perceptual Motor Skills 99(3 Pt 1): 990-2.

2. Abel, E. L. and M. L. Kruger (2006). “The Healthy Worker Effect in Major League Baseball Revisited.” Research in Sports Medicine 14: 83-87.

3. Coren, S. and D. F. Halpern (1993). “A replay of the baseball data.” Perceptual Motor Skills 76(2): 403-6.

4. Fudin, R., L. Renninger, et al. (1993). “Sinistrality and reduced longevity: Reichler’s 1979 data on baseball players do not indicate a relationship.” Perceptual Motor Skills 76(1): 171-82.

5. Hicks, R. A., C. Johnson, et al. (1994). “Do right-handers live longer? An updated assessment of baseball player data.” Perceptual Motor Skills 78(3 Pt 2): 1243-7.

6. Kalist, D. E. and Y. Peng (2007). “Does Education Matter? Major League Baseball Players and Longevity ” Death Studies 31(7): 653-670.

7. MetLife (1975). “Longevity of major league baseball players.” Statistical Bulletin of the Metropolitan Life Insurancy Company 56: 2-4.

8. Saint Onge,J., Rogers, R., Krueger, P (2008). “Major League Baseball Players’ Life Expectancies.” Social Science Quarterly 89(3): 817-830.

9. Waterbor, J., P. Cole, et al. (1988). “The mortality experience of major-league baseball players.” New England Journal of Medicine 318(19): 1278-80.

Read Full Post »

Occupational epidemiology seeks to identify hazardous exposures in the workplace using various methods of research. The gold standard for such research is the occupational cohort study, in which workers are enrolled and followed throughout their career in an industry and beyond, if possible.

Some of the major challenges for any cohort study are selection bias and incomplete follow-up. This is indeed a poignant issue in occupational epidemiology as incomplete record keeping can make cohort enumeration difficult and workers can easily be lost when they leave employment.

To avoid these problems, an ideal occupational cohort would need to have several key characteristics:

1. A complete listing of all the workers who ever were in the industry;

2. Complete work-history information for each of these individuals, including hire and separation dates and time spent at various positions;

3. Complete vital-status for all members until death;

4. Extensive accrued follow-up time spanning a long period;

5. A solid and reliable exposure measure;

6. Be low-cost to collect, process, and analyze

It is not often that something in reality offers us a close approximation to the ideal, but for once we have one: Major League Baseball.

Examining baseball through the lens of the criteria listed above, we see the following:

1. A complete listing of all the workers who were ever in the industry.
All Major League players are represented in the data, no matter how much or how little they played. There are numerous records of players who, in the last game of the season, got to bat one time, then never played again.

2. Complete work-history information for each of these individuals, including hire and separation dates and time spent at various positions.
Data on Major League Baseball has complete seasonal performance information dating back to 1871.

3. Complete vital-status for all members until death.
The Society of American Baseball Research has a Biographical Research Committee, staffed by over 100 volunteers, whose sole mission it is to maintain accurate biographical data on Major League ballplayers, including date and place of birth and death. The committee researches player deaths using the Social Security Death Index, news media reports, and contact with surviving family members.

4. Extensive accrued follow-up time spanning a long period.
Between 1900 and 1999 alone, Major League Baseball players accrued over 311,000 person-years of observation time.

5. A solid and reliable exposure measure.
While there have been slight changes in the rules of the game, the basic physical requirements and duties of the player positions have not changed in at least 100 years. Number of games played, number of plate appearances, and number of innings pitched all make very reliable measures of exposure.

6. Be low-cost to collect, process, and analyze.
Baseball data (including biographical data) are available from several websites, depending on the sorts of information and level of detail one seeks. These sources offer free downloads.

Now of course no cohort is perfect, and MLB data does have some serious weaknesses to match its strengths.

One of the biggest concerns from an epidemiological perspective is that once players leave the Major Leagues, their occupations are generally unknown, save for those players who go on to coach or manage in the Majors. A related point is that in the late 1800s and early 1900s baseball players were not yet true professionals – that is to say, almost all of them had off-season jobs. This opens the cohort up to untold exposures that may dramatically alter the effects of having been a professional athlete. Similarly, as only mortality and not morbidity data are known, there could be serious confounding from other medical conditions that are impacting survival estimates.

In spite of these problems, many studies have been published using data from Major League Baseball. My next post will review and explore some of these studies.

Read Full Post »