0
Articles |

Survival in Academy Award–Winning Actors and Actresses FREE

Donald A. Redelmeier, MD; and Sheldon M. Singh, BSc
[+] Article and Author Information

From University of Toronto, the Clinical Epidemiology and Health Care Research Program, Sunnybrook and Women's College Health Sciences Centre, and the Institute for Clinical Evaluative Sciences, Toronto, Ontario, Canada.


Acknowledgments: The authors thank Susan Campbell for data entry; Robert Tibshirani and Jerry Lawless for statistical insights; and Peter Austin, Ahmed Bayoumi, Chaim Bell, Victor Fuchs, David Juurlink, David Naylor, Miriam Shuchman, Leonard Syme, and John-Paul Szalai for commenting on drafts of this manuscript.

Grant Support: Dr. Redelmeier was supported by a career scientist award from the Ontario Ministry of Health, the de Souza Chair in Clinical Trauma Research of the University of Toronto, and the Canadian Institute for Health Research. Mr. Singh was supported by the Jane and Howard Jones Bursary at the University of Toronto.

Requests for Single Reprints: Donald A. Redelmeier, MD, Sunnybrook and Women's College Health Sciences Centre, Room G-151, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada; e-mail, dar@ices.on.ca.

Current Author Addresses: Dr. Redelmeier: Sunnybrook and Women's College Health Sciences Centre, Room G-151, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada.

Mr. Singh: Sunnybrook and Women's College Health Sciences Centre, Room G-106, 2075 Bayview Avenue, Toronto, Ontario M4N 3M5, Canada.

Author Contributions: Conception and design: D.A. Redelmeier, S.M. Singh.

Analysis and interpretation of the data: D.A. Redelmeier, S.M. Singh.

Drafting of the article: D.A. Redelmeier, S.M. Singh.

Critical revision of the article for important intellectual content: D.A. Redelmeier, S.M. Singh.

Final approval of the article: D.A. Redelmeier, S.M. Singh.

Provision of study materials or patients: S.M. Singh.

Statistical expertise: D.A. Redelmeier.

Obtaining of funding: D.A. Redelmeier, S.M. Singh.

Administrative, technical, or logistic support: D.A. Redelmeier.

Collection and assembly of data: D.A. Redelmeier, S.M. Singh.


Ann Intern Med. 2001;134(10):955-962. doi:10.7326/0003-4819-134-10-200105150-00009
Text Size: A A A

Social status is a consistent, powerful, and widespread determinant of death rates. The association between high status and low mortality has appeared throughout the world, has persisted for more than a century, and extends to diverse illnesses (14). Uncovering the mechanisms by which external factors (such as income and level of education) influence biological processes (such as the endocrine and immune systems) represents a major challenge for health scientists and a core issue for public policy. Research is difficult because many determinants of social status are closely interrelated, such as education with income. Animal models are unrealistic, aside from some primate studies, and randomized trials are impractical, aside from a few lottery winners.

Movie stars are an interesting group for the study of social status and health outcomes. First, performers can earn an enormous income without a substantial amount of education. Second, celebrity publicity is often boosted by more sustained promotion than is the attention given to politicians, singers, athletes, and other luminaries. Third, the lifestyles of movie stars can be notorious for extremes of competition, leisure, and excess. Fourth, they are highly visible public figures whose birthdays and deaths are regularly reported. Finally, big breaks to stardom are often haphazard and heavily dependent on chance. Indeed, some pundits suggest that being nominated for an Academy Award is due to talent whereas winning one is due to luck.

We wondered whether the Academy Awards might shed light on how social status affects all-cause mortality. We chose this event because it generates substantial attention; for example, in 1996, almost as many Americans watched the Academy Awards as voted in the presidential election (56). Moreover, the event is televised to more than 100 countries and has an estimated viewing audience of more than 1 billion people, making its broadcast one of the most widely shared current human experiences. Our theory was that winners would gain an important increase in their status but no increase in their formal education. The primary hypothesis was that the survival of winners would differ from that of less recognized performers.

Academy of Motion Picture Arts and Sciences

Membership in the Academy of Motion Picture Arts and Sciences is limited by invitation from the Board of Governors to those with movie distinctions, currently totals about 6000 persons, and has 13 branches (for example, an actors' branch that includes about 1000 persons). The annual awards selection process is complex and is described in detail elsewhere (http://www.oscars.org). In brief, each December the Academy compiles a list of films that are eligible for an award; each cast member in these films is eligible to be nominated for an acting award. In January, the list is sent to all Academy members and those in the actors' branch are invited to nominate five individuals in each of four acting categories. In February, the nominations are tabulated, the top five nominations in each category are identified, and all Academy members vote for one person in each category. The Academy Award goes to the person with the most votes.

Selection of Performers

We identified every person nominated for an Academy Award for acting. To do so, we obtained a full listing of all actors and actresses, along with the film in which they performed, from the Academy. The selection interval spanned from the inception of the Academy Awards to the present (72 years). For each performer, we also identified another cast member who performed in the same film as the nominee, was the same sex, and was born in the same era. This ensured that both were alive, working, prevailing in casting calls, winning good movie roles, and eligible for a nomination. In cases where several matches were possible, we picked a same-sex cast member by formally checking dates and choosing the one whose birth date was closest to that of the nominated performer.

Example of Matching Process

For clarity, we provide an arbitrary example of this matching process to illustrate the underlying method. Kate Winslet was nominated for the leading actress award in 1997 for her performance as the character Rose DeWitt Bukater in the movie Titanic. Five other women were cast members in that film, including Suzy Amis, who performed as the character Lizzy Calvert. Kate Winslet was born in 1975, and Suzy Amis was born in 1961; these two people had a 14-year difference in age. The other four women, Kathy Bates (born in 1948), Frances Fisher (born in 1952), Jenette Goldstein (born in 1960), and Gloria Stuart (born in 1910), all had an age difference greater than 14 years compared to Kate Winslet. Hence, Suzy Amis was selected as the match for Kate Winslet in this film.

Overall Matching Process

We repeated this matching process for all years and all four categories. No performer was excluded from analysis, and no performer was dropped because of missing data. Matches were not possible in some cases; for example, in 1951, Katharine Hepburn was nominated in a movie in which no other woman appeared. Otherwise, the matching process was uncomplicated and complete. In the matching process, we did not attempt to balance ethnicity, past experience, or future accomplishment of the performers. As a consequence, the person who performed opposite the nominee could previously have achieved or subsequently achieve greater recognition. Such potential misclassification might cause analyses to underestimate the differences attributable to winning an Academy Award.

Classification of Success

Many performers were eligible for inclusion on more than one occasion; for example, Katharine Hepburn won four Academy Awards during her career. We counted each person only once by categorizing performers according to their highest achievement. The three groups were termed “winners” (those who were nominated for and won at least one Academy Award), “nominees” (those who were nominated but never won an Academy Award), and “controls” (those who were never nominated and never won). For example, Jack Nicholson was classified as a winner because he had three wins, Richard Burton was a nominee because he was nominated seven times but never won, and Lorne Greene was a control because he was never nominated. Statistical tests based on counting performances rather than performers gave more extreme results and are not shown.

Determination of Death Rates

We collected data on each person's date of birth and death from the Internet through two databases: the All Movie Guide (http://www.allmovie.com) and the Internet Movie Database (http://www.imdb.com). Each source covers more than 100 000 movies, is updated continually, and undergoes extensive public scrutiny. Data were checked by consulting written publications, and conflicts were resolved by accepting information from printed sources over that found on the Internet (710). No birth dates were missing. Causes of death were sought by using the same methods and by inquiry to the National Film Information Service. In addition, we checked Internet sources that listed people who have sometimes been mistakenly rumored dead. People who were not reported dead were presumed to be alive.

Determination of Personal Characteristics

Additional data were retrieved by using methods similar to those described above, with the following exceptions. Determination of whether the person was born in the United States and whether the person had changed his or her name from the given name was made by using the All Movie Guide. Missing data were assumed to indicate the United States as the country of origin and no change in name. Ethnicity was determined by searching Internet sources and by viewing selected films. Although performers try to avoid being typecast, we classified each performer's main film genre according to that listed first by the All Movie Guide. Similarly, although the ratings given in film reviews are debatable, the All Movie Guide five-star ratings were considered to indicate high quality.

Setting Time-Zero

Research on the natural history of any condition requires identifying people at an early and uniform point in their course. Unstable definitions of “time-zero” might otherwise lead to distorted prognoses, an error called lead-time bias(1112). The baseline analysis in this study set time-zero as the performer's day of birth to conform to the accepted measure of longevity (13). Other analyses were conducted to test robustness. In the first of these, time-zero was set as the day on which each performer's first film was released. In the second, time-zero was set as the day of each performer's 65th birthday; therefore, all performers who died before 65 years of age were excluded. In the third analysis, time-zero was set as each performer's 50th birthday; all performers who died before 50 years of age were therefore excluded.

Reverse Causality

Survival analysis also requires avoiding artifacts related to survivor treatment-selection bias: That is, persons who are destined to live longer have more opportunity to gain special treatments, thereby potentially creating an illusory link between special treatments and longer survival (1416). One way to mitigate this bias is to use time-dependent covariates in a proportional hazards model, although doing so can produce a different bias in the opposite direction (1718). We analyzed survival both with and without a time-dependent step function for victory. In addition, we analyzed survival after adjusting for total films and total nominations in a person's career to see whether winning an Academy Award was distinct from other exposures that can accumulate over time.

Unmeasured Confounding

We used three strategies to test whether the survival associated with winning an Academy Award might be due to hidden confounding. First, we conducted analyses both with and without adjustments for baseline characteristics, on the rationale that if partial control based on available factors yielded only a small difference in estimates, then perfect control based on ideal factors would be less likely to yield a large difference in estimates. Second, we repeated all analyses by comparing winners with nominees and tested whether the survival difference persisted, on the rationale that nominees were intermediate between winners and controls in talent or other unknown factors. Third, we examined dose–response gradients by assessing survival in performers with multiple wins.

Statistical Analysis

The primary analysis compared mortality in Academy Award winners and controls. Survival was plotted by using the Kaplan–Meier method, life expectancy was estimated as the area under the curve, and comparisons were done by using the log-rank test. Regression analyses used the Cox proportional-hazards model to adjust for birth year, sex, ethnicity (white or nonwhite), birth country (United States or other), name change (yes or no), age at release of first film, and total films in career. Continuous covariates were coded as linear terms (models with quadratic and cubic terms yielded similar results and are not reported). The proportionality assumption was checked by inspection of log–log plots. Tests were done by using StatView software, version 5.0 (SAS Institute, Inc., Cary North Carolina), and SAS software, version 6.12 (SAS Institute, Inc.). All P values were two-tailed, and those less than 0.05 were considered statistically significant. Our data file is available on the Web site of the Institute for Clinical Evaluative Sciences (http://www.ices.on.ca).

Role of Funding Sources

The funding sources had no role in the design, conduct, or reporting of this study.

Overall, 1649 performers were nominated for an Academy Award or appeared opposite the nominated performer. The baseline characteristics of winners, nominees, and controls were similar (Table 1). In particular, the three groups did not differ greatly in birth year, sex, ethnicity, or country of birth, aside from a trend that nominees were born somewhat more recently than winners or controls. Fewer controls were listed as having a different name at birth, a finding perhaps related to lesser monitoring of this group. The median age at release of first film was 26 years, and almost all performers (98%) had started appearing in films by 49 years of age. The median age at first nomination was 35 years (identical for winners and nominees). Among winners, the median age at first award was 39 years, most (80%) had received an award by 49 years of age, and few (15%) had multiple wins.

Table Jump PlaceholderTable 1.  Baseline Characteristics

Each performer's career was assessed as the interval from their first to their most recent film credit (through the year 2000). On average, winners were in more total films than were nominees (58.9 vs. 47.4; P < 0.001). In contrast, nominees and controls had a similar number of total films (47.4 vs. 45.5; P > 0.2). Analyses of only films rated four stars or more revealed a similar pattern. For each group, the average performer was in about 1.5 films per year during their career. The most common film genre in each group was “drama,” and this was more frequent among winners than controls (82% vs. 72%; P = 0.003). Most winners and nominees received a first nomination within two decades of their first film (83% vs. 83%; P > 0.2). Among winners, information on level of education was available for 119 performers; half (61 of 119) had only a high school education (excluding honorary degrees).

A total of 772 performers had died by 28 March 2000 (median follow-up, 66 years from birth). A specific cause of death was listed for 556 performers and was not listed for 216 performers (Table 2). No major imbalances were seen among the three groups in identified causes of death. Ischemic heart disease was the most common cause of death and accounted for 23% of deaths overall (177 of 772 deaths). Injuries and poisoning occurred at all ages and accounted for 6% of the deaths overall (47 of 772 deaths). Of the 42 deaths from miscellaneous causes, 15 were due to postoperative complications and 8 were due to AIDS. Overall, almost all deaths (714 of 772) occurred after 50 years of age, and very few deaths (13 of 772) occurred within a decade of the performer's first film. Twenty performers were older than 90 years of age and were still alive at follow-up.

Survival was better among winners than among controls (Figure). The overall difference in life expectancy was 3.9 years (79.7 vs. 75.8 years; P = 0.003). The difference was similar for men and women (3.8 vs. 4.1 years; P > 0.2) but was greater for performers born in or after 1910 than for those born before or in 1909 (4.1 vs. 1.7 years; P = 0.015). The difference in life expectancy between winners and controls was 5.9 years (53.2 vs. 47.3 years; P < 0.001) in analyses based on survival after release of the first film, 2.5 years (83.0 vs. 80.5 years; P = 0.018) in analyses that excluded performers who died before 65 years of age, and 2.3 years (79.4 vs. 77.1 years; P = 0.028) in analyses that excluded performers who died before 50 years of age.

Grahic Jump Location
Figure.
Survival in Academy Award–winning actors and actresses (solid line) and controls (performers who were never nominated) (dotted line), plotted by using the Kaplan–Meier technique.P

Analysis is based on log-rank test comparing 235 winners (99 deaths) with 887 controls (452 deaths). The total numbers of performers available for analysis were 1122 at 0 years, 1056 at 40 years, 762 at 60 years, and 240 at 80 years.  = 0.003 for winners vs. controls.

Grahic Jump Location

The generally lower mortality hazard was equal to about a 28% relative reduction in death rates in winners (95% CI, 10% to 42%). Adjustment for birth year, sex, and ethnicity yielded similar results (Table 3). Accounting for birth country, name change, age at release of first film, and total films in career also made no large difference. Excluding performers who died before 50 years of age and those who won an award after 50 years of age yielded a relative reduction of 25% (CI, 2% to 42%), which decreased to 18% (CI, −7% to 37%) after adjustment for birth year, sex, and ethnicity. Analyses using time-dependent covariates, in which winners were counted as controls until the time of first victory, yielded a relative reduction of 20% (CI, 0% to 35%). Analyses excluding performers with multiple wins yielded a relative reduction of 25% (CI, 5% to 40%).

Table Jump PlaceholderTable 3.  Analysis of Death Rates

Additional analyses were done to evaluate the 762 performers who received at least one Academy Award nomination. Life expectancy was better for winners than for nominees (79.7 vs. 76.1 years; P = 0.013). This was equal to a 25% relative reduction in death rates (CI, 5% to 41%). Adjustment for demographic and professional factors yielded similar results, as did calculations based on time from first nomination rather than time from birth (relative reduction in death rate, 24% [CI, 3% to 40%]). Among winners and nominees, very few deaths occurred within a decade of their first nomination (4% vs. 5%; P > 0.2). Life expectancy was 2.7 years higher for performers with multiple wins than for those with single wins (82.0 vs. 79.3), but the difference was not statistically significant (P = 0.093). The award category (leading vs. supporting role) made no significant difference in survival.

Performers with long careers sometimes received more nominations than did those with short careers. The overall association was about 0.02 nomination (CI, 0.016 to 0.024 nomination) per year of career. The winners accumulated a total of 639 nominations, of which 362 were defeats. The nominees accumulated a total of 717 nominations. We found no association between number of defeats and reduced death rates, either for winners (relative reduction in death rate, 5% [CI, −6% to 15%]) or nominees (relative reduction in death rate, 2% [CI, −14% to 16%]). Analyses of winners and nominees together showed that each win was associated with a 22% relative reduction in death rates (CI, 5% to 35%), whereas each nomination otherwise was not associated with a significant reduction in death rates (relative reduction in death rate, 3% [CI, −6% to 11%]).

Other factors were also linked to longevity. Women lived longer than men (77.9 vs. 75.4 years; P < 0.001). Performers born in more recent decades also had reduced death rates (relative reduction in death rate, 6% [CI, 2% to 10%]). No other baseline characteristic (Table 1) was significantly related to survival. Similarly, total films in career—an indirect measure of income—was unrelated to death rates among winners (relative reduction in death rate, 0.1% [CI, −0.3% to 0.6%]), nominees (relative reduction in death rate, 0.3% [CI, −0.1% to 0.6%]), and controls (relative reduction in death rate, 0.0% [CI, −0.2% to 0.2%]). Age at first award was related to longevity, in that later wins led to smaller gains; for example, life expectancy beyond 50 years of age was greater for those who won in their 30s compared with those in their 40s (32.7 vs. 26.6; P = 0.007).

We found that winning an Academy Award was associated with a large gain in life expectancy for actors and actresses. The apparent survival advantage amounted to about 4 extra years of life (CI, 1.6 to 6.2 years), could not be explained by simple birth demographics, and was evident even though victory predated death by about four decades. Survival among performers who were nominated but did not win was about the same as that among performers with no nominations. Survival among performers with many nominations was no better than among those with single nominations, unless more nominations generated more wins. Our observations were not easily attributed to occupation, income, talent, random chance, measurement error, or reverse causality. Instead, the results suggest that success confers a survival advantage.

Several explanations might account for the increased survival of Academy Award–winning actors and actresses. Movie stars are often subjected to a personal scrutiny that far exceeds their dramatic achievements. They often need to preserve their image by continually avoiding disgraceful behaviors and maintaining exemplary conduct. They may be surrounded by managers and others who are invested in the person's reputation and can enforce high standards of behavior. They have personal chefs, trainers, nannies, or other staff that make it easy to follow the ideals of lifestyle. Furthermore, a movie star may have more control, ability to avoid stress, self-efficacy, resources, admirers, motivation, and access to special privileges than others in society. The full mechanism of the apparent survival benefit among successful actors and actresses is not known. Untangling the explanations is further complicated because some stars also engage in superstitious and deleterious behaviors.

Causal inferences should take into account possible confounding. Factors might develop before, persist for decades after, and be unaltered by the other effects of winning. These as-yet unidentified factors contribute to both victory and longevity but not to nomination. Such factors are equally important in men and women, are more intense in recent eras, and are unrelated to total films in a career. They predict who will win an Academy Award, will not change with repeated nominations, and do not differ for those in supporting rather than leading roles. Ambition, resilience, time preference, social support, work stress, environmental pollutants, or childhood experience (all of which have been suggested to play a role in survival) do not easily satisfy these conditions, but such factors are not impossible. A factor that was present in 80% of winners and 20% of controls would explain our findings if it created a 6.5-year survival difference.

Our study had two main limitations. First, information on many personal details, such as level of education, is not available for all performers. More biographical work is needed, especially because such factors as smoking and alcohol intake account for only a modest proportion of the social inequities in population mortality (19). Biographical work needs to include more performers than just extraordinary ones, because omitting performers with three or four Academy Awards still showed a 27% difference in survival (CI, 10% to 42%). Second, for some people the sting of defeat is more intense than the joy of victory (20). However, this asymmetry is unlikely to be the only explanatory factor, given that the average lifespan of controls was still much higher than that of the general U.S. adult population during the interval (21). Indeed, the results are surprising because performers sometimes understate their age, which would cause our data to underestimate their survival.

Winning an Academy Award can increase a performer's stature and may add to their longevity. The absolute difference in life expectancy is about equal to the societal consequence of curing all cancers in all people for all time (2223). Moreover, movie stars who have won multiple Academy Awards have a survival advantage of 6.0 years (CI, 0.7 to 11.3 years) over performers with multiple films but no victories. Formal education is not the only way to improve health, and strict poverty is not the only way to worsen health. The main implication is that higher status may be linked to lower mortality rates even at very impressive levels of achievement.

Marmot MG, Shipley MJ, Rose G.  Inequalities in death—specific explanations of a general pattern? Lancet. 1984; 1:1003-6. PubMed
CrossRef
 
Smith GD, Carroll D, Rankin S, Rowan D.  Socioeconomic differentials in mortality: evidence from Glasgow graveyards. BMJ. 1992; 305:1554-7. PubMed
CrossRef
 
Evans RG, Barer ML, Marmor TR, eds.  Why Are Some People Healthy and Others Not? The Determinants of Health of Populations. New York: A. De Gruyter; 1994:3-26.
 
Syme SL, Balfour JL.  Explaining inequalities in coronary heart disease. Lancet. 1997; 350:231-2. PubMed
CrossRef
 
Bureau of the Census. Statistical Abstract of the United States 1997. 117th edition. Washington, DC: U.S. Government Printing Office; 1997:287. Publication no. 003-024-08825-8.
 
ABC's Oscar ratings higher than last year. The Associated Press. 28 March 2000.
 
Walker J, ed.  Halliwell's Filmgoer's and Video Viewer's Companion. 12th ed. Toronto: Harper Perennial; 1997.
 
Truitt EM.  Who Was Who on Screen. 3rd ed. New York: RR Bowker; 1984.
 
Katz E.  The Film Encyclopedia. New York: Thomas Crowell; 1979.
 
Fuller G, Lloyd A.  The Illustrated Who's Who of the Cinema. London: Orbis; 1983.
 
Pelikan S, Moskowitz M.  Effects of lead time, length bias, and false-negative assurance on screening for breast cancer. Cancer. 1993; 71:1998-2005. PubMed
CrossRef
 
Tunnell RD, Millar BW, Smith GB.  The effect of lead time bias on severity of illness scoring, mortality prediction and standardised mortality ratio in intensive care—a pilot study. Anaesthesia. 1998; 53:1045-53. PubMed
CrossRef
 
Taylor R.  Measurement of population mortality. Kerr C, Taylor R, Heard G Handbook of Public Health Methods. New York: McGraw-Hill; 1998; 57-9.
 
Gail MH.  Does cardiac transplantation prolong life? A reassessment. Ann Intern Med. 1972; 76:815-7. PubMed
CrossRef
 
Cox DR, Oakes D.  Time dependent covariates.  Analysis of Survival Data. London: Chapman and Hall; 1984; 112-41.
 
Glesby MJ, Hoover DR.  Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996; 124:999-1005. PubMed
CrossRef
 
Katz MH.  Multivariable Analysis: A Practical Guide for Clinicians. Cambridge, UK: Cambridge Univ Pr; 1999; 173-5.
 
Fisher LD, Lin DY.  Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health. 1999; 20:145-57. PubMed
CrossRef
 
Lantz PM, House JS, Lepkowski JM, Williams DR, Mero RP, Chen J.  Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA. 1998; 279:1703-8. PubMed
CrossRef
 
Medvec VH, Madey SF, Gilovich T.  When less is more: counterfactual thinking and satisfaction among Olympic medalists. J Pers Soc Psychol. 1995; 69:603-10. PubMed
CrossRef
 
Bureau of the Census. Statistical Abstract of the United States 1996: The National Data Book. 116th ed. Washington, DC: U.S. Government Printing Office; 1996:88. Publication no. 003-024-08809-6.
 
Olshansky SJ, Carnes BA, Cassel C.  In search of Methuselah: estimating the upper limits to human longevity. Science. 1990; 250:634-40. PubMed
CrossRef
 
Detsky AS, Redelmeier DA.  Measuring health outcomes—putting gains into perspective [Editorial]. N Engl J Med. 1998; 339:402-4. PubMed
CrossRef
 

Figures

Grahic Jump Location
Figure.
Survival in Academy Award–winning actors and actresses (solid line) and controls (performers who were never nominated) (dotted line), plotted by using the Kaplan–Meier technique.P

Analysis is based on log-rank test comparing 235 winners (99 deaths) with 887 controls (452 deaths). The total numbers of performers available for analysis were 1122 at 0 years, 1056 at 40 years, 762 at 60 years, and 240 at 80 years.  = 0.003 for winners vs. controls.

Grahic Jump Location

Tables

Table Jump PlaceholderTable 1.  Baseline Characteristics
Table Jump PlaceholderTable 3.  Analysis of Death Rates

References

Marmot MG, Shipley MJ, Rose G.  Inequalities in death—specific explanations of a general pattern? Lancet. 1984; 1:1003-6. PubMed
CrossRef
 
Smith GD, Carroll D, Rankin S, Rowan D.  Socioeconomic differentials in mortality: evidence from Glasgow graveyards. BMJ. 1992; 305:1554-7. PubMed
CrossRef
 
Evans RG, Barer ML, Marmor TR, eds.  Why Are Some People Healthy and Others Not? The Determinants of Health of Populations. New York: A. De Gruyter; 1994:3-26.
 
Syme SL, Balfour JL.  Explaining inequalities in coronary heart disease. Lancet. 1997; 350:231-2. PubMed
CrossRef
 
Bureau of the Census. Statistical Abstract of the United States 1997. 117th edition. Washington, DC: U.S. Government Printing Office; 1997:287. Publication no. 003-024-08825-8.
 
ABC's Oscar ratings higher than last year. The Associated Press. 28 March 2000.
 
Walker J, ed.  Halliwell's Filmgoer's and Video Viewer's Companion. 12th ed. Toronto: Harper Perennial; 1997.
 
Truitt EM.  Who Was Who on Screen. 3rd ed. New York: RR Bowker; 1984.
 
Katz E.  The Film Encyclopedia. New York: Thomas Crowell; 1979.
 
Fuller G, Lloyd A.  The Illustrated Who's Who of the Cinema. London: Orbis; 1983.
 
Pelikan S, Moskowitz M.  Effects of lead time, length bias, and false-negative assurance on screening for breast cancer. Cancer. 1993; 71:1998-2005. PubMed
CrossRef
 
Tunnell RD, Millar BW, Smith GB.  The effect of lead time bias on severity of illness scoring, mortality prediction and standardised mortality ratio in intensive care—a pilot study. Anaesthesia. 1998; 53:1045-53. PubMed
CrossRef
 
Taylor R.  Measurement of population mortality. Kerr C, Taylor R, Heard G Handbook of Public Health Methods. New York: McGraw-Hill; 1998; 57-9.
 
Gail MH.  Does cardiac transplantation prolong life? A reassessment. Ann Intern Med. 1972; 76:815-7. PubMed
CrossRef
 
Cox DR, Oakes D.  Time dependent covariates.  Analysis of Survival Data. London: Chapman and Hall; 1984; 112-41.
 
Glesby MJ, Hoover DR.  Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996; 124:999-1005. PubMed
CrossRef
 
Katz MH.  Multivariable Analysis: A Practical Guide for Clinicians. Cambridge, UK: Cambridge Univ Pr; 1999; 173-5.
 
Fisher LD, Lin DY.  Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health. 1999; 20:145-57. PubMed
CrossRef
 
Lantz PM, House JS, Lepkowski JM, Williams DR, Mero RP, Chen J.  Socioeconomic factors, health behaviors, and mortality: results from a nationally representative prospective study of US adults. JAMA. 1998; 279:1703-8. PubMed
CrossRef
 
Medvec VH, Madey SF, Gilovich T.  When less is more: counterfactual thinking and satisfaction among Olympic medalists. J Pers Soc Psychol. 1995; 69:603-10. PubMed
CrossRef
 
Bureau of the Census. Statistical Abstract of the United States 1996: The National Data Book. 116th ed. Washington, DC: U.S. Government Printing Office; 1996:88. Publication no. 003-024-08809-6.
 
Olshansky SJ, Carnes BA, Cassel C.  In search of Methuselah: estimating the upper limits to human longevity. Science. 1990; 250:634-40. PubMed
CrossRef
 
Detsky AS, Redelmeier DA.  Measuring health outcomes—putting gains into perspective [Editorial]. N Engl J Med. 1998; 339:402-4. PubMed
CrossRef
 

Letters

NOTE:
Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).

Comments

Submit a Comment
The greater longevity of Oscar winners: partly an immortal time' artifact
Posted on April 24, 2006
Marie-Pierre Sylvestre
McGill University
Conflict of Interest: None Declared

To the Editor,

The large survival advantage "“ almost 4 years -- for Academy Award"“winning actors and actresses over their less successful peers [1] continues to receive attention. We write to point out that the statistical method used to derive the statistically significant survival difference gave the Oscar winners an unfair advantage, and to help readers recognize or avoid similar artifacts in other research reports.

The report was based on 235 Oscar winners; 527 other nominees who never won; and 887 "controls" who were never nominated, selected from performers of the same sex and approximately same age who played in the movies for which the nominees were nominated. In the primary analysis, survival was measured from birth, but other definitions of "time-zero" were also used. In all but one of the Kaplan-Meier, Log-Rank and Cox proportional hazards analyses reported, each performer was classified as a "winner" or "not a winner" One reported analysis employed "winner" as a time-dependent covariate, to reflect the fact that all started out as "non -winners", but that some changed status over time.

In the (more emphasized) authors' principal comparison, the Kaplan- Meier curves showed a 3.9 year advantage for the winners. The Cox model, with "winner" as a fixed-in-time covariate , yielded mortality rate reductions ranging from 28% (no adjustment) to 23% (adjustment for seven other covariates), all with 95% confidence limits above 0%. The one reported set of analyses that treated each performer's status (as winner or loser) as dynamic (time-dependent) yielded a reduction of 20%; the lower limit of the CI was 0%, i.e., the reduction was "just significant" at the conventional P=0.05 level. The abstract and the Figure focused on the 3.9 years and the 28% reduction in death rates, which were obtained without adjustment, and without taking into account that a performer's status changed with time.

The analyses that classified those who ultimately won as "winners from the outset" gave them an inbuilt survival advantage, by crediting the years prior to winning towards survival subsequent to winning an Oscar. These "˜immortal' years [2,3] were in fact a requirement for membership in the winners' group: winners had to survive long enough to win (more than 79 years in the two most extreme cases: see Figure). Conversely, actors who did not win had no minimum survival requirement, and some died even before some winners had won, i.e., before some "longevity contests" could begin. For example, 145 "non-winners" had already died by age 65, before 15 of the winners had even won; these unfair pairings (e.g., Burton vs. Burns) were implicitly included in the overall longevity contest between the two groups, and contributed to the apparent survival advantage of the winners, even if winning brought no survival benefit.

To estimate the longevity benefits of winning an Oscar, the comparison should begin at the time that each actor first won, and the "˜remaining-longevity' contest should only include those alive at the same age as the winner was when the winner won. A winner may legitimately be included in comparisons (risksets) prior to winning, but as a "non- winner".

An analysis where the status of a performer who ultimately won is treated as a winner throughout, even in the risksets prior to winning, produces "˜immortal time' bias. As the figure illustrates, a longevity that is measured from a time-zero that precedes the performer's Oscar win (e.g., an individualized one such as the day each performer's first film was released, or a common one such as each performer's 0th or 50th birthday, as in [1]) will necessarily contain some immortal time; there is no such immortality guarantee for those who do not win. Similarly, the matching process, involving a performer "who played opposite" a nominee, ensured that a "control" was alive when a person who ultimately won was nominated, but not necessarily at the time that winner won (the comparison of the 235 nominees who won vs. the 527 other nominees did not involve a matching process).

The authors did report one analysis where each performer's status was updated in each risk set. Annals plans to publish Our methods are described more fully in a technical appendix and our results are presented in a table that will be published by Annals. All of our analyses treat each performer's status as dynamic.

In our re-analyses, which take the immortal time, as well as the covariates sex and year of birth, into account, the point estimate of the actuarial advantage is approximately 1 year, and not statistically significantly different from 0 (the 95% CI is compatible with 0). The estimated percentage mortality reduction is also correspondingly smaller.

We also directly estimated the magnitude of the "immortal time" bias. In the winners vs. nominated comparison, we estimated that not accounting for "˜immortal time' produced an artifactual longevity advantage of 0.8 years (mortality rate ratio 0.94). In the winners vs. controls, not accounting for the immortal time "“ now more substantial -- between the year of a winning actor's first film and the year (s)he first won produced an artificial longevity advantage of 1.7 years and a mortality rate ratio of 0.87. In 1843, William Farr (4) described the statistical artifact created by classifying persons by their status at the end of follow-up, and analyzing them as if they had been in these categories from the outset. He used as examples the greater longevity of persons who reached higher ranks within their professions (bishops vs. curates; judges vs. barristers; generals vs. lieutenants). Despite textbook warnings (2,5,6), analyses that overlook this subtle bias are still common today.

In some longevity comparisons, e.g., (1,7,8), the consequences of an incorrect conclusion are minor. In the evaluation of the time-extension benefits of therapy (3,9,10), the consequences are more serious. How therefore to detect potential immortal time bias? We suggest that when reports compare two "groups", such as winners vs. nominees, one should carefully examine when and how persons entered a "group". Did being in (or moving to) a group have a time-related requirement? Was the classification based on the status at time-zero, or later? If later, was this accounted for? Is the term "status", which implies it could change, more appropriate than the term "group", where, as in a clinical trial, membership is fixed from the outset? Is it sufficient to classify the person just once, or do we need to re-classify the person-moments (the person at different times)? Showing the timelines may help. And, of course, readers and commentators should be doubly cautious whenever they encounter statistics that seem too extreme to be true.

References

1. Redelmeier DA, Singh SM. Survival in Academy Award-winning actors and actresses. Ann Intern Med. 2001 May 15;134(10):955-62.

2. Walker AM. Observation and Inference: An Introduction to the Methods of Epidemiology. Chestnut Hill, MA : Epidemiology Resources Inc., 1991. pp 59-60, and p161.

3. Suissa S. Effectiveness of inhaled corticosteroids in chronic obstructive pulmonary disease: immortal time bias in observational studies. Am J Respir Crit Care Med. 2003 Jul 1;168(1):49-53. Epub 2003 Mar 27.

4. Farr W. Vital statistics : a memorial volume of selections from the reports and writings of William Farr / with an introduction by Mervyn Susser and Abraham Adelstein. Metuchen, N.J. : Scarecrow Press, 1975.

5. Colton, T. Statistics in Medicine. Little, Brown and Co., Boston, 1974.

6. Hill A.B. A short textbook of medical statistics. London: Hodder & Stoughton, 1977, p278.

7. Rothman KJ. Longevity of jazz musicians: flawed analysis. Am J Public Health. 1992 May;82(5):761.

8. Hanley JA, Carrieri MP, Serraino D. Statistical fallibility and the longevity of popes: William Farr meets Wilhelm Lexis. Int J Epidemiol. 2006 Mar 16; [Epub ahead of print]

9. Gail MH. Does cardiac transplantation prolong life? A reassessment. Ann Intern Med. 1972 May; 76(5):815-7.

10. Glesby MJ, Hoover DR. Survivor treatment selection bias in observational studies: examples from the AIDS literature. Ann Intern Med. 1996;124:999-1005.

Conflict of Interest:

None declared

Lingering Concerns about Immortality Bias
Posted on June 6, 2006
Donald Redelmeier
Toronto
Conflict of Interest: None Declared

To the editor,

Sylvestre et al. correctly comment that survival statistics are fallible. Our primary analysis was based on the Kaplan-Meier method because life expectancy is the preferred metric in medical decision analysis [1]. Our article also provided 40 other secondary analyses to explore different models since no one statistic is ideal. Sylvestre et al. argue that the multivariate-adjusted Cox proportional hazards model with a time-varying step function is preferred over our primary analysis approach, do not discuss the limitations of such models, and intimate that other models give an unfair advantage. This position disagrees with us and with other reviews involving our work [2, 3].

We agree that time-varying functions are valuable for addressing a change in status from winning. One drawback with such models can be in assuming the same hazard for all winners following first win; for example, Jodi Foster (first win age 25) and Judi Dench (first win age 62) are assigned identical hazards from age 63 until death. However, we found earlier wins were associated with greater advantages, contrary to this assumption. Adding fixed covariates that additionally model age (linear or quadratic) is no simple solution because the likelihood of winning is no simple function of age. The models also have limited power on small datasets, assume no unmeasured heterogeneity, and rarely capture complex trajectories (eg, multiple films, nominations, and wins) [4, 5, 6, 7].

We thank many scientists for analyses of our database. We have also done an update to March 29, 2006 and observed 122 more individuals and 144 more deaths since first publication. Our primary unadjusted analysis shows a smaller survival advantage of 3.6 years (79.7 vs 76.1, p = 0.005). Applying Model 1 of the Sylvestre technical appendix so that winners are treated in a time-varying manner yielded a 8% mortality reduction (95% confidence interval: -14 to +26, p = 0.455). Modifying Model 1 so that both winners and non-winners are treated in a time-varying manner yielded a 15% mortality reduction (95% confidence interval: -6 to +31, p = 0.140). These estimates overlap earlier results. Apparently, the survival advantage depends on the analytic method chosen.

The statistical debate concerns inbuilt survival advantages that yield an immortality bias. We provided methods for addressing this bias, observed multiple findings suggesting this bias was not large in our cohort, and estimated the hidden confounding that would need to be postulated. We found no survival advantage comparing individuals with many nominations to individuals with no nominations, for example, contrary to estimates of a large immortality bias. Moreover, we presumed individuals not reported dead were alive, which is a different type of immortality bias that causes almost all of our analyses and Sylvestre et al.'s analyses to underestimate survival differences.

Donald A. Redelmeier Sheldon M. Singh dar@ices.on.ca

REFERENCES

1. Sox HC, Blatt MA, Higins MC, Marton KI. Medical decision making. Toronto: Butterworths 1988, p 182-4.

2.Redelmeier DA, Singh SM. Longevity of screenwriters who win an academy award: longitudinal study. BMJ. 2001;323:1491-6.

3. Redelmeier DA, Singh SM. Association between mortality and occupation among movie directors and actors. Am J Med. 2003;115:400-3.

4. Fisher LD, Lin DY. Time-dependent covariates in the Cox proportional-hazards regression model. Annu Rev Public Health. 1999;20:145 -57.

5. Therneau TM, Grambsch PM. Modeling survival data: extending the Cox model. New York: Springer, 2000, p 231-2.

6. Kalbfleisch JD, Prentice RL. The statistical analysis of failure time data. Hoboken, New Jersey: John Wiley and Sons Inc, 2nd Edition 2002, p 196-208.

7. Allison PD, Survival analysis using SAS: A practical guide. Cary NC: The SAS Institute Inc, 2004, p 111-84.

Conflict of Interest:

None declared

Submit a Comment

Summary for Patients

Social Status and Life Expectancy in an Advantaged Population: A Study of Academy Award–Winning Actors

The summary below is from the full report titled “Survival in Academy Award–Winning Actors and Actresses.” It is in the 15 May 2001 issue of Annals of Internal Medicine (volume 134, pages 955-962). The authors are DA Redelmeier and SM Singh.

Read More...

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.

Toolkit

Want to Subscribe?

Learn more about subscription options

Advertisement
Related Articles
Topic Collections
PubMed Articles
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.
(Required)
(Required)