Correspondence
Survival after Age 80
N Engl J Med 1996; 334:537-538February 22, 1996
- Article
To the Editor:
Manton and Vaupel (Nov. 2 issue)1 suggest that the white population of the United States has lower mortality between the ages of 80 and 100 years than the populations of Japan, Sweden, France, and England. They cite one of our papers as supporting the view that “published mortality rates are reasonably reliable for U.S. whites up to the age of 100.”1 The paper they cite does not assess the quality of U.S. data. Two other recent publications find U.S. data on mortality in old age to be seriously flawed.2,3
The data presented by Manton and Vaupel themselves contain important inconsistencies. These authors present data from the Medicare B files of the Social Security Administration for 1987 indicating that the probability of survival among women between the ages of 95 and 100 was 0.232 (their Table 1). In contrast, their extinct-cohort–based probability of survival in the same age range for the cohort born in 1890–1894 is 0.328. This cohort reached the age of 95 between 1985 and 1990, and thus the figures pertaining to it should be virtually identical to the value computed for 1987 if all data are accurate. In fact, for women ranging in age from 95 to 100 in the four countries other than the United States, the mean absolute difference between the 1987 value and the value for the cohort born in 1890–1894 is only 0.004. For the United States, the difference is 0.096. The results for men are similar but less dramatic. Extinct-cohort estimates, the principal basis of the analysis by Manton and Vaupel, appear to provide too optimistic a picture of survival for white Americans. The most likely reason is a net overstatement of ages at death.3,4
The Social Security data presented in the article, whose source is Kestenbaum,5 are undoubtedly the most reliable of the U.S. data presented. They show that the United States has an advantage with respect to mortality, although it is very small. Among women, the 1987 Social Security data show an advantage of 0.7 year in life expectancy at the age of 80, relative to the average life expectancy in the other four countries. The extinct-cohort method yields an estimated average advantage of 1.8 years (Table 1 of the article).1 For men the advantage is 0.4 year on the basis of the Social Security data, but 1.2 years when calculated by the extinct-cohort method. Furthermore, Manton and Vaupel present the Social Security data incorrectly. The figures for ages ranging from x to x + 5 years actually pertain to ages ranging from x - 0.5 to x + 4.5 years.5 This displacement by one half-year spuriously raises the figures presented for life expectancy in the United States by about 0.2 year (by our calculation), a large enough bias to raise the life expectancy of U.S. men above that of Japanese men.
There is considerable uncertainty about the existence and extent of any U.S. survival advantage at older ages.
5 ReferencesSamuel H. Preston
Irma T. Elo
University of Pennsylvania, Philadelphia, PA 19104-62961
Manton KG ,Vaupel JW . Survival after the age of 80 in the United States, Sweden, France, England, and Japan. N Engl J Med 1995;333:1232-1235
Full Text | Web of Science | Medline2
Kannisto AU. Development of oldest-old mortality, 1950-1990: evidence from 28 developed countries. Odense, Denmark: Odense University Press, 1994.
3
Coale AJ ,Kisker EE . Defects in data on old-age mortality in the United States: new procedures for calculating mortality schedules and life tables at the highest ages. Asian Pac Popul Forum 1990;4:1-314
Rosenwaike I ,Logue B . Accuracy of death certificate ages for the extreme aged. Demography 1983;20:569-585
CrossRef | Web of Science5
Kestenbaum B . A description of the extreme aged population based on improved Medicare enrollment data. Demography 1992;29:565-580
CrossRef | Web of Science | Medline
To the Editor:
By 1914 the subjects studied by Manton and Vaupel were 20 to 34 years old. Many of the fittest of their generation died, and others suffered from wounds, the effects of gas, or illnesses. Of 10 million British men 15 to 44 years of age, about 23 percent were casualties. Any after-effects would affect more men than women and would diminish in time as the susceptible died. France, England, and Sweden (a neutral country) are similar nations that did not have the immigration seen in America or the rapid change seen in Japan. If the shadow of war extends to old age, France and England may show wider gaps than Sweden in survival between women and men. Table 1 of the article by Manton and Vaupel shows that, overall, the sex difference in the percentage of people with a life expectancy of 5 years at the age of 80 is about 12.5 percent for France, 11.2 percent for England, and 7.4 percent for Sweden. The figures converge with age, so that at 95 the gaps are 4.0, 5.6, and 5.1 percent, respectively.
T.H. Hughes-Davies, F.R.C.P.
Breamore Marsh, Fordingbridge, Hampshire SP6 2EJ, United KingdomTo the Editor:
We believe that the observed differences in life expectancy may be more related to the effects of survivorship than the authors realize. Survival after the age of 80 is estimated in people who reach that age, who the authors acknowledge are likely to be more robust survivors in countries with higher mortality rates at younger ages. As the authors note, mortality in the United States was higher below the age of 65 and lower after the age of 80 than in Sweden, France, England, or Japan. In a more extreme example, survival after the age of 80 among men in Mali has been estimated to be about 9.3 years, as compared with 7.0 years among U.S. men, whereas life expectancy at birth for male infants is only 55 years in Mali, as compared with 72 years in the United States.1 The apparent survival advantage for elderly men in Mali certainly does not reflect better health services and more effective medical care. Are there data to rule out such an effect of survivorship in the comparisons presented by Manton and Vaupel?
1 ReferencesAnthony M. Sallar, M.P.H.
Robert S. Hogg, Ph.D.
Martin T. Schechter, M.D., Ph.D.
University of British Columbia, Vancouver, BC V6T 1Z3, CanadaAuthor/Editor Response
The authors reply:
To the Editor: Hughes-Davies suggests that injuries received in World War I increased differences in survival between men and women, with the differences decreasing with age. Mortality among the English and the French (combatants) and among the Swedish (noncombatants) is consistent with this hypothesis, suggesting that it is worthy of further study.
We agree with Sallar et al. that selection affected differences in mortality. The Japanese cohorts studied had higher mortality at younger ages than the U.S. cohorts. The higher mortality at older ages among the Japanese may be due to problems in caring for the elderly.1
Preston and Elo raise several questions. We cited two references when we suggested that data on U.S. whites were “reasonably reliable” to the age of 100. Condran et al.2 examined the consistency between death certificates and census reports of age in the United States (Table 4, 1960), Sweden, Japan, and England. They also suggest, on the basis of evidence gained from direct checks, that “inconsistency between records increase[s] with age and individual's ages are generally more likely to be misreported in censuses than in death registration data.” In their Table 4, ages reported on U.S. death certificates were relatively lower for persons above 85 than ages reported in the 1960 Census — a general pattern found in other comparisons with the Census. The oldest age we used in 1962 was 82. In an article referred to by Condran et al. as an “outstanding study of reports of ages at death,”2 Rosenwaike and Logue3 suggested that the use of extinct-cohort methods could improve distributions of age at death and drew this conclusion: “Findings of the present investigation — at least for the white population — dispute the belief that ages reported on death records for persons 85 years of age and over are of questionable credibility. On the contrary, it appears that death certificate ages for those 85 to 99 years in New Jersey and Pennsylvania in 1970 agree remarkably well with an independent check.”3 The oldest age we used in 1970 was 90; in 1980, it was 100. Kestenbaum's4 matched-record data for 1987 (for two states) suggest a close correspondence between ages reported on death certificates and those in Medicare data on persons up to the age of 100. Although matched-record data for the period of a year may indicate the quality of data, data for 1987 (or 19885) may not reflect trends in mortality, especially among the elderly and according to sex. Thus, several studies suggest that data on the mortality of U.S. whites are of reasonable quality for the age ranges we used over time. However, even a convergence of U.S. and Japanese survival at the age of 80 suggests the loss of an advantage in life expectancy of three to four years among Japanese infants at birth.
5 ReferencesKenneth G. Manton, Ph.D.
James C. Vaupel, Ph.D.
Duke University, Durham, NC 27708-04081
Ikegami N ,Campbell JC . Medical care in Japan. N Engl J Med 1995;333:1295-1299
Full Text | Web of Science | Medline2
Condran GA ,Himes CL ,Preston SH . Old-age mortality patterns in low-mortality countries: an evaluation of population and death data at advanced ages, 1950 to the present. Popul Bull U N 1991;30:23-603
Rosenwaike I ,Logue B . Accuracy of death certificate ages for the extreme aged. Demography 1983;20:569-585
CrossRef | Web of Science4
Kestenbaum B . A description of the extreme aged population based on improved Medicare enrollment data. Demography 1992;29:565-580
CrossRef | Web of Science | Medline5
Life tables for the United States Social Security area 1900-2080. Actuarial study 107. Washington, D.C.: Social Security Administration, August 1992. (SSA publication no. 11-11536.)
- Citing Articles (2)
Citing Articles
1
France Mesle, Jacques Vallin. (2006) Diverging Trends in Female Old-Age Mortality: The United States and the Netherlands versus France and Japan. Population and Development Review 32:1, 123-145
CrossRef2
Glenn V. Ostir, Kenneth J. Ottenbacher, Kyriakos S. Markides. (2004) Onset of Frailty in Older Adults and the Protective Role of Positive Affect.. Psychology and Aging 19:3, 402-408
CrossRef
