This paper was presented at the 10th Congress of the International Association of Biomedical Gerontology in Cambridge, UK  (September 2003).
This manuscript is now published soon in the Annals of the New York Academy of Sciences, 2004, vol. 1019, pp.513-517
[SBN 1-57331-496-X, Item code 0413 ]. 

Does Exceptional Human Longevity Come With High Cost of Infertility?

Testing the Evolutionary Theories of Aging

Natalia S. Gavrilova, Leonid A. Gavrilov,
Victoria G. Semyonova, Galina N. Evdokushkina

Center on Aging, NORC and the University of Chicago, 1155 East 60th Street, Chicago, IL 60637-2745, USA

1. Introduction

This introductory section is written as an abstract summarizing the idea, methods, findings and conclusions of the study. The purpose of this study is to test the prediction of evolutionary theory of aging that human longevity comes with the cost of impaired reproductive success (higher infertility rates, see [1]). Our validation study is based on the analysis of particularly reliable genealogical records for European aristocratic families. This dataset is appealing to use for two reasons:
(1) it has high data accuracy and completeness;
(2) confounding effects of socio-economic status are minimized in this socially elite group.
The dataset is comprised of 3,723 married women born in 1500-1875 and belonging to the upper European nobility. Every case of childlessness was cross-checked using at least two different sources. Data analyses were based on logistic regression model using childlessness as a dependent (outcome) variable, and woman lifespan, calendar year of birth, age at marriage, husband's age at marriage and husband's lifespan as dependent (predictor) variables. We found that woman's exceptional longevity does not increase her chances to be infertile. It appears that the previous reports of high infertility among long-lived women (up to 50% of infertility rate, see [1]) are related to data incompleteness, caused by under-reporting of children. Indeed, data cross-checking revealed that at least in 32% cases the allegedly 'childless' women did in fact had children. Thus, the concept of heavy infertility cost for human longevity is not supported by data, when these data are carefully cross-checked, cleaned and reanalyzed. Additional relevant information is available at our scientific website (

2. The Importance of Data Quality Control

Previous analysis of childlessness among aristocratic women1 was made on the assumption of data completeness. However, when claims are made that many long-lived women are childless [1], it is important to crosscheck the data and to make sure that the lack of children is real, rather than caused by data incompleteness.

An obvious step is to cross-check the initial dataset with other data sources. For example, we examined 335 claims of childlessness in the Bloore's dataset used by Westendorp and Kirkwood. When we cross-checked these claims with other professional sources of data, we found that at least 107 allegedly childless women (32%) did have children. Thus, at least 32% of childlessness claims proved to be wrong ("false negative claims").

This example demonstrates that extreme caution should be exercised when claims for common childlessness among long-lived women are made. The incompleteness of genealogies can itself generate a spurious increase in the prevalence of allegedly childless women among those who live long lives. This happens because children are often not mentioned in particularly obscure, side branches of genealogical trees (remote relatives). It is also known that long-lived people have more chances of being mentioned in incomplete genealogies, because of their longer paper trail in various archives generated during their long life. Thus, incompleteness of genealogies generates two types of biases - underreporting of children and inflated prevalence of long-lived people, thereby producing a spurious rise in claimed childlessness with increased lifespan.

Incomplete reporting of children may seriously affect and compromise scientific studies of human fertility. For example, Westendorp and Kirkwood reported: "None of the six women who were born before 1700 and who reached the exceptional age of 90 years and over had more than two children" [1,p.745]. Our data cross-checking with other data sources revealed that in fact none but one of these women had less than three children.

Among these six women was Antoinette de Bourbon (1493-1583) who allegedly had only one child according to the Bloore's database. Study of other data sources revealed that this well-known person (grandmother of Mary Stuart, Queen of Scots) had as many as twelve children! This fact is well known to professional genealogists and is even reported in The Catholic Encyclopedia (Vol. VII, House of Guise, Robert Appleton Company, 1909). Thus, if we compute an average number of children for women lived 90-99 years with corrected data for Antoinette de Bourbon alone, their average number of progeny would be even higher than average number of progeny for shorter-lived women.

This example demonstrates that genealogical data should be carefully checked against multiple genealogical and historical sources before using them in the scientific studies and making strong conclusions.

3. Results and Discussion

This section describes the results obtained with cross-checked, corrected data. Table 1 presents the dependence of the frequency of childlessness as a function of women lifespan (univariate analysis).

The data obtained by other researchers are also presented in the same table for comparison. Note extremely high proportion of childless women in data published by Westendorp and Kirkwood [1]. On the other hand, German data [2] as well as our data for aristocratic women are consistent with each other and do not demonstrate any increase in childlessness for long-lived women. Our estimates of childlessness also are consistent with estimates of childlessness among the British Peerage reported by Thomas Hollingsworth in his fundamental historical study [3].

Results presented in Table 1 were obtained using univariate analyses, which do not take into account many important explanatory variables. In order to avoid the omitted variable bias and to study the true relationship between childlessness and longevity, we need to take into account many other explanatory variables, which influence infertility rate. Therefore we applied multivariate logistic regression with childlessness as a dependent binary variable and calendar year of birth, female age at marriage, husband's age at marriage, female lifespan and husband's lifespan as predictor variables.

The main result of our study is presented in Figure 1. This figure shows odds of being childless as a function of female lifespan, adjusted for other important confounding variables. The odds of childlessness are particularly high, when women's lifespan is too short (below age 30), which is not surprising. What is really important is that the chances of being childless do not demonstrate any increase for long-lived women (lifespan 90+ years). This result confirms findings from our univariate analyses (Table 1) as well as from other studies [2,4], which demonstrated that long-lived women do not have higher rate of childlessness even when controlled for other important confounding variables.

Our study does not support the previous published claims that human longevity comes at a high cost of infertility. This conclusion may have both theoretical significance (testing some evolutionary theories of aging), as well as practical implications for the future of life extension. It helps to relax concerns over a question: "Is it morally acceptable to extend human longevity at the cost of infertility?" Some authors already raised their concerns on the unintended consequences of life span extension: "... increasing longevity through genetic manipulation of the mechanisms of aging raises deep biological and moral questions. These questions should give us pause before we embark on the enterprise of extending our lives" [5]. This study helps to alleviate some concerns on these issues.


This study was made possible thanks to a generous support from the National Institute on Aging (NIH, USA), and a stimulating working environment at the Center on Aging, NORC/University of Chicago. We would like to thank members of the Science Advisory Board, SAB ( for useful comments on our work made at the SAB discussion group.


  1. Westendorp, R.G.J. & Kirkwood, T.B.L. 1998. Human longevity at the cost of reproductive success. Nature 396: 743-746.
  2. Lycett, J.E. et al. 2000. Longevity and the cost of reproduction in a historical human population. Proc. Roy. Soc. London, 267: 31-35.
  3. Hollingsworth, T.H. 1964. The demography of the British Peerage. Population Studies, suppl., 18: 3-107.
  4. Korpelainen, H. 2000. Fitness, reproduction and longevity among European aristocrats and rural Finnish families in the 1700s and 1800s. Proc. R. Soc. Lond. B 267: 1765-1770.
  5. Glannon, W. 2002. Extending the human life span, Journal of Medicine and Philosophy, 27: 339-354.

Table 1. Proportion of childlessness by women's age at death. Comparison of our dataset with similar data for the historical German population [2] and data for the British aristocracy [1].

Age at Death,

Proportion of Childless Women in  Different Datasets

Gavrilovs dataset  on European upper nobility

Lycett et al.,(2000)
German data

Westendorp and Kirkwood (1998),
British aristocracy

































Figure 1. Childlessness odds ratio as a function of female lifespan. Net effects are adjusted for female calendar year of birth, female age at marriage, husband's lifespan, and husband's age at marriage. Multivariate regression analysis of 3,723 European aristocratic families.