Lifetime Data Analysis

, Volume 1, Issue 3, pp 255–273 | Cite as

Frailty models for survival data

  • Philip Hougaard


A frailty model is a random effects model for time variables, where the random effect (the frailty) has a multiplicative effect on the hazard. It can be used for univariate (independent) failure times, i.e. to describe the influence of unobserved covariates in a proportional hazards model. More interesting, however, is to consider multivariate (dependent) failure times generated as conditionally independent times given the frailty. This approach can be used both for survival times for individuals, like twins or family members, and for repeated events for the same individual. The standard assumption is to use a gamma distribution for the frailty, but this is a restriction that implies that the dependence is most important for late events. More generally, the distribution can be stable, inverse Gaussian, or follow a power variance function exponential family. Theoretically, large differences are seen between the choices. In practice, using the largest model makes it possible to allow for more general dependence structures, without making the formulas too complicated.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    R. Aaberge, O. Kravdal, and T. Wennemo, “Unobserved heterogeneity in models of marriage dissolution,” Discussion paper no. 42, Central Bureau of Statistics, Norway, 1989.Google Scholar
  2. 2.
    O. O. Aalen, “Two examples of modelling heterogeneity in survival analysis,”Scand. J. Statist, vol. 14 pp. 19–25, 1987a.Google Scholar
  3. 3.
    O. O. Aalen, “Mixing distributions on a Markov chain,”Scand. J. Statist. vol. 14 pp. 281–9, 1987b.Google Scholar
  4. 4.
    O. O. Aalen, “Heterogeneity in survival analysis,”Statist. Med. vol. 7 pp. 1121–37, 1988.Google Scholar
  5. 5.
    O. O. Aalen, “Modelling heterogeneity in survival analysis by the compound Poisson distribution,”Ann. Appl. Prob. vol. 2 pp. 951–72, 1992.Google Scholar
  6. 6.
    O. O. Aalen, “Effects of frailty in survival analysis,”Statistical Methods in Medical Research vol. 3 pp. 227–43, 1994.Google Scholar
  7. 7.
    P. K. Andersen, and O. Borgan, “Counting process models for life history data: A review,”Scand. J. Statist. vol. 12 pp. 97–158, 1985.Google Scholar
  8. 8.
    P. K. Andersen, O. Borgan, R. D. Gill, and N. Keiding,Statistical Models Based on Counting Processes, Springer Verlag, 1993.Google Scholar
  9. 9.
    S. K. Bar-Lev and P. Enis, “Reproducibility and natural exponential families with power variance functions,”Ann. Statist. vol. 14 pp. 1507–22, 1986.Google Scholar
  10. 10.
    J. Burridge, “Empirical Bayes analysis of survival time data,J.R. Statist. Soc. B vol. 43 pp. 65–75, 1981.Google Scholar
  11. 11.
    D. Clayton, “A model for association in bivariate life tables and its application in epidemiological studies of familial tendency in chronic disease incidence,”Biometrika vol. 65 pp. 141–51, 1978.Google Scholar
  12. 12.
    D. Clayton and J. Cuzick, “Multivariate generalizations of the proportional hazards model (with discussion),”J.R. Statist. Soc. A vol. 148 pp. 82–117, 1985.Google Scholar
  13. 13.
    D. R. Cox, “Regression models and life tables (with discussion),”J.R. Statist. Soc. B vol. 34 pp. 187–220, 1972.Google Scholar
  14. 14.
    M. Crowder, “A multivariate distribution with Weibull connections,”J.R. Statist. Soc. B vol. 51 pp. 93–107, 1989.Google Scholar
  15. 15.
    C. Elbers and G. Ridder, “True and spurious duration dependence: the identifiability of the proportional hazard model.”Rev. Econ. Stud. vol. XLIX pp. 403–9, 1982.Google Scholar
  16. 16.
    R. Ellermann, P. Sullo, and J. M. Tien, “An alternative approach to modeling recidivism using quantile residual life functions,”Operations Research vol. 40 pp. 485–504, 1992.Google Scholar
  17. 17.
    J. E. Freund, “A bivariate extension of the exponential distribution,”J. Am. Statist. Assoc. vol. 56 pp. 971–7, 1961.Google Scholar
  18. 18.
    C. Genest and MacKay, “Copules Archimediennes et familles de lois bidimensionnelles dont les marges sont donnees,”Canadian J. Statist. vol. 14 pp. 145–59, 1986.Google Scholar
  19. 19.
    E. J. Gumbel, “Bivariate exponential distributions,”J. Am. Statist. Assoc. vol. 55 pp. 698–707, 1960.Google Scholar
  20. 20.
    G. Guo, “Use of sibling data to estimate family mortality effects in Guatemala,”Demography vol. 30 pp. 15–32, 1993.Google Scholar
  21. 21.
    G. Guo and G. Rodriguez, “Estimating a multivariate proportional hazards model for clustered data using the EM algorithm. With an application to child survival in Guatemala,”J. Am. Statist. Assoc. vol. 87 pp. 969–76, 1992.Google Scholar
  22. 22.
    P. Hougaard, “Life table methods for heterogeneous populations: Distributions describing the heterogeneity,”Biometrika vol. 71 pp. 75–84, 1984.Google Scholar
  23. 23.
    P. Hougaard, “Survival models for heterogeneous populations derived from stable distributions,”Biometrika vol. 73 pp. 387–96, 1986a (Correction, vol. 75 pp. 395).Google Scholar
  24. 24.
    P. Hougaard, “A class of multivariate failure time distributions,”Biometrika vol. 73 pp. 671–8, 1986b. (Correction, vol. 75 pp. 395).Google Scholar
  25. 25.
    P. Hougaard, “Modelling multivariate survival,”Scand. J. Statist. vol. 14 pp. 291–304, 1987.Google Scholar
  26. 26.
    P. Hougaard, “Fitting a multivariate failure time distribution,”IEEE Transactions on Reliability vol. 38 pp. 444–8, 1989.Google Scholar
  27. 27.
    P. Hougaard, “Modelling heterogeneity in survival data,”J. Appl. Prob. vol. 28 pp. 695–701, 1991.Google Scholar
  28. 28.
    P. Hougaard, B. Harvald, and N. V. Holm, “Measuring the similarities between the lifetimes of adult Danish twins born between 1881–1930,”J. Am. Statist. Assoc. vol. 87 pp. 17–24, 1992a.Google Scholar
  29. 29.
    P. Hougaard, B. Harvald, and N. V. Holm, “Assessment of dependence in the life times of twins,”Survival Analysis: State of the Art (J. P. Klein and P. K. Goel, eds.) pp. 77–97, Kluwer Academic Publishers, 1992b.Google Scholar
  30. 30.
    P. Hougaard, B. Harvald, and N. V. Holm, “Models for multivariate failure time data, with application to the survival of twins,”Statistical Modelling (P. G. M. van der Heijden, W. Jansen, B. Francis and G. U. H. Seeber, eds.) pp. 159–173, Elsevier Science Publishers, 1992c.Google Scholar
  31. 31.
    P. Hougaard, P. Myglegaard, and K. Borch-Johnsen, “Heterogeneity models of disease susceptibility, with application to diabetic nephropathy,”Biometrics vol. 50 pp. 1178–88, 1994.Google Scholar
  32. 32.
    T.P. Hutchinson and C.D. Lai,The Engineering Statistician's Guide to Continuous Bivariate Distributions, Rumsby Scientific Publishing: Adelaide, 1991.Google Scholar
  33. 33.
    H. Joe, “Parametric families of multivariate distributions with given margins,”J. Mult. Anal. vol. 46 pp. 262–82, 1993.Google Scholar
  34. 34.
    B. Jørgensen, “Statistical properties of the generalized inverse Gaussian distribution,”Lecture Notes in Statistics vol. 9, Springer-Verlag: Heidelberg, 1981.Google Scholar
  35. 35.
    B. Jørgensen, “Exponential dispersion models,”J.R. Statist. Soc. B vol. 49 pp. 127–62, 1987.Google Scholar
  36. 36.
    J. P. Klein, “Semiparametric estimation of random effects using the Cox model based on the EM algorithm,”Biometrics vol. 48 pp. 795–806, 1992.Google Scholar
  37. 37.
    J. P. Klein, M. Moeschberger, Y. H. Li, and S. T. Wang, “Estimating random effects in the Framingham heart study,”Survival Analysis: State of the Art (J. P. Klein and P. K. Goel, eds.) pp. 99–120, Kluwer Academic Publishers, 1992.Google Scholar
  38. 38.
    T. Lancaster, “Econometric methods for the duration of unemployment,”Econometrica vol. 47 pp. 939–56, 1979.Google Scholar
  39. 39.
    L. Lee, “Multivariate distributions having Weibull properties,”J. Mult. Anal. vol. 9 pp. 267–77, 1979.Google Scholar
  40. 40.
    M.-L. T. Lee and G. A. Whitmore, “Stochastic processes directed by randomized time,”J. Appl. Prob. vol. 30 pp. 302–14, 1993.Google Scholar
  41. 41.
    J.-C. Lu, “Least squares estimation for the multivariate Weibull model of Hougaard based on accelerated life test of system and component,”Commun. Statist.-Theory Meth. vol. 19 pp. 3725–39, 1990.Google Scholar
  42. 42.
    J.-C. Lu and G. K. Bhattacharyya, “Some new constructions of bivariate Weibull models,”Ann. Inst. Statist. Math. vol. 42 pp. 543–59, 1990.Google Scholar
  43. 43.
    A. W. Marshall and I. Olkin, “A multivariate exponential distribution,”J. Am. Statist. Assoc. vol. 62 pp. 30–44, 1967.Google Scholar
  44. 44.
    S. A. Murphy, “Consistency in a proportional hazards model incorporating a random effect,”Ann. Statist. vol. 22 pp. 712–31, 1994.Google Scholar
  45. 45.
    G. G. Nielsen, R. D. Gill, P. K. Andersen, and T. I. A. Sørensen, “A counting process approach to maximum likelihood estimation in frailty models,”Scand. J. Statist. vol. 19 pp. 25–43, 1992.Google Scholar
  46. 46.
    D. Oakes, “A model for association in bivariate survival data,”J.R. Statist. Soc. B vol. 44 pp. 414–22, 1982.Google Scholar
  47. 47.
    D. Oakes, “Bivariate survival models induced by frailties,”J. Am. Statist. Assoc. vol. 84 pp. 487–93, 1989.Google Scholar
  48. 48.
    D. Oakes and A. Manatunga, “Fisher information for a bivariate extreme value distribution,”Biometrika vol. 79 pp. 827–32, 1992.Google Scholar
  49. 49.
    A. Pickles, R. Crouchley, E. Simonoff, L. Eaves, J. Meyer, M. Rutter, J. Hewitt, and J. Silberg, “Survival models for development genetic data: Age of onset of puberty and antisocial behaviour in twins,”Genetic Epidemiology vol. 11 pp. 155–70, 1994.Google Scholar
  50. 50.
    C. S. Rocha, “Survival models for heterogeneity using the non-central chi-squared distribution with zero degrees of freedom,” Notas e Comunicações, Centro de estatistica e aplicações da universidade de Lisboa, 1994.Google Scholar
  51. 51.
    J. A. Tawn, “Bivariate extreme value theory; Models and estimation,”Biometrika vol. 75 pp. 397–415, 1988.Google Scholar
  52. 52.
    D. C. Thomas, B. Langholz, W. Mack, and B. Floderus, “Bivariate survival models for analysis of genetic and environmental effects in twins,”Genetic Epidemiology vol. 7 pp. 21–35, 1990.Google Scholar
  53. 53.
    M. C. K. Tweedie, “An index which distinguishes between some important exponential families,”Statistics: Applications and New Directions. Proceedings of the Indian Statistical Institute Golden Jubilee International Conference (J. K. Ghosh and J. Roy, eds.) pp. 579–604, 1984.Google Scholar
  54. 54.
    J. W. Vaupel, K. G. Manton, and E. Stallard, “The impact of heterogeneity in individual frailty on the dynamics of mortality,”Demography vol. 16 pp. 439–54, 1979.PubMedGoogle Scholar
  55. 55.
    J. W. Vaupel and A. I. Yashin, “The deviant dynamics of death in heterogeneous populations,”Sociological Methodology (N. B. Tuma, ed.) pp. 179–211, Jossey-Bass Publishers, 1985.Google Scholar
  56. 56.
    J. T. Wassell and M. L. Moeschberger, “A bivariate survival model with modified gamma frailty for assessing the impact of interventions,”Statist. Med. vol. 12 pp. 241–8, 1993.Google Scholar
  57. 57.
    G. A. Whitmore and M.-L. T. Lee, “A multivariate survival distribution generated by an inverse Gaussian mixture of exponentials,”Technometrics vol. 33 pp. 39–50, 1991.Google Scholar
  58. 58.
    A. I. Yashin, K. G. Manton, and E. Stallard, “Dependent competing risks: a stochastic process model,”J. Math. Biol. vol. 24 pp. 119–40, 1986.Google Scholar
  59. 59.
    A. I. Yashin, K. G. Manton, and E. Stallard, “The propagation of uncertainty in human mortality processes operating in stochastic environments,”Theoretical Population Biology vol. 35 pp. 119–41, 1989.Google Scholar

Copyright information

© Kluwer Academic Publishers 1995

Authors and Affiliations

  • Philip Hougaard
    • 1
  1. 1.Novo Nordisk, Novo AlleBagsvaerdDenmark

Personalised recommendations