Lifetime Data Analysis

, Volume 1, Issue 4, pp 347–359 | Cite as

Fitting Weibull duration models with random effects

  • Carl Morris
  • Cindy Christiansen


Duration time models often should include correlated failure times, due to clustered data. These random effects hierarchical models sometimes are called “frailty models” when used for survival analyses. The data analyzed here involve such correlations because patient level outcomes (the times until graft failure following kidney transplantation) are observed, but patients are clustered in different transplant centers. We describe fitting such models by combining two kinds of software, one for parametric survival regression models, and the other for doing Poisson regression in a hierarchical setting. The latter is implemented by using PRIMM (Poisson Regression and Interactive Multilevel Modeling) methods and software (Christiansen & Morris, 1994a). An illustrative example for profiling data is included withk=11 kidney transplant centers andN=412 patients.


EM frailty hierarchical models kidney transplants medical profiling parametric survival models PRIMM 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    M. Aitkin and D. Clayton, “The fitting of exponential, Weibull, and extreme value distributions to complex censored survival data using GLIM,”Applied Statistics vol. 29(2) pp. 156–163, 1980.Google Scholar
  2. 2.
    J. E. Anderson and T. A. Louis, N. V. Holm and B. Harvald, “Time-dependent association measures for bivariate survival distributions,”Journal of the American Statistical Association vol. 87(419) pp. 641–650, 1992.Google Scholar
  3. 3.
    C. L. Christiansen and C. N. Morris, “Hierarchical Poisson regression modeling,” Technical Report, Harvard Medical School, Department of Ambulatory Care and Prevention, 1994a.Google Scholar
  4. 4.
    C. L. Christiansen and C. N. Morris, “README document,” June 7, 1994, a 7-page tutorial for PRIMM, available through anonymous ftp to, 1994b.Google Scholar
  5. 5.
    C. L. Christiansen and C. N. Morris,Fitting and Checking a Two-level Poisson Model: Modeling Patient Morality Rates in Heart Transplant Patients, Marcel Decker, D. Berry and D. Stangl, editors (to appear, 1996).Google Scholar
  6. 6.
    D. Clayton, “A Monte-Carlo method for Bayesian inference in frailty models,Biometrics vol. 47 pp. 467–485, 1991.Google Scholar
  7. 7.
    B. Efron and C. Morris, “Stein's estimation rule and its competitors—an empirical Bayes approach,”Journal of the American Statistical Association vol. 34(38) pp. 117–130, 1973.Google Scholar
  8. 8.
    Health Resources and Services Administration, “Report of Center-Specific Grant and Patient Survival Rates.” 5 vol. report, U.S. Department of Health and Human Services, 1992.Google Scholar
  9. 9.
    P. Hougaard, “Survival models for heterogeneous populations derived from stable distributions,”Biometrika vol. 73(2) pp. 387–396, 1986.Google Scholar
  10. 10.
    W. J. Huster, R. Brookmeyer and S. G. Self, “Modelling paired survival data with covariates,”Biometrics vol. 45 pp. 145–146, 1989.Google Scholar
  11. 11.
    W. James and C. Stein, “Estimation with Quadratic Loss,” inProceedings of the Fourth Berkeley Symposium on Mathematical Statistics and Probability, vol. 1, University of California Press, Berkeley, 1961, pp. 361–379.Google Scholar
  12. 12.
    J. F. Lawless, “Regression methods for Poisson process data,”Journal of the American Statistical Association, vol. 82(399) pp. 808–815, 1987.Google Scholar
  13. 13.
    D. V. Lindley and A. F. M. Smith, “Bayes estimates for the linear model,”Journal of the Royal Statistical Society Ser. B, vol. 34 1–41, 1972.Google Scholar
  14. 14.
    C.-H. Liu and D. B. Rubin, “The ECME algorithm: A simple extension of EM and ECM with faster monotone convergence,”Biometrika vol. 81(4) pp. 633–648, 1994.Google Scholar
  15. 15.
    P. McCullagh and J. A. Nelder,Generalized Linear Models, second edition, Chapman and Hall, 1989.Google Scholar
  16. 16.
    C. N. Morris, “Approximating posterior distributions and posterior moments,” inBayesian Statistics 3, J. M. Bernardo, M. H. DeGroot, D. V. Lindley and A. F. M. Smith (eds.), Oxford University Press, pp. 327–344, 1988.Google Scholar
  17. 17.
    C. Morris, E. Norton and X. H. Zhou, “Parametric duration analysis: A case study of its use applied to nursing home data,” accepted for publication inCase Studies in Biometry, John Wiley, ENAR, Chapter 12 (16 pp.), 1993.Google Scholar
  18. 18.
    G. B. Nielsen and R. D. Gill, P. K. Andersen and T. I. A. Sorensen, “A counting process approach to maximum likelihood estimation in frailty models,”Scandinavian Journal of Statistics vol. 19 pp. 25–43, 1992.Google Scholar
  19. 19.
    D. Oakes, “Bivariate survival models induced by frailties,”Journal of the American Statistical Association vol. 84(406) pp. 487–493, 1989.Google Scholar
  20. 20.
    A. Pickles and R. Crouchley, “A comparison of frailty models for multivariate survival data,”Statistics in Medicine vol. 14 pp. 1447–1461, 1995.Google Scholar
  21. 21.
    D. Sinha, “Semiparametric Bayesian analysis of multiple event time data,”Journal of the American Statistical Association vol. 88 pp. 893–979, 1993.Google Scholar
  22. 22.
    A. A. Skolnick, “Are there too many US transplantation centers? Some experts suggest fewer, cheaper, and better,”JAMA vol. 271(14) pp. 1062–1063, 1994.Google Scholar
  23. 23.
    L. Tierney,LISP-STAT, An Object-Oriented Environment for Statistical Computing and Dynamic Graphics, John Wiley & Sons, Inc., 1990.Google Scholar
  24. 24.
    J. T. Wassell, “A bivariate survival model with modified Gamma frailty for assessing the impact of interventions,”Statistics in Medicine vol. 12 pp. 241–248, 1993.Google Scholar
  25. 25.
    L. J. Wei, “The accelerated failure time model: A useful alternative to the Cox regression model in survival analysis,”Statistics in Medicine vol. 11 pp. 1871–1879, 1992.Google Scholar

Copyright information

© Kluwer Academic Publishers 1995

Authors and Affiliations

  • Carl Morris
    • 1
  • Cindy Christiansen
    • 2
  1. 1.Department of StatisticsHarvard UniversityCambridge
  2. 2.Department of Ambulatory Care and PreventionHarvard Medical SchoolBoston

Personalised recommendations