Lifetime Data Analysis

, Volume 24, Issue 4, pp 601–604 | Cite as

Commentary: Alignment of time scales and joint models

  • Kwun Chuen Gary Chan

Some may have an impression that survival analysis is a mature field in which models and methods are well developed. The thought-provoking paper by Dempsey and McCullagh challenges our conventional thinking on how survival and health outcomes data shall be analyzed. I congratulate the authors for publishing this intriguing piece of work. In the following, I would like to offer discussions on some of their interesting findings, as well as my observations and thoughts on this important and interesting topic.

Two processes or just one?

The authors devoted a considerable amount of effort to explain that when death is an absorbing state of the health process, then the time to death is a deterministic function of the process. Therefore, it is more natural to consider the health process and time to death as a single process instead of two separate, but correlated, processes. This is certainly an interesting point of view and one that is unarguable when the health process is brain electrical...



I would like to thank Prof. Niels Keiding for inviting me to be a discussant of Prof. McCullagh’s talk at 2017 Conference on Lifetime Data Science, and subsequently invited me to be part of this discussion. My research is partially supported by the US National Institutes of Health and National Science Foundation.


  1. Cai Q, Wang MC, Chan KCG (2017) Joint modeling of longitudinal, recurrent events and failure time data for survivor’s population. Biometrics 73(4):1150–1160MathSciNetCrossRefGoogle Scholar
  2. Chan IS, Neaton JD, Saravolatz LD, Crane LR, Osterberger J (1995) Frequencies of opportunistic diseases prior to death among hiv-infected persons. AIDS 9:1145–1152CrossRefGoogle Scholar
  3. Chan KCG, Wang MC (2010) Backward estimation of stochastic processes with failure events as time origins. Ann Appl Stat 4(3):1602–1620MathSciNetCrossRefzbMATHGoogle Scholar
  4. Chan KCG, Wang MC (2017) Semiparametric modeling and estimation of the terminal behavior of recurrent marker processes before failure events. J Am Stat Assoc 112(517):351–362MathSciNetCrossRefGoogle Scholar
  5. Chan KCG, Chen YQ, Di CZ (2012) Proportional mean residual life model for right-censored length-biased data. Biometrika 99(4):995–1000MathSciNetCrossRefzbMATHGoogle Scholar
  6. Cox DR, Isham V (1980) Point processes. CRC Press, LondonzbMATHGoogle Scholar
  7. Dempsey W, McCullagh P (2018) Survival models and health sequences. Lifetime Data Anal.
  8. Efron B (2002) The two-way proportional hazards model. J R Stat Soc Ser B 64(4):899–909MathSciNetCrossRefzbMATHGoogle Scholar
  9. Lin DY, Wei LJ, Yang I, Ying Z (2000) Semiparametric regression for the mean and rate functions of recurrent events. J R Stat Soc Ser B 62(4):711–730MathSciNetCrossRefzbMATHGoogle Scholar
  10. Liu L, Wolfe RA, Kalbfleisch JD (2007) A shared random effects model for censored medical costs and mortality. Stat Med 26(1):139–155MathSciNetCrossRefGoogle Scholar
  11. Lunney JR, Lynn J, Foley DJ, Lipson S, Guralnik JM (2003) Patterns of functional decline at the end of life. J Am Med Assoc 289(18):2387–2392CrossRefGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  1. 1.Department of BiostatisticsUniversity of WashingtonSeattleUSA

Personalised recommendations