Estimating the parameters of multi-state models with time-dependent covariates through likelihood decomposition
Introduction
Multi-state models have already been widely used in medicine. These models are of great interest in studying chronic diseases with several grades and/or treatments [1]. When the number of states is large, when back and forth transitions between states are allowed, and when time-dependent covariates (such as age) have to be taken into account, the resulting model can become quite complex. We faced all these difficulties when we designed a model to describe the successive treatments offered to patients with end-stage renal disease (ESRD).
ESRD is a life-threatening condition but, in the sixties, renal replacement therapies (RRTs) radically changed patients׳ prognoses and led to longer survivals, especially in renal transplant patients. Currently, ESRD patients may experience several switches between various RRTs [2], [3], and, in several countries, registries have been established to collect data on these switches [4], [5], [6], [7]. These switches can be described using the rates of transition between different RRTs.
Various solutions have been proposed to solve the specific problems of multi-state models in medical research [1], [8], [9]. These solutions are extensions of survival data analyses applied to settings with multiple transitions and competing risks [10], [11], [12], [13], [14], [15], [16]. In fact, several authors succeeded in using these solutions to analyse ESRD datasets, but their models used small numbers of transitions (only the main RRTs were studied), and they considered only age as a categorical variable (i.e., the estimates of the parameters were made for age classes) [17], [18], [19].
In France, 10 different RRT modalities are used and all of them should be considered to model the successive treatments a patient may undergo; this requires the estimation of a high number of transition rates. Moreover, the transitions to renal transplantation or death depend on the patients׳ age which should be integrated into the model.
The data recorded in the French ESRD registry [4] have two features that helped us overcome these difficulties. First, the patients׳ treatment courses are independent (that is, the switch of a patient from one treatment to another does not depend on the treatments the other patients are undergoing). Second, the exact times of all transitions from any treatment to another are known. These two features make it possible to decompose a multi-state model into sub-models whose likelihoods can be separately maximised [8], [10].
In this paper, we explain the way we solved a complex problem with a simple implementation of likelihood decomposition. This decomposition greatly facilitates the theoretical analysis and the numerical solution. Afterwards, considering short time intervals during which the time-dependent covariates are supposed constant led to the estimation of these covariate effects on the transition rates.
Section snippets
REIN registry
Several studies on ESRD patients are longitudinal studies with possible state changes [17], [18], [20]: the RRT may change over time during the disease. The three main RRTs are haemodialysis (HD), peritoneal dialysis (PD), and renal transplantation (TX). These RRTs may be subdivided further according to the technique, the type of facility, the type of assistance, and the type of renal graft.
The French Renal Epidemiology and Information Network (REIN) registry [4] collects data on patients with
Application on a simple model
First, the method was used with only three RRTs (Fig. 1). In this application, each transition rate was considered dependent on age and the effect of age on the transition rate from TX to PD considered equal to the effect on the transition rate from TX to HD. Table 2 presents the estimated parameters obtained with this model. As expected, the parameters relative to age were positive and significant at 5% for the transition rates to death whereas they were negative and significant at 5% for the
Discussion and conclusion
The follow-up of the treatment courses of ESRD patients requires the use of complex models with high numbers of states to allow for all RRTs [3]. In this study, estimating the transition rates with individual data and considering effects of time-dependent covariates with the available tools in R software [21], [26] was quite complicated to program. The alternative method presented here simplified the handling of the data and was easily reproducible.
The method is based on the decomposition of
Conflict of interest statement
The authors declare that they have no conflicts of interest.
Financial support
Agence de la Biomédecine and Haute Autorité de Santé.
Acknowledgements
We thank the Agence de la Biomédicine and the Haute Autorité de Santé for their financial support as well as the nephrologists and the professionals who collected and controlled the quality of REIN registry data.
We also thank Jean Iwaz, Ph.D. and scientific advisor, for several comments and suggestions that improved this manuscript.
References (33)
Compartmental models and competing risk
Math. Biosci.
(1994)- et al.
Markov Chain Monte Carlo simulation for projection of end stage renal disease patients in Greece
Comput. Methods Programs Biomed.
(2012) - et al.
The analysis of asthma control under a Markov assumption with use of covariates
Stat. Med.
(2003) - et al.
Modality options for renal replacement therapy: the integrated care concept revisited
Hemodial. Int.
(2006) - et al.
An evaluation of an integrative care approach for end-stage renal disease patients
J. Am. Soc. Nephrol.
(2000) - et al.
The renal epidemiology and information network (REIN): a new registry for end-stage renal disease in France
Nephrol. Dial. Transplant.
(2006) - S.P. McDonald, L. Excell, B. Livingston, ANZ Data Registry report, Australia and New Zealand Dialysis and Transplant...
- V.S. Stel, M.W. van de Luijtgaarden, C. Wanner, K.J. Jager, The 2008 ERA-EDTA Registry Annual Report-a précis, NDT...
- US Renal Data System, USRDS 2011 Annual Data Report Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the...
A Markov model for analysing cancer markers and disease states in survival studies
Biometrics
(1986)
Multi-state models and outcome prediction in bone marrow transplantation
Stat. Med.
Multi-state models for event history analysis
Stat. Methods Med. Res.
Multi-state models in epidemiology
Lifetime Data Anal.
Multi-state models: a review
Lifetime Data Anal.
Multi-state models for the analysis of time-to-event data
Stat. Methods Med. Res.
Cited by (4)
Trajectoires des patients en suppléance
2022, Nephrologie et TherapeutiquePredicting the sequential behavior of mobile Internet users based on MSM model
2020, International Journal of Market ResearchImproving survival in end-stage renal disease: A case study
2019, Statistical Methods in Medical ResearchRestricted mean survival time over 15 years for patients starting renal replacement therapy
2017, Nephrology Dialysis Transplantation
- 1
On behalf of REIN registry.