Semi-parametric frailty models are widely used to analyze clustered survival data. In this article, we propose the use of the hierarchical likelihood interval for individual frailties. We study the relationship between hierarchical likelihood, empirical Bayesian, and fully Bayesian intervals for frailties. We show that our proposed interval can be interpreted as a frequentist confidence interval and Bayesian credible interval under a uniform prior. We also propose an adjustment of the proposed interval to avoid null intervals. Simulation studies show that the proposed interval preserves the nominal confidence level. The procedure is illustrated using data from a multicenter lung cancer clinical trial.
HougaardPAnalysis of multivariate survival dataNew York: Springer, 2000.
2.
VaidaFXuR. Proportional hazards model with random effects. Stat Med2000; 19: 3309–3324.
3.
GrayRJ. A Bayesian analysis of institutional effects in multicenter cancer clinical trial. Biometrics1994; 50: 244–253.
4.
LegrandCDucrocqVJanssenP. A Bayesian approach to jointly estimate centre and treatment by centre heterogeneity in a proportional hazards model. Stat Med2005; 24: 3789–3804.
5.
KomarekALesaffreELegrandC. Baseline and treatment effect heterogeneity for survival times between centers using a random effects accelerated failure time model with flexible error distribution. Stat Med2007; 26: 5457–5472.
6.
CarlinBPLouisTABayes and Empirical Bayes methods for data analysis2nd edLondon: Chapman and Hall,2000.
7.
LeeYHaID. Orthodox BLUP versus HL methods for inferences about random effects in Tweedie mixed models. Stat Comput2010; 20: 295–303.
8.
LeeYNelderJA. Hierarchical generalized linear models (with discussion). J R Stat Soc, Ser B1996; 58: 619–678.
9.
LeeYNelderJAPawitanYGeneralised linear models with random effects: unified analysis via h-likelihood, Boca Raton, FL: Chapman and Hall2006.
10.
LeeYNelderJA. Likelihood inference for models with unobservables: another view (with discussion). Stat Sci2009; 24: 255–293.
11.
MengX-L. What's the H in H-likelihood: A holy grail or an achilles' heel? (with discussion). Bayesian Stat2011; 9: 473–500.
12.
VuHTVKnuimanMW. A hybrid ML-EM algorithm for calculation of maximum likelihood estimates in semiparametric shared frailty models. Comput Stat Data Anal2002; 40: 173–187.
13.
FayREHerriotRA. Estimates of income for small places: an application of James-Stein procedure to Census data. J Am Stat Assoc1979; 74: 269–277.
14.
MorrisCN. Mixed model prediction and small area estimation. Test2006; 15: 72–76.
15.
LiHLahiriP. An adjusted maximum likelihood method for solving small area estimation problems. Journal of Multivar Anal2010; 101: 882–892.
16.
EfronBTibshiraniTJAn introduction to the bootstrapLondon: Chapman and Hall, 1993.
17.
HendersonCR. Best linear unbiased estimation and prediction under a selection model. Biometrics1975; 31: 423–447.
18.
BoothJGHobertJP. Standard errors of prediction in generalized linear mixed models. J Am Stat Assoc1998; 93: 262–272.
19.
MorrisCN. Parametric empirical Bayes inference: theory and application. J Am Stat Assoc1983; 78: 47–59.
20.
BjørnstadJF. Predictive likelihood principle: a review (with discussion). Stat Sci1990; 5: 242–265.
21.
CarlinBPGelfandAE. Approaches for empirical Bayes confidence intervals. J Am Stat Assoc1990; 85: 105–114.
22.
LairdNMLouisTA. Empirical Bayes confidence intervals based on bootstrap samples. J Am Stat Assoc1987; 82: 739–750.
23.
KassRESteffeyD. Approximate Bayesian inference in conditionally independent hierarchical models (parametric empirical Bayes models). J Am Stat Assoc1989; 84: 717–726.
24.
HaIDLeeYMacKenzieG. Model selection for multi-component FMs. Stat Med2007; 26: 4790–4807.
25.
RondeauVMichielsSLiquetB. Investigating trial and treatment heterogeneity in an individual patient data meta-analysis of survival data by means of the penalized maximum likelihood approach. Stat Med2008; 27: 1894–1910.
26.
HaIDLeeYSongJK. Hierarchical likelihood approach for FMs. Biometrika2001; 88: 233–243.
27.
BreslowNE. Discussion of Professor Cox's article. J R Stat Soc, Ser B1972; 34: 216–217.
28.
HaIDLeeY. Comparison of hierarchical likelihood versus orthodox best linear unbiased predictor approaches for FMs. Biometrika2005; 92: 717–723.
29.
RipattiSPalmgrenJ. Estimation of multivariate FMs using penalized partial likelihood. Biometrics2000; 56: 1016–1022.
30.
HaIDLeeY. Estimating FMs via Poisson hierarchical generalized linear models. J Comput Graph Stat2003; 12: 663–681.
31.
HaIDSylvesterRLegrandC. Frailty modelling for survival data from multi-centre clinical trials. Stat Med2011; 30: 2144–2159.
32.
LeeYJangMLeeW. Prediction interval for disease mapping using hierarchical likelihood. Comput Stat2011; 26: 159–179.
33.
HaIDNohMLeeY. Bias reduction of likelihood estimators in semi-parametric FMs. Scand J Stat2010; 37: 307–320.
34.
MorrisCNChristiansenC. Fitting Weibull duration models with random effects. Lifetime Data Anal1995; 1: 347–359.
35.
BenjaminiYYekutieliD. False discovery rate-adjusted multiple CIs for selected parameters (with discussion). J Am Stat Assoc2005; 100: 71–93347–359
36.
EttingerDSFinkelsteinDMAbeloffMD. A randomized comparison of standard chemotherapy versus alternating chemotherapy and maintenance versus no maintenance therapy for extensive-stage small-cell lung cancer: a phase III study of the Eastern Cooperative Oncology Group. J Clin Oncol1990; 8: 230–240.
37.
PlummerM. Penalized loss functions for Bayesian model comparison. Biostatistics2008; 9: 523–539.
