An approach to joint analysis of longitudinal measurements and competing risks failure time data

Abstract

SUMMARYJoint analysis of longitudinal measurements and survival data has received much attention in recent years. However, previous work has primarily focused on a single failure type for the event time. In this paper we consider joint modelling of repeated measurements and competing risks failure time data to allow for more than one distinct failure type in the survival endpoint which occurs frequently in clinical trials. Our model uses latent random variables and common covariates to link together the sub-models for the longitudinal measurements and competing risks failure time data, respectively. An EM-based algorithm is derived to obtain the parameter estimates, and a profile likelihood method is proposed to estimate their standard errors. Our method enables one to make joint inference on multiple outcomes which is often necessary in analyses of clinical trials. Furthermore, joint analysis has several advantages compared with separate analysis of either the longitudinal data or competing risks survival data. By modelling the event time, the analysis of longitudinal measurements is adjusted to allow for non-ignorable missing data due to informative dropout, which cannot be appropriately handled by the standard linear mixed effects models alone. In addition, the joint model utilizes information from both outcomes, and could be substantially more efficient than the separate analysis of the competing risk survival data as shown in our simulation study. The performance of our method is evaluated and compared with separate analyses using both simulated data and a clinical trial for the scleroderma lung disease.

Recreational drug use and T lymphocyte subpopulations in HIV-uninfected and HIV-infected men

“…Furthermore, only 5% of the HIV-seropositive study visits were missing CD4 or CD8 T cell measurements, which is unlikely to cause substantial bias. To evaluate the degree of sensitivity of our results to the effects of potential non-random loss to follow-up, we performed joint modeling for longitudinal CD4 cell measurements and survival data using loss-to-follow up as events [Elashoff et al, 2007;Henderson et al, 2000]. This showed that our results were not sensitive to this potential source of bias.…”

Effects of 1-Year Treatment with Cyclophosphamide on Outcomes at 2 Years in Scleroderma Lung Disease

“…For assessment of the 24-month data, a longitudinal model was used for analysis of FVC % predicted and TLC, TDI, and skin score, adjusted for: baseline values; worst fibrosis score from the baseline HRCT scan; and nonignorable values missing due to death, treatment failure, or drop-out (29). Comparison of the two treatment groups at 9, 12, 15, 18, 21, and 24 months was performed by Huber's robust regression analysis (30) after imputing the missing observations using multiple imputation (31,32).…”

Robust inference for longitudinal data analysis with non-ignorable and non-monotonic missing values

“…A potential improvement of our model is to incorporate different missing mechanism models for different causes of missing data to better describe the missing mechanism. For example, Elashoff et al (2007), Elashoff et al (2008) and Li et al (2009) specify different survival functions for time to different dropout causes. A clear and detailed documentation of missing reasons and dropout causes is imperative for such analysis.…”

“…Finally, shared parameter models use latent variables, such as random effects, to account for the dependence between the outcome variable and missingness. For example, Elashoff et al (2007Elashoff et al ( , 2008 used a joint modelling approach to analyse the lung function data in a Scleroderma study in the presence of non-ignorable dropouts. They used a cause-specific hazard frailty model for competing risk failure times, and a linear mixed effects model for the outcome variable.…”