The Japanese Foundation for Cancer Research (JFCR) has provided a large database on gastric cancer for open use comprised of data on survival times after surgery and related covariates including prognostic factors. Data analysis based on the generalized hazards model incorporating the B-spline function (GHMBS) was performed using a dataset (JFCR dataset) composed of 9,631 cases within this database. A model selection method was adopted for clarifying the meaning of estimated parameters, because the GHMBS was comprised of the proportional hazards model (PHM) and accelerated failure time model (AFTM) as submodels. A preliminary simulation experiment to examine the performance of the model selection method based on the GHMBS was conducted under the condition that multiple covariates were considered, where one was the target covariate and the other was covariate for adjustment. After validation of the method by this simulation experiment, the method was applied to the JFCR dataset to estimate the period effect for prolonging survival time with an adjustment for the stage effect. The analysis revealed that the PHM was suitable for the period effect, while a mixture of the PHM and AFTM was for the stage effect and treatment for gastric cancer made steady progress from the 1950s through the 1990s.
Positivity of anti-thyroglobulin antibody (TgAb) is one of the markers of chronic thyroiditis (Hashimoto disease). From 2000 to 2003, a thyroid disease prevalence study was conducted at the Radiation Effects Research Foundation, in Hiroshima and Nagasaki. Utilizing the study's results, we show that via EM algorithm log-transformed TgAb level is compatible with a two-component mixture normal distribution, with the smaller normal distribution corresponding to the TgAb negative group but the larger distribution not necessarily corresponding to the TgAb positive group. A subject is determined to be TgAb positive if TgAb level is greater than a given cutoff. We compared the cutoff values from population-based methods and the laboratory method. The population-based methods consist of a simple method, a receiver operating characteristic (ROC) curve method, and a minimum misclassification rate (MMR) method. The simple method is used to determine positivity from only TgAb negative populations. Since the ROC curve and MMR methods are valid only when TgAb positivity and negativity are known but the simple method is valid only when TgAb negativity is known, the simple method was deemed useful for determining the cutoff in our data. In comparison with the simple, population-based method, we show that the cutoff from the laboratory method is appropriate and that the TgAb positive rates from various methods are approximately equal. With the two-component mixture normal distribution in TgAb level, our simple population-based method for determination of cutoff is another more practical example of handling the clinical measurement than the method given in Thompson et al. (Applied Statistics 1998).
Recruitment of patients is one of main issues of clinical trials, but there are neither standard statistical methods of monitoring nor those of decision-making. Methods for predicting future accrual of patients and re-estimating recruitment duration in a midpoint of trial seem useful. However, there is no method that considers a variability of recruitment rate in participating site (center effects). In this study, we propose a method for re-estimation of recruitment duration with adjustment of center effects in multicenter clinical trials utilizing an empirical Bayes approach and a normal approximation of posterior predictive distribution using Edgeworth expansion. We applied our methods to the randomized clinical trial of adjuvant therapy for breast cancer. In the application, the observed number of patients in the end of trial can be predicted closer with adjustment of center effects than without adjustment, because the effect of the increase of participating sites was modeled explicitly with adjustment of center effects.