ijsmr
|
|
Abstract: Generalized linear models based on Poisson regression are commonly applied to count data for area morbidity outcomes, focused on modelling the conditional mean of the response as a function of a set of risk factors. Mean regression models may be sensitive to outliers and provide no information on other distributional features of the response. We consider instead a Poisson lognormal hierarchical approach to quantile regression of spatially configured count data, allowing for observed risk factors and spatially correlated unobserved risk factors. This technique has the advantage that a profile of the relative outcome risk across quantiles can be obtained, including estimates of uncertainty (e.g. the uncertainty attaching to 2.5% or 5% relative risk quantiles). An application involves counts of emergency hospitalisations for self-harm for 6791 small areas in England. Known risk factors are area deprivation, a measure of social fragmentation and a measure of rural status. It is shown that impact of these predictors varies between quantiles, and that hierarchical quantile regression generally produces narrower risk intervals, except for outlier areas, and leads to a higher number of areas being classed as high risk. Keywords: Hierarchical quantile regression. Relative risk. Risk intervals. Elevated risk. Self-harm. /td> |
|
|
Abstract: Motivated by HIV retention, we present an application of the smooth test of goodness-of-fit under right-censoring to time to first occurrence of a recurrent event. The smooth test applied here is an extension of Neyman’s smooth test to a class of hazard functions for the initial distribution of a recurrent failure-time event. We estimate the baseline hazard function of time-to-first loss to follow-up, using a Block, Borges and Savits (BBS) minimal repair model of the data (n = 2,987,72% censored). Simulations were conducted at various percentages of censoring to assess the performance of the smooth test. Results show that the smooth test performed well under right-censoring. Keywords: BBS model, Hazard function, Loss to follow-up, Neyman’s smooth test, Recurrent events, Retention in HIV care. |
|
|
Abstract: The ‘stepped-wedge cluster randomised trial’ (SW-CRT) harbours promise when for ethical or practical reasons the recruitment of a control group is not possible or when a staggered implementation of an intervention is required. Yet SW-CRT designs can create considerable challenges in terms of methodological integration, implementation, and analysis. While cross-sectional methods in participants recruitment of the SW-CRT have been discussed in the literature the cohort method is a novel feature that has not been considered yet. This paper provides a succinct overview of the methodological, analytical, and practical aspects of cohort SW-CRTs.We discuss five issues that are of special relevance to SW-CRTs. First, issues relating to the design, secondly size of clusters and sample size; thirdly, dealing with missing data in the fourth place analysis; and finally, the advantages and disadvantages of SW-CRTs are considered. An Australian study employing a cohort SW-CRT to evaluate a domiciliary aged care intervention is used as case study. The paper concludes that the main advantage of the cohort SW-CRT is that the intervention rolls out to all participants. There are concerns about missing a whole cluster, and difficulty of completing clusters in a given time frame due to involvement frail older people. Cohort SW-CRT designs can be successfully used within public health and health promotion context. However, careful planning is required to accommodate methodological, analytical, and practical challenges. Keywords: Clinical trials, Stepped wedge design, missing data, sample size, Cluster randomized trial. |
|
|
Abstract: Melanoma of the skin is the fifth and seventh most commonly diagnosed carcinoma in men and women, respectively, in the USA. So far, gene signatures prognostic for overall and distant metastasis-free survival, for example, have been promising in the identification of therapeutic targets for primary and metastatic melanoma. But most of these gene signatures have been selected using statistics that depend entirely on the parametric distributions of the data (e.g. t-statistics). In this study, we assessed the impact of relaxing the parametric assumptions on the power of the models used for gene selection. We developed a semi-parametric model for feature selection that does not depend on the distributions of the covariates. This copula-based model only assumed that the marginal distributions of the covariates are continuous. Simulations indicated that the copula-based model had reasonable power at various levels of the false discovery rate (FDR). These results were validated in a publicly-available melanoma dataset. Relaxing parametric assumptions on microarray data may yield procedures that have good power for differential gene expression analysis. Keywords: Copula, False discovery rate, Melanoma, Microarray, Power. |
|
|
Abstract: In this study, we validate the smooth test of goodness-of-fit for the proportionality of the hazard function in the two-sample problem in cancer survival studies. The smooth test considered here is an extension of Neyman’s smooth test for proportional hazard functions. Simulations are conducted to compare the performance of the smooth test, the data-driven smooth test, the Kolmogorov-Smirnov proportional hazards test and the global test, in terms of power. Eight real cancer datasets from different settings are assessed for the proportional hazard assumption in the Cox proportional hazard models, for validation. The smooth test performed best and is independent of the number of covariates in the Cox proportional hazard models. Keywords: Cancer, Cox proportional hazards model, Global test, Neyman’s smooth test, Two-sample problem. |