China 
Africa 
Explore chapters and articles related to this topic
Structural Equation Modeling with Longitudinal Data
Published in Douglas D. Gunzler, Adam T. Perzynski, Adam C. Carle, Structural Equation Modeling for Health and Medicine, 2021
Douglas D. Gunzler, Adam T. Perzynski, Adam C. Carle
Longitudinal SEMs allow for the analysis of multiple repeated measures outcomes simultaneously and inclusion of both time-invariant and time-varying covariates. Time-invariant covariates have a single value for each individual in a data set and do not depend on the exact time of observation. Time-varying covariates have values that change across time. Race and sex at birth are typically treated as time-invariant variables in all models (cross-sectional or longitudinal). However, systolic blood pressure has a single value in a cross-sectional data set, but would likely have different values across multiple repeated measures for a single individual.
Real-World Evidence from Population-Based Cancer Registry Data
Published in Harry Yang, Binbing Yu, Real-World Evidence in Drug Development and Evaluation, 2021
Although the statistical methods are developed for cause-specific survival data where the responses are OS or cause-specific survival rate, various statistical methods have been adapted to the relative survival for population-based cancer survival data. The models may be classified into two groups: regression relative survival without cure and relative survival models with cure. Regression relative survival models are used to estimate the effect of covariates on survival and to predict survival in future calendar years. Cure survival models are used to estimate the proportion of subjects who are cured statistically and who have similar survival to the general cancer-free population. Most relative survival regression methods model the excess hazard of a cancer diagnosis [26, 27], which is defined in Eq. (3.3). For grouped survival data, Hakulinen and Tekanen [26] proposed using the binomial regression model with a complementary log-log link. Methods to estimate relative survival using individual data and the full likelihood approach have been developed by Dickman et al. [27] and Esteve et al. [28]. Recently, Nelson et al. [29] and Royston and Parmar [30] proposed flexible parametric models for relative survival data by fitting restricted cubic splines on the log cumulative excess hazard scale. The main advantages of these models are the ability to model time on a continuous scale and the possibility to incorporate time-varying covariates. Other methods have been recently developed to model relative survival with time-dependent effects [31–33].
Missing Data
Published in Craig Mallinckrodt, Geert Molenberghs, Ilya Lipkovich, Bohdana Ratitch, Estimands, Estimators and Sensitivity Analysis in Clinical Trials, 2019
Craig Mallinckrodt, Geert Molenberghs, Ilya Lipkovich, Bohdana Ratitch
Ancillary variables can be included in likelihood-based analyses by either adding the ancillary variables as covariates or as additional response variables to create a multivariate analysis. However, the complexity of multivariate analyses and the features of most commercial software make it easier to use ancillary variables via MI. With separate steps for imputation and analysis, post-baseline, time-varying covariates – possibly influenced by treatment – can be included in the imputation step of MI to account for missingness but then not included in the analysis step to avoid confounding with the treatment effects, as might be the case in a likelihood-based analysis.
Fully nonparametric survival analysis in the presence of time-dependent covariates and dependent censoring
Published in Journal of Applied Statistics, 2023
David M. Ruth, Nicholas L. Wood, Douglas N. VanDerwerken
This can happen, for example, when estimating survival without a transplant for people on the liver transplant waiting list, because the sickest patients (i.e. those closest to failure) are prioritized for transplant, which is a common censoring mechanism. In this context, patient medical histories constitute time-varying covariate information. With this application in mind, we develop a Kaplan–Meier-type estimator which can incorporate time-dependent covariates and avoids the bias from having dependent failure and censoring times. Through simulation we demonstrate improvements over traditional Kaplan–Meier and Cox models. We then apply the proposed technique to estimate without-transplant survival using a de-identified data set of liver waitlist candidates from the Scientific Registry of Transplant Recipients (SRTR).
Sexual Obligation and Perceived Stress: A National Longitudinal Study of Older Adults
Published in Clinical Gerontologist, 2021
We controlled for both time-varying and time-invariant sociodemographic covariates which were suggested to be associated with both stress and sexual life (Aggarwal et al., 2014; Cohen & Williamson, 1988; DeLamater, 2012; Ezzati et al., 2014; Laumann et al., 1999; Payne et al., 2014). Time-varying covariates included age (in years, centered at the mean), marital status (0 = married or cohabiting and 1 = unmarried), self-rated health (from 1 = poor to 5 = excellent, centered at the mean), and employment status (0 = not currently working and 1 = currently working). Time-invariant covariates (all measured at Wave 1) included race/ethnicity (non-Hispanic white (reference), non-Hispanic black, Hispanic, and others), education (less than high school (reference), high school degree, some college but no degree, and college degree or above) and gender (0 = men and 1 = women).
Epilepsy after severe traumatic brain injury: frequency and injury severity
Published in Brain Injury, 2020
Hanna Siig Hausted, Jørgen F. Nielsen, Lene Odgaard
To assess injury severity as a prognostic factor of PTE, we graphed the CIP of PTE stratified by injury severity (Figure S1). Because these graphs indicated no difference between injury groups during the first month after injury we split the follow-up period at 28 days. Finally, we calculated subhazard ratios (SHRs) for PTE during the early period (≤28 days) and the later period (>28 days) using multivariable competing risk regression analyses adjusting for sex and age (18–30 years, 31–60 years, and >60 years) (29). Each measure of injury severity was analyzed in separate models. The proportional sub hazards assumptions were assessed in models in which covariates were also included as time-varying covariates. To explore the counter-intuitive tendency in the early period that less severe injuries had higher risk of PTE compared to more severe injuries, we graphed CIPs stratified by age based on the competing risk regression models.