China
Africa
Explore chapters and articles related to this topic
Study Design
Published in Marcello Pagano, Kimberlee Gauvreau, Heather Mattie, Principles of Biostatistics, 2022
Marcello Pagano, Kimberlee Gauvreau, Heather Mattie
Stratified randomization ensures that an important confounder such as age or sex is more evenly distributed between treatment groups than might happen by chance alone. Using this technique, randomization is performed separately within strata, for subjects with and without the confounder. (Recall the discussion of stratified sampling in Chapter 21.) As an example, in the study of patients undergoing liver transplantation, randomization was stratified by whether or not the subject had been diagnosed with hepatitis C. In small studies, stratified randomization can improve statistical power. It has less benefit in large studies, where random assignment is more likely to ensure an even distribution of baseline characteristics across treatment groups.
Cluster Randomized Trials
Published in Susan Halabi, Stefan Michiels, Textbook of Clinical Trials in Oncology, 2019
The Korean Health Study [2] provides an example of the use of stratified randomization in a cluster randomized trial. This study evaluated a church-based intervention to improve hepatitis B virus serological testing among Korean Americans in Los Angeles. Fifty-two Korean churches were stratified by size (small, medium, large) and location (Koreatown versus other) and randomized to intervention or control conditions within the six strata. This ensured balance between the intervention and control arms on size and location. Church location was considered potentially prognostic because of acculturation differences among participants attending churches inside versus outside Koreatown. Church size was considered potentially prognostic because of the potential for competing activities and resource differences at larger churches.
Clinical Development Plan and Clinical Trial Design
Published in Mark Chang, John Balser, Jim Roach, Robin Bliss, Innovative Strategies, Statistical Solutions and Simulations for Modern Clinical Trials, 2019
Mark Chang, John Balser, Jim Roach, Robin Bliss
Like simple randomization, a stratified randomization is a randomization with a fixed allocation probability. When there are important known covariates (confounding factors), stratified randomization is usually recommended to reduce treatment imbalances. For a stratified randomization, the target population is divided into several homogenous strata, which are usually determined by some combinations of covariates (e.g., patient demographics or patient characteristics). In each stratum, a simple randomization is then employed.
A Tenodesis-Induced-Grip exoskeleton robot (TIGER) for assisting upper extremity functions in stroke patients: a randomized control study
Published in Disability and Rehabilitation, 2022
Hsiu-Yun Hsu, Kang-Chin Yang, Chien-Hsien Yeh, Yu-Ching Lin, Keng-Ren Lin, Fong-Chin Su, Li-Chieh Kuo
There are some additional limitations to this study. First, although this was a randomized control trial with a clinically applicable dose, the training for those patients in the control group was not exclusively focused on rehabilitation of the hand and wrist but included repetitive skill building for tasks of daily living in general. This may have led to less improvement in the hand and wrist function of the participants in the control group. A second limitation was the differences in the baseline characteristics of the study conditions and this, in turn, decreased the statistical power of the study. Thus, the obtained results should be generalized with caution. It is also worth noting that the training paradigm of the TIGER is only composed of continuous passive and functional modes but lacks the training paradigm of an assist-as-needed to provide assistance only as much as it is needed to complete the motor or functional task is the major limitation of the device. In the future, stratified randomization could be employed in the study design to prevent an imbalance in the baseline conditions between treatment groups and thus minimize the variations in treatment responsiveness. In future research, the test for the control system of the TIGER such as motor driving force test for validating mechanical safety shall be carried out to ensure a safer training process. Finally, the design will integrate the Internet of Things with the TIGER system to allow for investigation of the effects of applying the TIGER system as a home-based treatment for chronic stroke patients who have limited rehabilitation resources.
Online learning versus workshops: a rank minimized trial comparing the effect of two knowledge translation strategies designed to alter knowledge, readiness to change, and self-efficacy with respect to rehabilitation outcome measures
Published in Disability and Rehabilitation, 2022
Mike Szekeres, Joy C. MacDermid
Subjects were allocated using minimization at each site, an allocation method that placed participants in intervention groups to minimize the differences across key predictors [16,17]. Minimization can be useful in small clinical trials where stratified randomization is not feasible because it can improve data utilization, simplify the statistical analysis, and reduce the risk of selection bias. The rationale is that minimization across key predictors helps to balance prognostic variables and result in more valid comparisons and has been shown to be superior to randomization for small trials [18]. Participants were recruited for each site and once they consented and cleared eligibility was added to the site list used for minimization matches. Pre-test scores, years of practice, practice area (urban/rural), and practice type (PT/OT) were used as the key predictors. At each site, the pool of participants was allocated minimizing difference by creating pair groupings based on professional training (PT/OT), matching area practice and then most similar pre-test scores, and, finally, by minimizing years of practice. When this process was complete, subjects at each site were informed of their assignment. Once allocation was complete no changes were made. These matches were performed by hand by an independent researcher.
Effects of brief self-exercise education on the management of chronic low back pain: A community-based, randomized, parallel-group pragmatic trial
Published in Modern Rheumatology, 2021
Hiroshige Jinnouchi, Ko Matsudaira, Akihiko Kitamura, Hironobu Kakihana, Hiroyuki Oka, Mina Hayama-Terada, Kazumasa Yamagishi, Masahiko Kiyama, Hiroyasu Iso
Eligible participants who provided informed consent and fulfilled the inclusion criteria were randomly assigned to 1 of 2 groups in a 1:1 ratio: an intervention group (brief-See) or control group (material-based education). We employed stratified randomization in terms of age (65 years old or older/younger), sex (female/male), pain intensity (NRS, 7 or higher/lower), and the STarT Back subgroup (low risk/medium to high risk). The allocation sequence was performed by randomization of staff who were not involved in the intervention and baseline assessment. The intervention therapists were informed of the results of patient allocation before the initial intervention. Self-administered questionnaires were applied for all assessment measures, which were submitted by mail or were collected by visiting the participants. The staff responsible for collating data were different from the intervention therapists, and they ensured that there were no missing values.