China
Africa
Explore chapters and articles related to this topic
Cluster Randomized Trials
Published in Susan Halabi, Stefan Michiels, Textbook of Clinical Trials in Oncology, 2019
The main impact of stratification in a CRT is to reduce the between-cluster variance component, [8]. Because clusters in different conditions are now compared within strata, is replaced with the variance among clusters within strata, which we denote . The variance of the treatment effect estimator becomes and the total number of clusters required is see Crespi [32]. Alternatively, we can define as the correlation between cluster-level means within strata or matched pairs, equal to , and the formula becomes Thus the design effect is reduced by a factor of . These formulas apply to studies with matched pairs and with larger strata.
D
Published in Filomena Pereira-Maxwell, Medical Statistics, 2018
The effect of study design features such as clustering on the power of a statistical test, i.e. on its ability to detect real effects as statistically significant. The design effect may be defined as the factor by which a sample size must be multiplied in order for a statistical test to retain a given level of power. Design effects are often applied to sample size calculations for clustered designs, as the effective sample size will be reduced. Larger sample sizes are usually required to make up for the fact that clusters, not individuals, are the study units. The magnitude of a design effect depends on the extent of clustering, as measured by the intraclass correlation coefficient (ICC), and on the intended average cluster size. The larger the ICC and average cluster size, the larger the magnitude of the design effect. See BLAND (2015), KIRKWOOD & STERNE (2003), MACHIN & CAMPBELL (2005) and ELDRIDGE & KERRY (2012) for further discussion and formulae.
Weighting and Complex Sampling Design Adjustments in Longitudinal Studies
Published in Jason T. Newsom, Richard N. Jones, Scott M. Hofer, Longitudinal Data Analysis, 2013
Shayna D. Cunningham, Nathalie Huguet
In other words, the design effect gauges the loss or gain in precision of a sampling design. In the case of a complex sample design, the design effect indicates the combined effects of using the sample weight, stratification, and clustering. A design effect greater than 1 indicates that the variance of a statistic from a particular design is greater than that from a comparable SRS design. Relating this notion of the design effect to the sample size, the effective sample size can be defined as the actual sample size divided by the design effect. When the design effect is greater than 1, the effective sample size is smaller than the actual size.
Musculoskeletal disorders and risk factors among heavy load carriers in Yaounde city, Cameroon
Published in International Journal of Occupational Safety and Ergonomics, 2022
William Mbang Bian, Jerson Mekoulou Ndongo, Wiliam Richard Guessogo, Edmond Ebal Minye, Peguy Brice Assomo Ndemba, Georges Gassina, Samuel Honoré Mandengue, Abdou Temfemo
Following information gathered at the association of labourers specialized in offloading/loading heavy loads from trucks, the number of registered carriers stood at 500. The number of participants selected as the representative sample population was obtained using the formula 14]. Based on the 88.4% prevalence reported by Alegbeh et al. [15] among dockers in the autonomous port of Lomé with a consistency threshold of 1.96 and a margin of error of 5%, the expected proportion of MSDs (p) was set at 0.6, the margin of error (y) at 0.05, the confidence level (tp = 1.96) at 95% and N = 500 as the number of estimated handlers in the city. The design effect gave a sample size (n) of 213 as the minimum representative population.
Circulation of dengue serotype 1 viruses during the 2019 outbreak in Dar es Salaam, Tanzania
Published in Pathogens and Global Health, 2021
Gaspary O. Mwanyika, Leonard E. G. Mboera, Sima Rugarabamu, Mariam Makange, Calvin Sindato, Julius J. Lutwama, Janusz T. Paweska, Gerald Misinzo
The sample size was calculated using a formula described by Arya and Antonisamy [20], assuming 20.9% seroprevalence of DENV infection in Dar es Salaam region [8], and an error rate of 10%. A design effect of three was used to correct for the variability between study districts. The study districts were regarded as clusters and a design effect of three was chosen to obtain an effective sample size of 191 subjects adequate to detect the expected effect. The clinicians recruited patients with dengue-like illness and fever (temperature ≥ 38°C), presenting with at least one of the following clinical signs: retro-orbital pain, rash, arthralgia, malaise, signs of persistent vomiting, severe hemorrhage and organ failure. Febrile patients with bacterial infections and those who were unwilling to participate in the study were excluded. All dengue cases were categorized clinically either as dengue with/without warning signs and severe dengue according to the World Health Organization classification scheme [21].
Five-Year Changes of Anterior Corneal Indices in Diabetics versus Non-Diabetics: The Shahroud Eye Cohort Study
Published in Current Eye Research, 2019
Hassan Hashemi, Soheila Asgari, Shiva Mehravaran, Mohammad Hassan Emamian, Akbar Fotouhi
In this report, we compared changes in the apical corneal thickness (ACT), the minimum corneal thickness (MCT), the corneal thickness in the 4 mm (average of 12 points at 30 degree intervals), the 6 mm (average of 4 points at 90 degree intervals), the 8 mm (average of 20 points at 18 degree intervals), the total corneal volume in the 10 mm diameter, maximum keratometry (Kmax), and minimum keratometry (Kmin) between diabetic and non-diabetic participants. The diagnosis of DM was based on fasting blood sugar and hemoglobin A1c as described previously.14 Inclusion criteria included having acceptable corneal images (displayed as OK by the device) in both phases, and exclusion criteria were a history of any eye surgery or the diagnosis of keratoconus, glaucoma, corneal opacity, dystrophy and vascularization at any time. Since the correlation between fellow eyes for all indices in the second phase of the study were between 0.879 and 0.931, only right eye data were used in the analyses. Repeated measures analysis of variance was used to compare the 5-year thickness changes between diabetic and non-diabetic groups with controlling for age, gender, pterygium, and smoking. Adjustments for the design effect due to cluster sampling were applied.