Premium
Novel joint selection methods can reduce sample size for rheumatoid arthritis clinical trials with ultrasound endpoints
Author(s) -
Allen John C.,
Thumboo Julian,
Lye Weng Kit,
Conaghan Philip G.,
Chew LiChing,
Tan York Kiat
Publication year - 2018
Publication title -
international journal of rheumatic diseases
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.795
H-Index - 41
eISSN - 1756-185X
pISSN - 1756-1841
DOI - 10.1111/1756-185x.13185
Subject(s) - medicine , sample size determination , post hoc analysis , confidence interval , rheumatoid arthritis , ultrasonography , clinical trial , ultrasound , post hoc , statistics , surgery , radiology , mathematics
Objectives To determine whether novel methods of selecting joints through (i) ultrasonography (individualized‐ultrasound [ IUS ] method), or (ii) ultrasonography and clinical examination (individualized‐composite‐ultrasound [ ICUS ] method) translate into smaller rheumatoid arthritis ( RA ) clinical trial sample sizes when compared to existing methods utilizing predetermined joint sites for ultrasonography. Methods Cohen's effect size (ES) was estimated ( E S ^ ) and a 95% CI (E S ^ L ,E S ^ U ) calculated on a mean change in 3‐month total inflammatory score for each method. Corresponding 95% CI s [ n L (E S ^ U ) ,n U (E S ^ L ) ] were obtained on a post hoc sample size reflecting the uncertainty inE S ^ . Sample size calculations were based on a one‐sample t ‐test as the patient numbers needed to provide 80% power at α = 0.05 to reject a null hypothesis H 0 : ES = 0 versus alternative hypotheses H 1 : ES =E S ^ , ES =E S ^ L and ES =E S ^ U . We aimed to provide point and interval estimates on projected sample sizes for future studies reflecting the uncertainty in our studyE S ^ S . Results Twenty‐four treated RA patients were followed up for 3 months. Utilizing the 12‐joint approach and existing methods, the post hoc sample size (95% CI ) was 22 (10–245). Corresponding sample sizes using ICUS and IUS were 11 (7–40) and 11 (6–38), respectively. Utilizing a seven‐joint approach, the corresponding sample sizes using ICUS and IUS methods were nine (6–24) and 11 (6–35), respectively. Conclusions Our pilot study suggests that sample size for RA clinical trials with ultrasound endpoints may be reduced using the novel methods, providing justification for larger studies to confirm these observations.