Premium
Assessing the trade‐offs between crossover and parallel group designs in sleep research
Author(s) -
BERRY CHARLES C.,
MOORE POLLY,
DIMSDALE JOEL E.
Publication year - 2006
Publication title -
journal of sleep research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.297
H-Index - 117
eISSN - 1365-2869
pISSN - 0962-1105
DOI - 10.1111/j.1365-2869.2006.00546.x
Subject(s) - crossover , crossover study , polysomnogram , obstructive sleep apnea , sleep (system call) , research design , polysomnography , psychology , placebo , computer science , medicine , statistics , apnea , mathematics , artificial intelligence , psychiatry , anesthesia , alternative medicine , pathology , operating system
Summary Sleep researchers invariably struggle with decisions regarding the optimal design for their studies. Whether such studies involve treatment for insomnia, obstructive sleep apnea, or any other sleep disorder, questions arise regarding the respective trade‐offs between a parallel group and a crossover design. This study analyzed the variance structure of commonly measured polysomnographic variables in an effort to describe the statistical impact of these alternate designs. The study examined the effects of opioids on sleep and employed multiple crossovers between placebo, MS‐contin, and methadone using a double‐blind, randomized crossover design. Thirty‐seven healthy subjects were studied. Four of the subjects were unable to complete the protocol for a variety of reasons, and polysomnogram data was unavailable for one subject. Data from 37 subjects provide the basis for this analysis. Despite dropouts, the crossover study was approximately four times as efficient as the parallel group design in terms of being able to recognize differences in deep sleep across these conditions. Other polysomnographic variables also favored the crossover design to varying extents. Despite the operational complexity of a crossover design, the statistical efficiency of this approach makes it a preferable approach for designing intervention studies in sleep research.