Reporting of key methodological and ethical aspects of cluster trials in hemodialysis require improvement: a systematic review

Table 2 Reporting of study characteristics

Component	Number of studies (%) (N total = 31)
Trial design
Parallel arm	30 (97%)
Stepped-wedge design	1 (3%)
Types of randomized clusters
Hemodialysis centers	17 (55%)
Hemodialysis shifts or sessions	12 (39%)
Providers or professionals	1 (3%)
Nephrology units^₱	1 (3%)
Method of random allocation
Completely randomized design (unrestricted randomization)	8 (26%)
Stratified design	4 (13%)
Pair-matched design	4 (13%)
Split-cluster (i.e., shifts within a hemodialysis center)	11 (35%)
Covariate-constrained	1 (3%)
Not reported	3 (10%)
Number of clusters per trial [median (25th, 75th percentile)]¥	28 (12, 43)
Number of patients per trial [median (25th, 75th percentile)]^₳	228 (120, 1723)
Number of patients per cluster [median (25th, 75th percentile)]^€,Ϫ	20 (8, 32)

^₱It is not clear how a “nephrology unit” was defined
Estimate is based on ^¥32, ^₳29, and ^€28 trials. Missing data may have been a result of not reporting or the study being a protocol with no final information on the number of clusters/patients being available
^ϪFor each study, we estimated the average cluster size by dividing the total number of patients recruited by the number of clusters (e.g., 200 patients recruited in a trial/10 clusters = 20 patients per cluster). We then took the median of the calculated average of patients per clusters from each trial

ISSN: 1745-6215