Hi, I'm planning a 2-period cross-over clinical study with measures taken at baseline, end of first period, end of wash-out and end of second period. Measures of the primary variable are taken semi-continuously via an electronic device that each patient carries on himself. These devices are subject to malfunction, in which case the absence of data would lead to the exclusion from the analysis of an otherwise valid patient. I was wondering whether multiple imputation could be of help in this case, but I have some doubts linked to the crossover design of the study.
Does anyone have previous experience of application of multiple imputation methods to crossover studies? any suggestion would be appreciated. Thanks in advance Roberto Ferrara
