We would like to ask your advise regarding the missing data review. We are focussing on missing GOS values only. We would like to clarify the distinction of pairwise and listwise deletion.
We have seen that the majority of studies report predictors on all patients enrolled, however many lose patients to follow up. They then carry out analyses of outcomes using the number of patients where outcome is available only. Is this considered pairwise deletion of data?
Conversely, there are studies in whom completeness of GOS data was an inclusion criterion. Therefore they did not even report predictors on those patients with missing GOS. We consider this listwise deletion.
We would welcome your opinion.