Table 1 Responses from Cochrane review group co-editors who provided reasons for missing data in their reviews

Pain, Palliative Care and Support Care

‘I think this is about right [amount of missing data]. Our scope is large and these problems are more common when there are no received methods (e.g. rehabilitation, palliative care, complex interventions) and less common where there is a history of methods development and consensus (e.g., headache).

Dementia and Cognitive Impairment

‘Would suggest that it may not be possible in our CRG to specify outcomes only by the condition without reference to the type of intervention. For example, some interventions in dementia may aim to modify the course of the disease, and cognitive outcomes should clearly be part of a minimum outcome set in such trials; for others, such as certain psychosocial interventions (e.g. touch and massage), modifying cognitive decline would be an unreasonable aim, and collection of cognitive data would be unduly burdensome for patients and researchers’.

Oral Health

‘Our reviews are very diverse so we would need to develop many different core outcome groups’.

Neuromuscular Diseases

‘Neuromuscular Disease is very diverse, even sometimes within one disease. For instance some patients with Motor Neuron Disease will have lots of spasticity and complications related to that, others flaccidity and others bulbar involvement. Variations in outcomes chosen depend upon the focus of the intervention and the researcher. We try to keep our outcomes fairly loosely defined - e.g. any measure of walking ability - especially in physio/fatigue/strength training and other qualitative research areas - but even then there are some fairly abstract outcomes measured that don’t fit in clearly’.

Renal

‘The problem is the heterogeneity of our scope. Unlike some other CRGs we are an organ-related CRG (and then cover many orphan reviews). We cover everything from UTI (urinary tract infections) to transplantation’.

Cystic Fibrosis and Genetic Disorders

‘One set of outcomes is not always appropriate for every trial in cystic fibrosis, e.g. for a nutritional intervention body mass index may be appropriate. For a respiratory therapy it may be spirometry’.

Heart

‘The issue of missing data for primary outcomes may reflect the biomedical bias in many heart disease trials - so if we have 20 RCTs in a review with a clinical primary outcome specified, it is very likely that only a few of the trials will have reported on the clinical outcome but many more will have reported on physiological outcomes..Our scope is much wider than many review groups, so I can envisage difficulties coming up with a core outcome set’.

Wounds

‘Our reviews cover a variety of conditions and interventions (probably most groups do) and relate to topics with quite under-developed research cultures (nursing, dermatology, surgery) so the high levels of missing data in our field doesn’t surprise me’.