Suggested answers

Randomisation and allocation concealment

Assignment to groups was determined by central off-site randomisation and so could not be influenced by the investigators, nurses or participants. Allocation concealment was therefore adequate.

Blinding

This trial is an example of a complex, procedural intervention where complete blinding is not possible. Attempts to blind participants may be made either by designing an appropriate placebo or partially concealing (within ethical boundaries) the purpose of trial from participants. Neither was done in this trial. Obviously, in a procedural trial it is not possible to blind the health care workers (unlike a drug trial when an identical placebo will blind the carer as well as the participant). Outcome assessment (determining rates of death and rates and cause of admission) was blind to treatment allocation. A potential problem with outcome assessment is the use of hospital and health department records to determine death and readmission, with the possibility of records missed. A more appropriate method of assessing outcome would have been for the investigators to formally follow-up participants at given time points.

Here is a good time to make note of a common problem in assessing the adequacy of blinding within randomised controlled trials. Often you will see a trial, and indeed validity scale, referring to the trial being "single blind", "double blind" and "triple blind". The problem with this phrasing is that, while single blind obviously refers to only one party being masked from the intervention, double blind two parties etc, it does not make it clear who it is without the knowledge. A triple blinded trial is usually referring to intervention being masked from participants, care providers and outcome assessors, but what of double blind? Some people interpret this as participants and care providers, others as participants and outcome assessors. And what of single blind? As you can see, a much better way of describing blinding within the methods of a trial, and in any scale you chose to use to assess its validity, is to specify which party or parties were unaware of the intervention.

Loss to follow up

165 patients were randomised and 156 completed the trial. A description is provided of those who failed to complete (see Figure 1). From Table 3, we can see intention-to-treat was used as the column heading, showing 81 participants in usual care and 84 in intervention group. This value is the number randomised.

Now return to Module 7 and incorporate what you have learned into your data extraction form.