Marion Campbell: If you wish to use a patient-reported outcome measure in your trial, it is important to choose your measure wisely to ensure that you won’t observe ‘floor’ or ‘ceiling’ effects in your data.

Quoting Marion Campbell, Professor of Health Services Research at the University of Aberdeen, on X/Twitter:

“If you wish to use a patient-reported outcome measure in your trial, it is important to choose your measure wisely to ensure that you won’t observe ‘floor‘ or ‘ceiling’ effects in your data. Floor/ceiling effects happen when a significant percentage of your trial population scores the worst values (floor effect) or alternatively the best possible values (ceiling effect).

The consequence of floor/ceiling effects is that the measure may not be able to discriminate between the trial groups or identify a difference between interventions (For example, if lots of people always score the best values, how can any improvement/difference be observed?).

This can severely impact the potential of your trial to undertake a valid comparison – if the measurement outcome cannot discriminate appropriately. There is no hard and fast rule to conclude that floor or ceiling effects exist, although some rules of thumb exist. For example, in orthopaedic research, floor/ceiling effects are usually deemed to happen if >15% of participants score the best/worst scores.

The chance of high/low values being observed is a mix of the measure and the population to which you are applying them. For example, a measure will have more likelihood of low values in a critically ill population and high levels in a healthy population.

As such, you need to ensure the measure you choose is appropriate for your trial population & that its properties are well understood in your particular context. Prior feasibility work is essential if this is not known in your field.

Written some years ago, the National Institute of Health and Care Research report written by Ray Fitzpatrick et al remains a key resource here. It outlines the key elements that you need to consider when choosing an appropriate patient-reported outcome measure for use in clinical trials.”

Source: Marion Campbell/Twitter