Table 4 A new common sense for patient-reported outcome assessment^a

Do not rely on instrument databases for PRO identification and selection.

HRQL consists of symptoms, functions and limited aspects of the impact of these.

HRQL is very different from QoL.

The needs-based model of QoL is the most widely employed in medical research.

True QoL has rarely been measured in clinical studies and trials.

The content of QoL measures must be derived from relevant patients.

PROMs must be simple to administer, complete and score.

Simple two-point response formats are preferable to multiple response formats [43].

All PROMs used in clinical trials should be disease-specific.

Generic PROMs do not allow the impact of different diseases on patients to be compared.

Population norms for PROMs are invalid.

Think twice before selecting generic measures such as the EQ-5D to determine utility estimates, as they have limited psychometric quality.

QoL is a unidimensional construct.

Data collected using PROMs must be shown to be unidimensional.

Scores on subscales can rarely be added together to give a total score.

High reliability (reproducibility) is crucial to the accuracy of PROMs.

Forward-backward translation is a flawed methodology, creating unnecessary work.

Think carefully before using PROMs developed in the Western world in Asia and Africa.

Evidence is required of the scalability, reproducibility and construct validity of all language versions of PROMs used in a clinical trial.