If these factors tend to suggest the need for a brief questionnaire, the investigator is usually well served by an anticipated maximum time to completion of about 20 minutes. If a longer packet is feasible, an anticipated maximum duration of 45-60 minutes is usually appropriate, particularly if the methodology allows questionnaire completion over more than one sitting.

Completion of the questionnaire packet over more than one sitting is one of several techniques that may be built into the protocol in the effort to maximize the completion rate. If the questionnaire is a self-report, it is appropriate to permit the interviewer to read part or all to the patient, if fatigue or reading problems would otherwise prevent completion. If completion at home can be allowed, it should be aided by enclosure of a stamped self-addressed envelope.

In the absence of prior experience with the questionnaire packet in the study population, it is usually prudent to build into a survey a pilot phase during which the questionnaire packet is given to a small number of representative patients (usually about 10-20). Patients who participate in the pilot can be asked for feedback about the questionnaires. Length can be adjusted, and specific questions can be deleted or changed on the basis of this information.

It is inevitable that some questionnaires will be returned with missing data. It is appropriate to establish reasonable guidelines for the handling of this eventuality. There is no one best approach. If there are important subscales, e.g. for depression or sleep, it is appropriate to stipulate the number of missing items from the scale that would lead to deletion of the subscale score. This usually depends on the number of items in the scale. For the questionnaire overall, it may be appropriate to designate 15% incompletion as the threshold for withdrawing the patient from the totals.