This study reports the effectiveness of a simple logistic intervention in reducing the non-completion of QOL assessment. The intervention consisted of simply changing the timing of distribution of questionnaires to participating centers from the first/second to the fifth week after treatment initiation. This simple intervention clearly improved the questionnaire response rate. We therefore plan to adopt this distribution procedure in future JPLSG protocols.
The new procedure is also expected to reduce the burden on physicians, who no longer need to keep the QOL questionnaire for a period of weeks but can instead administer it to patients almost immediately after receipt. Inconvenience and lack of manpower are barriers to introducing PRO assessment [9]. Our results clearly indicate that physician misplacement of the questionnaires or forgetting to conduct the QOL assessment is a major factor in non-completion. Determining the feasibility of different procedures for each center is important, as is consideration for manpower at the QOL office.
Changing the procedure in the middle of the clinical trial did not seem to cause confusion in the centers. In contrast, the QOL office received only a few reminders from some of the centers stating, “We have not received the QOL questionnaire set for a patient. Have you already sent it?” In response, the QOL office immediately sent the questionnaire set to the inquiring center. No obvious adverse effects arose from the procedure change.
It is possible that not all of the estimated effect of the intervention was directly due to the change in distribution procedure. For example, the change may have increased some physicians’ interest in QOL assessment. However, we do not think that awareness of QOL (i.e. understanding of the importance of QOL assessment) is easily changed. We announced this change in procedure only once, and informed staff only about what would be changed (timing of distribution), not why the change was being made (i.e. to improve the response rate). It is therefore unlikely that this announcement and change in procedure increased interest in QOL assessment at the participating centers. Instead, we propose that the estimated effect was due to the change in distribution procedure.
This intervention did not strongly appeal or intrusively compel the centers to administer the QOL questionnaire to patients. Thus, it functioned as a ‘nudge’ intervention [10]. The change in procedure likely induced an unconscious change in physicians’ behavior. Our findings indicate that even a casual change in clinical trial design can lead to a significant improvement.
In the current setting, utilization of electronic data collection systems is feasible. At the time we introduced QOL assessment into clinical trials around 2010, however, we were unable to determine whether QOL assessment would be routinely included in future clinical trials at all JPLSG centers and, given the high initial cost of electronic systems, selected a paper-based system instead. With our current implementation of QOL assessment in several trials, however, the total cost of an electronic system in supplies and manpower would be lower than the present paper-based system, and implementation of an electronic data collection system for JPLSG is now underway. With these systems, the initial distribution of system information to patients should be done by a physician or research coordinator at each center. Therefore, the implications of this study concerning distribution process and timing are still relevant.
Some limitations of this study warrant mention. First, we only reported the effect of the intervention on the response rate to the first QOL questionnaire. A previous paper reported that while it was possible to improve the response rate, sustaining the improvement is difficult [11]. Second, the reason for the decrease in response rate with time is unclear. Reported factors for non-completion of QOL assessment [2] do not explain the decrease. Third, we did not employ a randomized design, and the possibility of confounding by unknown confounding factors cannot be eliminated.
These limitations notwithstanding, we found that a seemingly minor change in the distribution timing of questionnaires resulted in a clinically significant improvement in first QOL assessment in a clinical trial of pediatric cancer.