Analyzing multiple learning effects in health care using multilevel modeling: Application to radiotherapy at an early stage of innovation

Abstract

Objectives:Learning effects may have considerable influence on the performance of new health technologies, thereby on cost-effectiveness and ultimately on resource allocation. In the area of radiotherapy, equipment is becoming increasingly costly and the analysis of learning effects is complex given that sequential treatments are necessary, with multiple sessions for each patient. Our study aimed at analyzing learning effects in radiotherapy at an early stage of innovation.Methods:We used multilevel analysis to separate out the different learning effects of the new technique. Statistical analysis of observational data collected in a French National prospective survey was performed using an individual growth model. Intrapatient learning was modeled at level 1, and two types of interpatient learning were considered at level 2, regarding possible influences of professional experience on (i) the duration of each patient's first session in a given setting and (ii) the rate of change of session duration over time for a given patient. Conventional radiotherapy was also considered for comparison.Results:Our results demonstrate a substantial type-1 interpatient learning effect and an even higher intrapatient learning effect. No type-2 interpatient learning was at work: professional experience did not impact intrapatient learning. Moreover, some intrapatient learning was also reported with conventional radiotherapy and was not significantly modified by innovation. Session duration was in any case strongly influenced by disease.Conclusions:Because professionals highly underestimated the learning phenomenon, assessment of learning cannot be based on professional statements and it requires careful analysis of observational data.