Memory-Assisted Global Impression of Change (MAGIC)

Abstract

Background: The United States Food and Drug Administration is developing a series of patient-focused drug development guidance documents regarding the collection of patient experience data, including methods for understanding treatment benefit from the patient perspective. The goal of this research was to investigate the concern that the global impression of change scale is subject to recall error and thus not optimal for use as an anchor for estimating meaningful within-person change thresholds. We explored whether memory assistance for recalling baseline status would make a difference in how study participants diagnosed with Parkinson's disease (PD) responded to a patient global impression of change (PGIC) and patient global impression static (PGIS) item.

Methods: The research was completed as a secondary objective of a non-interventional 28-day (± 4 days) study among persons with Parkinson's disease and associated motor fluctuations. At baseline, participants completed the PGIS and then recorded a voice message to their future self in which they spoke about how their PD had affected their "day-to-day" activities over the preceding few days. At the final visit, the PGIC and PGIS were completed, after which participants listened to their memory assistance voice recording, and then completed both items for a second time to calculate a memory-assisted global impression static and change scores (MAGIS and MAGIC, respectively). Spearman correlations (ρ) were examined for the pre- and post- memory assistance evaluations. The degree of agreement pre- and post-memory assistance was quantified using the Shrout & Fleiss intraclass correlation coefficient (ICC [2,1]). An ICC(2,1) ≥ 0.7 served as the pre-specified criterion of acceptability for both the ρ and ICC(2,1) values.

Results: Participants in the analytic sample were mean age 68.7 and mostly white (91.7%) and male (69.4%). The average length of time since PD diagnosis was 6.5 years. Correlations between the PGIS and MAGIS were ρ = 0.88; correlations between PGIC and MAGIC were ρ = 0.86. The estimated ICC(2,1) for both the PGIS/MAGIS and PGIC/MAGIC exceeded target success criterion of ICC(2,1) ≥ 0.70.

Conclusion: Our results show that the MAGIS/MAGIC methodology is feasible and that memory assistance did not substantially alter the PGIS/PGIC scores at the final visit.

Keywords: Anchor-based; Meaningful within-person change; Parkinson’s disease; Patient global impression of change; Patient global impression static item.