Evaluating treatment response to intramuscular steroids in rheumatoid arthritis: exploring the potential of electronic patient-generated health data

Background: The increasing availability of mobile health devices presents exciting opportunities for remote collection of high frequency electronic patient-generated health data (ePGHD). This novel type of data has the potential to provide detailed insights into disease activity outside of usual clinical settings. Assessment of treatment response, which can be hampered by infrequency of appointments and the influence of recall bias, is a promising and novel application of ePGHD. This challenge is well-illustrated by drugs with short treatment effects, such as intramuscular steroid injections. Patients are unlikely to accurately recall their treatment response at follow-up visits which often take place several months after administration. Retrospective assessment means that response may be over- or under-estimated. High-frequency ePGHD, such as daily patient-reported symptoms collected via smartphone app between clinic appointments, may provide an opportunity to bridge this gap. However, the potential of ePGHD remains untapped due to the absence of established methodological approaches for analysis of this type of data, and the absence of established definitions for what entails treatment response using ePGHD.

Objectives: This study aims to explore the feasibility of evaluating treatment response to intramuscular steroid therapy in a case series of patients with rheumatoid arthritis tracking daily symptoms using a smartphone app.

Methods: We report a case series of patients who collected ePGHD through the REmote Monitoring Of Rheumatoid Arthritis (REMORA) smartphone app for daily remote symptom tracking. Symptoms were tracked on a 0-10 scale. We described their longitudinal pain-scores before and after intramuscular steroid injections. A baseline pain-score was calculated as a mean pain-score in the 10 days prior to the injection. This was compared to the pain-scores in the days following the injection. Response was defined as any improvement from the baseline score on the first day following the injection. The response end time was defined as the first date when the pain-score exceeded the pre-steroid baseline.

Results: We included six patients who, between them, received nine steroid injections. Average pre-injection pain-scores ranged 3.3-9.3. Using our definitions, seven injections demonstrated response. Of the responders, duration of response ranged from 1-54 days (median 9); average pain-score improvement ranged from 0.1-5.3 (median 3.3); maximum pain-score improvement ranged from 0.1-7.0 (median 4.3).

Conclusions: This case series demonstrates the feasibility of using ePGHD to evaluate treatment response and is an important exploratory step towards developing more robust methodological approaches for analysis of this novel data type. Issues highlighted by our analysis include the importance of accounting for one-off data points, varying response start times and confounders such as other medications. Future analysis of ePGHD across a larger population is required to address issues highlighted by our analysis and to develop meaningful consensus definitions for treatment response in time-series data.