Background/Purpose: Arthritis symptoms reported by patients have been anecdotally associated with weather changes, but large-scale, systemic evaluations are few in number. A variety of parameters associated with weather that might underlie arthritis-related pain and related symptoms have been inconsistently reported.

Methods: Patients participating in the ArthritisPower registry and contributing data via a Smartphone or computer App from the continental U.S. were eligible for analysis. Geolocation (latitude/longitude) was extracted from the Smartphone’s physical location or computer IP address. Various weather parameters (e.g. temperature, humidity, wind speed/direction, barometric pressure) were obtained from the nearest National Oceanic and Atmospheric Administration (NOAA) weather station based on patient’s geolocation. Various restrictions in the maximal allowable distance to the nearest weather station (e.g. <25 miles) were evaluated. Patient disease activity by the RAPID3, and patient reported outcomes (PROs) including pain interference, fatigue and physical function measured by the NIH PROMIS instruments (using computer adaptive testing) were obtained from the registry, and associated with NOAA weather data at that same time (to the nearest hour) and location, and at the same location 24 hours before and after each patient observation. Cross-sectional correlation between various weather parameters and PROs were quantified as r values using Pearson correlation coefficients. A “cold front” definition was proposed based on the confluence of longitudinal change over 3 days in relative humidity, wind direction, barometric pressure, and dew point.

Results: At the time of this analysis, 1334 unique patients contributed 2425 PRO observations with linkable NOAA weather data. Mean(SD) age was 53.9(10.3) years, 91% women, 90% white. In terms of various arthritis conditions represented in ArthritisPower, 45% had rheumatoid arthritis, 10% psoriatic arthritis, 9% ankylosing spondylitis, and 62% osteoarthritis (with or without a concomitant inflammatory arthritis). Many of the correlations between various weather parameters and PROs were statistically significant (p < 0.001) albeit numerically weak (all r values < 0.2). For patients contributing any PRO data at the time of an evolving cold front using the proposed definition, patient symptoms were not different as measured by various PROs (Table). 

Conclusion: Weather is quantitatively related to patient’s arthritis symptoms. Additional work is ongoing to refine specific weather parameters and their associations with PROs in order to provide potentially actionable information to patients and their healthcare providers.

Table: Association between Patient Symptoms and the Presence of a Cold Front based on Geolocation and Linked Weather Data (n=2425)