Background A physician global estimate of patient status (DOCGL) was developed to quantify inflammatory activity in rheumatoid arthritis (RA) clinical trials. However, DOCGL may be affected by joint damage and/or distress (in fibromyalgia, depression, etc). One approach to document the possible impact of these problems on DOCGL is to add 3 physician visual analog subscale (VAS) estimates for inflammation, damage, and distress. These subscales have been shown to be useful in patients with diagnoses other than RA (1) but inter-rater reliability has not been analyzed.
Objectives To analyze inter-rater reliability between senior rheumatologists and trainees on 4 VAS estimates for overall DOCGL, inflammation (DOCINF), damage (DOCDAM) and distress (DOCSTR), in patients with various rheumatic diagnoses.
Methods Patients seen in routine care were assigned 4 physician VAS estimates for overall DOCGL, and levels of inflammation or reversible symptoms (DOCINF), organ damage or irreversible symptoms (DOCDAM), and distress or symptoms not explained by inflammation or damage (DOCSTR). VAS estimates were assigned independently by a senior rheumatologist and a rheumatology trainee for the same patient at the same visit. Mean differences, correlations, and possible discordance of ≥2units/10 between estimates of the senior rheumatologist and the trainee were analyzed.
Results VAS estimates by the 2 physicians were analyzed in 64 patients with different rheumatic diseases, including osteoarthritis (16%), RA (14%), fibromyalgia (14%), and systemic lupus erythematosus (13%). Mean differences of scores assigned by the senior rheumatologists versus trainees were <0.43/10, less than 5% of the total scales, slightly lower for DOCINF, and slightly higher for the 3 other subscales (p<0.001) (Table). Mean estimates of both physicians for damage and distress were higher than for inflammation by 1.1 to 1.6 units (Table). Correlations of all 4 VAS between rheumatologists and trainees were significant (p<0.001) (Table). More than 70% of the estimates were concordant for DOCGL (75%), DOCINF (78%), and DOCDAM (70%), while concordance was somewhat lower for DOCSTR (57%) (Table).
Conclusions Good inter-rater agreement between two physicians is seen for 4 VAS estimates for overall global assessment, inflammation, damage, and distress. Mean scores for damage and distress were higher than for inflammation, indicating the complexity of rheumatology care. Quantitative scores can add to documentation of patient status and to support of clinical decisions for doctors, patients, and payers.
Bull Hosp Jt Dis (2013). 2015 Jul;73(3):178–84.
Disclosure of Interest T. Pincus Shareholder of: Health Report Services, Inc, I. Castrejon: None declared, J. Chua: None declared, A. Kugasia: None declared, J. Schmukler: None declared, S. Weinberg: None declared, J. Block: None declared