Background There is a growing interest in the use of patient-reported outcome measures in clinical trials and in clinical practice. The 2008 EULAR/ACR collaborative recommendations for rheumatoid arthritis (RA) included patient global assessment (PaGl), pain and fatigue as key patient-reported outcome measures (ref.). However, although associations between PaGl, pain and fatigue have been examined on the group level, studies focusing on the agreement between these patient-reported measures in individual patients are missing. A better understanding of how tight the measures are bounded in individuals may improve our ability to deal with them in the daily clinic.
Objectives To examine associations on the group level and agreements on the individual patient level between PaGl, pain and fatigue as scored on visual analogue scales (VAS) in the daily clinic by patients with active RA.
Methods Data on 221 RA patients with active and uncontrolled disease planned to initiate treatment with a biological agent were extracted from the Danish registry for biological treatment in rheumatology (DANBIO). Data included PaGl, pain and fatigue assessed on VAS-scales (0-100) and age, swollen and tender joints counts, physician global assessment, CRP and HAQ-DI. The predictability of PaGl, pain and fatigue was examined using multiple regression analysis. Agreements between PaGl, pain and fatigue on the individual patient level was examined by Bland-Altman analyses yielding 95% lower and upper limits of agreement (LLoA and ULoA) between intra-individual assessments. The difference between the scores on the group level was expressed as the bias.
Results Mean age was 57±14 years, mean DAS28-CRP 5.0±0.9 and mean PaGl 63.6±22.6. Women (n=171) had higher levels of fatigue than men (63±23 vs. 52±21, p<0.005) and the difference remained statistically significant (p<0.05) after correcting for potential covariates. PaGl was most strongly predicted by pain and fatigue, pain by PaGl and fatigue, and fatigue by PaGl and pain (beta 0.18 - 0.69, p<0.05-0.0001). More objective measures were not or far less predictive. When repeating the regression analyses with each of the three VAS scores as dependent variables but after exclusion of the two other VAS scores as independent variables, HAQ-DI was the strongest predictor of both PaGl, pain and fatigue (beta 0.39 - 0.45, p<0.0001). On the individual level, LLoA and ULoA [bias] for PaGl vs. pain was -19.1 and 29.5 [5.2], for PaGl vs. fatigue -22.8 and 28.6 [2.9], and for fatigue vs. pain -29.2 and 33.8 [2.3]. LLoA and UloA remained constant over the whole range of the VAS-scales.
Conclusions PaGl, pain and fatigue scores were most strongly explained by each other, not by more objective clinical measures. Although PaGl, pain and fatigue were highly and independently inter-correlated and the mean scores were very close on the group level, disagreements between the scores were substantial on the individual patient level. The findings emphasize the complexity of understanding patient-reported outcome measures and their diverging interplay across individuals.
Aletaha D et al. Reporting disease activity in clinical trials of patients with rheumatoid arthritis: EULAR/ACR collaborative recommendations. Arthritis Rheum 2008; 59: 1371-7.
Disclosure of Interest None declared