Background Patient reported outcomes (PRO) represent an important part in disease activity and outcomes assessment in rheumatoid arthritis (RA) patients. The Patient Global Assessment (PGA) represents one of the most essential PROs, being comprised in many disease activity indices and the new remission criteria. In a seven day test-retest study a smallest detectable difference (SDD) for PGA of 27mm was found. However, since RA disease activity may fluctuate to an unknown extent, it is unclear how estimates of SDD are influenced by the investigated time interval, and to what extent SDD may differ according to the baseline assessment of PGA.
Objectives We aimed to investigate if SDD estimates change in different test-retest intervals, and if they can be better specified for patients based on their current measurement of PGA.
Methods Forty RA patients from our outpatient clinic were asked to report their PGA on a daily basis over a period of 28 days in this prospective observational study. We calculated variability of PGA comparing the initial measurement to ones performed after 1 to 27 days. We calculated reliability using the intra-class correlation coefficient (ICC), which reports the amount of measurement variability attributable to true differences (and not to measurement or random error). The SDD was conventionally calculated as the standard deviation of the difference between two measurements multiplied by 1.95. The latter is very helpful for interpretation in daily practice. In a final step, we divided the patient population into tertiles by baseline PGA to investigate differences in SDD based on the starting point.
Results Data of forty patients (85% female, 63% rheumatoid factor positive, mean SDAI: 13.9--±8.7, mean disease duration 13.2±8.2) was used for analyses in this study. The mean ICC between day 1 and each consecutive day was 0.631 (SD=0.127) and ranged between 0.42 and 0.91 (Figure). The SDD ranged between 13.4mm and 33.3mm and averaged to 25.5±7.5 (Figure). As expected, higher reliability (ICC) coincides with a smaller SDD, but in neither of the two a temporal trend was found, indicating that the calculation of SDD is more generic and not interval dependent. Ranges for the three tertiles by baseline PGA were 0 to 11 (mean=5.6mm), 15 to 28 (mean=20.4mm), and 29 to 74 (mean=48mm). SDDs were significantly different across these three groups, resulting in (mean with 95% confidence interval; in mm) 16.0 (12.6 to 19.4), 21.1 (18.0 to 24.2), and 32.3 (27.8 to 36.8), respectively.
Conclusions We demonstrated here that reliability and SDD of PGA are generic, and not dependent on the time between the measurements. An overall 25mm cut-off was found for claiming true change in PGA in general, but much smaller changes are relevant in subsets of patients. Particularly numerous patients who fail to reach remission just because of the PGA may potentially be considered as remitters if PGA variability is considered.
Disclosure of Interest None Declared