Back to 2026 ePosters
Correlations Between Patient-Reported Outcome Measures and the Curtis Radiographic Classification for Thumb Carpometacarpal Osteoarthritis
Bradley C Nelson, MS
1,2, Kavya K Sanghavi, MPH
2, Aviram M. Giladi, MD, MS
3; Kenneth R. Means, Jr., MD
3(1)Georgetown University School of Medicine, Washington, DC, (2)Curtis National Hand Center, Baltimore, MD, (3)The Curtis National Hand Center, Baltimore, MD
Introduction:
Available data indicate a general lack of correlation between the available thumb carpometacarpal osteoarthritis (CMC OA) radiographic classifications and patient-reported symptoms. We developed the Curtis classification as an extension of the Eaton-Glickel system by incorporating radiographic measurements via modern digital imaging technology (Figure 1) and found improved reliability (Table 1). In the current study, we investigated whether the Curtis classification stages-and each of its radiographic parameters-correlated with CMC OA patient?reported outcome measures (PROMs).
Materials and Methods:
We performed a retrospective analysis of 85 patients from our non-operative CMC OA registry. Eligible patients (ages 35-85) had no prior CMC surgery and no non-surgical treatment within 12 months prior to enrollment. Inclusion required adequate radiographs within one year of baseline PROM capture. PROMs included the Brief Michigan Hand Questionnaire (bMHQ), three visual analog/numerical rating scales (VA/NRS) for different pain aspects, PROMIS Pain Interference (PI) v1.1, and PROMIS Global Health (GH), from which Global Mental Health (GMH), Global Physical Health (GPH), and health-related quality of life per EuroQoL are calculated. A board-certified hand surgeon and a trained medical student graded the images. Spearman's rho was used to assess correlations between Curtis stage, along with its constituent parameters, and PROMs.
Results:
Inter- and intra-rater reliability for Curtis staging was substantial to near-perfect (? = 0.69 - 0.90). We included 106 CMC radiographs from the 85 non-operative registry patients in the final dataset (median age 61, 76% female). Final Curtis staging distribution was: Stage 1 (n=5), Stage 2 (n=31), Stage 3A (n=36), Stage 3B (n=23), Stage 4 (n=11). All correlations between Curtis stage (Table 2), along with its radiographic constituents, and PROMs were weak or very weak (rho <0.30) and not statistically significant (
p > 0.05).
Conclusion:
Despite excellent reliability, Curtis classification and its radiographic elements did not meaningfully correlate with baseline PROMs, supporting existing evidence that radiographic severity is poorly linked to clinical impact in CMC OA.
Back to 2026 ePosters
