POS0120 REDUCING SELECTION BIAS WHILE MAINTAINING PRECISION THROUGH AN INTEGRATED ANALYSIS: 2-YEAR LONGITUDINAL ANALYSIS OF IMAGING OUTCOMES IN THE SPONDYLOARTHRITIS CAUGHT EARLY COHORT
Annals of the Rheumatic Diseases, vol. 84, pp. 414–415
Abstract
Background: Imaging outcomes in ongoing cohort studies are typically evaluated across multiple reading rounds by several readers. Most often, only the scores from the most recent reading round are used, disregarding valuable data from previous rounds and potentially introducing selection bias by excluding participants lost to follow-up. An alternative approach involves using an integrated analysis that incorporates scores from multiple readers and rounds. While this method provides a more truthful reflection of the data, it may introduce additional variability, potentially reducing the precision of the estimates. Objectives: To assess if an integrated analysis, with use of available imaging data over time from different reading rounds, affects the precision of estimates of change in patients with early axSpA, using a completers analysis as reference. Methods: Patients with chronic back pain (CBP, ≥3 months; ≤2 years; onset <45 years) from the SPondyloArthritis Caught Early (SPACE) inception cohort were diagnosed by a rheumatologist as axSpA or non-axSpA CBP after two-years of follow-up. Patients with a ‘definite' (level of confidence about the diagnosis ≥7 at last visit) or ‘most likely' (level of confidence <7 at last visit and a consistent diagnosis in the last 2 consecutive visits) diagnosis of axSpA were included in this analysis.(Marques et al. 2024) MRIs and radiographs of the sacroiliac joints (SIJ) and spine obtained at baseline, 1 and 2 years of follow up were scored for inflammatory and/or structural lesions by at least 2 and a maximum of 3 readers in two reading rounds: in round A, baseline and 1 year were scored; in round B, baseline, 1 and 2 years were scored. Each outcome was analysed per reader. Change over time was analysed with multilevel generalized estimating equations (considering the reader level), with ‘time' as explanatory variable. Two approaches were used for the analyses: i) ‘integrated analysis' (including all patients with ≥1 score from ≥1 reader from ≥1 reading round); ii) ‘completers-only analysis' (including all patients with ≥1 score from ≥1 reader from reading round B at 2-year follow-up). The second and most commonly use approach was considered the reference. Different time transformations (e.g. linear, quadratic, cubic) were tested to investigate which yielded the lowest quasi-likelihood under the independence model criterion (QIC), and thus the best model fit. In cases of comparable fit, the model with the simplest format of time was chosen. Results: In total, we analyzed 332 patients with axSpA (age 30 (SD 8) years; 53% males) of which 279 (84%) patients were included in the ‘completers-only analysis'. The estimated change over time of the imaging outcomes is shown in Tables 1 and 2. Binary and continuous variables for inflammatory lesions decreased over time in the MRI-SIJ and the MRI-spine, while structural lesions increased over time. Models with a linear representation of time showed either the best fit or were very comparable to the ones with the best fit and were therefore used for all outcomes. Comparing both types of analysis, the ‘integrated analysis' included more patients but with similar levels of precision of the change estimates than the ‘completers-only analysis' for both the binary and continuous variables. Particularly for binary outcomes, a comparable incidence was obtained by the integrated analysis with more precision compared to the completers-only analysis, e.g. sacroiliitis according to ASAS criteria [estimated % change per year (95% CI): -5.58 (-8.30; -2.85) vs -5.74 (-8.52; -2.96)]. Conclusion: In this 2-year longitudinal study, the use of an integrated analysis allows for the inclusion of all available data, thereby reducing selection bias in determining the population for analysis, while maintaining precision of estimates of change in patients with early axSpA. REFERENCES: [1] Marques ML et al. Ann Rheum Dis 2024;83:589–598. Acknowledgements: NIL . Disclosure of Interests: Liese de Bruin: None declared, Floris van Gaalen Novartis, MSD, AbbVie Bristol Myers Squibb and Eli Lilly, Stichting vrienden van Sole Mio, Stichting ASAS, Novartis, UCB, Manouk de Hooge UCB Pharma, Miranda van Lunteren: None declared, Mary Lucy Marques Novartis, Novartis, Monique Reijnierse: None declared, Roberta Ramonda: None declared, Inger Jorid Berg: None declared, Carl Turesson Abbvie, Bristol Myers-Squibb, Nordic Drugs, Novartis, Pfizer and Roche, Abbvie, Nordic Drugs, Robert Landewé AbbVie, UCB, Novartis, Galapagos, Jansen pharma, AbbVie, UCB, Galapagos, Jansen, Désirée van der Heijde AbbVie, Alfasigma, ArgenX, BMS, Elly-Lilly, Grey-Wolf Therapeutics, Janssen, Novartis, Pfizer, Takeda, UCB Pharma, Sofia Ramiro Eli Lilly, Novartis and UCB, AbbVie, Eli Lilly, Galapagos/Alfasigma, Janssen, MSD, Pfizer, UCB, Sanofi, AbbVie, Galapagos/Alfasigma, MSD, Novartis, Pfizer, UCB. © The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.
Authors 11
-
Leiden University Medical Center
Affiliation as printed
Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands
-
Affiliation as printed
Ghent University Hospital, Department of Rheumatology, Ghent, Belgium
-
Leiden University Medical Center
Affiliation as printed
Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands
-
Leiden University Medical Center · Unidade Local de Saúde de Coimbra
Affiliation as printed
Coimbra Local Health Unit, Department of Rheumatology, Coimbra, Portugal
Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands
-
Leiden University Medical Center
Affiliation as printed
Leiden University Medical Center, Department of Radiology, Leiden, Netherlands
-
Affiliation as printed
University of Padova, Rheumatology Unit, Department of Medicine-DIMED, Padova, Italy
-
Affiliation as printed
Diakonhjemmet Hospital, Center for treatment of Rheumatic and Musculoskeletal Diseases (REMEDY), Oslo, Norway
-
Affiliation as printed
Lund University, Rheumatology, Department of Clinical Sciences Malmö, Malmö, Sweden
-
Zuyderland Medisch Centrum · Amsterdam University Medical Centers
Affiliation as printed
Amsterdam University Medical Center, Department of Rheumatology, Amsterdam, Netherlands
Zuyderland Medical Center, Department of Rheumatology, Heerlen, Netherlands
-
Leiden University Medical Center
Affiliation as printed
Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands
-
Leiden University Medical Center · Zuyderland Medisch Centrum
Affiliation as printed
Leiden University Medical Center, Department of Rheumatology, Leiden, Netherlands
Zuyderland Medical Center, Department of Rheumatology, Heerlen, Netherlands
Cited by 0 stored of 0
No patents citing this paper on Lens.org (checked 2026-10-11).