Background:Hand osteoarthritis (OA) is a chronic, progressive disease, commonly affecting middle aged women. OA at the interphalangeal joints (IPJs) or the thumb base are considered different disease subsets (1). Few studies have investigated individual risk factors for IPJ OA progression (2). Prediction models can be used to calculate overall disease risk from multiple risk factors. This can guide prevention and treatment options.Objectives:Develop and internally validate a prediction model for IPJ OA progression.Methods:Data from the Chingford 1000 Women Study (Chingford Study), the largest population-based cohort worldwide assessing hand OA, was used. It is representative of the middle-aged female population in the United Kingdom (3). At baseline, 1,003 women aged 45 to 64 years’ old were recruited, and 693 measurements taken. Hand radiographs were taken at baseline and after ten years, read using the Kellgren-Lawrence (KL) atlas (inter-observer correlation: ≥0.7 (4)).For the current study, participants must have had OA (KL ≥2 in ≥1 IPJ) on baseline hand radiographs. Participants with KL 4 in all 16 IPJs at baseline were excluded. Risk factors from the Chingford Study at baseline were selected by biological plausibility, literature evidence (2), and hand surgeons‘ consensus (5): age (years), occupation (manual versus non manual), OA in ≥1 thumb base (KL ≥2 versus KL<2), body mass index (BMI) (kg/m2), family history of hand OA (yes versus no). The outcome was defined on an ordinal scale for the number of IPJs (up to >5 IPJs) with OA progression (increase by KL ≥1), at ten years’.The prediction model was developed using a penalized proportional odds logistic regression. Odds ratios (95% confidence intervals) were reported for each risk factor. The model was internally validated using 2,000 bootstrap iterations. Model performance was assessed for discrimination (C-statistic), and calibration (C-slope). 3.5% of data was missing, and complete case analysis was used.Results:699 women had baseline hand radiographs: 38 were unreadable, 459 had no IPJ OA. Seven participants had missing data (occupation: 5, BMI: 1, family history: 1) and were excluded. 195 participants were included this study. Median age at baseline was 59 (interquartile range: 8) years.181 (92.8%) participants had OA progression at 10 years (Figure 1). Thumb base OA (odds ratio: 1.32 (0.93 to 1.88)) was most strongly associated with IPJ OA progression (Table 1). C-statistic was 0.57, and calibration slope was 1.38 for the optimism-corrected model.Table 1.Odds ratios for risk factorsRisk factorOdds ratio (95% confidence interval)Age (years)1.02 (0.99 to 1.06)Occupation (manual versus non manual)0.88 (0.60 to 1.29)Thumb base OA (Kellgren-Lawrence grade ≥2 versus <2)1.32 (0.93 to 1.88)Family history of hand OA (yes versus no)1.03 (0.72 to 1.45)Body mass index (kg/m2)1.04 (0.99 to 1.09)OA: OsteoarthritisConclusion:More stringent cut-offs for OA progression would be clinically useful. It was only weakly possible to predict which participants with IPJ OA would progress. This suggests that other risk factors, such as gender, ethnicity and genetics, may be predominant.Figure 1.Hand interphalangeal joints with osteoarthritis progression (Kellgren-Lawrence grade ≥1) at 10 years’ follow upReferences:[1]Kloppenburg M, et al. Research in hand osteoarthritis: time for reappraisal and demand for new strategies. Ann Rheum Dis. 2007;66(9):1157-61.[2]Shah K, et al. Risk factors for the progression of finger interphalangeal joint osteoarthritis: a systematic review. Rheumatol Int. 2020;40(11):1781-1792.[3]Hart DJ, Spector TD. The relationship of obesity, fat distribution and osteoarthritis in women in the general population: the Chingford Study. J Rheumatol. 1993;20:331-335.[4]Hart DJ, et al. Reliability and reproducibility of grading radiographs for osteoarthritis of the hand. Br J Rheum. 1993;32:S1.[5]Shah K, et al. Delphi consensus of risk factors for development and progression of finger interphalangeal joint osteoarthritis. J Hand Surg Eur Vol. 2019;44(10):1089-1090.Acknowledgements:We would like to thank all of the participants of The Chingford 1000 Women Study, Professor Tim Spector, Dr Deborah Hart, Dr Alan Hakim, Maxine Daniels, Alison Turner, James van Santen and Julie Damnjanovic for their time and dedication.Disclosure of Interests:Karishma Shah: None declared, Garrett Bullock: None declared, Alan Silman: None declared, Dominic Furniss: None declared, Nigel Arden Consultant of: Receives personal fees from Pfizer/Lily for consultancy outside the scope of this work, Grant/research support from: Receives grant from Merck outside the scope of this work, Gary Collins: None declared
External validation and updating of a prediction model for the diagnosis of gestational diabetes mellitus
