POS1379 A NOVEL MEASURE OF END-STAGE KNEE OSTEOARTHRITIS REDUCES THE DURATION AND SAMPLE SIZE REQUIRED FOR OBSERVATIONAL STUDIES AND TRIALS

Background Total knee replacement (TKR) has been used as an outcome measure in research into the causes and possible treatments for knee osteoarthritis (KOA). However, because KOA progresses slowly, and because TKR has a low incidence, research using TKR as an outcome measure necessitates long duration and/or large sample sizes. Moreover, TKR is influenced by multiple factors (such as education and income) besides the progression of KOA. Objectives We defined a novel outcome measure that signifies end-stage KOA (esKOA); and determined whether esKOA was sensitive enough to detect the effect of an exposure that is known to have a modest effect on reducing TKR, namely weight loss. Methods A knee was considered to have esKOA if any of the following two conditions were met: 1) moderate, intense, or severe KOA symptoms (i.e., the sum of the pain and disability scores on the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) ≥ 12) and severe radiographic knee osteoarthritis (RKOA), defined as a Kellgren and Lawrence Grade (KLG) of 4); or 2) intense or severe KOA symptoms (i.e., the sum of the pain and disability scores on the WOMAC ≥ 23) and frequent knee pain (i.e., knee pain on most days of one or more months in the past 12 months) and mild or moderate RKOA (KLG = 2 or 3). We used data from two prospective cohort studies: the Osteoarthritis Initiative (OAI) and the Multicenter Osteoarthritis Study (MOST). We analyzed the data in two ways: as an observational study; and as an emulated trial. In the emulated trial, participants who lost ≥ 5% of their weight between baseline and 1 - 1.25 years (weight loss group) were matched with participants who gained ≥ 5% of their weight in that same period (weight gain group), using 1:1 nearest-neighbour matching on propensity scores. In both analyses, we used a multilevel mixed-effects generalized linear model. In the observational study, we investigated the association of weight loss between baseline and the following time points with esKOA and TKR at these time points: 1 - 1.25 years; 2 - 2.5 years; and 4 - 5 years. In the emulated trial, we compared the odds of incidence of esKOA and TKR between the weight loss and weight gain group at the following time points: 2 - 2.5 years; and 4 - 5 years. Results The observational study included 7107 participants (58.4% female, mean ± SD age and BMI 61.4 ± 8.8 years and 29.2 ± 5.1 kg/m2 at baseline, and an incidence of esKOA of 2.9, 6.8, and 10.4% at 1 - 1.25 years, 2 - 2.5 years, and 4 - 5 years, respectively, and a corresponding incidence of TKR of 0.1, 0.5, and 1.6%. While weight loss was associated with a reduced adjusted odds ratio (aOR) for both esKOA and TKR at 4 - 5 years (for 5% weight loss: 0.85 [95% CI 0.79 - 0.92] for esKOA and 0.79 [0.67 - 0.93] for TKR), weight loss was only associated with a reduced aOR for esKOA - and not TKR - at the earlier time point of 2 - 2.5 years (for 5% weight loss: 0.80 [0.72 - 0.90] for esKOA and 1.05 [0.76 - 1.45] for TKR). At 1 - 1.25 years, there was no association between weight loss and esKOA or TKR. The sample size required to detect a 50% reduction in the odds of esKOA was 6% to 13% of the sample size required for that of TKR (236 versus 3990 at 2 - 2.5 years; 162 versus 1286 at 4 - 5 years). In the emulated trial, compared to the weight gain group (367 participants), the weight loss group (also 367 participants) had significantly lower odds of esKOA but not TKR at 4 - 5 years (0.43 [0.22 - 0.84] for esKOA and 0.39 [0.06 - 2.67] for TKR). There was no difference between groups in the odds of esKOA or TKR at the earlier time point of 2 - 2.5 years in the emulated trial. Conclusion Given that our novel measure of esKOA could detect an association with weight loss at a time point 1.5 - 3 years earlier than TKR in an observational study, and in a sample size that was too small to detect an association with TKR at 4 - 5 years in an emulated trial, esKOA is recommended as an outcome measure for observational studies and trials investigating causes and possible treatments for KOA. Our powerful novel measure of esKOA enables shorter and smaller – hence cheaper – studies, which can boost the research on effective treatment for KOA. Acknowledgements We acknowledge the provision of datasets and/or research tools from two cohort studies: the Osteoarthritis Initiative (OAI) study and the Multicenter Osteoarthritis Study (MOST. Disclosure of Interests Zubeyir Salis: None declared, Jeffrey Driban Consultant of: Consultant for Pfizer Inc and Eli Lilly and Company., Timothy McAlindon Consultant of: Consultant for Remedium-Bio, Anika, Chemocentryx, Grunenthal, Kolon Tissue Gene, Novartis, BioSplice, Organogenesis, and Pfizer Inc., Amanda Sainsbury-Salis Speakers bureau: Received presentation fees and travel reimbursements from Eli Lilly and Co, the Pharmacy Guild of Australia, Novo Nordisk, the Dietitians Association of Australia, Shoalhaven Family Medical Centres, the Pharmaceutical Society of Australia, and Metagenics, and serving on the Nestlé Health Science Optifast VLCD advisory board from 2016 to 2018.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.