Ab1378 Divergence of Case Distribution of Major Osteoporotic Fracture and Separation With the Cut-Off Index of Its Risk in Patients With Rheumatoid Arthritis

Background: Rheumatoid arthritis (RA) is the only disease listed as an independent risk for osteoporotic fracture. However, the risk weight varies depending on accompanying conditions. Objectives: We tried to evaluate risk weight in patients with RA using two ways: a risk factor extraction technique based on the odds ratio and sensitivity-added specificity, and to compare the case distributions. Additionally, we also tried to assess the effectiveness of anti-osteoporotic drugs for the conditions. Methods: RA patient's dataset that was used previously for evaluating the risk factors for incident major osteoporotic fracture (MOF) [1] was used in the study. Risk factors in clinical background and during follow-up were picked up and evaluated the risk weight according to the risk weight exponential assessed with a Cox regression analysis. Factors that were significantly higher or lower were extracted. Statistical significance was determined with a p-value of 0.05. Cut-off indices (COI) were determined for each extracted factor using a receiver operating characteristic (ROC) analysis. A chi-square test for the incidence of MOF was examined with separation from the COI. The first highest risk factor (HRF) was determined in two ways: the highest odds ratio (OR) or the most value of sensitivity-added specificity (S&S), and it was picked up. After dividing with the COI of the first HRF, the second procedures used for choosing the second HRF were performed in each way. These were repeated until no significant factors existed in each divided patient group. After the final separation, a chi-square test for the incidence of MOF according to the risk positivity in each way was examined, and the odds ratios and sensitivity-added specificities were compared between the two ways. Results: A total of 278 patients were studied. Of those, the sex distribution was 39 males and 239 females. The mean age, disease duration of RA, and observation length were 74.1 years old, 7.6 years, and 51.4 months. These prevalent MOFs counted 140 at baseline, and the incident MOF developed in 44. The first HRF was the presence of prevalent MOF in the way of S&S, whereas gender classification was determined with the OR way. The second HRF in the way with S&S was gender classification for the first HRF negative group and Vxy of the Joint Index Vector (JIV), which is the novel disease activity scoring system developed by Nishiyama [2], for the positive group, whereas the presence of prevalent MOF for the first HRF negative group and Vxy of the JIV for the positive group in the OR way. Picked-up HRFs commonly in both ways were gender classification, the presence of MOF, Vxy, the presence of lifestyle-related diseases, pain score using a visual analog scale, glucocorticoids (GCs) administration during follow-up, and total dose of GCs. Final case distribution in the S&S way was 134/3 for the MOF negative/positive in the risk-negative group and 100/41 in the risk-positive group. Whereas the OR way was 198/16 and 36/28 in the risk-negative and the risk-positive group. The odds ratios were 13.31 and 9.63 in the S&S and OR way, respectively. On the other hand, sensitivity-added specificity was 126.9% and 136.3% in the S&S way and the OR way, respectively. Conclusion: The expected value would alter as the approach way changes. However, the strong risk factors are identical. REFERENCES: [1] Yoshii I, et al. Ann Rheum Dis. 2033;81(2):296-9. DOI: 10.1136/annrheumdis-2021-221093. [2] Nishiyama S, et al. J Big Data. 2018;5:1-15. Acknowledgements: NIL. Disclosure of Interests: None declared.