jhsmr_17610339541761033954Somyot Chirasatitsinsomjot.c@psu.ac.thFaculty of Medicine, Prince of Songkla UniversityJournal of Health Science and Medical ResearchJ Health Sci Med Res2630-05592586-998101012024ChonlawatChaichanSirindaSritipsukhoSasinuchRutjanawechPaskornSritipsukhoThammanardCharernboonObjective: Long coronavirus disease (long COVID) represents a significant burden on healthcare systems and requires enhanced management strategies. There is a critical need for more comprehensive care and targeted healthcare services for affected populations. This study aimed to develop a clinical prediction scoring system for long COVID in patients recovering from COVID-19. Material and Methods: This prospective cohort study collected data at Thammasat University Hospital and the Thammasat Field Hospital during the Delta- and Omicron-variant-dominant epidemics. Phone interviews regarding long COVID symptoms were conducted with 2516 patients at 3 months post-infection. A stepwise logistic regression model was employed to develop the final predictive model for long COVID. Results: In total, 40.46% of patients exhibited long COVID symptoms 3 months after infection. Our model comprised 5 predictors: dyspnea, healthcare worker status, female gender, severity of acute illness, and variant dominant wave. With a sensitivity of 57.1% and a specificity of 67.3% at 3 months, the risk score exhibited an area under the receiver operating characteristic curve of 0.62 for long COVID prediction. The probability of long COVID for each risk score point was also reported. The Hosmer–Lemeshow test (p-value=0.49) indicated good model calibration, with closely aligned observed and expected frequencies. Conclusion: The predictive risk score demonstrated satisfactory accuracy in identifying COVID-19 patients at high risk of developing long COVID 3 months post-infection. 1017202520251262http://creativecommons.org/licenses/by-nc-nd/4.010.31584/jhsmr.20251262https://www.jhsmr.org/index.php/jhsmr/article/view/1262https://www.jhsmr.org/index.php/jhsmr/article/viewFile/1262/1458https://www.jhsmr.org/index.php/jhsmr/article/viewFile/1262/1458