Optimizing the Yield of Abnormal Preoperative Chest Radiographs in Elective Non-cardiothoracic Surgery: Development of a Risk Prediction Score and External Validation
Issued Date
2023-01-01
Resource Type
ISSN
03642313
eISSN
14322323
Scopus ID
2-s2.0-85170072715
Journal Title
World Journal of Surgery
Rights Holder(s)
SCOPUS
Bibliographic Citation
World Journal of Surgery (2023)
Suggested Citation
Seangleulur A., Thakkinstian A., Supaopaspan W., Kwankua A., Sukkasem W., Kunawudhi A., Soonthornkes N., Limpavitayaporn P., Sirisreetreerux P., Saiphoklang N., Attia J., McKay G., Phongkitkarun S., Okascharoen C. Optimizing the Yield of Abnormal Preoperative Chest Radiographs in Elective Non-cardiothoracic Surgery: Development of a Risk Prediction Score and External Validation. World Journal of Surgery (2023). doi:10.1007/s00268-023-07146-7 Retrieved from: https://repository.li.mahidol.ac.th/handle/20.500.14594/90027
Title
Optimizing the Yield of Abnormal Preoperative Chest Radiographs in Elective Non-cardiothoracic Surgery: Development of a Risk Prediction Score and External Validation
Other Contributor(s)
Abstract
Background: Guideline recommendations for preoperative chest radiographs vary to the extent that individual patient benefit is unclear. We developed and validated a prediction score for abnormal preoperative chest radiographs in adult patients undergoing elective non-cardiothoracic surgery. Methods: Our prospective observational study recruited 703 adult patients who underwent elective non-cardiothoracic surgery at Ramathibodi Hospital. We developed a risk prediction score for abnormal preoperative chest radiographs with external validation using data from 411 patients recruited from Thammasat University Hospital. The discriminative performance was assessed by receiver operating curve analysis. In addition, we assessed the contribution of abnormal chest radiographs to perioperative management. Results: Abnormal preoperative chest radiographs were found in 19.5% of the 703 patients. Age, pulmonary disease, cardiac disease, and diabetes were significant factors. The model showed good performance with a C-statistics of 0.739 (95% CI, 0.691–0.786). We classified patients into four groups based on risk scores. The posttest probabilities in the intermediate-, intermediate-high-, and high-risk groups were 33.2%, 59.8%, and 75.7%, respectively. The model fitted well with the external validation data with a C statistic of 0.731 (95% CI, 0.674–0.789). One (0.4%) abnormal chest radiograph from the low-risk group and three (2.4%) abnormal chest radiographs from the intermediate-to-high-risk group had a major impact on perioperative management. Conclusions: Four predictors including age, pulmonary disease, cardiac disease, and diabetes were associated with abnormal preoperative chest radiographs. Our risk score demonstrated good performance and may help identify patients at higher risk of chest abnormalities.