Manual Versus Hammer Percussion Frequency-Domain Acoustic Screening for Hip Fracture: A Comparative Study

Introduction Delayed diagnosis of hip fracture (HF), one of the most common fractures encountered in clinical practice, is associated with serious complications or adverse outcomes. However, these fractures are frequently missed on radiography. Auscultatory percussion, a simple screening method that compares percussion sounds between lower limbs using a stethoscope, has limitations owing to its subjective evaluation, which can lead to unstable diagnostic accuracy. Therefore, this study aimed to verify whether objective frequency analysis of percussion sounds could overcome this issue to achieve high-accuracy diagnosis. We further compared the diagnostic performance of manual percussion versus percussion with a tendon hammer. Materials and methods This case-control study enrolled 40 patients with HFs (fracture group) and 20 hospitalized patients without fractures (control group). All patients underwent percussion of the pubic symphysis using manual percussion and a tendon hammer, and sounds were recorded from both patellae. The absolute value of the sound pressure difference between the sides was calculated across 512 frequency bands (0-24,000 Hz). Diagnostic performance was evaluated using group comparisons (Mann-Whitney U test), receiver operating characteristic (ROC) analysis with bootstrap internal validation, decision curve analysis (DCA), and multivariate logistic regression analysis. Results Manual percussion demonstrated the highest diagnostic performance at the 2718.75-Hz band, with an optimism-corrected area under the curve (AUC) of 0.923 (95% confidence interval: 0.845-0.979). At a cutoff value of 1.28 dB, the sensitivity and specificity were 97.5% and 69.8%, respectively, yielding an extremely low negative likelihood ratio of 0.04. Hammer percussion also showed good performance in the 421.88-Hz band (AUC: 0.861), although the difference in diagnostic performance between the two methods was not statistically significant. DCA confirmed the clinical utility of both methods. In multivariate analysis, the sound pressure difference for both methods was a significant independent predictor of fracture. Furthermore, exploratory subgroup analysis suggested the diagnostic performance of manual percussion was stable across key patient subgroups. Conclusion These results indicate that objective frequency-domain analysis of percussion sounds, particularly with manual percussion, represents a simple and effective screening tool for ruling out HFs, given its high sensitivity and excellent negative likelihood ratio. This study revealed new scientific insight that the diagnostically effective frequency band depends on the physical properties of percussion. These findings overcome the challenges of conventional subjective diagnostic methods and could contribute to the future development of accessible, objective, and non-invasive fracture diagnosis technologies.