This study aimed to assess the accuracy of smartphone-aided diagnosis of scoliosis by a trained nurse compared with scoliometer-based diagnosis by a spine surgeon.
Summary of Background Data.
Many assessments have been developed to estimate the reliability of smartphone-aided measurements in diagnosing scoliosis. However, clinical studies assessing the accuracy of smartphone-aided diagnosis with radiographs or scoliometers are scarce.
A total of 2702 grade 7 students (mean age 13.56 yrs, range 13–15) at 10 middle schools were first screened with a smartphone by a trained nurse from the orthopedics department. Approximately half a year later, most of the students underwent a chest x-ray examination as part of a compulsory medical examination. Students with suspicious findings in either the first screen or the chest x-ray were recommended to a scoliosis clinic for single-blind tests, such as a forward bending test (FBT) and an analysis of the angle of trunk rotation (ATR) with a scoliometer, performed by an experienced spine surgeon. Finally, the Cobb method was conducted with full-spine radiographs to serve as the gold standard.
The agreement between the first screening by the nurse and the second test by the spine surgeon was low in cases with a Cobb angle <10° (κ = 0.128 [0.04–0.22], P = 0.035) and fair in cases with a Cobb angle >10° (κ = 0.349 [0.19–0.50], P < 0.001). The results of receiver operating characteristic (ROC) curve analysis also suggested that these two tests were similar in their ability to diagnose scoliosis. However, when the Cobb angle cutoff was adjusted to 15°, the latter had markedly better diagnostic ability than the former. Overall, the sensitivity of the smartphone screening was not acceptable for recognizing scoliosis.
This study revealed that smartphone-aided screening for scoliosis is risky.
Level of Evidence: 3