Referral criteria for school scoliosis screening: assessment and recommendations based on a large longitudinally followed cohort.

Study Design: This study was a retrospective cohort study.
Objective: To examine the criteria recommended in the literature for the school-based scoliosis screening program in Hong Kong.
Summary of Background Data: School-based screening for scoliosis has been a controversy. Objectors to the policy were concerned about the high over-referral and false-positive rates. Recommendations were then made for improvement, but the feasibility of these recommendations has not been studied.
Methods: The cohort consisted of students in Grade 5 in 1995/1996 or 1996/1997 who underwent scoliosis screening in Hong Kong. Participants who had an angle of trunk rotation (ATR) ≥15°, 2 or more moiré lines, or presented significant clinical signs were referred for radiography. Screening histories and radiography records before the age of 19 years were extracted. The accuracy measures for different combinations of screening tests were examined.
Results: There were 115,178 students in the cohort, of which 3228 (2.8%) were referred for radiography. Among the 1406 students who displayed a curve ≥20° during screening, 257 (18.3%) were boys and 336 (23.9%) were identified as 16 years or older, ruling out the suggestion of screening only 10-year-old girls. The sensitivity and positive predictive value for the current referral criteria were 88.1% and 43.6%, respectively. The sensitivity would drop substantially if the use of moiré topography (39.8%) or clinical signs (55.5%) were discarded. With the inclusion of these 2 tests, the clinical effectiveness measures were robust to the cutoff for ATR, unless it was set below 10°.
Conclusion: Selectively screening only premenarche girls was not feasible, as this screen would have missed a significant proportion of children with significant curvature. No refinement of the current protocol was necessary, although boys could be screened beginning at 12 years of age. The tandem use of ATR, moiré topography, and clinical signs was recommended for future studies.