Carman et al. asked how large a Cobb angle change must be before a clinician can confidently call it true progression rather than measurement noise. Five observers measured scoliosis and kyphosis radiographs twice each, and tolerance limit statistics were used to define reliable progression thresholds.
Every time you review a scoliosis follow-up film, the number on the report means less than you think. Before this paper, a 5-degree Cobb increase was widely cited as evidence of progression — used to justify bracing, escalate treatment, or trigger surgical planning.
Carman showed that threshold is statistically indefensible: it has a 30% chance of being noise. Use 10 degrees as your minimum bar for calling true scoliosis progression, and 11 degrees for kyphosis.
When the same observer re-measures the same film on two occasions, that single reader's inconsistency accounts for nearly all the error. Not the difference between your read and a colleague's. Getting a second opinion does not solve the problem; getting a better measurement technique does.
This paper is the statistical foundation behind why natural history studies, bracing trials, and SRS outcome standards define progression at 5-6 degrees for screening triggers but require larger thresholds for research endpoints and treatment decisions.
Carman et al. asked how large a Cobb angle change must be before a clinician can confidently call it true progression rather than measurement noise. Five observers measured scoliosis and kyphosis radiographs twice each, and tolerance limit statistics were used to define reliable progression thresholds.
Every time you review a scoliosis follow-up film, the number on the report means less than you think. Before this paper, a 5-degree Cobb increase was widely cited as evidence of progression — used to justify bracing, escalate treatment, or trigger surgical planning.
Carman showed that threshold is statistically indefensible: it has a 30% chance of being noise. Use 10 degrees as your minimum bar for calling true scoliosis progression, and 11 degrees for kyphosis.
When the same observer re-measures the same film on two occasions, that single reader's inconsistency accounts for nearly all the error. Not the difference between your read and a colleague's. Getting a second opinion does not solve the problem; getting a better measurement technique does.
This paper is the statistical foundation behind why natural history studies, bracing trials, and SRS outcome standards define progression at 5-6 degrees for screening triggers but require larger thresholds for research endpoints and treatment decisions.