I read with interest the recent report by Uchiyama and associates of their magnetic resonance imaging study of superior oblique (SO) muscle cross-section in patients with unilateral SO palsy. These authors compared coronal plane magnetic resonance imaging (MRI) in 50 cases, and in 45 control subjects. The authors calculated the ratio of SO cross-section between the paretic and normal sides. The figure provided by these authors shows a very impressive scatter of data for SO palsy, yet very small 95% confidence intervals are superimposed on the data. For SO palsy in total, the 95% confidence intervals for the ratio are stated as ranging from 0.66 to 0.80. Nevertheless, all but 3 of their 49 unilateral cases lie outside of the 95% confidence interval. Of course, such an occurrence is a statistical impossibility, because the idea of a 95% confidence interval is to encompass the great bulk of all observations. Readers will probably be confused by the data as presented. It appears that the authors are actually plotting 95% confidence intervals for the sample mean of the ratio for a group of patients, rather than 95% confidence interval for individual patient observations. That latter value, which is equal to the sample mean ±1.96 standard deviations, gives a confidence interval ranging from 0.01 through 1.31. Because clinicians make clinical decisions based on individual patient observations, such a wide confidence interval provides little insight in the absence of a wider perspective on current controversies concerning SO function. Uchiyama and associates’ stated confidence intervals should not be used for diagnostic purposes.
The title of Uchiyama and associates’ article reflects the authors’ recognition that the ratio of paretic SO to contralateral SO cross-section varies widely. Their article might have made the further inference that the clinical criteria leading to the diagnosis of SO palsy could be associated with almost any MRI morphologic features of the SO ranging from complete absence through apparent significant hypertrophy. The authors’ finding supports the contention that the clinical criteria for SO palsy generally are nonspecific and seemingly argue that more objective measures, such as MRI, would be more useful in elucidating the mechanisms of cyclovertical strabismus in our patients. MRI has proven increasingly useful in demonstrating the physiologic function and pathologic changes in the SO. For example, a recent experimental study in monkeys demonstrated that intracranial trochlear neurectomy results in rapid, significant atrophy and elongation of the SO. It therefore seems that SO denervation reliably causes muscle atrophy, but not all cyclovertical strabismus with the 3-step hypertropia pattern traditionally associated with SO palsy actually is associated with SO atrophy. Presumably, masquerading conditions exist that may be identified by imaging studies and may be treated differently than SO palsy because of trochlear neuropathy. Although a valuable contribution, Uchiyama and associates’ study does not yet define the distinction between true and simulated SO palsy.