Hertel exophthalmometry: reliability and interobserver variation

To quantify and analyze the interobserver variation in Hertel exophthalmomety and to evaluate the impact of experience on it.

In a population-based epidemiologic study, Hertel exophthalmometry was performed on 1063 randomly selected normal subjects (stratified sampling) by an oculoplastic attending surgeon (A) and a third-year ophthalmology resident (R). Both observers were masked to the reading of the other. The Chi-square test, two-tailed paired sample t-test, and two-tailed paired independent t-test were used to analyze the data (SPSS Release 9.0, Chicago).

There were 463 (43.5%) females and 600 (56.5%) males. The age ranged from 6 to 70 years (mean +/- SD = 20.3 +/- 10.9). The mean right eye protrusion was 14.8 mm for both 'A' and 'R'. The mean left eye protrusion was 14.6 mm and 14.7 mm, respectively. There was good correlation (r = 0.80) and no statistically significant difference (0.1 < P < 0.6) for the ocular protrusion values (right, left, and average) measured by 'A' and 'R'. However, the percentage agreement (within +/-1 mm limits of acceptance) was about 60%. Comparison of the first 530 subjects versus the remaining cases showed no improvement in agreement with the senior observer.

Hertel exophthalmometry is a reliable method for measuring the ocular protrusion. However, as with most clinical measures, there is a negligible interobserver variation that seems to be unavoidable.