A clinical evaluation of the non-contact tonometer.

Paired non-contact tonometer readings were taken on patients at the Boston City Hospital Eye Clinic before and after routine vision care that included Goldmann tonometry. The variability of NCT readings was determined and comparisons made between readings obtained with the NCT and Goldmann instruments. The regression equation relating the pre-Goldmann NCT findings (N1), to the Goldmann (G), was N1 = 1.01G + 1.97; and the equation relating the post-Goldmann NCT findings (N2), to the Goldman was N2 = .94G + 1.70. The N1 mean was 2.06 mm. higher than the mean G, while the N2 mean was .58 mm. higher than the mean G. The correlation coefficients and standard deviation of the differences for the two comparisons were, .88, 3.91, and .93, 3.07, respectively. The Goldmann findings therefore were seen to agree more closely with the NCT findings taken after them. The linear regression and correlation coefficients between the NCT and the Goldmann were generally in good agreement with those of prior studies, although the standard deviations of the differences between findings were larger in this study. This result is not surprising since no attempt was made to train Goldmann operators to criterion or to limit variability induced by differing observer criteria or by the use of only one Goldmann operator. The operators of the Goldmann tonometer had varying degrees of training as residents in ophthalmology. Interestingly, the agreement between pre-Goldmann NCT and post-Goldmann NCT findings was not good. This suggested that the taking of the Goldmann findings themselves may have had a significant effect on the tonometric readings and that the considerable time that frequently took place between the two sets of NCT readings could have affected the value. The standard deviation of the differences between pre- and post-Goldmann readings on the same patient was 3.27 as compared to that for paired readings which was 2.28. The data also showed a much better agreement between the NCT and the rechecked Goldmann findings than between the initial Goldmann findings and the rechecked Goldmann findings, indicating that where discrepancies existed, it was more likely that they were due to the Goldmann instrument.