Repeatability and agreement in the measurement of horizontal fusional vergences.

PURPOSE This study was designed to determine the repeatability of fusional vergence ranges measured using the rotary prisms in the phoropter and in free space using the prism bar. The level of agreement between the two methods was also investigated. METHODS In two separate sessions, negative and positive fusional vergence ranges (NFV and PFV, respectively) were measured at distance and near in 61 young adults (mean age 19.74, S.D. 2.5 years) who were unfamiliar with the methods used. Base-in and base-out blur, break and recovery points were sequentially determined. Both sets of measurements were obtained by the same examiner. At each distance, NFV was determined first and then PFV. The repeatability of the tests and agreement between measurements made with the phoropter rotary prisms and the prism bar were estimated by the Bland and Altman method. RESULTS For both the phoropter rotary prisms and prism bar, NFV measurements showed better repeatability than PFV at both near and distance. Mean differences recorded for the NFV break and recovery points were non-significant (under 0.5Delta), while those observed for PFV were generally greater than 2Delta. When agreement between the two tests was assessed, it was found that break points were higher when determined using the phoropter rotary prisms, while recovery points were generally higher for the prism bar method. In clinical terms, according to the expected values of the NFV and PFV, agreement between the two techniques can be described as fair, because although mean differences were never greater than 5.5Delta, 95% agreement intervals were as wide as +/-8.00Delta for NFV and +/-13.19Delta for PFV. CONCLUSIONS The two methods used to measure fusional vergences showed fairly good inter-session repeatability for measuring NFV but repeatability was reduced for PFV measurements. The level of agreement observed between the two methods was such that their interchangeable use in clinical practice is not recommended.