A comparison of computerized audiometry by Ansi, Bekesy fixed-frequency, and modified ISO procedures in an industrial hearing conservation program.

Abstract : Three audiometric procedures were programmed at .5, 1, 2, 3, 4, and 6 kHz and presented on test and retest to 50 persons: (1) a version of the Amer. Nat'l. Standards Inst. (ANSI) guidelines (the '10-dB down - up 5 dB' rule) incorporating time-window response validation, (2) a slight modification of the procedure recommended by the Internat'l. Standards Organization (ISO), and (3) the NSMRL Mark I Bekesy-type system (a) incorporating numerous quality controls, (b) providing not only an estimate of Hearing Threshold Level (HTL) to the nearest dB but providing also an estimate of the confidence to be placed in each HTL, and (c) continuously adapting its parameters to interact with the behavior of the testee. Test-retest reliability was high (r = .83-.98) except that the ANSI procedure was less so at .5, 1, and 2 kHz (r = .72-.78). Test and retest group-mean HTLs never changed by as much as 2 dB. Individual test-retest differences of 5 dB or less were found in 90.7% of cases for the Mark I, compared with 65.5% reported by Cluff (1980) using a standard Grason-Stadler 1703 Bekesy audiometer. Estimated time to complete a two-ear audiogram on the average S was 10.5 + or - 1.07 min for the ANSI, 9.25 + or - 0.63 min for the ISO, and 3.35 + or - 0.35 min for the Mark I; average savings for individual Ss with the Mark I vs the ANSI procedure was 7.27 min (range 4.3-20.2 min). It was shown that HTLs furnished by a computerized adaptive Bekesy system can profitably be reported to the nearest 2.5 dB.