Noise Health Home 

[Download PDF]
Year : 1999  |  Volume : 1  |  Issue : 2  |  Page : 73--77

On quality assurance in occupational hearing conservation programs: An evaluation based on ANSI S12.13-1991

Stig Arlinger, Ulla Ivarsson 
 Division of Technical Audiology, Department of Neuroscience and Locomotion, Linköping University Hospital, S-581 85 Linköping, Sweden

Correspondence Address:
Stig Arlinger
Division of Technical Audiology, Department of Neuroscience and Locomotion, Linköping University Hospital, S-581 85 Linköping

How to cite this article:
Arlinger S, Ivarsson U. On quality assurance in occupational hearing conservation programs: An evaluation based on ANSI S12.13-1991.Noise Health 1999;1:73-77

How to cite this URL:
Arlinger S, Ivarsson U. On quality assurance in occupational hearing conservation programs: An evaluation based on ANSI S12.13-1991. Noise Health [serial online] 1999 [cited 2023 Mar 21 ];1:73-77
Available from:

Full Text


The purpose of an occupational hearing conservation program is to preserve hearing, i.e. to prevent hearing loss caused by occupational noise exposure to occur. Like many other activities, occupational hearing conservation programs need systematic quality assurance. The concept of quality in this context refers to how successful the program is in preventing noise­induced hearing loss. In the US, American National Standards Institute published in 1991 a draft standard "Evaluating the effectiveness of hearing conservation programs" for trial use, comment, and criticism. This document proposes comparisons of successive audiograms from repeated hearing tests as the basis for such evaluation.

An ideal variable for assessment of the quality of a hearing conservation program should be sensitive to both random and systematic changes in hearing threshold levels. Random changes are primarily related to random sources of errors in the pure tone audiometric test procedure as well as to a varying degree of temporary threshold shifts. Systematic changes may be due to true permanent changes in hearing sensitivity, to errors in the calibration of the audiometer being used or to systematic changes in the test procedure. The ANSI document proposes several alternative variables which could be used for such a systematic evaluation:

One is the percent worse sequential, %W, i.e. the incidence of hearing threshold levels showing a shift of 15 dB or more at any test frequency in the range 500 - 6000 Hz in either ear between two sequential audiograms. This variable has the drawback of being sensitive to changes in only one direction.

Another is the percent better or worse, %BW, the percent of the population which shows a shift of 15 dB or more toward either better or worse threshold level at any test frequency and either ear. This is the most attractive variable since it is sensitive to both random and systematic changes in both directions.

The third variable is the standard deviation of difference in hearing threshold levels, averaged between the ears, either at separate frequencies or as average over 3 or 4 frequencies. This is probably the least useful variable since it is not sensitive to systematic changes unless it is combined with corresponding mean values.

The ANSI document suggests as input data results from annually repeated audiometry on a constant, restricted population. A certain minimum number of audiograms should be available for each individual and the population should have a minimum size, proposed to be 30 people. We wanted to test the usefulness of these variables by applying them on a large database in order to assess some special questions that might affect their usefulness.


In our hearing conservation programme we have a database (Arlinger & Ivarsson, 1985) presently covering the results from approximately 150.000 audiometric tests performed on about 75.000 people since the early nineteen seventies. In addition to personal and employment data, each file contains data on noise exposure and use of hearing protectors.

For each of the variables mentioned, the ANSI document proposes criterion ranges for rating hearing conservation program effectiveness in terms of values which represent acceptable, marginal and unacceptable effectiveness, respectively. There were three specific questions which we wanted to look into with regard to whether the measures of variability proposed are affected by them:

- The age distribution of the subjects tested might affect the results. Older people show larger variation and greater spontaneous change over time in hearing thresholds than do younger people.

- The interval between tests might likewise affect

- the ANSI standard prescribed annual tests but often tests are repeated with longer intervals.

- The group size - the ANSI standard proposes at least 30 people which is still a relatively small group from a statistical point of view.


In our database we had more than 22.000 individuals who were represented by two or more audiograms taken with intervals in the range between one and five years. This subset was used for the analysis. These individuals could be separated into subgroups depending on whether they had declared that they were exposed to noise or not in their work-place and whether they were using hearing protectors or not.


We found a considerable effect of interval on all variables tested. [Figure 1] shows the variable %Worse between first and second audiograms as a function of interval from 1 to 5 years on subjects who are not exposed to occupational noise. The horizontal lines represent the ANSI criteria for acceptable and unacceptable values, referred to one-year intervals. [Figure 2] show the corresponding outcome for the variable % Better or Worse and [Figure 3] for the standard deviation for average hearing threshold levels at 3, 4 and 6 kHz. The overall conclusion is that interval between successive audiograms is a parameter which does affect the variables.

We looked into the effect of group size by starting with the subgroup consisting of about 7.600 people from whom we had at least two audiograms obtained with 3 years interval. Step by step we created smaller subgroups down to around 60 people by random selection based on birth dates. [Figure 4] shows what could be expected, i.e. a clear effect of group size: As the group size becomes smaller there is an increasing difference between maximum and minimum values for the variable calculated, in the example shown here the % Worse variable.

Finally we looked at the effect of age by splitting the same, those with at least two audiograms with 3 year interval, into three age groups: subjects aged less than 35, between 35 and 50, and more than 50 years. For the whole group as well as the two subgroups non-exposed and exposed but using hearing protectors we found a clear age effect on all variables. [Figure 5] shows the outcome on the variable % Better or Worse.


The outcome of our study indicates serious limitations in the quality measures proposed by the ANSI draft standard. We found significant effects on the variables proposed from duration of the interval between successive hearing tests as well as by the age of the subjects being tested. In conclusion, the proposed strategy for assessing the effectiveness of hearing conservation programs does not seem to be neither easily applicable nor sufficiently reliable. The best way still seems to be access to be a comparison with a reliable national database B, as suggested in ISO 1999. This means a reference population which has been selected by epidemiologically reliable methods as representative for the population in the country under consideration with regard to everything which may affect hearing except that no occupational noise exposure should have occurred. Ideally, such a reference data base should provide data not only on prevalence of hearing loss versus age and gender, but also on incidence.


This study was supported by the Swedish Work Environment Fund.[3]


1ANSI S12.13-1991. Evaluating the effectiveness of hearing conservation programs. Draft American Standard, Acoustical Society of America, New York.
2Arlinger S & Ivarsson U 1985. A data bank on hearing and use of hearing protection among 45.000 workers in small industries. Proc Inter-Noise 85, Wirtschaftsverlag NW, Bremerhaven, 1375-8.
3ISO 1999, 1990. Acoustics . Determination of occupational noise exposure and estimation of noise­induced hearing impairment. International Organization for Standardisation, Geneva.