Field Comparison Study: Agreement Between Soundtrace Audiograms and Prior Vendor Audiograms

Study Overview

This study examined how employees' audiograms performed using the Soundtrace system compared to their most recent historical audiograms completed by external occupational hearing testing providers. These prior tests were performed using a wide range of equipment, booths, mobile vans, and technician practices. The goal of the analysis was to understand the level of agreement between Soundtrace results and the thresholds recorded by previous providers under real-world conditions.

The dataset includes a large, diverse population: 5,961 employees analyzed, 4,988 valid matched comparisons (after applying a 5% outlier filter), 116 employers represented, more than 225 calibrated Soundtrace audiometers, and more than 280 individuals trained by Soundtrace using the protocol established by the Soundtrace Professional Supervisor.

These characteristics make the dataset larger and more operationally diverse than most published comparisons of audiometric systems.

Primary Results (5% Outlier Removal)

The 5% trimmed dataset is presented as the primary result because it reflects common practice in method-comparison studies, where a small number of extreme values are removed to represent typical performance while still preserving nearly all of the data.

Summary of PTA (Pure-Tone Average) Differences: Mean difference (Soundtrace – Prior Vendor) of –2.42 dB, standard deviation of 3.70–4.25 dB, approximately 76% of results within ±5 dB, and approximately 97% within ±10 dB. Left and right ears showed similar patterns.

Sources of Variability

Differences between Soundtrace audiograms and historical vendor audiograms can occur for several reasons. These include different equipment types and headphone models used across vendors, differences in testing environments (sound booths, vans, offices), variability in technician training and instructions, subject factors such as attention, fatigue, or temporary hearing fluctuation, and the time interval between the two tests — often one year or longer.

Because the dataset includes thousands of employees tested under varied real-world conditions, these factors introduce natural spread in the comparison results.

For transparency, results are shown at three outlier-handling levels: 0% (no removal), 5%, and 10%.

Comparison to Published Benchmarks

Large audiometric studies — including clinical booth retests, multi-clinic comparisons, and occupational surveillance datasets — typically report mean differences of 0 to –5 dB, SD values of 3–6 dB, ±5 dB agreement of 60–80%, and ±10 dB agreement of 90–98%.

The Soundtrace results fall fully within these expected ranges, despite much larger scale (nearly 5,000 matched tests), wide variation in prior vendor methods, over 280 different test administrators, and over 225 Soundtrace devices in simultaneous use.

This scale of data exceeds the size of most published audiology comparison studies and allows for reliable estimation of real-world variability.

Conclusion

This analysis shows that Soundtrace audiometric results are consistent with expected patterns of variability when compared to historical audiograms from external occupational hearing testing providers. The direction and size of the mean difference, the standard deviations, and the ±5 dB and ±10 dB agreement rates align with values reported in clinical and occupational audiology literature.

Given the large scale of the dataset — spanning thousands of employees, over 225 Soundtrace audiometers, more than 280 trained test administrators, and 116 employers — the results provide a stable estimate of real-world agreement.