Choosing hearing protectors: why the NRR is only a start

This page explains what the rules and common practice call for. Every site is different, and what a survey or a filing needs at your facility is something we work out with you after we see it.

The Noise Reduction Rating (NRR) printed on the package is a laboratory figure, measured in a test booth on trained subjects with someone supervising the fit. In real work, protectors usually provide less attenuation (noise reduction) than that.

Both OSHA and NIOSH require you to derate the NRR, meaning mark it down, before you use it to estimate what a worker is actually exposed to.

Fit, not rating, is usually the limiting factor. A plug with an NRR of 33 that is seated poorly can provide less attenuation than a plug with an NRR of 25 that is seated properly.

What the standard actually requires

The 85 dB target is set worker by worker. Once a worker has an STS, the target for that person changes, so protector selection is not a single decision for the whole plant. Some workers have a stricter target.

"Variety" is not met by two sizes of the same plug. OSHA's noise standard requires a real choice because comfort drives use. A worker who finds a protector uncomfortable takes it off, and removing it for even a small part of the shift sharply reduces the protection the worker gets. The unprotected minutes dominate the average.

→ Standard threshold shift (STS): when an audiogram shows a drop in hearing

The problem with a single number

A protector attenuates different amounts of noise at different frequencies, and the NRR compresses all of that into one laboratory figure. Two separate issues follow.

Issue one: a laboratory fit is not a shop-floor fit. Subjects fitted under supervision in a quiet booth get protection that workers on a production floor generally do not. Both OSHA and NIOSH adjust for this, and their adjustments differ.

Issue two: one number cannot describe noise that is spread across many frequencies. This is the larger of the two issues, and no derating corrects it. We cover it under "Why the frequency of the noise matters" below.

Derating the NRR, and the calculation Appendix B requires

Appendix B of OSHA's noise standard is mandatory. For A-weighted readings, the weighting built into a normal decibel measurement (the A in dBA), Appendix B sets this calculation:

Estimated A-weighted exposure under the protector = the worker's TWA (in dBA) − (NRR − 7)

The 7 decibels are subtracted because the NRR is derived on a different scale, called C-weighted, while exposure is measured on the A-weighted scale. The subtraction accounts for that difference and is part of the method, not a safety factor.

The safety factors are applied on top of it, and the two agencies apply them differently:

Two details are often reversed. First, OSHA's derating does not depend on the type of protector. It is half the attenuation, whatever the protector is. Second, under the NIOSH scheme, formable foam plugs take the smallest derating of any earplug, not the largest. Premolded and custom plugs take the 70% reduction.

A worked example: a foam plug with an NRR of 33, on a worker whose TWA is 95 decibels (dBA).

Each row, step by step:

• Appendix B alone. Subtract 7 from the NRR of 33, then subtract the remainder from 95. The result is 69.
• OSHA. Subtract 7 from the 33 and halve the remainder, then subtract that from 95. The result is 82.
• NIOSH. Take 50% of the 33, which is the derating for foam plugs, then subtract 7. Subtract the remainder from 95 and the result is 85.5.

The same protector on the same worker produces three answers that span more than 16 decibels. The most conservative answer sits at the 85 dB target for a worker with an STS, so the method you use determines whether you can show that such a worker is protected.

Plugs and muffs worn together do not add their ratings. For plugs worn under muffs, take the higher of the two NRRs and add 5 decibels, then run the same calculation on that figure. A plug with an NRR of 33 under a muff with an NRR of 25 is treated as an NRR of 38, not 58.

Why the frequency of the noise matters

Hearing protectors do not attenuate all frequencies evenly. Most attenuate high-frequency sound well and low-frequency sound considerably less, which follows from the mass of the protector and the way it seals.

The result is that the NRR consistently overstates protection where the noise is low-frequency. A protector with an NRR of 30 against the standard laboratory noise may provide far less attenuation where most of the sound energy is below 500 hertz (Hz).

This shows up around:

• Large fans, blowers and air handling equipment
• Compressors and vacuum pumps
• Diesel engines and heavy mobile equipment
• Rotating machinery, mills and crushers
• Furnaces and combustion equipment

If your noise is mostly low-frequency and you selected protectors on the NRR alone, the actual protection is probably lower than the paperwork suggests.

→ Octave band analysis: measuring the frequency content of noise

The NIOSH long method

The long method does not use a single number for the noise or for the protector. It uses readings by frequency band for both, known as octave band data.

The steps, in outline:

1. Measure the noise in octave bands, usually 125 Hz through 8000 Hz, at the worker's location.
2. Take two figures for each band from the protector's published test data: the mean attenuation (the average amount it blocks) and the standard deviation (how much that varies).
3. Subtract a conservative figure for each band. That figure is the mean attenuation minus a multiple of the standard deviation, so the estimate covers a protective share of wearers and not only the average one.
4. Apply the A-weighting correction to each band.
5. Combine the band levels with a logarithmic sum, which is how decibels are added. The total is the estimated A-weighted level under the protector.

This is more work than subtracting a single number, and it is more accurate. It checks this protector against this noise, band by band, not against a standard laboratory noise that may look nothing like your plant.

Two protectors with the same NRR can differ by several decibels in actual protection where the noise is low-frequency. The long method is how you see that before you settle on one protector for the whole site.

The trade-off is the equipment. The long method needs octave band measurements of the noise, which take different equipment and a longer site visit than personal noise dosimetry does. For general manufacturing with broadband noise (noise spread across many frequencies), the single number with a proper derating is good enough.

The extra time is worth spending in three cases:

• Your noise is mostly low-frequency.
• You sit close to a limit.
• You have workers on the 85 dB target.

→ Octave band analysis · → Noise surveys

What actually improves protection

In rough order of effect:

Fit testing. A fit-test system measures the attenuation one worker gets from one protector, and a poor fit can be corrected on the spot. OSHA's noise standard does not require it, but it deals directly with fit, which is the limiting factor.

Choice. The standard requires it, and it works. Ear canals vary a great deal, so a plug that seals one ear may seal the next one poorly.

Training that includes the roll, pull and hold. Roll the foam plug to a thin crease, pull the outer ear up and back, then insert the plug and hold it until it expands. Skip a step and the plug sits in the outer ear canal, where it does very little. It is a two-minute demonstration with a large effect.

Not over-protecting. A worker who cannot hear speech, alarms or their own machine will find a way around the protector. Reducing a 92 dBA workplace to 60 dBA under the muff creates a different hazard and works against consistent use.

When you may not need this

If your survey shows no worker at or above a TWA of 85 dBA, protectors are voluntary and none of this applies.

If you are modestly above that and your noise is broadband, a competent safety manager can handle this without a consultant, using derated NRRs and a good fit-training routine.

Bring someone in for these cases:

• Your noise has an unusual frequency content.
• Hearing protection, not engineering controls, is your means of compliance.
• You have workers with an STS and need to show the 85 dB target is met.

Our part is measuring the noise, including the octave band measurements the long method needs. Fit testing, training and running a hearing conservation program are not our work, and they are outside the four fields we work in.

→ Hearing conservation programs · → Personal noise dosimetry · → How many employees to sample for noise