Highlights
– In a single-blinded crossover RCT of 56 older adults, 77% expressed a preference for either fast or slow compression; preferences were split roughly equally between the two modes.
– The only consistent predictor of preference was the degree of hearing loss: participants with greater hearing loss preferred slow-acting compression; the boundary was a 4-frequency average (better ear) of about 35 dB HL.
– Compression speed produced no measurable differences in speech recognition in quiet or in noise, and neither cognitive scores nor prior hearing-aid experience predicted preference or objective benefit.
Background
Dynamic range compression (DRC) is a fundamental signal-processing feature in modern hearing aids. Compression reduces the gain applied to incoming sounds as level increases, with the aim of restoring audibility for soft sounds while preventing uncomfortable loudness for intense sounds. The time constants that define how quickly the compression responds to changes in level are classified as fast-acting (short attack/release times) or slow-acting (longer attack/release times). The choice of compression speed has been hypothesized to influence both subjective preference and objective performance: fast compression may maximize audibility for dynamic speech, whereas slow compression better preserves temporal and amplitude contrasts that can support speech intelligibility and sound quality.
Study design
Windle, Dillon, and Heinrich carried out a single-blinded, crossover randomized controlled trial to determine whether older adults fitted with hearing aids prefer slow or fast compression and whether preference or outcomes are influenced by cognitive ability, degree of hearing loss, or hearing-aid experience. Fifty-six participants aged 56–85 years with symmetrical mild-to-moderate sensorineural hearing loss were recruited from an NHS hearing assessment clinic; both new and experienced hearing-aid users were included.
Each participant trialed hearing aids for two months in each of two settings: slow-acting compression and fast-acting compression. Order was randomized. Outcome measures included aided speech recognition in quiet and in noise (measured after fitting with each compression setting), a battery of cognitive tests, patient-reported hearing-aid outcomes, and a direct question about preference for the first or second fitting at the end of the trial. The trial was single-blinded (participant-blinded) and used real-world acclimatization periods rather than brief laboratory exposures.
Key findings
Preference and prediction
– Seventy-seven percent of participants stated a preference for one of the compression speeds; the remaining 23% reported no preference.
– Preferences were approximately evenly divided between fast and slow compression. The only measured factor that predicted compression-speed preference was hearing threshold: participants with larger hearing losses were more likely to prefer slow-acting compression.
– The authors identified an approximate decision boundary at a four-frequency pure-tone average (PTA) of 35 dB HL in the better ear. Using a pragmatic rule — default to fast compression if PTA < 35 dB HL, and to slow compression if PTA ≥ 35 dB HL — would have matched a user’s stated preference in about 80% of cases.
Objective outcomes
– There was no significant effect of compression speed on objective speech recognition in quiet or in noise. Speech-understanding scores, when measured in controlled conditions after acclimatization to each setting, did not differ between fast and slow compression settings.
Other potential predictors
– Cognitive test scores did not predict compression-speed preference or objective benefit. Likewise, previous hearing-aid experience did not reliably predict which compression speed participants preferred or which produced better speech outcomes.
Additional observations and limitations
– The trial noted a strong association between preference and the order of fittings: participants tended to prefer the second fitting irrespective of whether it was fast or slow compression. This order effect suggests either an accommodation effect (participants becoming more comfortable with their device) or an expectancy/recency bias, and it complicates interpretation.
– There was also a positive association between hearing thresholds and prior hearing-aid experience (participants with larger losses tended to be more experienced users), which could introduce confounding when interpreting preference drivers.
Clinical interpretation and practical implications
This trial provides pragmatic, clinically actionable information. First, compression speed per se did not change measurable speech-understanding outcomes in controlled tests after a real-world acclimatization period. That aligns with prior clinical impressions that compression speed often affects perceived sound quality rather than gross speech intelligibility measures in idealized test conditions.
Second, participants’ subjective preference correlates with the degree of hearing loss, not cognitive status. A simple heuristic — fast default for milder losses (PTA < 35 dB HL) and slow default for greater losses (PTA ≥ 35 dB HL) — would match preferences in about 4/5 patients. This is a useful starting point for clinicians programming hearing aids, particularly in busy services where rapid, evidence-based default settings are valuable.
However, the strong order effect — preference for the second fitting — highlights the need for caution. Patient preference is malleable and can be influenced by recency, adaptation, or counseling. Therefore, clinicians should treat default settings as a starting point and plan a short follow-up with real-world listening checks and structured guidance to refine settings.
Mechanistic considerations
Why might greater hearing loss favor slow compression? One explanation is that more severe sensorineural loss often co-occurs with reduced spectral and temporal resolution; slow compression preserves amplitude contrasts and temporal envelopes that convey speech cues important for those with degraded peripheral encoding. Conversely, people with milder loss may value increased audibility of softer consonants that fast compression can restore, and they may tolerate the temporal smearing that can accompany rapid gain changes.
The absence of a cognitive interaction in this sample argues against the commonly held expectation that cognitive status strongly moderates compression-speed benefit in older adults. That said, cognitive effects may be subtle, task-dependent, and influenced by the particular cognitive tests used; larger studies with specific measures of working memory and processing speed in complex listening tasks may be needed to definitively exclude interactions.
Study strengths and limitations
Strengths include a randomized crossover design, ecologically valid two-month acclimatization periods for each setting, and inclusion of both new and experienced hearing-aid users typical of clinical practice. The single-blinded design minimized some biases, and the pragmatic approach enhances external validity for routine audiology services.
Limitations include a modest sample size (n=56), potential residual confounding between hearing threshold and hearing-aid experience, and the robust order effect that suggests incomplete blinding or adaptation biases. The trial used specific devices and compression implementations; results might differ with other algorithms or multichannel designs. Finally, objective outcomes were limited to standard speech-recognition tests; measures of listening effort, real-world communication ability, or ecological momentary assessments could yield additional insights.
Recommendations for clinicians
– Consider using a pragmatic default rule: fast compression for better-ear PTA < 35 dB HL and slow compression for PTA ≥ 35 dB HL. This is likely to match patient preference in roughly 80% of cases.
– Treat defaults as starting points: schedule follow-up to assess subjective satisfaction, sound quality, and real-world function. Use structured A/B comparisons after sufficient acclimatization if patients report problems.
– Be mindful of potential order/recency effects: when trialing settings, counterbalance exposure or allow washout periods and obtain preference data after equivalent real-world use.
Research gaps and future directions
Further work should examine the stability of preference over longer periods, test-device generalizability across manufacturers and multichannel compression schemes, and incorporate objective measures of listening effort (e.g., pupillometry, dual-task paradigms) and ecological outcome measures (daily diaries, smartphone-based sampling). Larger samples stratified by cognitive profile and degree of hearing loss can clarify subtle interactions that this study could not detect.
Funding and trial registration
Funding and clinicaltrials.gov or other registry details were not included in the summarized material above. Readers should consult the original publication for full disclosures and trial registration information: Windle R, Dillon H, Heinrich A. Ear Hear. 2025; DOI: 10.1097/AUD.0000000000001716. PMID: 40936168.
References
1. Windle R, Dillon H, Heinrich A. Preference and Outcomes for Fast Versus Slow Compression in Hearing Aids for Older Adults: A Randomized Control Trial. Ear Hear. 2025 Sep 12. doi: 10.1097/AUD.0000000000001716. Epub ahead of print. PMID: 40936168.
2. National Institute for Health and Care Excellence (NICE). Hearing loss in adults: assessment and management. NICE Guideline [NG98]. 2018. Available from: https://www.nice.org.uk/guidance/ng98
3. Dillon H. Hearing Aids. 2nd ed. Thieme; 2012.
Summary
The Windle et al. randomized crossover trial offers clinically useful, evidence-based guidance: compression-speed preference in older adults is predicted primarily by hearing-loss severity, not cognition or past hearing-aid use, and compression speed does not systematically change speech-recognition scores in quiet or noise after acclimatization. A simple PTA-based default (35 dB HL boundary) will align with most patients’ preferences, but clinicians should confirm preferences after real-world use and remain alert to order and adaptation effects.
