Indoor sound and occupant outcomes in mechanically ventilated offices: a threshold inference field study
Amit Kant Kaushik, Yousef AlHorr, Rubina Islam, Mohammed ArifPurpose
To test whether indoor sound shows a non-monotonic dose-response with occupant health and productivity in mechanically ventilated offices, using formal threshold inference.
Design/methodology/approach
A total of 365 sensor-survey observations from 42 occupants across 12 mechanically ventilated office zones in Doha, Qatar; confirmatory analyses pre-specified in an internal pre-analysis plan. Segmented logistic regression with occupant-clustered errors, Davies' supremum-Wald existence test, profile-likelihood intervals and Bayesian random-intercept sensitivity.
Findings
A composite of noise-pathway symptoms (headache, anxiety, low energy) inflects in the 44–52 dB(A) band (Davies' p = 0.038; point estimate 47.75 dB(A)), with a corroborating regime change in the acoustic productivity rating at 49 dB(A). It holds under IEQ adjustment, in 4 of 7 alternative composites and under Bayesian sensitivity; negative controls do not.
Research limitations/implications
Findings apply to hot-arid mechanically ventilated offices and are cross-sectional. Exposure is hour-averaged A-weighted level only, so the acoustic mechanism is not identified. An underpowered symptom-class interaction leaves the noise-pathway interpretation resting on component-level negative-control behaviour, and significance depends on clustering by occupant or zone.
Practical implications
The band offers preliminary evidence relevant to provisions bearing on over-quiet conditions in WELL v2, BREEAM Hea 05 and ISO 22955:2021, which rest on privacy inference, not symptom data. Acoustic operation is better expressed as a target band than a quieter-is-better rule.
Originality/value
First application of Davies' supremum-Wald threshold inference to multi-domain indoor environmental quality field data, giving a transferable pattern for dose-response research.