Questionnaires record what a person recalls and understands about their COPD at the moment they complete them. PROMs have been central to COPD care for decades. The CAT (COPD Assessment Test), mMRC dyspnoea scale, and St George's Respiratory Questionnaire are validated, widely used, and genuinely useful measures of subjective experience. By design, however, each records how the patient felt then, interpreted through that patient's own frame of reference.
That interpretation leaves predictable gaps. Knowing where patient-reported data is dependable, where it fails systematically, and what physiological measurements contribute is not an attack on PROMs. It is necessary for using any monitoring method appropriately.
Where PROMs Are Strong
PROMs provide important information about several areas of COPD management. Only the patient can directly describe functional limitation, exercise tolerance, sleep quality, and psychological burden. No spirometric or acoustic measure is equivalent to "I stopped going to the corner shop because I could not manage the return walk." That is clinically relevant information, and physiology alone cannot infer it.
PROMs are also useful for tracking direction over time when one patient completes the same instrument at intervals. A steadily rising CAT score or worsening mMRC grade represents a meaningful trajectory signal. Within-patient change is generally more dependable than the absolute score, and trajectory is often what clinicians need most.
Reporting Bias in Chronic Disease
Research in chronic respiratory disease shows a consistent pattern: people with slowly progressive illness adjust their internal reference point. A patient with GOLD 3 COPD whose FEV1 has declined gradually over five years may call their breathlessness "moderate" because it has become normal for them, not because the objective situation is moderate. This adaptation can lead PROMs to underestimate functional limitation when disease duration is longer.
A second issue is timing. COPD exacerbations do not necessarily appear in the words patients use to complete questionnaires. Early exacerbation involves physiological changes, including a higher respiratory rate, altered breath sound characteristics, a changed I:E ratio, and greater work of breathing. Patients may describe this as "a bit more tired" or "not quite right", rather than identify a respiratory change. When symptoms are clear enough to prompt a diary entry or call to the surgery, physiological deterioration may already be several days old.
Acoustic Data's Added Value
Acoustic monitoring does not measure subjective experience. It measures how the airway physically behaves during breathing. The two signals are distinct, and their divergence can be informative.
During a COPD exacerbation, airway inflammation and increased secretions change the acoustic qualities of breath sounds. Wheeze, crackle frequency, and the inspiratory to expiratory sound-duration ratio can shift before the patient consciously notices deterioration. This does not question the patient's awareness. It reflects the limits of proprioception: people are remarkably poor at sensing gradual physiological change until it produces a functional effect.
In our early pilot work, we have seen daily acoustic summaries begin to shift four to five days before patients reported any symptom change in check-in responses. They were not withholding information and genuinely did not feel substantially different. Their airways were behaving differently nonetheless.
Both Sources Are Needed
Acoustic monitoring is not intended to replace patient-reported data. Its value is providing a separate channel that can detect early physiological change in ways questionnaire responses may not.
Consider a practical example. A COPD patient's CAT score remains 18 for four months, while her acoustic summary shows wheeze duration gradually increasing. Without the acoustic signal, the unchanged CAT score suggests stable disease. With both measures, the mismatch warrants clinical review. The combined picture is more informative than either signal alone.
The reverse pattern can also help clinically. Rising CAT scores alongside stable acoustic patterns may point to worsening anxiety, depression, or a non-respiratory health problem rather than airway deterioration. Stable acoustic data does not exclude a problem, but it helps direct clinical attention appropriately.
Listening Quality Matters
Acoustic monitoring has its own sources of noise, both literal and figurative. Ambient sound, microphone position, and adherence to recording conditions all influence signal quality. A kitchen recording with a radio playing and a family conversation will differ from one made in a quiet room in the morning. Data-quality protocols and consistency checks are as important as the algorithm processing the signal.
We do not claim that acoustic data gives a clear, unambiguous physiological reading in every setting. Its systematic limitations differ from those of self-report. When both sources are properly quality-controlled, clinicians have more useful information to assess.
Designing Better Monitoring Programmes
PROM-only programmes are not necessarily wrong, but they depend on a sensor system with recognised blind spots. Replacing PROMs entirely with physiological monitoring would lose the functional and experiential aspects of disease that only patients can describe.
A coherent design combines a brief daily symptom check-in, taking under a minute, with passive or brief acoustic capture. The first follows subjective experience, while the second adds physiological information without extra patient burden. Together they identify more of the early deterioration pattern, and they measure different rather than duplicate signals.
The design challenges include combining these measures well, calibrating them to each patient's baseline rather than population averages, and creating clinical response workflows for the combined signal rather than either component alone. Early-stage respiratory monitoring programmes are still working through these questions. Evidence for the best configuration is emerging, so claims that the problem is definitively solved deserve caution.