Domiciliary pulse-oximetry at exacerbation of chronic obstructive pulmonary disease: prospective pilot study

Exacerbations of chronic obstructive pulmonary disease (COPD) are deteriorations in respiratory health that punctuate the course of this prevalent condition [1]. Patients with COPD also experience day-to-day symptom variation. Distinguishing this from exacerbation is challenging but important as access to prompt treatment at exacerbation is associated with earlier recovery [2], while treating symptom variation as exacerbation might result in unnecessary exposure to therapy. There is real clinical need for an objective way to confirm exacerbation onset, which would also be important for validating these events in clinical trials. Current attention has focused on symptom scores to confirm exacerbation [3], but symptoms (even when scored) are, by definition, subjective. Despite exacerbations being associated with increases in local and systemic inflammation [4], no biomarker has been shown to reliably differentiate stable COPD from exacerbation [5].

Exacerbations are known to be associated with changes in physiological measures, such as peak-expiratory flow (PEF) [6], but absolute changes are small and could not reliably detect exacerbations when used in isolation [7]. We hypothesised that changes in heart rate and oxygen saturation (SpO₂), readily measurable using pulse oximetry, might assist in the differentiation of stable COPD from exacerbation. Despite the wide availability of such devices, and their use in tele-monitoring programmes [8], there are no existing data examining day-to-day variation of oximetry variables in stable COPD, or the changes in heart rate and SpO₂ observed through the onset and recovery of exacerbations.

We have performed a pilot study of daily domiciliary pulse oximetry in a patient sample from the London COPD cohort. Using prospective diary card data allowed us to precisely define the onset of exacerbation symptoms. We have therefore been able to describe the extent of normal variation in heart rate and SpO₂ on symptom free days, the disease and demographic characteristics that affect this baseline variation, and the time-course of changes with exacerbation. We go on to report the ability of such data to differentiate exacerbation from both stable disease and day-to-day symptom variation.

We have provided a preliminary demonstration that changes in heart rate and oxygen saturation, measured using domiciliary pulse oximetry and incorporated into a composite score, are able to differentiate exacerbation onset from stable COPD with reasonable sensitivity and specificity. In addition, we have defined the extent of daily variation in oximetry variables in stable, ambulatory patients with COPD, and demonstrated changes in these indices immediately prior to, at onset and through the recovery of an exacerbation. Our results and methodology will be of immediate interest to those employing domiciliary oximetry to monitor COPD and, with further validation, have the potential to facilitate objective confirmation of exacerbation using simple technology, enabling access to prompt therapy and verification of events in clinical trials.

Our study has a number of strengths. By prospectively recording daily symptom changes we have been able to differentiate symptom free days from day-to-day symptom variation, and by using validated criteria for exacerbation we have been able to differentiate symptom variation from exacerbation onset. In total, the data reported in this manuscript comprises 2705 days (7.4 patient-years) recording symptoms, oximetry and PEF in 31 patients with well-characterised COPD. These patients are typical of those in whom monitoring and early therapy at exacerbation may be desirable, having moderate and severe disease and a previous history of exacerbation.

Exacerbations are important events that punctuate the natural history of COPD. Patients experiencing frequent exacerbations have a more rapid decline in lung function [11, 14], poorer quality-of-life [9], reduced physical activity [12] and greater mortality [15]. Exacerbations are therefore associated with a considerable burden of health-care costs [16] such that preventing exacerbations is a key goal of COPD therapy. Prompt therapy is associated with more rapid exacerbation recovery [2] and there is therefore the need to identify exacerbations early in their course. This has resulted in attempts to objectively confirm exacerbations, using tools such as the EXACT-PRO [3], but quantifying symptoms does not address the subjectivity of symptoms themselves. Symptom scores may also be prone to ceiling effects, and exacerbation symptoms must be differentiated from the day-to-day symptom variations that characterise COPD. Exacerbations remain difficult events to detect as there is currently no diagnostic biomarker [5]. This can result in over-treating symptom variation as exacerbation, and delaying exacerbation therapy when deteriorations are mis-attributed to symptom variation. This may be particularly important during high-risk periods for recurrent exacerbations, in the weeks immediately following a first event [17]. The distinction between stable disease and exacerbation is not just important for patients, but is also relevant in the conduct of clinical trials where the ability to objectively confirm these events would be a major development.

We have demonstrated changes in heart rate, SpO₂ and PEF immediately prior to, at onset and during exacerbations of COPD. A composite score of change in heart rate and SpO₂ performed moderately well in distinguishing day-to-day symptom variation from COPD exacerbation. This is a potentially important development. Oximetry is cheap and reliable, and electronic data capture make calculation of composite scores practical. With knowledge of a patient's baseline heart rate and SpO₂, calculation of the score can distinguish symptom variation from exacerbation onset by the second day of symptom changes, with a cut-off that can be chosen to maximise sensitivity or specificity. In this way, inappropriate antibiotics might be avoided, and early therapy might be facilitated. Key to our approach was the careful collection of data from each patient at baseline, and expression of subsequent changes as multiples of that patient's own, individual day-to-day variation.

Although pulse oximetry is widely used in tele-monitoring programs [8], and by patients themselves, we are not aware of any previous work defining the extent of day-to-day variation in heart rate and oxygen saturation in patients with moderate and severe COPD. Indeed, the European Commission Communication on Telemedicine has highlighted the need for robust methodological studies in this area [18], such as ours. Use of oximeters supports the current emphasis on management of long-term conditions through self-management and care closer to (or in) the patient's home. Even in the absence of telemedicine, our data could be incorporated into a written self-management plan. For example, we have shown that there would be only a 2.5% chance of a patient having an oxygen saturation <2% of their baseline mean on a stable day, and a reduction in saturation of this magnitude, or rise in heart rate >10 min^-1, might be used to prompt re-assessment of respiratory health. Our technique, employing simple, readily available technology is much more practical than the domiciliary measurement of airway or systemic biomarkers which in any case have been unable to differentiate exacerbation from the stable state [5].

There are some limitations of our approach. We present our findings as preliminary but believe these data will be of immediate utility to those monitoring patients with domiciliary oximetry, enabling subsequent validation of our technique in larger cohorts. To establish an individual patient's baseline, our method requires rigorous data collection, at the sensitivity limit of currently available technology. However, with tele-health such methods are increasingly practical. We did not include patients with respiratory failure and it may be that oxygen saturation changes at exacerbation in patients on the steeper part of the oxygen dissociation curve are greater. Patients in atrial fibrillation were also excluded. We do not have data to inform on how long day-to-day variation in the baseline state remains stable. We have previously reported that PEF declines (slowly) with time [11] and we are not able to state how frequently the stable monitoring period would need to be repeated. Our data do not inform on the ability of pulse oximetry to distinguish exacerbation of COPD from other causes of symptom deteriorations in these patients (such as pneumonia), and it is quite likely that such pathologies would also result in heart rate and SpO₂ changes. Exacerbation of COPD remains a clinical diagnosis of exclusion and the clinician (and patient) must remain aware that other conditions may mimic or complicate exacerbations [19]. We note that there were fewer unreported exacerbations in this study than we have previously detected from this cohort. Finally, it is important to remark that in some patients (17.5% in this study) it was not possible to ascertain baseline readings on 30 symptom free days, despite prolonged monitoring, and this might limit applicability to a proportion of patients in clinical practice.

The Oximetry Score was able to differentiate exacerbation onset from day-to-day symptom variation with reasonable sensitivity and specificity. Our data are the first to report that day-to-day symptom variation in COPD is associated with small, but demonstrable changes in physiology. Addition of PEF to the oximetry variables was not associated with a better composite score, despite demonstrating detectable changes in PEF at exacerbation (as previously reported [6]). This may be because it is more difficult to assess smaller changes in PEF, on a scale where 10l min^-1 is the minimal detectable difference. In addition, the PEF changes lag those in heart rate and SpO₂ and were maximal at four days into the exacerbation (symptom day five). One attraction of adding PEF is that this is a marker of airway involvement, believed to be the primary site of pathology during exacerbations. In contrast SpO₂ reflects ventilation-perfusion relationships, and changes in heart rate a variety of mechanisms including increased β₂-agonist use. We do not have data on β₂-agonist use, though this would not be expected to significantly affect changes in oxygen saturation or PEF. Adding PEF increased the performance of the score differentiating exacerbation onset from all other days, regardless of symptoms, but PEF is also effort dependent and given the difficulties in assessing PEF reliably, and the small changes involved, an approach restricted to heart rate and SpO₂ would seem to be more attractive. This is in contrast to the validity of PEF during exacerbations of asthma [20].

Both scores, and individual components showed increased variation immediately preceding exacerbation onset, with changes that were greatest in magnitude 2-4 days following the start of exacerbation symptoms (at about the time that patients commenced therapy). The magnitude of the changes in heart rate and PEF were greater than the magnitude of change in SpO₂, such that many exacerbations are associated with only small (or no) change(s) in SpO₂. Much of the initial signal in the composite scores may therefore be attributed to alterations in heart rate. As initial changes were small, it seems likely that exacerbation detection should remain dependent on patients first identifying changes in symptoms and our data suggest that oximetry can assist in differentiating exacerbation symptoms from day-to-day variation. However, we have also provided evidence that the composite scores can identify exacerbation onset from all other days in the absence of symptom assessment. The Oximetry-PEF score performed best in this instance. Oximetry monitoring may therefore also be important for encouraging assessment of 'unreported' (and therefore untreated) exacerbations [6, 21], events which contribute to the negative impact of exacerbations on COPD health status.

In summary, this is the first report to provide data on the day-to-day variation of heart rate and oxygen saturation in ambulatory patients with stable moderate and severe COPD. We have further described how this information might be used to facilitate confirmation of COPD exacerbation onset, desirable in facilitating access to prompt treatment, and providing objective confirmation of these events in the context of clinical trials. These data will be of immediate use to those employing domiciliary pulse oximetry monitoring in COPD, and there is now the need to replicate these findings in separate, larger cohorts with the aim of using oximetry to facilitate and guide exacerbation therapy.