Oxygen desaturation during a 6 min walk test is a sign of nocturnal hypoxemia

1Reasearch Department, Mount Sinai Hospital Center; 2 Faculty of Medicine, School of Physical and Occupational Therapy, McGill University, Montreal, Quebec Correspondence and reprints: Dr Norman Wolkove, Mount Sinai Hospital Center, 5690 Cavendish Boulevard, Cote St-Luc, Montreal, Quebec H4W 1S7. Telephone 514-369-2222 ext 1060, fax 514-369-2225, e-mail nwolkove.sinai@ssss.gouv.qc.ca; marc.baltzan@clinepi.mcgill.ca Some patients with chronic obstructive pulmonary disease (COPD) experience significant hypoxemia. For those with a resting partial pressure of oxygen in arterial blood (PaO2) <55 mmHg, supplemental oxygen administration has been shown to be beneficial, particularly in improving exercise tolerance and survival (1,2). However, less well recognized is that some individuals with COPD may desaturate only during sleep and become substantially more hypoxemic at night than they are during the day. Although nocturnal oximetry can readily identify such individuals, this requires overnight monitoring and is time consuming. Furthermore, it is unclear which patients require such a study. It would be useful if simple daytime parameters could help predict which patients are more likely to exhibit significant nocturnal desaturation (SND). In the present study, we sought to determine whether any parameters routinely measured in patients with COPD could be helpful in predicting the occurrence of significant sleeprelated oxygen desaturation. In particular, we aimed to determine whether measuring arterial oxygen saturation (SpO2) during a 6 min walk test (6MWT) could provide additional information that might enable us to predict which patients were more likely to experience SND. The main finding from an initial study was applied to a second population to confirm its predictive capacity for the identification of patients who experience SND. Methods study population The data for the exploratory study were obtained by conducting a chart review of patients with a history of COPD who had been admitted for pulmonary rehabilitation at Mount Sinai Hospital (Montreal, Quebec) between March 2006 and May 2007. Subjects in the validation study were selected retrospectively (July 2008 and July 2009) based on results from a 6MWT in which they desaturated to or below 88%. Approximately two age-matched controls were included per subject. The control patients had COPD but did not desaturate to or below 88%. Patients with a history of obstructive sleep apnea were excluded, as were individuals who were on long-term supplemental oxygen. The present study was approved by the Mount Sinai Hospital Center Ethics Committee (approval number 052007).


Oxygen desaturation during a 6 min walk test is a sign of nocturnal hypoxemia
Adrienne S Scott MS 1 , Marcel A Baltzan MD FRCPC 1 , Ryan Chan BSc 2 , Norman Wolkove MD FRCPC FCCP S ome patients with chronic obstructive pulmonary disease (COPD) experience significant hypoxemia.For those with a resting partial pressure of oxygen in arterial blood (PaO 2 ) <55 mmHg, supplemental oxygen administration has been shown to be beneficial, particularly in improving exercise tolerance and survival (1,2).However, less well recognized is that some individuals with COPD may desaturate only during sleep and become substantially more hypoxemic at night than they are during the day.Although nocturnal oximetry can readily identify such individuals, this requires overnight monitoring and is time consuming.Furthermore, it is unclear which patients require such a study.It would be useful if simple daytime parameters could help predict which patients are more likely to exhibit significant nocturnal desaturation (SND).In the present study, we sought to determine whether any parameters routinely measured in patients with COPD could be helpful in predicting the occurrence of significant sleeprelated oxygen desaturation.In particular, we aimed to determine whether measuring arterial oxygen saturation (SpO 2 ) during a 6 min walk test (6MWT) could provide additional information that might enable us to predict which patients were more likely to experience SND.The main finding from an initial study was applied to a second population to confirm its predictive capacity for the identification of patients who experience SND.

Methods study population
The data for the exploratory study were obtained by conducting a chart review of patients with a history of COPD who had been admitted for pulmonary rehabilitation at Mount Sinai Hospital (Montreal, Quebec) between March 2006 and May 2007.Subjects in the validation study were selected retrospectively (July 2008 and July 2009) based on results from a 6MWT in which they desaturated to or below 88%.Approximately two age-matched controls were included per subject.The control patients had COPD but did not desaturate to or below 88%.Patients with a history of obstructive sleep apnea were excluded, as were individuals who were on long-term supplemental oxygen.The present study was approved by the Mount Sinai Hospital Center Ethics Committee (approval number 052007).

study design
All patients underwent routine spirometry that was performed according to published guidelines (3).Arterial blood gases were not analyzed routinely but were performed at the discretion of the attending physician.A 6MWT was conducted before initiation of the rehabilitation program using published recommendations (4).This was performed in a hospital corridor, and the total distance travelled was recorded in As scott, MA Baltzan, R Chan, N Wolkove.oxygen desaturation during a 6 min walk test is a sign of nocturnal hypoxemia.Can Respir J 2011;18(6):333-337.
BACkgRouNd/oBJeCtives: Patients with chronic obstructive pulmonary disease (COPD) may experience sleep disordered breathing with nocturnal desaturation.An exploratory study was performed to determine whether any commonly measured clinical parameters were useful in predicting nocturnal desaturation in patients with COPD.A validation study was subsequently performed to confirm the utility of the parameter identified in the exploratory study as most useful in this regard.Methods: A total of 103 (exploratory cohort) and 200 (validation cohort) consecutive patients with COPD admitted for pulmonary rehabilitation were evaluated.Standard outcome measures including nocturnal oximetry and the 6 min walk test (6MWT) on room air with continuous pulse oximetry were assessed.Patients with sleep apnea or those undergoing long-term oxygen therapy were excluded.Results: In the exploratory study, the mean (± SD) patient age was 70±9.9 years, with forced expiratory volume in 1 s of 0.76±0.34L, which was 36±16% of predicted.Body mass index, arterial oxygen tension, oxygen saturation by pulse oximetry at rest and during the 6MWT all demonstrated significant correlations with percentage of time spent with a saturation <90%.When the lowest pulse oximetry during the 6MWT was ≤88%, 10 of 21 patients demonstrated a saturation <90% for at least 30% of sleep time.This measure yielded a positive likelihood ratio of 3.77 (95% CI 1.87 to 7.62) compared with those who did not reach this threshold value.The validation study confirmed similar detection characteristics.CoNClusioNs: Results from the present study suggest that monitoring oxygen saturation changes during a 6MWT is useful in helping to identify COPD patients who may experience significant nocturnal desaturation.
metres.The pulse oximetry saturation (ie, SpO 2 ) was monitored using a finger sensor (Onyx II, Nonin Medical Inc, USA) and the values obtained were recorded before and immediately on completion of the 6MWT.If the patient stopped during the test, the SpO 2 at that point was recorded.In addition, at these times, patients were asked to rate their perceived dyspnea on a 10-point Borg scale.During overnight oximetry, SpO 2 was continuously recorded for a minimum of 6 h using a pulse oximeter and finger sensor (920M Plus oximeter, Respironics Inc, USA) while the patient slept in his/her hospital bed.A hard-copy tracing was then produced from which mean and lowest overnight saturation was obtained, as well as the percentage of valid recorded time spent at saturation <90%.As previously defined, a patient was characterized to have SND if ≥30% of overnight recorded time was spent with an oxygen saturation <90% (5).Body mass index (BMI) was calculated by dividing body weight (kg) by height (m 2 ).

statistical analysis
Descriptive statistics (means, SDs, counts and frequencies in per cent) were used to describe patients' baseline characteristics.Pearson correlations were also calculated to evaluate the relationship between continuous baseline variables and percentage of time spent at <90% saturation.A contingency table and ROC analysis was performed for the prediction of significant desaturation.Statistical significance was set at a two-sided level of 0.05, and 95% CIs were calculated.

Results subjects
The overall flow of participants and organization of the study protocols are summarized in Figure 1.The most common reasons for exclusion were an additional diagnosis of obstructive sleep apnea (45%), use of oxygen (26%) and insufficient data (29%).The descriptive characteristics of both the exploratory and validation cohorts are presented in Table 1.Both populations contained a high proportion of patients who were classified as having severe COPD.

outcomes in the 6MWt and nocturnal oximetry (exploratory study)
Results of the 6MWT are summarized in Table 2. Patients generally demonstrated a reduced walk distance for their age and moderate dyspnea on exertion.Nocturnal oximetry data are detailed in Table 3. Twenty of 103 patients (20%) met the criteria for SND (Figure 1).

Correlation and contingency analysis (exploratory study)
BMI, PaO 2 and SpO 2 before and after the 6MWT all demonstrated significant correlations with the percentage of time spent with SpO 2 <90% during sleep (Table 4).The prevalence of SND was 48% (10 of 21) when the postwalk saturation was ≤88% compared with 12% (10 of 82) when the postwalk saturation was >88%.The critical threshold of 88% for desaturation during the 6MWT was supported by ROC analysis (Figure 2).The area under the curve statistic was 0.68 (95% CI 0.54 to 0.82).A positive likelihood ratio for SND was 3.77 (95% CI 1.87 to 7.62) in patients who desaturated with SpO 2 ≤88% (Table 5).Of the patients who maintained a saturation ≥95% during exercise, only 5% (two of 44) experienced SND.The greatest sum of sensitivity (50%) and specificity (87%) was at a pulse oximetry threshold of ≤88%, also suggesting that this value represented the optimum threshold for detection (6).The threshold SpO 2 of ≤88% during a 6MWT was retained as the primary predictor of SND to validate in a subsequent cohort.

evaluation of the primary predictor (validation study)
The 200 patients used in the validation analysis were obtained as shown in Figure 1, with their characteristics also noted in Table 1.Contingency analyses (Table 5) using the same cut-off of 88% derived in the exploratory study, proved to be operational and was still predictive of SND.The ROC analysis supported the use of the SpO 2 threshold ≤88% as a predictor of SND (Figure 3).The area under the curve in the validation population was 0.73 (95% CI 0.65 to 0.80).In our initial (ie, exploratory) study, we found that desaturation following a 6MWT was useful in helping to predict which patients with COPD had SND.Almost one-half of subjects who had a 6MWT with a lowest SpO 2 ≤88% during their walk, had SND according to our definition.This represented a positive likelihood of 3.77 compared with those who did not reach this threshold value.The findings were consistent in our validation study, in which 36 of 63 (57%) patients with a 6MWT SpO 2 ≤88% had SND.Because the 6MWT is frequently used to assess patients, we suggest that SpO 2 also be monitored during this evaluation.Using the above threshold level, individuals with the highest likelihood of having SND can be identified, thus enabling physicians to use overnight oximetry in a population in which the probability of finding SND is high.
We chose to focus particularly on the 6MWT because it has been established as an informative parameter in patients with COPD (7,8).The 6MWT distance has been shown to be a better predictor of mortality than forced expiratoy volume in 1 s (FEV 1 ) in these individuals (9,10).Casanova et al (7) found that oxygen desaturation during the 6MWT can help predict which patients may experience a poor outcome independent from the distance walked.Garcia-Talavera et al ( 8) also advocated recording SpO 2 during the 6MWT, and found that patients with COPD who desaturated during the first minute of their walk were those most likely to desaturate during activities of daily living or during sleep.
Although we believed that desaturation during the 6MWT was most useful in predicting SND, we found that several other baseline measurements also correlated with nocturnal desaturation to levels <90%.These included a reduced PaO 2 and/or SpO 2 at rest before the 6MWT.Nocturnal desaturation also correlated with BMI.These relationships have been noted by other authors, but their predictive value has been variable.Fletcher et al (11) found a low PaO 2 and a high partial pressure of carbon dioxide in arterial blood (PaCO 2 ) to be related to -but not very predictive of -nocturnal desaturation.Bradley et al (12), however, identified awake SpO 2 and PaCO 2 as independent and predictive variables related to nocturnal desaturation.Similarly, Zanchet and Viegas (13) found that the only variable capable of predicting nocturnal desaturation was daytime SpO 2 .A limitation of these studies was their small sample size.Nevertheless, these studies and ours reinforce the intuitive notion that individuals with low baseline oxygen saturation are more likely to desaturate at night because they are already closer to whichever threshold for defining these events is used.In addition, because they are starting at a lower baseline level, they are near the steep portion of the oxygen dissociation curve, where small changes in PaO 2 can lead to relatively large changes in SpO 2 .Baseline lung function was not useful in predicting nocturnal desaturation in our study, a finding consistent with the literature (14,15).Heijdra et al (16) were an exception, finding that in addition to gas exchange, FEV 1 and maximal inspiratory muscle strength correlated with nocturnal desaturation.Zanchet and Viegas (13) highlighted the difference between correlates and predictors of nocturnal desaturation in patients with COPD.Although they found that the FEV 1 /FVC ratio correlated positively and significantly with SpO 2 during sleep, the only variable capable of predicting nocturnal desaturation was daytime SpO 2 .
The prevalence of nocturnal desaturation among patients with COPD is unknown.Obviously, the lower the baseline SpO 2 in the population studied the higher the prevalence of SND is likely to be.Little et al (17) observed nocturnal desaturation in patients with SpO 2 ≤93% but in none with SpO 2 ≥95%.Others have refuted these results, reporting, for example, patients with daytime saturation as low as 91% or 92% who do not have SND (13).Although unusual, we found several patients in our series with daytime saturation ≥95% who did have nocturnal desaturation.In one of the earliest attempts to address this issue, Fletcher et al (18) found that 27% of patients with an awake PaO 2 of at least 60 mmHg experienced REM-associated desaturation.Subsequently, several small studies of patients with COPD reported the prevalence of SND to be approximately 50% in patients who had some degree of daytime hypoxemia (13,14).Similarly, after screening more than 800 patients with COPD, Lewis et al (19) found nocturnal desaturation in 29 of 59 patients with a daytime SpO 2 <95%.They then assumed that the remaining large group with a daytime SpO 2 ≥95% would not desaturate at night (which may not be accurate) and, thus, estimated the overall incidence of nocturnal desaturation to be 4.8%.Our study of 103 patients represents one of the largest in the literature attempting to address the issue of nocturnal desaturation in individuals with COPD.It is also unique in that it is the largest unselected group to be studied in such a way.We found that 20 of 103 (19.4%) patients in our study had SND.This figure, intermediate between the values of Thomas et al (14), Zanchet and Viegas (13) and the conservative estimate of Lewis et al (19), probably reflects the fact that we included a wide range of individuals with COPD.In our study, for example, approximately 43% (n=44) of our patients demonstrated a daytime SpO 2 ≥95% at one extreme, while another 28% (n=29) had a baseline SpO 2 ≤92%.
Several mechanisms have been postulated to explain the development of nocturnal desaturation in patients with COPD.These include increased bronchial tone alveolar hypoventilation, ventilation perfusion mismatch, increased upper airway resistance and diminished responses to hypoxia and hyperapnea (20,21).
The significance of nocturnal desaturation in patients with COPD remains unclear.Although two major trials have proven that supplemental oxygen therapy can reduce mortality among patients with COPD and persistent hypoxemia, the evidence supporting a similar benefit for patients who desaturate only at night is less obvious (1,2).It is known that transient episodes of significant hypoxemia can produce vasoconstriction, pulmonary hypertension and arrythmias (22,23).However, it has not been firmly established that patients with isolated nocturnal desaturation necessarily develop sustained pulmonary hypertension and have an increased risk of mortality.In one of the few studies to address this issue, Fletcher et al (18) concluded that nocturnal desaturation in patients with COPD was associated with shorter survival.A trend toward increased survival in 35 oxygen-treated versus 38 nontreated subjects was noted, but a statistically significant difference would likely have required more subjects.In a review of available studies in the literature, Cranston et al (24) concluded that home oxygen therapy did not appear to improve survival in patients who experienced desaturation at night only.Similarly, although many patients with COPD report poor quality or interrupted sleep, the limited data in the literature have not shown that nocturnal desaturation is associated with impairment in quality of life, sleep quality or daytime function (19,25).However, the paucity of information available in this regard suggests that larger studies are needed if we are to more completely assess the significance of isolated nocturnal desaturation in patients with COPD.
Presently, a three-year, multicentre, placebo-controlled, randomized trial of nocturnal oxygen therapy (CANOX) has been designed and started by Lacasse et al in Canada (26).Patients to be studied are those with COPD and nocturnal desaturation only.This study will hopefully provide meaningful data to determine whether nocturnal oxygen therapy provided for a period of three years decreases mortality, improves quality of life and/or delays the prescription of long-term oxygen therapy in these patients with COPD.
ACkNoWledgeMeNts: Author contributions: Ms Scott: contrib- uted to design, study coordination, data collection, data analysis, results interpretation, literature review, manuscript writing and revisions.Dr Baltzan: contributed to design, data collection, data analysis, results interpretation, manuscript writing and revisions.Dr Wolkove: contributed to design, results interpretation, literature review, manuscript writing and revisions.Mr Chan: contributed to the collection of data and analysis.
FiNANCiAl/NoNFiNANCiAl disClosuRes: The authors have reported to the Canadian Respiratory Journal that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.
Funded by the Mount Sinai Hospital Research Fund.Dr Norman Wolkove has no conflict of interest to disclose.Adrienne S Scott has no conflict of interest to disclose.Dr Marcel A Baltzan has no conflict of interest to disclose.Ryan Chan has no conflict of interest to disclose.

Figure 3 )Figure 2 )
Figure 3) ROC curve.Pulse oximetry after the 6 min walk test in the validation cohort.AUC Area under the curve