Sleep disordered breathing (SDB), which includes. Bilevel Positive Airway Pressure Worsens Central Apneas During Sleep*

Bilevel Positive Airway Pressure Worsens Central Apneas During Sleep* Karin G. Johnson, MD; and Douglas C. Johnson, MD Study objectives: While most patients with sleep-disordered breathing are treated with continuous positive airway pressure (CPAP), bilevel positive airway pressure (BLPAP) is often used. Having observed that BLPAP therapy increased central apneas in some of our patients undergoing sleep studies, we conducted this study to evaluate the effects of BLPAP. Design: Retrospective analysis of all sleep studies performed in an outpatient sleep center that used BLPAP over a 2-year period. We assessed the incidence and frequency of events during rapid eye movement (REM) sleep and non-rem sleep during baseline conditions, CPAP, and BLPAP. Desaturations, hypopneas, obstructive apneas, and central events, including periodic breathing (PB), Cheyne-Stokes respiration (CSR), and non-csr central apneas were evaluated. Patients: Ninety-five of the 719 patients who underwent sleep studies met inclusion criteria. Eighty of the 95 patients treated with BLPAP were also treated with CPAP. Results: BLPAP was more likely to worsen than improve CSR (p 0.002), non-csr central apneas (p < 0.001), and CSR or PB (p < 0.001). CSR (p 0.03) and non-csr central apneas (p 0.01) were more likely to worsen with BLPAP (24% and 23%, respectively) than with CPAP (11% and 8%). Central events (p 0.04) and CSR (p 0.009) were more likely to worsen during BLPAP in patients with baseline CSR or PB (62% and 48%, respectively) than develop in those without baseline CSR or PB (34% and 18%). Higher BLPAP differences worsened central events in 28% of patients, while 7% improved (p 0.02). During REM sleep, central apneas improved, while hypopneas and obstructive apneas worsened (p < 0.001). Conclusions: BLPAP often increases the frequency of CSR and non-csr central apneas during sleep. Since CSR has adverse effects on cardiac function and sleep, it is important to consider this possible adverse effect of BLPAP. (CHEST 2005; 128:2141 2150) Key words: bilevel positive airways pressure; Cheyne-Stokes respiration; continuous positive airways pressure; periodic respiration; polysomnography; positive pressure respiration; sleep, rapid eye movement; sleep apnea, central; sleep apnea syndromes Abbreviations: BLPAP bilevel positive airway pressure; CHF congestive heart failure; CPAP continuous positive airway pressure; CSR Cheyne-Stokes respiration; EPAP expiratory positive airway pressure; IPAP inspiration positive airway pressure; OSA obstructive sleep apnea; PB periodic breathing; PSV pressure support ventilation; REM rapid eye movement; SDB sleep-disordered breathing Sleep disordered breathing (SDB), which includes both obstructive and central apneas and hypopneas, is very common and is associated with morbidity and mortality. Approximately 7% of adults have moderate-to-severe obstructive sleep apnea (OSA), 20% have mild OSA, and 5% 65 years old have *From the Rhode Island Hospital (Dr. K. Johnson), Providence, RI; and Spaulding Rehabilitation Hospital (Dr. D. Johnson), Boston, MA. This work was performed at the Spaulding Rehabilitation Hospital. Manuscript received November 17, 2004; revision accepted March 28, 2005. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Douglas C. Johnson, MD, Spaulding Rehabilitation Hospital, 125 Nashua St, Boston, MA 02114; e-mail: djohnson5@partners.org central sleep apnea. 1 Nearly 50% of patients with congestive heart failure (CHF) have SDB. 2 SDB is usually treated with continuous positive airway pressure (CPAP). An alternative treatment is bilevel positive airway pressure (BLPAP), which detects the patient s inspiration to trigger the change to the higher pressure (inspiration positive airway pressure [IPAP]), and switches to the lower pressure (expiratory positive airway pressure [EPAP]) when inspiration ends. BLPAP can be set to provide a backup rate, which changes to the higher pressure if the patient does not initiate a breath within a specified time. We observed that many patients referred for outpatient sleep studies acquired or had more frequent central apneas during BLPAP. Central apneas occur when there is an absence of central ventilatory motor output. Central apneas www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2141

during sleep cause repetitive arousals, increase catecholamine levels, increase BP, disrupt sleep, worsen CHF, and are associated with increased mortality in patients with CHF. 3 Cheyne-Stokes respiration (CSR) is a subset of central apneas in which there is a gradual decrease in breath size followed by an apnea, then a gradual increase in breath size. CSR is common among patients with CHF and stroke. Central apneas also include non- CSR central apneas that often occur following arousals or with sleep onset or changes in sleep state. The standard treatment of both OSA 4 and CSR 5 is CPAP, which improves OSA by holding the airway open and improves CSR by several mechanisms. BLPAP has been reported as effective as CPAP for treating OSA 6 and central apneas. 7 We were concerned that BLPAP could worsen central apnea in many patients. Although there has been at least one case report 8 of more frequent central apneas with BLPAP, we are unaware of any large studies. This retrospective study evaluates patients who received BLPAP to manage their SDB. Materials and Methods All sleep studies performed in the sleep center from January 2002 to December 2003 were reviewed. The review included 852 studies on 719 subjects. Each study began by evaluating tolerance of an interface using CPAP while awake. If sufficient respiratory events were present after a 2-h baseline period, then CPAP was begun and titrated to eliminate the respiratory events. If the patient did not tolerate CPAP because of difficulty breathing against the pressure or had persistent respiratory events with CPAP, then BLPAP was attempted. If hypoxia developed (oxygen saturation 4% below awake baseline) that did not improve with CPAP, BLPAP was then attempted prior to adding supplemental oxygen. BLPAP was also used for patients with known respiratory muscle weakness or previously determined to need BLPAP. If there were significant central apneas with BLPAP, then a back-up rate of 10/min was begun or CPAP resumed. Near the end of the study period, an exhalation pressure relief device (C-Flex; Respironics; Murrysville, PA) was used if there were significant central events with BLPAP. Our current practice is to use the pressure relief device prior to BLPAP. The pressure relief device detects the patient s exhalation, and makes it easier to breathe out than CPAP by lowering the pressure for approximately one third of a second at the start of exhalation. The C-flex device has three comfort settings, which allows one to adjust the amount of pressure drop. Our protocol for setting BLPAP pressures was to set EPAP at the level found during CPAP titration that eliminated obstructive apneas, or to 4 cm H 2 O if there were no obstructive apneas, and then increase EPAP if the inspiratory efforts did not consistently trigger IPAP. The protocol for setting IPAP pressure was to start 3cmH 2 Oor4cmH 2 O higher than EPAP, and then titrate higher to eliminate hypopneas or improve saturations. Standard criteria were used to stage sleep 9 and identify apneas and hypopneas. 4 For periods of BLPAP with a backup rate, apneas were identified by the chest and abdominal signal. The central and obstructive apnea indexes are the number of respective apneas per hour of sleep, with mixed apneas counted in both indexes. Each study was reviewed for diagnosis. OSA had more than five obstructive apneas or hypopneas per hour. CSR had crescendo-decrescendo alterations in respiratory effort and tidal volume separated by periods of central apnea (Fig 1). Periodic breathing (PB) had periodic increases and decreases in respiratory effort without central apneas (Fig 2). Non-CSR central apneas were central apneas that were not associated with crescendo-decrescendo alterations in respiratory effort, and often occurred with sleep onset or after arousal (Fig 3) All patients receiving BLPAP were identified (n 127). Patients were excluded for 10 min of sleep on BLPAP (excluding 18 patients), a difference of 4cmH 2 O between inspiratory and expiratory pressures (n 7), lack of both baseline and CPAP (n 4), tracheostomy (n 1), and failure to trigger BLPAP (n 2). With these criteria, a study group of 95 patients remained. Of the study group, 10 patients were treated with BLPAP with only a backup rate, and 3 patients were treated with BLPAP both with and without a backup rate. CPAP data were absent for 15 patients, and 1 patient had no baseline data. The study group was evaluated for respiratory events during baseline, CPAP, and BLPAP. The indexes of obstructive apneas, central apneas, mixed apneas, hypopneas, and desaturations were determined for each condition using the periods with the highest CPAP or IPAP levels that totaled 1 h. Our analysis proceeded as follows: (1) The period with highest CPAP was identified. If the sleep time was 1 h, this was the only period included in the analysis. (2) If the sleep time was 1 h, the period with the next highest CPAP was also included. (3) This process continued until there was at least 1 h of sleep time or until all the CPAP periods were included. For BLPAP, we used a similar analysis to determine the periods with the highest IPAP to a total of 1hof sleep. BLPAP periods with a pressure difference of 4cmH 2 O were excluded because we wanted there to be a clear difference between CPAP and BLPAP. We also determined the periods with the highest pressure difference (IPAP minus EPAP) with at least 10 min of sleep so we could test the hypothesis that higher pressure differences worsened central events. The studies were reviewed to determine the presence and effect of treatment on respiratory events. Each study was evaluated for presence of rapid eye movement (REM) periods during baseline, CPAP, and BLPAP, and whether there were no events, an increase, no change, or a decrease in obstructive apneas, central apneas, and hypopneas in REM vs non-rem sleep. Statistical Analysis Mean SD values are reported. A two-tailed Fisher Exact Test was used to compare events between groups with or without baseline CSR or PB. 2 test was used to determine if the numbers of patients improving vs worsening respiratory events differed. Results Of the 719 patients, 638 patients had primary OSA, including 14 patients with CSR and 47 patients with non-csr central apneas. Eighteen patients had OSA and CSR as co-equal diagnoses; 14 patients had primary CSR, including 9 patients with OSA. The 95 patients treated with BLPAP included 77 patients with primary OSA, including 2 patients with 2142 Clinical Investigations

Figure 1. CSR, with crescendo-decrescendo alterations in respiratory effort and airflow, separated by periods of central apnea. There were no arousals during this period. Oxygen saturations (O2 Sat) are highest during the periods of apnea. CSR and 1 patient with non-csr central apnea. Six patients had OSA and CSR as co-equal diagnoses. Six patients had primary CSR, including four patients with OSA. Two patients had OSA and non-csr central apnea as co-equal diagnoses. Two patients had periodic limb movements with OSA, one patient had non-csr central apnea with OSA, and one patient had amyotrophic lateral sclerosis. Thirty-nine patients had periodic limb movements, including 10 patients with periodic limb movements as a primary diagnosis. Medical conditions included hypertension (n 47), obstructive lung disease (n 29), diabetes (n 19), irregular heart rate (n 16), CHF (n 11), stroke (n 6), and pulmonary hypertension (n 3). Their mean age was 57.8 13.8 years, and mean body mass index was 36.0 9.0. There were 54 men and 41 women. Oxygen saturation was 90% for 17.5 21.6% of the time. For the periods analyzed, the average CPAP was 11.0 3.6 cm H 2 O, IPAP was 15.9 4.8 cm H 2 O, EPAP was 9.4 4.4, and IPAP minus EPAP was 5.9 3.1 cm H 2 O. For the period with the greatest IPAP minus EPAP value, the IPAP was 15.8 4.2 cm H 2 O, EPAP was 8.3 4.0 cm H 2 O, and IPAP www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2143

Figure 2. Periodic breathing, with crescendo-decrescendo alterations in respiratory effort and airflow without central apneas. There were no arousals during this period. Oxygen saturations are highest during the periods of hypopnea. See Figure 1 legend for expansion of abbreviation. minus EPAP was 8.0 2.4 cm H 2 O. Reasons for using BLPAP included difficulty tolerating CPAP (n 28, 3 with backup), hypoxia persisting with CPAP (n 22), persistent respiratory events with CPAP (n 17, 1 with backup), hypoxia and events continuing on CPAP (n 9, 1 with backup), central apneas on CPAP (n 13, 6 with backup), BLPAP at home (n 6), and known respiratory muscle weakness (n 1). At baseline, 14 patients had CSR, 18 patients had PB, 21 patients had either CSR or PB, and 3 patients had non-csr central apnea without CSR or PB. At baseline, 23 patients had a central apnea index 5/h, and 15 patients had a central apnea index 10/h. These baseline characteristics did not differ based on reported medical condition. Central apneas, including both CSR and non-csr central apneas, and PB were more frequent (worse) during BLPAP than during baseline. Twenty-three patients had worse CSR, 65 patients were un- 2144 Clinical Investigations

Figure 3. Non-CSR central apnea, with postarousal central apneas. There are two arousals (shaded on EEG tracing) related to obstructive hypopneas, then a central apnea, then two closely timed arousals related to the central apnea, and then a second central apnea. The arousals were associated with increased heart rate. See Figure 1 legend for expansion of abbreviation. changed, and 6 patients improved (p 0.002). Twenty-two patients had worse non-csr central apneas, 71 patients were unchanged, and 1 patient improved (p 0.001). Twenty-seven patients had worsening of either CSR or PB, 61 patients were unchanged, and 6 patients improved (p 0.001). Patients with CSR or PB at baseline were more likely to acquire or have worsened central events, in particular CSR, with BLPAP (Table 1). BLPAP was more likely than CPAP to worsen CSR (p 0.03) and non-csr central apneas (p 0.01) [Fig 4]. While more patients improved CSR and non-csr central apneas with CPAP than BLPAP, these differences were not significant. BLPAP increased central apneas compared to baseline and CPAP (Fig 5), with 24% of patients having a higher ( 5/h) central apnea index with BLPAP than CPAP, vs 10% having a lower index www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2145

Table 1 Patients With Worsened Central Events During BLPAP Compared to Baseline* Event Type CSR/PB at Baseline (n 21) No CSR/PB at Baseline (n 73) Total (n 94) p Value CSR 10 (48) 13 (18) 23 (24) 0.009 PB 3 (14) 6 (8) 9 (10) Non-CSR central apnea 7 (33) 15 (21) 22 (23) Combined central events 13 (62) 25 (34) 38 (40) 0.04 *Data are presented as No. (%) of patients. Patients with CSR/PB at baseline compared to those without CSR/PB at baseline. Only p values 0.2 are shown. Combined central events include CSR, PB, and/or non-csr. (p 0.03). A backup rate with BLPAP did not abolish central events, with 6 of 12 patients having a higher ( 5/h) central apnea index compared to baseline and 4 of 13 patients having a higher index compared to CPAP. Both CPAP and BLPAP improved obstructive apnea (p 0.001 and p 0.05, respectively), hypopnea (p 0.001 and p 0.001), and desaturation indexes (p 0.001 and p 0.001) compared to baseline. BLPAP and CPAP had a similar percentage of patients improving (reducing) obstructive apnea index ( 5/h) [40% and 37%, respectively], but more patients worsened obstructive apnea index ( 5/h) with CPAP (19%) than BLPAP (4%) compared to baseline. Comparing BLPAP to CPAP, 24% had a lower obstructive apnea index with BLPAP and 10% had a higher index (p 0.07). There were similar improvements in hypopneas and desaturations with BLPAP and CPAP. The hypopnea index decreased 5/h in 70% of patients with BLPAP and 62% with CPAP, while it increased 5/h in 18% and 21%, respectively. The desaturation index (falls on oxygen saturation 3%/h) decreased 5/h in 69% of patients with BLPAP and 59% with CPAP, while it increased 5/h in 27% and 21% of patients, respectively. Two patients were treated with the exhalation pressure relief device. Both patients had OSA and CSR at baseline, had worse CSR on BLPAP, and improved with exhalation pressure relief. During REM sleep compared to non-rem sleep, central apneas improved (decreased) [p 0.001] in most patients, while hypopneas and obstructive apneas worsened (increased) [p 0.001; Fig 6, Table 2]. The only patient with more central apneas during REM sleep had worse obstructive apneas leading to more postarousal central apneas. Higher BLPAP differences worsened central apneas. Of the 43 patients who could be evaluated at pressure differences 4cmH 2 O, 12 patients (28%) worsened, 3 patients (7%) improved, and 28 patients (65%) showed no change (p 0.02). Recommended treatment included 57 patients for BLPAP (25 patients with BLPAP only, 14 patients with BLPAP and oxygen, 18 patients for BLPAP and a backup rate [9 patients with oxygen]); 35 patients for CPAP (26 patients with CPAP only, 9 patients also with oxygen); and 3 patients for exhalation pressure relief. Of the 92 patients with OSA, BLPAP was recommended for 54 patients. Thus, of the entire group of 666 OSA patients, BLPAP was recommended for 8%. Discussion Figure 4. Effect of BLPAP (n 94) and of CPAP (n 80) on CSR and non-csr central apnea, compared to baseline. While CPAP is the standard treatment for OSA and CSR, BLPAP is often used. This study was undertaken to address our concern that BLPAP may cause adverse effects in some patients. We found that BLPAP increased central apneas including CSR and non-csr central apneas compared to baseline and CPAP. CSR occurs when instability of the ventilatory control system leads to oscillations in Pco 2,Po 2, and breathing. The onset of CSR often occurs with sleep onset or following an arousal. This is probably due to a lower respiratory drive with a higher apneic threshold during sleep than when awake, or during sleep than arousal. When Pco 2 drops below the apneic threshold, central apnea occurs, 10 resulting in a 2146 Clinical Investigations

Figure 5. Effect of treatment on central apnea index. Groups include BLPAP vs baseline (n 93), CPAP vs baseline (n 78), and BLPAP vs CPAP (n 69). *Patients improved compared to those who worsened. delayed increase in Pco 2, which increases the drive to breathe, which in turn decreases the Pco 2, resulting in another apnea. A central apnea may also cause an arousal, which can result in deeper breaths and subsequent fall in Pco 2, resulting in another central apnea. CSR is more common during sleep than when awake. During sleep, the ventilatory response to carbon dioxide changes, with hypocapnia having a more potent inhibitory effect on ventilation during non-rem sleep. 11 Reduction in lung volumes in the supine position lowers gas stores, which contributes to PB because each apnea and respiration leads to a larger change in alveolar Pco 2. CSR is associated with CHF, stroke, and ascent to high altitude. CSR occurs in approximately 50% of patients with CHF, 2,12,13 and is associated with an increased morbidity and mortality. 14 CHF often results in delayed circulation time, increased respiratory drive secondary to interstitial edema, and lower lung volume, factors that make CSR more likely. Treatment of CSR with CPAP in patients with CHF improves heart function and transplant-free mortality. 3 Patients with stroke often have CSR, 15 likely related to increased respiratory drive due to reduced cortical inhibition of the central respiratory center. High altitude increases respiratory drive due to hypoxia. 16 We are unaware of much literature on non-csr central apnea. Figure 6. Effect of REM compared to non-rem sleep on central apneas (n 27), obstructive apneas (n 35), and hypopneas (n 109) among patients with events. Central apneas improved during REM, while obstructive apneas and hypopneas worsened. This combines the data from baseline, CPAP, and BLPAP conditions. www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2147

Table 2 Effect of REM Sleep on Respiratory Events* Event Type/Setting Improved Same Worse No Events p Value Central apnea Baseline 3 (75) 0 (0) 1 (25) 35 CPAP 3 (50) 3 (50) 0 (0) 36 BLPAP 11 (65) 6 (35) 0 (0) 48 0.001 Obstructive apnea Baseline 0 (0) 1 (5) 19 (95) 19 0.001 CPAP 0 (0) 2 (33) 4 (67) 36 0.12 BLPAP 1 (11) 2 (22) 6 (67) 56 0.12 Hypopnea Baseline 1 (3) 7 (19) 28 (78) 3 0.001 CPAP 1 (3) 8 (26) 22 (71) 11 0.001 BLPAP 4 (10) 10 (24) 28 (67) 23 0.001 *Data are presented as No. (%) of patients with change in frequency in events from review of polysomnogram. Patients improved compared to those who worsened; only p values 0.2 are shown. The only patient with more central events during REM sleep had worse obstructive events leading to more frequent postarousal central apneas. Hypopneas include central and obstructive. We found that worse CSR with BLPAP was more common among patients with baseline CSR or PB. Patients who start with an unstable respiratory system are particularly prone to become more unstable during BLPAP. There are several mechanisms by which BLPAP could increase the incidence and severity of central apneas. BLPAP increases tidal volume for a given respiratory effort, contributing to instability of ventilation and making it more likely that Pco 2 will fall below the apneic threshold. The finding that inhalation of carbon dioxide improves CSR 17 supports this hypothesis. Large tidal volumes can also cause central apnea through neurochemical inhibition even if normocapnia is maintained. 18 Morrell et al 19 studied the effects of BLPAP on six normal subjects during sleep and found that one patient had PB and another patient had CSR. Both CPAP and BLPAP could unmask CSR by improving obstructive respiratory events. Pressure support ventilation (PSV) is often used in the hospital setting to treat patients with respiratory distress. Since PSV is comparable to BLPAP, providing a higher inspiratory pressure with patient-initiated breaths, we would expect PSV to have the same worsening of CSR and non-csr central apnea during sleep. Meza et al 20 showed that most healthy subjects have PB during sleep and PSV of 5 to 10 cm H 2 O, which is similar to our IPAP minus EPAP values. Parthasarathy and Tobin 21 found that central apneas occur frequently among patients with respiratory failure treated with PSV, particularly those with CHF. During BLPAP treatment with larger pressure differences, we found that central apneas and PB were more likely to increase than decrease. Larger BLPAP pressure differences lead to larger tidal volumes, which lower Pco 2, thus worsening central apneas. Hommura et al 8 reported a case of worsening number and duration of apneas as the pressure difference increased in a patient with central sleep apnea. Our results and others are consistent with the worsening of central apneas with BLPAP caused by the pressure difference, and not simply the IPAP setting. Both CPAP and BLPAP effectively reduced obstructive apneas and hypopneas during non-rem and REM sleep. Our finding that BLPAP was somewhat more effective than CPAP for improving obstructive events was probably due to having higher IPAP than CPAP settings for the periods analyzed. Our finding that REM sleep consistently worsened obstructive apneas during baseline and during positive airway pressure was expected. Reduced muscle tone during REM sleep increases upper airway obstruction. Central apneas and periodic respiration occur rarely during REM sleep at baseline. 14,22 We found that BLPAP did not worsen central apneas during REM sleep as it did during non-rem sleep. This is likely due to sleep state-related differences in ventilatory control. During REM sleep, ventilatory control is largely independent of the usual chemoreceptor drive to breathe, 22 with significantly reduced respiratory drive to carbon dioxide, 23 absence of an apneic threshold, 24 and nonchemoreceptor activation of respiratory neurons. 25 The difference in response to BLPAP during non-rem and REM sleep supports our hypothesis that BLPAP worsens central apneas during non-rem sleep by lowering Pco 2 below the apneic threshold. We found that CPAP was effective treatment for OSA, less likely to worsen or induce CSR or PB than BLPAP, and more likely to improve CSR or PB than 2148 Clinical Investigations

BLPAP. CPAP improves CSR by several mechanisms. By increasing dead space 26 and carbon dioxide lung volume, CPAP helps keep carbon dioxide above the apneic threshold and reduce fluctuations in Pco 2. By increasing lung volume, CPAP increases lung oxygen stores reducing fluctuations in Po 2.By improving OSA, CPAP reduces the number of postarousal central apneas. By reducing cardiac afterload and preload, CPAP leads to improved cardiac function and decreased circulation time in patients with CHF. By reducing interstitial edema in patients with CHF, CPAP could reduce pulmonary vagal efferent stimulation leading to reduced ventilatory drive. 27 Nocturnal BLPAP is a well-established treatment for patients with chronic respiratory failure, improving daytime dyspnea, sleepiness, and Pco 2. 28 BLPAP with a backup rate is effective treatment for patients with primary central sleep apnea. Kohnlein et al 7 report equivalent effects of CPAP and BLPAP with a backup rate for treating patients with CHF and CSR. Willson et al 29 found that BLPAP with a backup rate set to eliminate respiratory efforts effectively treated CSR in patients with CHF. Teschler et al 30 studied patients with CHF and found comparable improvements in central apnea and arousal indexes with oxygen and CPAP, further improvements with BLPAP with a backup rate two below the baseline respiratory rate, and further improvements with adoptive servoventilation. We found, however, that CSR often persists during BLPAP with a backup rate. The backup rate and pressure settings we used were probably not sufficient to maintain Pco 2 below the apneic threshold. It is important to note that our findings were on the initial night of positive airway pressure therapy. It is possible there might be different responses to long-term therapy. BLPAP is often recommended for treatment of OSA in patients who have difficulty tolerating CPAP, and find it more comfortable to breathe against a lower expiratory pressure. 31 Our current practice is to use exhalation pressure relief for patients with difficulty exhaling with CPAP, prior to attempting BLPAP. In both patients included in this study, exhalation pressure relief effectively managed central apneas that were worsened on BLPAP. Another reason for treating OSA patients with BLPAP is the need for higher tidal volumes in patients with elevated Pco 2. BLPAP is also used at home for the treatment of OSA and other SDB, with Resta et al 32 reporting starting 18% of their OSA patients on BLPAP. Among our patients with OSA, we recommended BLPAP therapy for 8%, and expect this will drop now that the exhalation pressure relief device is available. It is important to recognize that BLPAP and PSV worsen central sleep apneas in some patients, placing them at risk for adverse cardiac effects and sleep disturbance. Based on our results and current literature, we believe that CPAP should be the primary treatment for OSA and CSR. For patients who cannot tolerate CPAP due to difficulty with exhalation, exhalation pressure relief should be tried first, then BLPAP. BLPAP without a backup rate may be effective for patients with OSA, but should not be used to treat CSR. BLPAP with a high backup rate (ie, more than or equal to the patient s baseline respiratory rate) or adoptive servoventilation are options to treat CSR, but have not been as fully studied as CPAP. Hospitalized patients receiving BLPAP or PSV should be observed for central sleep apneas. Our findings support Medicare guidelines that BLPAP should only be used to treat sleep apnea if the patient has failed to respond to CPAP. Our findings do not affect the standard practice of using BLPAP with a backup rate for patients with primary central sleep apneas with hypercapnia. In summary, we found that many patients acquired or had worsened CSR and non-csr central apneas when treated with BLPAP. Patients with baseline CSR or PB are more likely to have worsened CSR with BLPAP than those without baseline CSR or PB. CSR often persisted despite using a backup rate with BLPAP. Central apneas consistently improved during REM sleep. Since CSR has detrimental health effects, by the principle of do no harm, it is important to ensure that treatment for SDB does not create more problems. BLPAP should be used with caution to treat OSA, particularly if there is periodic respiration or CSR at baseline. References 1 Young T, Peppard PE, Gottlieb DJ. Epidemiology of obstructive sleep apnea: a population health perspective. Am J Respir Crit Care Med 2002; 165:1217 1239 2 Javaheri S, Parker TJ, Wexler L, et al. Occult sleep-disordered breathing in stable congestive heart failure. Ann Intern Med 1995; 122:487 492 3 Sin DD, Logan AG, Fitzgerald FS, et al. Effects of continuous positive airway pressure on cardiovascular outcomes in heart failure patients with and without Cheyne-Stokes respiration. Circulation 2000; 102:61 66 4 Loube DI, Gay PC, Strohl KP, et al. Indications for positive airway pressure treatment of adult obstructive sleep apnea patients: a consensus statement. Chest 1999; 115:863 866 5 Yan AT, Bradley TD, Liu PP. The role of continuous positive airway pressure in the treatment of congestive heart failure. Chest 2001; 120:1675 1685 6 Reeves-Hoche MK, Hudgel DW, Meck R, et al. Continuous versus bilevel positive airway pressure for obstructive sleep apnea. Am J Respir Crit Care Med 1995; 151:443 449 7 Kohnlein T, Welte T, Tan LB, et al. Assisted ventilation for heart failure patients with Cheyne-Stokes respiration. Eur Respir J 2002; 20:934 941 www.chestjournal.org CHEST / 128 / 4/ OCTOBER, 2005 2149

8 Hommura F, Nishimura M, Oguri M, et al. Continuous versus bilevel positive airway pressure in a patient with idiopathic central sleep apnea. Am J Respir Crit Care Med 1997; 155:1482 1485 9 Rechtschaffen A, Kales A. A manual of standardized terminology, techniques and scoring systems for sleep stages of human subjects. Los Angeles, CA: UCLA Brain Information Service/Brain Research Institute, 1968 10 Xie A, Skatrud JB, Puleo DS, et al. Apnea-hypopnea threshold for CO2 in patients with congestive heart failure. Am J Respir Crit Care Med 2002; 165:1245 1250 11 Schafer T. Variability of vigilance and ventilation: studies on the control of respiration during sleep. Respir Physiol 1998; 114:37 48 12 Findley LJ, Zwillich CW, Ancoli-Israel S, et al. Cheyne- Stokes breathing during sleep in patients with left ventricular heart failure. South Med J 1985; 78:11 15 13 Lofaso F, Verschueren P, Rande JL, et al. Prevalence of sleep-disordered breathing in patients on a heart transplant waiting list. Chest 1994; 106:1689 1694 14 Leung RS, Bradley TD. Sleep apnea and cardiovascular disease. Am J Respir Crit Care Med 2001; 164:2147 2165 15 Nachtmann A, Siebler M, Rose G, et al. Cheyne-Stokes respiration in ischemic stroke. Neurology 1995; 45:820 821 16 Khoo MC, Kronauer RE, Strohl KP, et al. Factors inducing periodic breathing in humans: a general model. J Appl Physiol Respir Environ Exerc Physiol 1982; 53:644 659 17 Leung RS, Floras JS, Lorenzi-Filho G, et al. Influence of Cheyne-Stokes respiration on cardiovascular oscillations in heart failure. Am J Respir Crit Care Med 2003; 167:1534 1539 18 Leevers AM, Simon PM, Dempsey JA. Apnea after normocapnic mechanical ventilation during NREM sleep. J Appl Physiol 1994; 77:2079 2085 19 Morrell MJ, Shea SA, Adams L, et al. Effects of inspiratory support upon breathing in humans during wakefulness and sleep. Respir Physiol 1993; 93:57 70 20 Meza S, Mendez M, Ostrowski M, et al. Susceptibility to periodic breathing with assisted ventilation during sleep in normal subjects. J Appl Physiol 1998; 85:1929 1940 21 Parthasarathy S, Tobin MJ. Effect of ventilator mode on sleep quality in critically ill patients. Am J Respir Crit Care Med 2002; 166:1423 1429 22 Dempsey JA, Smith CA, Przybylowski T, et al. The ventilatory responsiveness to CO 2 below eupnoea as a determinant of ventilatory stability in sleep. J Physiol 2004; 560:1 11 23 Berthon-Jones M, Sullivan CE. Ventilation and arousal responses to hypercapnia in normal sleeping humans. J Appl Physiol Respir Environ Exerc Physiol 1984; 57:59 67 24 Xi L, Smith CA, Saupe KW, et al. Effects of rapid-eyemovement sleep on the apneic threshold in dogs. J Appl Physiol 1993; 75:1129 1139 25 Orem J. Central respiratory activity in rapid eye movement sleep: augmenting and late inspiratory cells [erratum appears in Sleep 1995; 18:137]. Sleep 1994; 17:665 673 26 Khayat RN, Xie A, Patel AK, et al. Cardiorespiratory effects of added dead space in patients with heart failure and central sleep apnea. Chest 2003; 123:1551 1560 27 Panacek EA, Kirk JD. Role of noninvasive ventilation in the management of acutely decompensated heart failure. Rev Cardiovasc Med 2002; 3(suppl):S35 S40 28 Hillberg RE, Johnson DC. Noninvasive ventilation. N Engl J Med 1997; 337:1746 1752 29 Willson GN, Wilcox I, Piper AJ, et al. Noninvasive pressure preset ventilation for the treatment of Cheyne-Stokes respiration during sleep. Eur Respir J 2001; 17:1250 1257 30 Teschler H, Dohring J, Wang YM, et al. Adaptive pressure support servo-ventilation: a novel treatment for Cheyne- Stokes respiration in heart failure. Am J Respir Crit Care Med 2001; 164:614 619 31 Sanders MH, Kern N. Obstructive sleep apnea treated by independently adjusted inspiratory and expiratory positive airway pressures via nasal mask: physiologic and clinical implications. Chest 1990; 98:317 324 32 Resta O, Guido P, Picca V, et al. Prescription of ncpap and nbipap in obstructive sleep apnoea syndrome: Italian experience in 105 subjects; a prospective two centre study. Respir Med 1998; 92:820 827 2150 Clinical Investigations