REVIEW For reprint orders, please contact: [email protected]

Comparing methods of respiratory event detection during the treatment of obstructive sleep apnea Renewed focus on comparative effectiveness research presents a unique opportunity to develop optimal clinical management pathways for patients with obstructive sleep apnea. With this momentum comes the challenge of measuring treatment effect on sleep-disordered breathing, especially in large, multisite studies. In-laboratory polysomnography, the current gold standard sleep assessment of obstructive sleep apnea severity, is costly and imposes significant participant burden. Alternatives include home unattended sleep testing and overnight pulse oximetry recording. Research studies using positive airway pressure treatment have the additional option of using the information recorded by the patient’s positive airway pressure device to assess treatment effectiveness. Recent research has shown relatively good agreement between manual identification of residual respiratory events in overnight in-laboratory polysomnography and the automatic event detection utilized in positive airway pressure machines. In addition to assessing the effects of interventions on sleep disordered breathing, obstructive sleep apnea-related comparative effectiveness studies need to assess the impact of the interventions on patient burden, cost of therapy, timeliness of care, improved quality of life and other clinically relevant outcomes.

Barry G Fields*1 & Samuel T Kuna2 The University of Pennsylvania & the Philadelphia Veterans Affairs Medical Center, Penn Sleep Center, 3624 Market Street, Suite 205, Philadelphia, PA 19104, USA 2 Veterans Integrated Service Network 4 Regional Sleep Center, Philadelphia Veterans Affairs Medical Center (111P), 3900 Woodland Avenue, Philadelphia, PA 19104, USA *Author for correspondence: Tel.: +1 215 615 4875 Fax: +1 215 615 4874 [email protected] 1

Keywords: apnea–hypopnea index n automatic event detection n obstructive sleep apnea n polysomnogram n positive airway pressure

On 21 May 2012, the US-based Patient-Centered Outcomes Research Institute Board of Governors published its first ‘National Priorities for Research and Research Agenda.’ One of its five major funding priorities is research that assesses treatment options aimed at “creat[ing] a foundation of information for personalized decision-making” [1]. The National Heart, Lung and Blood Institute has identified lung diseases and sleep disorders as potential beneficiaries of this American-based Patient-Centered Outcomes Research Institute-supported research [2]. Studies of effectiveness among treatment options for patients with obstructive sleep apnea (OSA) are already numerous, and many more can be anticipated owing to these new government and institute-based initiatives. Given the expected growth in sleep-related comparative effectiveness research, several questions arise. For instance, what do investigators need to know about assessing OSA treatment effectiveness? How can they compare effectiveness among different treatment modalities? What effect parameters are most meaningful? Although comparing objective measures of OSA treatment – such as residual breathing disruption during sleep – will be the focus of this review, additional considerations when designing a comparative effectiveness study for OSA treatment include assessments of patient burden, cost of care, timeliness of therapy and improved quality of life. ■■ Defining & diagnosing OSA

OSA is characterized by the repetitive closure of the pharyngeal airway during sleep that is associated with oxygen desaturation and/or arousal. Obesity (BMI >30 kg/m2) is the greatest risk factor for OSA. Disease severity is categorized objectively as

10.2217/CER.12.58 © 2012 Future Medicine Ltd

1(6), 489–499 (2012)

part of

ISSN 2042-6305

489

REVIEW  

Fields & Kuna

the apnea–hypopnea index (AHI) – that is, the average number of apneas and hypopneas per hour of sleep. Apneas are defined as a total or near total obstruction of airflow for at least 10 s and hypopneas are defined as a period of shallow breathing for at least 10 s, which is terminated by either arousal from sleep or at least a 3% oxyhemoglobin desaturation [3,4]. By consensus, an AHI ≥5 events/h indicates mild OSA, an AHI ≥15 events/h indicates moderate OSA and an AHI ≥30 events/h indicates severe disease. Patients with OSA present with symptoms of loud snoring, choking or gasping during sleep, and excessive daytime sleepiness. Other symptoms include nocturia, depression, decreased libido and decline in memory and concentration. Patients with OSA are at increased risk of coronary artery disease, chronic heart failure, hypertension, stroke, and motor vehicle and industrial accidents [5–7]. Previous work has shown that at least 2% of middle-aged women and 4% of middle-aged men have OSA syndrome (OSA with daytime sleepiness) [8–10], and more recent epidemiological work shows a prevalence of up to 33% in a population of 20–80 year-olds [11]. This burgeoning public health issue places a significant burden on the healthcare system. OSA is typically diagnosed by overnight polysomnogram (PSG) performed in a sleep laboratory and attended by a PSG technologist. The following signals are usually recorded during this comprehensive evaluation: fourchannel electro­encephalogram; bilateral electro-oculograms; chin muscle activity; oral and nasal airflow detected via oro-nasal thermistor and nasal airway pressure; thoracic and abdominal movement; body position; snoring; one-lead electrocardiogram; and pulse oximetry. Commercially available portable monitors (PMs) can be used to perform unattended PSGs in the patient’s home, recording the same signals as the in-laboratory recording. Over the past decade, home sleep testing (HST) using type 3 PMs with a reduced number of signals has become increasingly popular for the diagnosis and management of OSA. Although these PMs differ in the number and type of signals they record, most lack the PSG channels used to differentiate between wakefulness and sleep (Table 1). The American Academy of Sleep Medicine recommends HST with PMs in certain high-risk populations in conjunction with

490

J. Compar. Effect. Res. (2012) 1(6)

a comprehensive clinical evaluation by a sleep specialist [12]. ■■ Treating OSA

Patients with OSA are treated based on disease severity, clinical symptoms and patient preference. Treatment options include weight loss, positional therapy, oral mandibular-advancement appliances, nasal and pharyngeal airway surgery, and positive airway pressure (PAP). PAP, the first-line and most widely studied treatment, includes continuous positive airway pressure (CPAP), autotitrating positive airway pressure (APAP or auto-CPAP) and bilevel positive airway pressure. In any of these modes, the PAP device delivers pressurized, filtered room air through a tube to a facial mask interface worn during sleep. Working as a pneumatic splint to prevent airway collapse, this highly efficacious treatment is capable of reducing the AHI to c­linically acceptable levels in almost all patients [13]. Most PAP machines identify and record residual respiratory events, including apneas, hypopneas, snoring and air leak from the circuit during machine use. The data can be downloaded to personal computer software via the device’s data chip, or transmitted by wired or wireless modem to the manufacturer’s website. Sleep specialists routinely use this objective information to assess treatment effectiveness; at least 17% of patients on ‘optimal’ PAP therapy show significant residual OSA on their home-based machine downloads, even after having undergone in-laboratory PAP titration [14,15]. However, interpretation based solely on reported residual AHI should be combined with other information provided by the PAP download, including the amount of air leak from the circuit and the presence or absence of snoring. A large air leak in the circuit (quantified slightly differently by each manufacturer) may occur owing to a poor mask fit and can render respiratory event reporting unreliable. Careful attention should also be paid to the presence of snoring on the download report. Persistent residual snoring suggests that the airway is compromised and that a higher a­irway pressure is needed. PAP units also provide precise, objective information on treatment adherence by recording mask-on-time – that is, periods when the breathing circuit is pressurized. Adherence to PAP is often a challenge given the relative inconvenience of having to use the device on a daily

future science group

Comparing methods of respiratory event detection during the treatment of obstructive sleep apnea 

basis and the potential stigma that accompanies its use. Estimates of PAP adherence in the general population vary widely; a recent study estimates that PAP nonadherence (defined as PAP usage of less than 4 h each night) ranges from 46 to 83% of all users [16]. In PAP intervention effectiveness trials, objective measures of adherence are essential to ensure that lack of improvement in clinical outcome is not due to inadequate adherence.

Table 1. Comparison of signals recorded between in-laboratory polysomnogram and most Type 3 portable monitors. Signal

PSG

EEG

X

Type 3 PM

EOG

X

Chin EMG

X

ECG

X

Nasal pressure

X

X

Snoring

X

X

Respiratory event detection during PAP treatment

Respiratory belts

X

X

■■ Computer-assisted manual respiratory event detection

Pulse oximetry

X

X

Body position

X

X

Leg movements

X

The most well-established ‘gold standard’ method of determining treatment effect on sleep-disordered breathing is to perform an inlaboratory, attended PSG. In patients who are being initiated on PAP treatment, sleep technologists modify pressure levels and modes throughout the night until they demonstrate an optimal setting that overcomes the patient’s upper airway obstruction, reducing their AHI to normal or near normal levels. Some research study designs require reassessment of treatment efficacy during and at the end of the intervention. In studies in which a device such as PAP, oral appliance or positional pillow is used for treatment, the PSG can be performed with the patient using that specific treatment. PSGs on PAP should be performed with the patient on the same pressure level and mode of pressure delivery that they use at home. Until recently, manual ‘scoring’ of respiratory events had been based on American Academy of Sleep Medicine (AASM) 2007 criteria [3]. Although a 2012 update has superseded some of those guidelines, their description remains germane given the large body of research retaining 2007 nomenclature and definitions. An apnea was defined as a drop in peak thermal sensor excursion by ≥90% of baseline for at least 10 s, a definition now broadened to include a drop in other airflow signals when thermal sensing is unavailable. Defining a ‘hypopnea’ has undergone more significant revisions. In 2007, the AASM recommended a hypopnea be scored if a nasal pressure signal excursion drop by ≥30% for at least 10 s is accompanied by a ≥4% oxyhemoglobin desaturation. However, they also offered an alternative definition, stating that a hypopnea can be scored if a nasal pressure signal excursion drop by ≥50% for at least 10 s is accompanied

future science group

REVIEW

EOG: Electrooculogram; PM: Portable monitor; PSG: Polysomnogram; X: The presence of.

by a ≥3% oxyhemoglobin desaturation or an EEG-identified arousal from sleep. To complicate matters further, some sleep laboratories (especially those conducting research) still utilize the AASM 1999 ‘Chicago Criteria’ – the most liberal of those mentioned – which defines hypopneas as any >50% decrease in airflow without requiring oxyhemoglobin desaturation or an arousal. Events with lesser airflow reduction can still be scored as hypopneas as long as there is >3% oxyhemoglobin d­esaturation or an associated arousal [17]. The 2012 AASM update attempts to simplify and better standardize hypopnea scoring by providing just one, recommended definition. A respiratory event may now be designated a hypopnea only if: first, peak nasal pressure (diagnostic study off of PAP) or flow (PAP titration) drops by ≥30% of baseline; second, the event lasts for ≥10 s; and third, there is a ≥3% oxygen desaturation from baseline or an arousal [4]. It is imperative that any clinical effectiveness investigators are aware of these recent changes and incorporate them into new trial protocols. Whether this change will lead to improved accuracy and inter-center agreement in AHI d­etermination remains to be seen. Ruehland et al. evaluated the impact of the different hypopnea criteria on AHI determination [17]. A total of 328 PSGs that had been scored using the Chicago Criteria were rescored using the AASM recommended and alternative criteria. A median AHI of 25.1  events/h (moderate OSA) using the Chicago Criteria was reduced to 14.9 events/h (borderline mild–­ moderate OSA) when using the alternative

www.futuremedicine.com

491

REVIEW  

Fields & Kuna

A ASM criteria, and to 8.3 events/h (mild OSA) using the recommended AASM criteria. Although this study focused only on diagnostic PSGs, the same scoring rules applied to PSGs on PAP that are performed to demonstrate respiratory event control. Therefore, sleep-related comparative effectiveness trials should specify a priori the scoring criteria used to score the sleep tests and report these methods in subsequent publications. One way multisite trials have attempted to control site-specific scoring variability is to have one sleep center perform all of the scoring. The level of agreement in sleep stage scoring is better among scorers within the same laboratory than between laboratories [18]. However, centralized scoring comes with its own challenges. It increases the cost of the project by requiring the budgeting of additional resources to the chosen site; it increases turnaround time, delaying timeliness of ana­lysis; and it requires measures be taken to assure secure data transfer. A recent study found that variation in AHI at five academic sleep centers scoring the same PSGs was most closely tied to differences in hypopnea scoring [18]. Agreement was very good when scorers at all centers used the 2007 recommended AASM criteria, with a progressive decrease in inter-site agreement when centers used the AASM alternative and Chicago Criteria. These findings suggest that multicenter comparative effectiveness trials may not need to expend resources on centralized respiratory event scoring, provided each site employs experienced scorers and applies the same AASM criteria for hypopnea detection. The ability of individual sites within a multicenter research study to score their own PSGs to diagnose In-laboratory PSG Home unattended PSG Feasibility Access

Home unattended sleep study (type 3 PM)

Information obtained Cost/revenue

Oximetry PAP data

■■ Automatic respiratory event detection

Figure 1. Continuum of currently available sleep testing modalities. PAP: Positive airway pressure; PM: Portable monitor; PSG: Polysomnogram.

492

patients with OSA and evaluate the effect of study intervention on OSA severity allows the data to be collected under more real-world conditions, thereby enhancing the generalizability of the research findings. Regardless of whether a clinical sleep research study is performed at one or many sites, and regardless of whether it uses a centralized scoring laboratory or allows scoring at each participating site, investigators should conduct a quality assurance program to monitor interscorer and intrascorer agreement. This can be accomplished by recycling de-identified recordings to the scorers throughout the course of data collection and ana­lysis. Intra-class correlation coefficients of the outcome measures of interest can be calculated to assess agreement within and between scorers and sites. In addition, these data can be used to detect a drift in scoring – that is, change in event detection – during the course of a multi-year project. It is likely that a comparative effectiveness research study of OSA management may need to determine the efficacy of treatment intervention during the course of the protocol and at the end of the intervention. The intervention might be PAP treatment but might also be treatment with an oral appliance, weight loss, upper airway surgery, a positional pillow, nasal airway expiratory splints and so on. In-laboratory PSG can be repeated at prespecified time points within the protocol, but it is expensive and imposes significant participant burden. Less expensive and more accessible, but less comprehensive, alternatives are available to assess treatment effects on sleep-disordered breathing. They range from home unattended PSGs to home sleep testing via type 3 PMs and overnight oximetry (Figure 1). An overnight oximetry recording would provide an oxygen desaturation index, the mean number of oxygen desaturation events/h of recording. This single channel recording has many limitations and its use should be restricted to patient populations with a high likelihood that a respiratory event will be associated with an oxygen desaturation, for example, obese adults and patients with cardiopulmonary disorders.

J. Compar. Effect. Res. (2012) 1(6)

If PAP is being used for treatment, the data recorded by the PAP unit can be collected to document treatment efficacy. However, automatic event detection (AED) software algorithms used

future science group

Comparing methods of respiratory event detection during the treatment of obstructive sleep apnea 

to delineate and quantify respiratory events differ markedly from those used in polysomnography. While PSGs furnish airflow, respiratory effort, pulse oximetry and EEG to assist in manual respiratory event scoring, PAP AED only uses an airflow signal. Some PAP machines are equipped with overnight pulse oximetry ports, further refining respiratory event detection. Accuracy of AED-generated respiratory indices, such as the AHI, is further limited by an inability to distinguish wakefulness from sleep. As noted above, the AHI on PSG is calculated as the number of respiratory events per hour of sleep, whereas, similar to Type 3 PMs, PAP devices report this index based on hours of use.

REVIEW

This limitation can underestimate the residual AHI and, therefore, o­verestimate ­t reatment effectiveness. APAP devices have changed the landscape of PAP treatment by adjusting PAP levels automatically over a programmable pressure range based on the presence or absence of: reductions in airflow (hypopnea), absence of airflow (apnea), snoring, inspiratory flow limitation and mask leak (Figure 2). The use of APAP units for treatment of OSA obviates the need to perform an in-laboratory PSG PAP titration study to determine the fixed pressure needed for CPAP treatment. Adopting this strategy in research studies can reduce participant burden and study-related

Pressure (cmH2O)

Mode: auto CPAP with C-flex

Auto CPAP

Min CPAP setting

90% pressure: 11.5

Max CPAP setting

20

Average CPAP pressure: 8.9

15 10 5 0

0

1

2

3

4

5

6

7

8

9

10

11

Sleep therapy flags

Indices 1.0% of night in PB CA: 0.1 OA: 1.5 H: 2.0 FL: 0.8 VS: 16.8 RE: 1.0 AHI: 3.6

PB CA OA H FL VS RE 0 1 Total leak (LPM) Normal mask fit

2

3

4

Breathing not detected

5

6

Large leak

7

8

9

10

11

Total leak

Min in large leak 0.0 mins

LL 120 100 80 60 40 20

% of night in large leak 0% of night Average leak 29.0 0

1

2

3

4

5

6

7

8

9

10

11

Figure 2. Typical summary data download from an autotitrating positive airway pressure device. (A) Variable positive airway pressure throughout the night predicated on residual respiratory event sensing; (B) hash marks indicating detection of various residual respiratory events with overall apnea–hypopnea index indicated (3.6 events/h); (C) mask leak indicator. AHI: Apnea–hypopnea index; CA: Clear airway apnea; CPAP: Continuous positive airway pressure; FL: Flow rate; H: Hypopnea; OA: Obstructed airway apnea; RE: Respiratory event; VS: Vibratory snore.

future science group

www.futuremedicine.com

493

REVIEW  

Fields & Kuna

cost. Recent Practice Parameters advocate use of APAP for initial treatment of OSA in some patient populations [19]. The APAP unit’s reliance on its AED to make pressure changes in real time has underscored the need to establish the accuracy of its AED algorithms, especially since those algorithms vary somewhat among production companies. One study in Veterans with OSA (n = 70) using the RemStar® Auto device (Philips-Respironics, PA, USA) found reasonable correlation between APAP-generated AHI and the AHI from a simultaneous, manually scored PSG (r = 0.74). This correlation was strongest among those patients with less residual OSA (AHI

Comparing methods of respiratory event detection during the treatment of obstructive sleep apnea.

Renewed focus on comparative effectiveness research presents a unique opportunity to develop optimal clinical management pathways for patients with ob...
2MB Sizes 0 Downloads 0 Views