| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Pulmonary Division, University Hospital of Zurich, Zurich, Switzerland.
Correspondence to: Konrad E. Bloch, MD, FCCP, Pulmonary Division, University Hospital of Zurich, Rämistrasse 100, 8091 Zurich, Switzerland; e-mail: pneubloc{at}usz.unizh.ch
Abstract
Objectives: Treatment of obstructive sleep apnea syndrome (OSA) is often delayed because polysomnography, the recommended standard diagnostic test, is not readily available. We evaluated whether the diagnosis of sleep apnea could be inferred from the response to a treatment trial with nasal continuous positive airway pressure (CPAP).
Design: Study on diagnostic accuracy.
Setting: Sleep-disorders clinic of a university hospital.
Patients: Seventy-six sleepy snorers consecutively referred for sleep apnea evaluation.
Interventions: CPAP treatment trial over 2 weeks as an initial diagnostic test in comparison with polysomnography, and treatment success over 
4 months.
Measurements and results: The main outcome was diagnostic accuracy of the CPAP trial. The trial result was positive if the patient had used CPAP for > 2 h per night and wished to continue therapy. This suggested sleep apnea. The trial was evaluated in terms of predicting an obstructive apnea/hypopnea index (AHI) > 10/h during polysomnography performed for validation, and in terms of identifying sleep apnea patients treated successfully over 4 months. Forty-four of 76 patients (58%) had sleep apnea as confirmed by an AHI > 10/h. The CPAP trial predicted sleep apnea with a sensitivity of 80%, a specificity of 97%, and positive and negative predictive values of 97% and 78%, respectively. In 35 of 76 sleep apnea patients (46%) with positive CPAP trial results, polysomnography could have been avoided. These patients were prescribed long-term CPAP therapy. After 4 months, 33 of 35 patients (94%) still used CPAP, and their symptoms remained improved. These patients were identified by the CPAP trial with positive and negative predictive values of 92% and 100%, respectively.
Conclusions: In a selected population, a CPAP trial may help to diagnose OSA, to identify patients who benefit from CPAP, and to reduce the need for polysomnography.
Key Words: continuous positive airway pressure • diagnosis • polysomnography • sleep apnea • treatment
The obstructive sleep apnea syndrome (OSA) is a common disorder affecting from 2 to 26% of adults, depending on criteria for syndrome definition, age, sex, and other factors that affect prevalence estimates.12 OSA causes excessive sleepiness, impairs quality of life, and represents a risk for traffic accidents and cardiovascular diseases.34 Polysomnography is considered the "gold standard" for the diagnosis of OSA,56 but it is expensive, is not readily available,7 and has other shortcomings. For example, measures derived from polysomnography, eg, the apnea/hypopnea index (AHI), correlate poorly with major consequences of OSA such as sleepiness and cognitive impairment,8 and do not reliably predict the response to the standard therapy for OSA, nasal continuous positive airway pressure (CPAP).910 Therefore, and in the face of large numbers of patients awaiting polysomnography and OSA therapy, alternatives to the conventional diagnostic and treatment algorithms are desirable.7
Since CPAP is the first-line therapy for OSA, has no serious side effects, and because satisfaction with CPAP during a 2-week trial correlates with severity of disease, and successful treatment after 1 year,11 we reasoned that the response to 2 weeks of empiric CPAP treatment might identify patients with OSA. Therefore, we evaluated a 2-week CPAP trial instead of polysomnography as the initial diagnostic test for patients with suspected OSA. Our hypothesis was that the diagnosis of OSA would be established without requirement for a polysomnography in patients presenting with typical symptoms who responded favorably to a CPAP trial. Patients without a clear benefit from a CPAP trial would require polysomnography to evaluate potential OSA, and alternative diagnosis.
The accuracy of the CPAP trial was evaluated in two ways: (1) by comparison to the results of polysomnography performed in all patients for validation purposes, and (2) by comparison to the clinical outcome of OSA patients after 4 months of treatment with CPAP following a positive trial result. We expected that the CPAP trial would provide effective and timely treatment for the most symptomatic OSA patients and reduce the need for polysomnography.
Materials and Methods
Patients
Successive patients referred by primary care physicians to the sleep-disorder center for evaluation of suspected OSA were enrolled if they were habitual snorers, complained of daytime sleepiness, had an Epworth sleepiness scale (ESS) score 8,1213 and gave informed consent to the study protocol, which was approved by the hospital ethics committee. Exclusion criteria were as follows: a contraindication for CPAP or auto-adjusting CPAP (ie, unstable congestive heart failure, significant lung disease, obesity hypoventilation syndrome)1415; significant nasal obstruction; a previous diagnosis of any sleep disorder and previous CPAP treatment; a diagnosis of an internal medical, neurologic, or psychiatric disorder explaining some of the symptoms; and inability to understand German.
Protocol and Measurements
Figure 1
illustrates the study protocol. Baseline evaluation was a medical history and physical examination, evaluation of sleepiness by the ESS,16 other symptoms by the questionnaire of Kump et al,17 and quality of life (Short-Form health-related quality-of-life questionnaire [SF-36].18 Vigilance was measured by a behavioral maintenance of wakefulness test (OSLER test; Stowood Scientific Instruments; Oxford, UK).1920 It consisted of 40-min sleep resistance challenges performed in a quiet, dark room at four occasions on 1 day. The mean sleep resistance time of the four tests was determined. During a driving simulator test (Steer Clear; Sheffield, South Yorkshire, UK),21 patients watched a computer screen showing steers crossing a road at irregular intervals over a 30-min test. Using a keyboard that controlled a car displayed on the screen, the patients had to try to avoid hitting the steers. The computer counted the percentage of cows hit.
|
At the end of the 2-week trial, patients returned to the clinic. The ESS was administered, and the perceived effect of CPAP treatment was assessed on a Likert scale extending from 1 (very disturbing, no benefit) to 5 (excellent benefit). CPAP use over the 2-week period and the AHI (events per hour of CPAP use) measured and stored by the CPAP device were downloaded.
The outcome of the CPAP trial was determined by the response to the following two principal questions: "are you willing to continue CPAP treatment?" and "was the measured average CPAP use > 2 h per night?"11 A positive answer to both questions (ie, a positive trial result) was considered suggestive for OSA, and long-term CPAP therapy could be established without the need for polysomnography (Fig 1). A negative response to one or both questions (ie, a negative trial result) was considered to require polysomnography and additional workup to evaluate an alternative diagnosis. To assess the effect of the CPAP trial in greater detail, the 15-point response scale illustrated in Figure 2
was also employed. We have selected a cutoff of > 2 h of minimal CPAP use during the 2-week trial since this was found to be predictive of successful therapy at 1 year in a previous study.11 In an additional analysis, diagnostic accuracy of the CPAP trial was evaluated with a minimal average nightly CPAP use required for a positive response of 4 h.
|
10 s525 without an oxygen desaturation criterion since the latter significantly depends on the type of pulse oximeter used.26 Central apnea was differentiated from obstructive apnea/hypopnea by the absence of chest-wall motion during the event.27 A patient was classified as having OSA or central sleep apnea syndrome depending on the more prevalent type of event. For central sleep apnea syndrome, > 10 central and < 10 obstructive events per hour were required. Polysomnographic studies were analyzed by an investigator not involved in the clinical care of the patients; this investigator did not reveal the results to the patients or to the attending physicians. Only after the outcome of the CPAP trial had been assessed by the physician and the patient were they were unblinded to the results of polysomnography.
In a second analysis, the ability of the CPAP trial to identify OSA patients with successful CPAP therapy over 4 months was evaluated. In patients with a positive trial result and OSA confirmed by polysomnography (obstructive AHI > 10/h) long-term CPAP therapy was prescribed. Four months after initiation of treatment, reassessment included the response scale (Fig 2), subjective treatment benefit, the ESS score, mean CPAP use and the AHI measured by the CPAP device over the last month, and a driving simulation test. Successful treatment was defined by 4 points on the response scale.
Data Analysis
Normally distributed data are presented as means (SD). Nonnormally distributed data and categorical variables are presented as medians (quartile ranges)2829; t tests or Mann-Whitney U tests were performed to compare means and medians. Repeated assessments were evaluated by analysis of variance; p < 0.05 was considered significant.
A mean number of > 10 obstructive apnea/hypopnea events per hour of sleep (AHI > 10/h) was defined as evidence of OSA. Sensitivity and specificity of the CPAP trial for identification of OSA, and the positive and negative predictive values were the primary outcome measures. Since the upper limit of normal for the AHI is not clearly established, diagnostic performance of the CPAP trial in identifying patients with > 5 and > 15 obstructive apnea/hypopnea events per hour were also analyzed. Diagnostic accuracy was assessed by receiver operating characteristic (ROC) curves.30
Results
Baseline Patient Characteristics
Of 195 successive patients referred for evaluation of sleep apnea, 76 patients (39%, 15 women and 61 men) fulfilled inclusion criteria and participated in the study. Seven patients (4%) refused participation, 10 patients (5%) had to be excluded because of inability to understand German, and 102 patients (52%) had one or several medical exclusion criteria (known sleep disorder or previous CPAP therapy, medical, neurologic or psychiatric disorder explaining some of the symptoms).
The 76 patients participating in the trial were 52 ± 12 years old. They were moderately overweight (body mass index [BMI], 31.7 ± 6.0 kg/m2) and had excessive sleepiness according to the selection criteria. The ESS score was 13.6 ± 3.7. The vitality domain of the SF-36 was reduced. There was a broad range of the AHI (0.8 to 119.3/h). Sleep resistance time was moderately reduced (OSLER test, 31.8 ± 9.7 min; range, 7.5 to > 40 min).
Diagnostic Performance of the CPAP Trial Compared to Polysomnography
All 76 patients completed the follow-up examination 2 weeks after recruitment. If an obstructive AHI > 10/h was assumed to define OSA, 44 of the 76 patients (58%) had the disease. The CPAP trial outcome assessed by answering the two questions, "are you willing to continue CPAP treatment?" and "was the average CPAP use > 2 h per night?" (ie, a cutoff level of 4 points on the response scale; Fig 2) had a favorable diagnostic performance: there were 35 true-positive, 31 true-negative, 9 false-negative, and 1 false-positive responses (Fig 1). This corresponded to a sensitivity of 80%, a specificity of 97%, and positive and negative predictive values of 97% and 78%, respectively. Patients with positive and negative CPAP trial results (cutoff 4 points on the response scale), and positive and negative polysomnography (cutoff obstructive AHI > 10/h), respectively, revealed similar baseline characteristics, with the exception of a slightly greater BMI and neck circumference in patients with an obstructive AHI > 10/h vs those with 
10/h (Table 1
). The reported side effects of CPAP therapy were minor and included skin irritation, dry mouth, and nasal irritation. The prevalence of side effects was similar in patients with positive and negative CPAP trial results.
|
Of the nine patients with false-negative results, only two patients were willing to continue CPAP treatment after repeated instruction and a trial over another 2 weeks with a fixed mask pressure. The seven remaining patients were informed on other treatment modalities for OSA. Three patients decided to proceed with weight reduction, and four patients decided against any further treatment.
In one patient, the CPAP trial result was false-positive, ie, the score on the response scale was 5, whereas the obstructive AHI during polysomnography was only 2/h. This patient had subjectively rated the effect of CPAP as a 3 on the 5-point Likert scale, and her ESS score had improved by only 1 point (from 17 to 16). This patient liked CPAP treatment mainly because it improved her socially disturbing snoring. Based on clinical assessment and polysomnography, a diagnosis of poor sleep hygiene was made.
Figure 3 displays the sensitivity and specificity of the CPAP trial for identification of OSA at various cutoff points of the response scale. Diagnostic accuracy expressed as the area under the ROC curve was 0.93 (95% confidence interval, 0.87 to 0.99) for the criterion obstructive AHI of > 10/h. The respective areas under the ROC curves for obstructive AHI of > 5/h and > 15/h were 0.92 (95% confidence interval, 0.85 to 0.98) and 0.86 (95% confidence interval, 0.77 to 0.95), respectively. Diagnostic accuracy was not significantly altered if the minimal mean hours of nightly CPAP use that defined a positive response was increased from > 2 h to > 4 h: the corresponding area under the ROC curve was 0.92 (95% confidence interval, 0.85 to 0.98, p = not significant vs 2 h of nightly CPAP use) for the criterion AHI > 10/h.
|
|
4 points on the response scale; Fig 2). The sensitivity and specificity were 100% and 93%, and the positive and negative predictive values were 92% and 100%, respectively. Evaluation of sensitivities and specificities at various cutoff points of the response scale for assessment of the CPAP trial revealed an area under the ROC curve of 0.99 (95% confidence interval, 0.98 to 1.00; Fig 3, bottom, B). For comparison, the accuracy of polysomnography in prediction of the same outcome (successful CPAP therapy at 4 months) is also illustrated by the corresponding ROC curve in Figure 3, bottom, B. An obstructive AHI > 10/h during polysomnography had a sensitivity of 100%, a specificity of 74%, and positive and negative predictive values of 75% and 100%, respectively. The proposed approach comprising the CPAP trial as the first diagnostic step would have allowed to correctly diagnose OSA in 35 of 76 patients (46%) without polysomnography. In addition, the CPAP trial was highly accurate in predicting OSA patients successfully using CPAP therapy for at least 4 months (ie, 33 of the 35 OSA patients, 94%).
Discussion
We evaluated a novel approach to the diagnosis of patients with clinically suspected OSA. This approach is based on observing the response to a CPAP treatment trial over 2 weeks as the initial diagnostic step rather than on polysomnography, the standard test. The outcome of the CPAP trial was assessed in a simple way by the response to two questions reflecting the benefit perceived by the patients, their propensity to continue treatment, and the objectively measured compliance with CPAP during the trial. The CPAP trial had a high diagnostic accuracy in predicting an increased AHI during polysomnography (Fig 3, top, A). It was even more accurate in identification of OSA patients with successful CPAP therapy over several months (Fig 3, bottom, B). Our data indicate that a 2-week CPAP trial helps to diagnose OSA, promptly relieves symptoms in a considerable proportion of patients (35 of 76 patients, 46%, in our study, who would otherwise await polysomnography untreated; Fig 1), and reduces the need for sleep studies. The CPAP trial runs against current thinking in which we usually diagnose OSA and justify treatment based on objective diagnostic testing, but it is based on the same sort of approach we use for other prevalent and treatable conditions such as certain infectious diseases, or gastroesophageal reflux disease.
We have evaluated performance of the CPAP trial against polysomnography, the current standard diagnostic test for OSA.31 The lack of a clearly established upper limit of the AHI, one of the limitations of polysomnography, did not affect our results: independent of the selected criterion for an abnormal polysomnography result (ie, AHIs > 5/h, >10/h, or > 15/h), the diagnostic accuracy of the CPAP trial was similarly high with areas under the ROC curves of 0.86 to 0.93 (Fig 3, top, A). Additional ROC analysis confirmed that the diagnostic accuracy of the CPAP trial was not altered if the minimal average nightly CPAP use required for a positive response was increased from 2 to 4 h. Therefore, the CPAP trial seems to be a robust means to identify OSA.
We further evaluated the utility of a CPAP trial for predicting a clinically relevant outcome, ie, successful CPAP therapy of OSA patients over at least 4 months. We found favorable positive and negative predictive values of 92% and 100%, respectively. In 33 of the 35 patients with a positive CPAP trial result, a high treatment adherence was objectively documented over at least 4 months, and they experienced a persistent major benefit in terms of improved sleepiness, quality of life, and driving simulation performance (Table 2). Therefore, a CPAP trial is a pragmatic way to identify OSA patients with "CPAP-responsive disease"32 who benefit from long-term CPAP treatment. This is consistent with a previous study11 suggesting that treatment satisfaction and CPAP use during a 2-week home trial predicted patients who complied with CPAP treatment at 1 year. In this regard, polysomnography was less accurate with positive and negative predictive values of 100% and 73%, respectively (Fig 3, bottom, B).
These results were achieved in patients with a high pretest probability of OSA, and may not be extrapolated to an unselected population. According to the selection criteria applied in this study, the proposed approach to the diagnosis of OSA was found to be applicable in 83 of 195 consecutively referred patients (43%) [ie, in the 76 participating patients plus the 7 patients who refused participation], whereas the remaining patients had either a medical condition (102 of 195 patients, 52%) such as heart failure or significant lung disease, or language problems (10 patients, 5%) that prevented application of a CPAP trial.
In patients with a negative CPAP trial result, polysomnography is mandatory according to the proposed algorithm to evaluate OSA and other sleep disorders not responding to CPAP (Fig 1). Nevertheless, even in some of the negative cases, ie, the false-negative results with an elevated AHI (9 of 40 patients with negative results, 30%), the CPAP trial provides useful information since CPAP would be prescribed as the first-line therapy according to standard practice.9 Patients with central sleep apnea were classified as true-negative since CPAP, although effective in some cases, is not clearly established as the standard therapy for this disorder. In patients with true-negative results (31 of 76 patients, 41%), a CPAP trial may or may not cause a delay in diagnosis and treatment, depending on the waiting time for a sleep study. Since waiting times for polysomnography of several months are not unusual as emphasized recently,7 it is unlikely that a 2-week CPAP trial performed directly after the initial consultation would significantly delay the diagnostic workup.
A potential problem with the proposed diagnostic strategy might arise from false-positive cases, ie, patients with a subjectively favorable effect of CPAP in the absence of polysomnographically documented breathing disturbances (one snorer in the current study). Whether this represented a placebo effect or a therapeutic effect of CPAP on subtle sleep-related breathing disturbances that were undetected by standard polysomnography (upper airway resistance syndrome)33 remains unanswered. A similar diagnostic uncertainty remains with two patients deemed true-negative and suffering from periodic limb movements that has been associated with upper airway resistance.34 While our study provides data on diagnostic performance of a CPAP trial, it was not designed to determine its economic implications, an aspect that warrants future evaluation.
The risks and costs of prescribing CPAP to occasional patients experiencing a placebo effect during a CPAP trial has to be weighted against the consequences of untreated OSA in a large number of patients awaiting polysomnography. Untreated OSA represents a risk for affected individuals and the community by the associated traffic accidents, high BP, and other cardiovascular complications. Furthermore, delaying the diagnosis deprives patients from improvement in sleepiness, quality of life, and optimal professional performance. These points strongly emphasize the need for development of strategies that facilitate and expedite the diagnosis and effective therapy in OSA patients in the face of the limited capacity for performing polysomnography. By providing diagnostic information and treatment at the same time, the proposed novel approach to the evaluation of suspected OSA fulfills this requirement.
Footnotes
Abbreviations: AHI = apnea-hypopnea index; BMI = body mass index; CPAP = continuous positive airway pressure; ESS = Epworth sleepiness scale; OSA = obstructive sleep apnea syndrome; ROC = receiver operating characteristic; SF-36 = Short-Form health-related quality-of-life questionnaire
Unconditional grant support was provided by the Lung Leagues of Zurich and Schaffhausen, Switzerland.
Received for publication May 4, 2005. Accepted for publication June 9, 2005.
References
This article has been cited by other articles:
|
|
 |
|
  B. Phillips Improving Access to Diagnosis and Treatment of Sleep-Disordered Breathing Chest, November 1, 2007; 132(5): 1418 - 1420. [Full Text] [PDF]
|
|
|
|
 |
|
 C. F. Ryan, A. T. Mulgrew, and N. T. Ayas Ambulatory Management of Obstructive Sleep Apnea without Polysomnography Ann Intern Med, September 4, 2007; 147(5): 350 - 351. [Full Text] [PDF]
|
|
|
|
|
|
R. Thurnheer and N. A. Collop Diagnosis of obstructive sleep apnea: alternatives to polysomnography. Chest, November 1, 2006; 130(5): 1625 - 1626. [Full Text] [PDF]
|
|
|
|
 |
|
 P. Sikka Response to a 2 week continuous positive airway pressure trial accurately identified patients with obstructive sleep apnoea syndrome Evid. Based Med., October 1, 2006; 11(5): 152 - 152. [Full Text] [PDF]
|
|
|
|
 |
|
 Diagnosis of Sleep Apnea by Response to a Therapeutic Trial Journal Watch (General), February 3, 2006; 2006(203): 2 - 2. [Full Text]
|
|
|
|
|
|
N. A. Collop Blue Light Special on CPAP, Aisle 11 Chest, January 1, 2006; 129(1): 6 - 7. [Full Text] [PDF]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
