An algorithm including body mass index, neck size, and excessive daytime sleepiness is a viable tool for predicting likely candidates for split-night protocols
The sleep-mediated narrowing or actual collapse of the upper airway constitutes the core pathophysiological event for most sleep-related breathing disorders. Apneas and hypopneas frequently result in episodes of hypoxemia as well as brief arousals and awakenings from sleep. The repeated occurrence of these events results in a spectrum of symptoms and signs that have come to be recognized in sleep medicine as significant because of the morbidity and mortality associated with them. Included in this symptom profile is the significant behavioral morbidity caused by sleep-disordered breathing.1 Specifically, excessive daytime sleepiness (EDS) constitutes an important presenting symptom among affected populations. In addition, it is known that obesity represents a strong risk factor in the presentation of sleep-related breathing disorders (SRBD), in particular as an increasing body mass index (BMI) and neck size (NS) are likely to predict worsening in the severity of the condition.2
The BMI and NS are easily quantifiable and for many serve to increase the suspicion of the disorder. More recently, self-report measures of daytime sleepiness have been validated and help identify individuals with significant sleepiness levels. In particular, the excessive daytime sleepiness scale of the sleep wake-activity inventory (SWAI) has been shown to be valuable in both clinical and epidemiological studies.3,4
The availability of pulse oximetry represents an additional source of information to determine the frequency and severity of oxygen fluctuations. In fact, pulse oximetry has been proposed as a potential screening tool for the identification of cases with SRBD, in particular as reasonable correlations have been demonstrated between oxygen desaturation events (>3%) and the polysomnographically derived apnea/hypopnea index.5 However, concerns about its relatively low sensitivity, or low specificity (depending on the algorithm being used for the interpretation of the pulse oximetry), have prompted caution on how to best use this clinical information prior to the sleep laboratory evaluation. However, it is increasingly recognized that pulse oximetry, when combined with other clinical data, can help identify those patients more likely to be affected by SRBD.6 The currently available pulse oximetry equipment is easily portable and many devices have sufficient memory to store an entire nights oxyhemoglobin saturation and pulse readings.
The purpose of this study was to identify the best possible algorithm to enable the identification of patients more likely to present with significant SRBD. The availability of such an algorithm could more reliably allow clinicians to expedite the polysomnographic evaluation as well as avoid delays in the implementation of a successful therapeutic modality. Specifically, it would enable the identification of patients better suited for split-night protocols.
The development of the algorithm was based on clinical experience, which has previously been documented in the literature.1-6 The following parameters were judged to be predictors of a positive polysomnographic diagnosis of SRBD: BMI of >32, NS of >17 inches, an EDS score on the SWAI of <50, and overnight pulse oximetry with an index of >3% desaturation events of >10. These parameters were defined a priori and prospective collection of data was done for all patients being evaluated at the clinic who also completed diagnostic overnight polysomnography. The data on BMI, NS, and EDS scale of the SWAI were gathered at the time of initial consultation. All initial consultations were done by, or under the supervision of, a board-certified sleep medicine specialist. Those patients who were judged to require polysomnographic evaluation were scheduled for sleep laboratory testing. All polysomnographic evaluations consisted of an 8-hour time in bed recording. Electrode placements included unipolar monitoring of the central and occipital electroencephalograms (EEG), electro-oculograms (EOG), and submental electromyogram (EMG). An electrocardiogram (ECG), a tibialis anterior EMG to monitor for leg movements, a position monitor, and a snoring microphone were also used. Respiration was monitored with a thermistor at the nose and mouth to detect airflow and by thoracic and abdominal strain belts to detect respiratory effort. Oximetry was recorded using a finger oximeter worn on the patients first or second digit. The data collected from the oximeter were downloaded and analyzed to determine the number of >3% desaturation events. All oximetry data included in this report were collected during the polysomnographic evaluation.
The data for the present report were collected on all consecutive patients evaluated for a possible diagnosis of SRBD at the clinic from July 1999 to April of 2000. A total of 434 patients were evaluated (296 males and 138 females). The presenting chief complaint of this group of patients was snoring (37%), excessive daytime sleepiness (30%), witnessed stop breathing episodes (27%), insomnia (5%), and other (1%). The average age was 48 ± 13 years and the groups BMI was 35 ± 8. The data set utilized in the present analysis included only those patients on whom all dependent variables were available.
The sensitivity, specificity, predictive value (PV) negative, and predictive value (PV) positive were calculated for the models that were hypothesized to be of potential interest at the outset of data collection. The models were: 1) BMI and NS; 2) BMI, NS, and EDS; 3) BMI, NS, and pulse oximetry; and 4) BMI, NS, EDS, and pulse oximetry (Table 1). The apnea/hypopnea index, which was derived from the polysomnographic evaluation, was used in order to determine the viability of these models. An apnea/hypopnea index of >10 was utilized as the cutoff to identify positive cases.
|BMI, NS||BMI, NS, EDS||BMI, NS, Ox||BMI, NS, EDS, Ox|
The results of this study corroborate the viability of using clinically derived algorithms to establish best practice management strategies for patients with SRBD. This is clearly indicated as the prevalence of these conditions continues to increase the demand for sleep laboratory services. Community-based studies have found that about 2% of women and 4% of men between 30 and 60 years old are affected by obstructive sleep apnea/hypopnea syndrome.7 Data on the prevalence of the disorder among other age groups have not been clearly established, but it is believed that at least a similar proportion of the general population meet minimal diagnostic criteria. Furthermore, the prevalence of upper airway resistance syndrome is not known but is likely to affect a higher proportion of the population.
These estimates of the prevalence of SRBD, and the growing public awareness of the importance of sleep disordered breathing, have resulted in an increased demand for sleep laboratory services. Many patients are usually evaluated for diagnostic purposes with a full night of polysomnography. Then, those patients who elect continuous positive airway pressure (CPAP) therapy are eligible for a second full night at the laboratory for CPAP titration purposes. The labor-intensive nature of the diagnostic and therapeutic sleep laboratory studies has resulted in patient waiting lists and delays in the diagnosis and/or treatment of the condition in many places. As a result of this situation, some experts have called for a combined diagnostic and treatment effort during the polysomnographic evaluation. This approach, known as the split-night protocol, represents a recognized alternative strategy to reduce health care costs and patient inconvenience. The split-night protocol calls for technicians to initiate CPAP titration studies during the second portion of the polysomnographic evaluation. This approach requires technicians to initiate CPAP titration if a diagnosis of obstructive sleep apnea/hypopnea is evident during the first portion of the night. While this strategy seems to be effective in some specialized centers,8 split-night protocols might be problematic for facilities with limited resources or more complex patient populations. In addition, split-night protocols might yield limited diagnostic information and incomplete CPAP titration.9 Also, it is not infrequent that the patient is more willing to consider CPAP titration after reviewing the diagnosis with their treating physician. In this context, the availability of a viable algorithm for the detection of patients more likely to have a positive polysomnographic test is highly desirable. Such an algorithm would enable the treating physician to predict those patients who are more likely to require early therapeutic intervention, ideally in the form of prompt CPAP titration and therapy. Thus, these patients can be oriented in advance not only about the likely diagnosis but also about the indications for CPAP therapy. Such a strategy is likely to result in a reduced number of CPAP failures and improved patient satisfaction and enable more efficient utilization of the sleep laboratory.
The elements that constitute the algorithms used in the present study are inexpensive, are easy to administer, and impose minimal discomfort on the patient.
Despite these features, however, one should be careful not to mistakingly use these algorithms for screening purposes, in particular as the sensitivity of all the proposed models is relatively low. The present results indicate that the algorithm including BMI, NS, EDS, and pulse oximetry represents a viable tool for the prediction of the most likely candidates for split-night protocols. It is proposed that those patients who are attending a sleep medicine consultation and have a BMI of >32, NS of >17 inches, and EDS score of <50 be given an ambulatory oximeter to take home for overnight recording. Those patients who have an index of over 10 desaturation events (3% or higher) are those with whom the physician should discuss the split-night protocol. Based on the results of this study, it would be predicted that this strategy will likely yield a very low number of patients for whom the diagnosis is not confirmed. The implementation of this algorithm is likely to result in potentially significant cost savings; the need for reevaluation would virtually be eliminated for those patients who are mistakenly chosen for a split-night protocol, thus enabling them to benefit from a full night of polysomnography. In addition, those patients identified as likely candidates for a split-night protocol can be prioritized in order to avoid delays for those who require early intervention.
Leon Rosenthal, MD, is a staff physician at Sleep Medicine Associates of Texas, Dallas. Formerly, he was the medical director of the Sleep-Disorders Center, Henry Ford Hospital, Detroit.
The data presented in this report were collected at the Henry Ford Sleep Disorders Center, Detroit. The collaboration of Peter Guido, MD, Ryan Day, Mary Lou Syron, and Joe Fortier made this work possible.
1. Day R, Gerhardstein R, Lumley A, Roth T, Rosenthal L. The behavioral morbidity of obstructive sleep apnea. Progress in Cardiovascular Diseases. 1999;41:341-354.
2. Rosenthal L, Bishop CH, Guido P, et al. The sleep/wake habits of patients diagnosed as having obstructive sleep apnea. Chest. 1997;11:1494-1499.
3. Rosenthal L, Roehrs T, Roth T. The sleep/wake activity inventory: a self-report measure of daytime sleepiness. Biol Psychiatry. 1993;34: 810-820.
4. Johnson E, Breslau N, Roth T, Roehrs T, Rosenthal L. Psychometric evaluation of daytime sleepiness and nocturnal sleep onset scales in a representative community sample. Biol Psychiatry. 1999;45:764-770.
5. Rosenthal L, Moyles T, Zorick F, et al. Hypoxemia and sleep fragmentation in sleep apnea. Sleep Research. 1988;17:24.
6. Jamieson A, Becker P. Snoring: its evaluation and treatment. Hospital Medicine. 1996.
7. Young T, Palta M, Dempsey J, et al. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328:1230-1235.
8. Sanders M, Costantino J, Strollo P, et al. The impact of spilt-night polysomnography for diagnosis and positive pressure therapy titration on treatment acceptance and adherence in sleep apnea/hypopnea. Sleep. 2000;23:17-24.
9. Strollo P, Sanders M, Costantino J, et al. Split night studies for the diagnosis and treatment of sleep-disordered breathing. Sleep. 1996;19:S255-S259.