Chung and colleagues are to be commended for simplifying the screening process for obstructive sleep apnea (OSA) by developing the STOP questionnaire using an apnea hypopnea index (AHI) derived during polysomnography as the “gold standard” for OSA.1Having recently used the AHI in evaluating high-resolution pulse oximetry as a perioperative screening tool for OSA in surgical patients, we would like to draw attention to several limitations of the AHI.

In their methods, Chung and colleagues score the AHI according to “standard criteria,” referencing the 1999 recommendations of the American Academy of Sleep Medicine Task Force on definitions and measurement techniques during polysomnography. Unfortunately, the American Academy of Sleep Medicine stated in the revised Manual for Scoring Sleep and Associated Events  as recently as 2007 that there is as yet no consensus on scoring the AHI.2Redline’s group has shown that the median value of AHI rendered by polysomnography can vary as much as 10-fold, depending on the definition of hypopnea used in scoring the AHI.3Manser and colleagues demonstrated that three different but acceptable scoring systems yielded significantly different disease prevalence and severity estimates.4Differences in signal averaging times of pulse oximeters used during polysomnography can also affect the AHI, and standardization of oximetry techniques during polysomnography remain lacking.5It is thus possible that Chung’s predictive parameters for disease severity (table 6) may “shift” significantly between severity classifications if correlated with AHI measured in other centers. This may partly explain the decline in specificity and positive predictive value of all three tools (STOP, Berlin, and American Society of Anesthesiologists Screen) with increasing severity of disease.6More clarity could be brought to the severity data by presenting nonoverlapping categories (5 < AHI < 15 < AHI < 30 < AHI) and including sample sizes. Were the low specificity and positive predictive value demonstrated a result of small sample size in the moderate and severe categories, or do these findings simply corroborate the low specificities from the other studies cited? While we agree that high sensitivity is the primary goal of a screening tool, specificity (and positive predictive value in a low-prevalence population) must also be considered in this setting, as extensive monitoring of patients mistakenly identified as being at higher risk of complications is expensive, and leads to reduced vigilance by postoperative caregivers.

Lastly, the inability to demonstrate an increase in postoperative complications in patients with severe OSA may not be entirely attributed to their postoperative intensive care monitoring, or to ambiguity in defining AHI thresholds. The AHI is simply the sum of the number of apneas and hypopneas during polysomnography, averaged per hour of total sleep time. It provides no information as to the events’ duration, magnitude or rate of desaturation, adequacy of ventilation recovery in response to apneas, or the level and stability of the arousal threshold during sleep. For example, Patient A shows 8 apneas per hour, a high arousal threshold in response to hypoxemia, a mean event duration of 50 s, and a rapid desaturation rate of 0.8% per second (falling to a mean Spo2nadir of 56%). This patient has an AHI of 8, and is deemed to have “mild” OSA. In contrast, patient B shows 35 hypopneas per hour, with a very low arousal threshold such that each hypopnea is reversed by an electroencephalographic arousal, without oxygen desaturation. Patient B is calculated to have an AHI of 35 (“severe” OSA). Yet patient A is at much higher risk of a postoperative respiratory complication than patient B, even though the AHI suggests the inverse. The AHI may lack the discriminating power to stratify risk in the postoperative environment, where, unlike the sleep laboratory, patients’ respiratory function is challenged by opiates, sleep deprivation, and rapid eye movement sleep rebound. Even sleep specialists are weary of relying on the AHI as a surrogate of disease severity.7 

The STOP questionnaire is a practical step forward in identifying patients with OSA (AHI > 5). It will require refinement and supplementation before we have a tool that can quantify risk, guide postoperative monitoring, and predict outcome in our expanding population of surgical patients with OSA.

*Medical University of South Carolina, Charleston, South Carolina.

Chung F, Yegnewswaran B, Liao P, Chung SA, Vairavanathan S, Islam S, Khajehdehi A, Shapiro CM: STOP questionnaire. A tool to screen patients for obstructive sleep apnea. Anesthesiology 2008; 108:812–21
Iber C, Ancoli-Israel S, Cheeson A, Quan SF: For the academy of Sleep Medicine. The AASM manual for the Scoring of Sleep and Associated Events. Rules, Terminology and Technical Specifications. 1st edition. Westchester, Illinois, American Academy of Sleep Medicine, 2007
For the academy of Sleep Medicine
Westchester, Illinois
American Academy of Sleep Medicine
Redline S, Kapur V, Sanders M, Quan SF, Gottlieb D, Rapoport D, Bonekat W, Smith PL, Kiley J, Iber C: Effects of Varying Approaches for Identifying Respiratory Disturbances on Sleep Apnea Assessment. Am J Respir Crit Care Med 2000; 161:369–74
Manser R, Rochford P, Pierce R, Byrnes G, Campbell D: Impact of Different Criteria for Defining Hypopneas in the Apnea Hypopnea Index. Chest 2001; 120:909–14
Saffar S, Ayappa I, Norman R, Krieger A, Walsleben J, Rapoport D: Choice of Oximeter Affects Apnea-Hypopnea Index. Chest 2005; 127:80–8
Chung F, Yegneswaran B, Liao P, Chung SA, Vairavanathan S, Islam S, Khajehdehi A, Shapiro C: Validation of the Berlin Questionnaire and American Society of Anesthesiologists Checklist as Screening Tools for OSA in Surgical Patients. Anesthesiology 2008; 108:822–30
Phillips B: Sleep Apnea and Public Health. Pulmonary Perspectives 2005; 22:1–4