Article Text

A quagmire for clinicians: when technological advances exceed clinical knowledge
Free
1. SUSAN REDLINE
1. MARK SANDERS
1. Department of Pediatrics, Medicine, and Epidemiology
2. Case Western Reserve University
3. Rainbow Babies and Children’s Hospital
4. 11100 Euclid Avenue
5. Cleveland
6. Ohio 44106-6003
7. USA
8. Department of Medicine and Anesthesiology
9. University of Pittsburgh School of Medicine
10. Veterans Affairs Medical Center
11. and Pulmonary Sleep Disorders Program
12. University of Pittsburgh Medical Center
13. Pittsburgh
14. Pennsylvania
15. USA

## Statistics from Altmetric.com

Obstructive sleep apnoea hypopnoea syndrome (SAHS) is currently estimated to affect between 2% and 25% of the adult population.1 2 Increasingly, data indicate that obstructive SAHS, if untreated, may result in both short and long term sequelae including daytime sleepiness, poor quality of life, neuropsychological impairment, hypertension, and cardio-cerebrovascular diseases.3 Its high prevalence and potentially substantial morbidity present challenges to the health care system and to individual care providers to diagnose and identify those individuals at greatest risk of obstructive SAHS related complications and those most likely to benefit from specific interventions. On the one hand, the costs associated with evaluation with the “gold standard” (overnight laboratory based multichannel polysomnography) could exceed $1500/patient. In the USA this cost alone could result in annual health care expenditures of >$18 billion if all adults with suspected SAHS were tested.4 On the other hand, the economic costs of untreated SAHS are substantial. These, however, are more difficult to estimate since they may include the costs associated with loss of work productivity, occupational and vehicular accidents, and potentially preventable hypertension and cardio-cerebrovascular diseases. Regarding the latter alone, it has been estimated that between $3 million and$2 billion spent on treatment of hypertension and cardiovascular diseases annually in the USA may be reduced by effective treatment of SAHS (estimates varying according to the estimated attributable risk).4 In times of escalating aggregate health care costs, how should the appropriate balance between costs and benefits be achieved?

One strategy to reduce the costs associated with using complex expensive technology to diagnose a condition associated with common symptoms (snoring and daytime sleepiness, found in >50% to >20% of the population, respectively3) is to use screening tests and/or diagnostic tests that are simpler and less costly than overnight laboratory based polysomnography. When using a highly sensitive screening test (high negative predictive value) only patients who test positive would proceed to the gold standard. On the other hand, use of a highly specific test (high positive predictive value) may require continued testing of those patients who screen negative, but might allow treatment to be applied only to those patients with positive results on the screening test without proceeding to further testing.

Evaluation of new technology and determination of test sensitivity and specificity has been performed predominantly by comparing the new tests with “conventional” laboratory based polysomnography. The premise of this work has been that obstructive SAHS is a disorder that is diagnosed specifically only after a critical threshold of apnoeas + hypopnoeas are exceeded by the “gold standard” polysomnographic evaluation. Evaluations of screening tests and new technology have therefore largely been based on comparing the number of “events” detected by overnight laboratory based polysomnography with the number of “events” detected by alternative tests. Indeed, third party payers who have required a specific apnoea + hypopnoea index (AHI) to justify reimbursement for specific treatments have endorsed this supposition.

Over the past 10 years a number of candidate tests/studies for screening and diagnostic purposes have been evaluated. These include limited channel and/or ambulatory polysomnography and single channel recordings, usually of oximetry. Evaluations of these studies have generally been encouraging, but use of these diagnostic modalities has not been widely endorsed because levels of prediction—when disease is defined on the basis of the number of “events”—while high, are imperfect. In addition, the overall economic benefit of such strategies has never been convincingly demonstrated.

In the current issue of Thorax Sériès and Marc5 report on the evaluation of a relatively new technology—namely, measurement of nasal pressure and flow from a simple nasal cannula, similar to that used to deliver oxygen, attached to a pressure transducer. Over the last five years this method of recording has gained much popularity because of its relative simplicity and because the signals obtained are generally clear and the changes in breathing pattern are easy to recognise. The method is intuitively appealing since the sensor detects patterns which reflect changes in flow and volume, parameters considered to be closely related to the underlying physiological disturbances of obstructive SAHS. This approach appears to be more physiologically grounded than the use of thermal sensors which measure changes in temperature at the nose and mouth.

Sériès and Marc first evaluated this technology by comparing the number of events detected by the sensor with the number of events detected by two more conventional approaches: scoring hypopnoeas by identifying breathing amplitude changes from (1) data obtained from inductance sensors and (2) data from a thermistor. Compared with each of the conventional approaches evaluated, a greater AHI was measured with the nasal cannula—4.5/h higher than the inductance method and 8.8/h higher than the thermistor method. In addition, 39% of events that appeared to be hypopnoeas by conventional methods appeared to be apnoeas by the nasal cannula method. When an AHI of >15/h was used to classify an individual as having obstructive SAHS, 22% of subjects who would have been classified as “normal” using the conventional (inductance) method were considered to have SAHS with the cannula method. Before interpreting these data it may be useful to consider other differences reported in the study.

Sériès and Marc also compared differences in the AHI determined by using the two other more conventional approaches. They found systematic differences of 4.3/h between the two more conventional methods, with more events being identified by the inductance method than by the thermistry method. Disease classification could therefore change significantly when a number of “events” is used to identify disease and when identification of such events, even in the context of a “gold standard” overnight polysomnographic study, has not been standardised because of variable use of different sensors for detecting breathing changes (as well as because of differences in interpreting such signals, which also has not been standardised). Such differences in AHI that result from varying commonly used and divergent approaches has, in fact, been recognised by others6 7 and suggest the need to re-evaluate the concept of “gold standard” as applied to polysomnography.

The findings by Sériès and Marc highlight the problems in current approaches for evaluating new technology and in the application of such technology in a rational fashion to diagnose obstructive SAHS efficiently. Overnight laboratory based polysomnography is not a “gold standard” reference test; rather, it is a fairly general approach for the measurement and interpretation of sleep and breathing disturbances with substantial latitude for varying the use and interpretation of specific sensors. The latter, in turn, may result in discrepancies in AHI estimates between laboratories, each using the “gold standard”, that may exceed many discrepancies reported between “standard” polysomnography and alternative tests. If, indeed, the conventional approaches are not “gold standards”, how does one evaluate new technology? Accordingly, it is difficult to interpret the results of the study by Sériès and Marc which found more “events” detected by the newer technology. Are the additional events found with the nasal cannula “false positives” or is it a more sensitive technique that more accurately identifies individuals at short and/or long term risk of adverse health effects related to obstructive SAHS?

Using the number of detected events (apnoeas + hypopnoeas) as the benchmark for evaluation of new technology is a premise that should be questioned. It has not been validated by clinical data showing a clear cut dose-response relationship between the number of such events and the occurrence of adverse clinical consequences. The development of adverse cardiovascular and neuropsychological effects secondary to SAHS is, in fact, generally thought to be related to complex and related phenomena that include exposure to hypoxaemia, hypercapnia, sleep fragmentation, intrathoracic pressure swings, and autonomic nervous system activation. However, the AHI has not been convincingly shown to predict specific biological responses or clinical outcomes.

In an attempt to address the physiological significance of their findings, Sériès and Marc show that the arousal index, a marker of sleep fragmentation, was increased in the group of patients who tested negatively by the conventional approach and positively by the nasal cannula approach. Although this suggests that such individuals may be at risk of daytime sleepiness, little is really known regarding their overall increased health risks or likelihood of benefiting from specific treatment.

These gaps in our knowledge have not diminished the enthusiasm of many experts in sleep disorders for using this technology. Sensors and software for nasal pressure measurements are now being incorporated into ambulatory and simple sleep monitoring devices. This is occurring in the face of data, such as those reported by Sériès and Marc, that suggest that this technology may not be suitable for 9% of patients because of underlying nasal obstruction, and may require frequent repositioning of the nasal cannula that may be difficult to perform in an ambulatory setting. Although the advantages of such monitoring approaches may far exceed these limitations, it is incumbent on both sleep researchers and industry to work together to assure that the newest technology is most appropriately used for clinical decision making.

It seems that much of the emphasis over the past 10 years in evaluating new diagnostic or screening tests for obstructive SAHS has been misplaced. Industry appropriately has continued to press for the adoption of more sophisticated technology. However, most work has centred on evaluating both old and newly emerging technology against a gold standard that itself is poorly standardised and from which we have yet to derive a definitive metric of the disease process. Recording techniques and measurement approaches vary considerably within the rubric of laboratory based polysomnography. Of even more concern is the fact that the superiority of any given laboratory approach to identify short and long term morbidities or to predict responsiveness to treatment over other approaches, including clinical evaluation, simple single channel screening (oximetry), and multi-channel ambulatory recording, has not been established. The emergence of exciting and physiologically based approaches for measuring the stresses associated with obstructive SAHS—such as the nasal pressure flow techniques and other techniques such as pulse transit time (which measures blood pressure or subcortical arousals8)—provides the challenge to evaluate systematically their abilities to identify efficiently, economically, and accurately clinically meaningful outcomes rather than to compare them with an imperfect gold standard. The work by Sériès and Marc and others working with the nasal pressure/cannula9 10 is important in delineating the comparability of data obtained with the newer sensors with more conventional approaches. However, future studies should also address the multiple lacunae that exist regarding the ability of new and old technologies to provide clinically and epidemiologically useful data. Sleep experts and industry need to form new partnerships that go beyond one dimensional assessments of “event” comparisons, and rather address the clinical usefulness of any given technique with regard to clinical predictive ability, patient acceptability, failure rates, and costs.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.