45.Kerns RL. Research in orthokeratology. Part I: Introduction and background. J Am Optom Assoc 1976;47(8):1047–1051.

46.Barr JT. Contact Lenses 2002: Annual Report. Contact Lens Spectrum. 2003(January).

See also BIOMATERIALS: POLYMERS; BLIND AND VISUALLY IMPAIRED,

ASSISTIVE TECHNOLOGIES; LENSES, INTRAOCULAR; VISUAL PROSTHESES.

CONTINUOUS POSITIVE AIRWAY PRESSURE

DAVID M. RAPOPORT

NYU School of Medicine

New York, New York

RON S. LEDER

Universidad Nacional Autonoma de Mexico

Mexico, Distrito Federal

INTRODUCTION

Beginning in the 1970s, positive-end expiratory pressure (PEEP) began to be added to the pressure applied during inspiration in patients undergoing mechanical ventilation. The rationale was that when a patient had loss of alveolar surfactant, the alveoli tended to collapse during expiration. ‘‘Holding them open’’ by offsetting the increased elastic recoil with a pressure that did not return to atmospheric at the end of expiration was beneficial to gas exchange because it prevented complete collapse with resultant shunting of blood past airless lung. This process was applied to both infant lungs (neonatal respiratory distress syndrome, RDS) and to adult lungs (adult respiratory distress syndrome, ARDS) with improved oxygenation as the main endpoint.

As PEEP was more widely used, it was observed that at the time of removal of respiratory support, some patients (especially newborns) benefited from PEEP for oxygenation despite being able to ventilate. This suggested that the strategy of providing a constant distending pressure to the lung (CPAP) during BOTH inspiration and expiration without increasing the pressure during the inspiratory phase (i.e., ventilation) provided some transient benefit during the period before extubation. In addition, it proved possible to apply CPAP via a nose or face mask after extubation, with continued benefit to the lung (Table 1).

CPAP IN OBSTRUCTIVE SLEEP APNEA/HYPOPNEA SYNDROME (OSAHS)

Basic Circuit and Rationale for Use

CPAP was first introduced in 1981 as a treatment for obstructive sleep apnea/hypopnea syndrome (OSAHS). The concept was initially proposed by Collin Sullivan (Australian Patent AU-B83901/82) as a pneumatic upper airway splint and later shown to work even in the presence of chronic respiratory failure (chronic hypercapnia) by David Rapoport (U.S. Patent 4,655,213). In this application of CPAP, the effect of interest is that of continuous positive airway pressure and not its effect on the lung (as with PEEP), although this is necessarily always present. The critical rationale is the

effect a positive pressure in the AIRWAY has on the collapsible upper airway (i.e., the posterior pharynx and hypopharynx), which is not relevant in the intubated patient. The pressure is applied via nose or mouth mask and distends the area that extensive physiologic work has shown to have a tendency to collapse during sleep (especially during the negative pressure of inspiration).

The original circuit proposed consisted of a nose mask to which was attached either a pressure-dissipating threshold valve or a restrictor that created a roughly constant backpressure due to a constant bias flow provided by a blower or other source of compressed air. Early in development, it became clear that fans and blowers had better characteristics than piston-type high-pressure compressors, due to their ability to deliver high flow rates to the mask with control via motor speed, little dependence of delivered pressure on the backpressure, low cost, and quieter operation.

The original concept described by Sullivan was that the CPAP (pressure) was needed continuously to ‘‘hold’’ the airway open against a natural tendency of the walls of the airway to collapse due to loss of active muscle tone during sleep and the suction caused by inspiration. There is a tendency for some degree of airway collapse during sleep in everyone. Patients with the obstructive sleep apnea/ hypopnea syndrome tend to collapse their airway to excess. In all cases, the collapse and airway obstruction occurs because of loss of tone in the airway muscles, whose role is to stiffen the walls against the suction created by breathing during inspiration. Although there has been much debate, most models of this process of collapse and its treatment with CPAP have suggested that the treatment pressure needs to be relatively constant at the point of collapse unless the patient changes body position, head and neck position, sleep state, or wakes up. The point of collapse is usually found to be at the back of the throat or at the level of the soft palate.

Leak Circuit Modification

Until 1985, CPAP was delivered by means of a restrictor or mechanical valve that was placed on the patient’s nose mask. This valve, through its design and its passive mechanical properties, held the pressure at a value that was as constant as the mechanics of the valve could achieve (a so-called ‘‘threshold’’ valve, which opens more to discharge air when pressure rises). It also provided a vent for exhaled CO2 and excess humidity, as it was located near the patient; a side-effect of the constant dissipation of pressure was venting of the circuit, including the exhaled gas from the patient. In 1985, Rapoport proposed that the valve used to set the pressure in the circuit could be removed from the mask to increase patient comfort. However, this required that the ‘‘venting’’ function (removal of exhaled CO2 from the circuit) be performed separately. The modification consisted of a small controlled leak deliberately introduced near the mask that did not significantly dissipate the pressure (previously this had been a large leak or a threshold valve). This modified circuit is the most widely used hose circuitry in both CPAP and noninvasive mask ventilation.

A further improvement was instituted in the mid-1980s, when it was observed that the use of a threshold valve was

optional. This was because the blower could be designed to have a sufficiently flat flow-to-pressure relationship at a given speed of rotation to maintain a near-constant pressure during the increased flow of inspiration and decreased flow of expiration and changing amount of mask leak. Since then, CPAP blowers have either been entirely passive (set at one blower speed for each prescribed CPAP) or had some type of speed control that adjusted the speed in response to sensed pressure feedback. A few devices still use a threshold valve, but these have tended to replace the passive mechanical valves with active electronically controlled stepper motor-driven valves.

Variations in Delivered Pressure

Because active control of pressure is necessitated by removal of the threshold valve from the mask, there has been gradually increasing attention to modifying the pressure contour provided to the patient interface. In particular, various techniques have been used to keep a particular pressure constant. Conceptually, two distinct targets for stabilization of the pressure are either pressure at the blower or pressure at the mask. Initially, CPAP systems were designed to have a constant pressure at the blower,

neglecting that this constant pressure at the blower would cause fluctuations at the patient mask (see above). More recently, attention has been directed to maintaining a constant pressure at other points in the circuit.

As air flows through a closed tube, it is driven by the pressure drop, which occurs progressively in the direction of flow. This implies that in any system with a nonzero resistance, there will always be a difference in pressure as one travels in the direction of flow along the tube. Specifically, as one travels from the blower along the tubing toward the patient’s most collapsible airway point, the airway pressure will fall from that set at the blower and will differ depending on the rate of flow through the system and on where it is measured. Pressure differences between points along this route also depend on the direction of airflow (inspiration and expiration) and the size of the bias flow (e.g., through intentional or unintentional leaks at the mask. Thus, during inspiration, pressure is always higher by some small amount at the machine end of the tubing than it is at the patient’s nose, and during expiration, it is often lower at the machine end of the tubing than at the nose if flow reverses. The amount of pressure difference between the machine end of the tubing and the patient depends on the resistance of the tubing connecting the two

and on the flow through the system, which is the sum of the patient’s breathing airflow and any leak that occurs at the mask.

As the purpose of the CPAP is to maintain a therapeutic pressure that prevents upper airway collapse, a strategy to control the variations in this pressure must be established. Different approaches have been taken by different devices intended to deliver what is called CPAP. In the earliest CPAP devices, the valve located at the mask controlled the pressure; this intrinsically adjusted for changes in leak and reversal of flow from inspiration to expiration, and the only requirement of the blower was to provide an excess (not necessarily constant) flow to the valve located near the patient. When the valve was removed from the mask, pressure control shifted away from the patient to a point in the circuit near the blower. At least under some conditions, pressure can differ considerably from the desired therapeutic pressure as felt by the patient. The following is a list of some strategies adopted by current CPAP devices to deal with this (in terms of the original therapeutic intent, constant pressure at the mask during inspiration is key):

1.The controller sets a constant pressure at the machine (constant blower speed). This pressure must be slightly in excess of the patient’s need; i.e., it must be sufficiently high to allow some fall during inspiration under maximal leak conditions, or the patient will be under-treated at this critical time in inspiration. This strategy necessarily implies that pressure at the patient will be in excess of the required therapeutic pressure at all other times, and this may contribute to patient discomfort.

2.The controller is driven by active feedback from the pressure as measured in the mask. This feedback will cause the blower to continuously vary pressure (at the blower) so as to maintain it constant at the mask. Either blower speed or valve opening may be varied, but pressure as sensed at the mask is the controlled variable. Until leak at the mask becomes enormous, this will be the closest to the original concept of a mask CPAP proposed by Sullivan and implied by the ventilator uses of PEEP and CPAP.

3.Control of pressure as exerted at the machine is based on assumptions about how the pressure will change as it travels along the tubing that connects the blower to the patient. Some devices assume a known pressure drop across the tubing and just add this to the desired therapeutic pressure. Other devices use the flow (or some estimate of flow such as blower speed) to calculate a predicted drop in pressure between machine and patient, creating a deliberate but variably higher pressure than prescribed—in an attempt to deliver the constant therapeutic target.

The above strategies handle changes in flow through the system (e.g., changing leaks), but they may not adequately address changes in backpressure during each breath related to breathing. This is because the intrinsic properties of blowers (fans) are such that at a fixed rotational

speed, these devices tend to produce a flow (not pressure) that is heavily influenced by backpressure (e.g., the tubing resistance and the difference in magnitude and direction of flow between inspiration and expiration). As a result, fans tend to produce a relatively constant pressure against a wide range of loads (because of the changes in delivered flow). Thus, blowers result in a pressure profile during breathing that is close in their behavior to that of a circuit with a threshold valve. The result of this pattern of response to varying backpressures is that setting a constant blower speed results in a system that, to a first approximation, maintains a pressure that is relatively constant at the blower, independent of the patient’s breathing pattern. However, blowers are not perfect in this regard. Blowers (fans), when kept at a constant speed within each breath, necessarily produce slight changes in delivered pressure (higher during expiration and lower during inspiration). Greater variability in breath size, and large leaks through the mask will all result in progressively greater pressure swings at the blower. Because of tubing resistance, even greater pressure changes will occur at the patient if the system is entirely passive. Specifically, pressure in the system and at the patient will fall during inspiration and rise during expiration to a value different from the treatment pressure.

The latest CPAP machines (U.S. Patent Application 2005/0188989) have begun to address these intrabreath pressure variations by modifying the pressure they deliver within individual breaths as a function of the instantaneous flow. The assumption is that this will improve pressure (exhalation) induced discomfort, which is reported by many users of CPAP, by limiting unnecessary rises in pressure above therapeutic during expiration. The simplest way to accomplish this pressure stabilization is to vary the blower speed in response to fluctuations detected in measured pressure. This type of control is a classic feedback system and is used to keep pressure constant at the blower by responding to any deviations or perturbations that occur in the desired constant pressure. Detected changes in pressure result in the controller changing the speed of the blower. Typically, pressure in the circuit varies as a result of changes in the patient’s breathing (inspiration vs. expiration) or changes in the leak from the system at the mask, both of which produce changes in the backpressure felt by the blower.

As pointed out, varying the blower speed within a breath in response to the sensed instantaneous flow can also be done to vary the blower pressure profile such that it is maintained constant (without measurement) at the mask. An alternative to this is to reinsert an active threshold valve at the blower that accomplishes a similar function based on sensing flow in the circuit or some other measured variable that allows prediction of pressure in the mask. Much like the original CPAP circuit, pressure control is provided by driving the blower to produce a pressure in excess of that needed, and diverting (‘‘bleeding off’’) some pressure in the system by variably opening the valve at a ‘‘threshold’’ pressure. However, instead of targeting a constant blower pressure, the valve is instructed to produce a pressure profile predicted to cause a constant mask pressure, by adjusting the opening and closing of the valve.

332 CONTINUOUS POSITIVE AIRWAY PRESSURE

‘‘BiLevel’’ PAP. Introducing a valve under microprocessor control provided an interesting opportunity to create patterns other than a constant pressure in the system. As soon as the pressure delivered to the patient begins to be significantly higher during inspiration than during expiration, however, this nonconstant pressure is fundamentally different from CPAP. In fact, this is similar to the behavior of artificial ventilation devices (ventilators). If the control system is made aware of when inspiration and expiration begin, the valve used in venting pressure can be adjusted to rapidly achieve higher and lower pressures in synchrony with the patient; this can assist or even fully replace patient breathing efforts and is the essence of mechanical ventilation. Whereas CPAP is the imposition of a control algorithm targeting a nearconstant pressure in the system or at least at the patient, ventilation (sometimes referred to as ‘‘bilevel ventilation’’) is the imposition of a nonconstant waveform of pressure on the output of the blower so as to raise inspiratory pressure above expiratory pressure at the patient level. However, as the pressure profile (constant or variable at the patient) depends only on the programming of the valve controller, much confusion exists in the literature about whether a ‘‘bilevel’’ device is being used for ‘‘CPAP’’ or assisted ventilation.

In concept, patients with obstructive sleep apnea have no ventilatory control abnormality once the airway is open. Thus, assistance with ventilation (once the airway is splinted open) is not indicated. The original proposal for bilevel ‘‘CPAP’’ was not targeted at ventilation, but to date, there has been little in the published literature to support its use for ‘‘comfort’’ in patients with OSAHS alone. However, as a noninvasive ventilator, bilevel devices are very effective and deliver what is essentially a combination of PEEP and pressure support ventilation. Their use is clearly indicated in chronically hypercapnic patients and in those with nocturnal hypoventilation. Not only is there little logic to the use of this type of device for intermittent obstructive apnea, but also recent publications have suggested that they can exaggerate central apnea—presumably because increasing breath size (pressure support) will increase plant gain in the patient’s respiratory control loop and tend to produce increased overshoot of the size of compensatory ventilatory efforts whenever there is instability of breathing, creating a classic ‘‘ringing’’ system.

Although the above discussion shows that bilevel ventilation is completely different in purpose and application from CPAP, current devices are such that they can deliver both modes with little change in their circuitry if they contain the means to rapidly change the pressure according to a prescribed algorithm. As a result, there continues to be confusion about what is being done when a physician prescribes a treatment for a patient. Clarification as to the algorithm being used by a setting on the machine requires a decision as to whether the device targets

A pressure that is constant at the blower (the controller removes fluctuations at the blower). When the pressure is measured at the patient, there will necessarily be small fluctuations throughout breathing.

Pressure will be lower at the patient during inspiration and higher during expiration than at the blower. This is passive CPAP.

A pressure that is constant at the patient (the controller removes fluctuations at the patient). To accomplish this, the pressure when measured at the blower will be slightly higher during inspiration and lower during expiration. This is the purest form of classic CPAP.

A pressure that is higher during inspiration than during expiration both at the blower and the patient. This type of pressure oscillation has as a purpose to actively assist the patient in magnifying his breathing efforts. The pressure changes assist the patient’s spontaneous muscular breathing efforts when these are weak. This is active ventilatory support.

In the last two above cases, the expiratory pressure has been lowered from the value it would have achieved during expiration if the system was left to behave passively in response to the patient’s breathing backpressure. The difference between the two algorithms above is not in the direction of change applied to the output expiratory pressure, but in the purpose for which it is lowered and the amount that pressure at the output of the blower is made to fall during expiration through active control. If the pressure at the blower is not lowered, i.e., forced to be constant, then the pressure as measured at the patient will rise during expiration. If the pressure at the blower is forced to fall slightly during expiration, the pressure may remain constant at the patient. Finally, if the pressure is forced to fall sufficiently at the blower during expiration, pressure will also fall at the patient during expiration. Unlike the first two algorithms, this pattern at the patient of a fall in pressure during expiration when compared with inspiration produces actual assistance to the patient’s breathing efforts, and it defines assisted ventilation; this type of ventilatory assistance is fundamentally different from CPAP, whose purpose is only to hold the airway open.

MONITORING/TITRATION ISSUES

Recording the Pressure

The clinical prescription of CPAP is usually given as a single therapeutic pressure value. Typically, this is derived from some type of titration in a recorded sleep study. As should be evident from the earlier discussion of pressure gradients, this prescription pressure should to be related to how pressure was measured during the titration, as well as to how it will be implemented by the patient’s CPAP machine, but this is often overlooked. If it assumed that the prescription is a generic one for a therapeutic pressure to be delivered in the mask, then the mask pressure should be the one measured during the titration study. However, most CPAP machines used in the laboratory do not provide this pressure as an easily available electronic output because they do not measure it. Instead they measure the pressure at the blower, which may differ by up to 1– 2 cm H2O from that at the patient and vary with respiration actively or passively. Furthermore, this gradient, as

discussed above, varies with the uncontrolled leak conditions at the mask and the amount and type of tubing circuitry, including whether a humidifier is in line. Furthermore, because many CPAP machines output an internally measured pressure as an analog or digital signal to facilitate laboratory recording during the sleep study, it is critical to know whether the actual mask pressure is being output or whether the output is a calculated estimate of pressure of a CPAP machine to that assumed to be present at the patient interface. In our laboratory, we prefer the actual measurement of pressure in the mask of the patient and provide this to the patient as his ‘‘prescription pressure.’’ This should be independent of the brand of CPAP chosen for chronic use by the patient in its relation to adequacy of pressure if measured in the mask.

Algorithm for Deciding on the ‘‘Therapeutic’’ Pressure

When a patient undergoes a ‘‘CPAP titration,’’ the pressure in the system during the period of monitoring is gradually increased until all evidence of upper airway obstruction disappears. Different laboratories titrate to different indices, but in principle most include trying to abolish evidence of both severe and partial obstruction as below:

Apneas (complete cessation of airflow caused by obstruction for at least 10 s) usually disappear first, so that at pressures above 8 to 10 cm H20, it is rare to find obstructive apneas. Central apneas (failure of respiratory effort to occur, but usually without obstruction) may appear, especially at higher pressures. These are usually distinguished from obstructive apneas by the absence of persistent respiratory movements (rib and abdomen movements) during the apnea.

Hypopneas (significant reductions in airflow lasting at least 10 s) tend to predominate once CPAP has been applied at low levels. These are easily identified as obstructive by the presence of a flattened inspiratory flow/time contour, which differs from the sinusoidal shape of normal inspiration and breaths with unobstructed reductions in effort (‘‘central’’ hypopneas). This ‘‘flow limited’’ behavior of obstructive hypopneas is explained by a Starling resistor model of the upper airway where dynamic collapse of the airway occurs due to the negative intraluminal pressure of inspiration. Transient appearance and disappearance of the flattened contour of groups of individual breaths indicates recurrent obstructive apnea and indicates the need for increased CPAP. Most laboratory titrations will strive to eliminate these events by raising CPAP.

Evidence of sustained elevated upper airway resistance (in contrast to discrete ‘‘events’’) may remain after all apneas and hypopneas disappear. This evidence can consist of stable snoring (upper airway vibration induced by unstable airway tissue), sustained runs of breaths with an inspiratory contour suggesting Starling behavior (‘‘flow limitation’’), or other direct measures of elevated airway resistance (e.g., direct measurement of intrathoracic pressure, from an esophageal catheter probe, divided by flow). It is currently often assumed that this evidence of high upper airway resistance must be completely relieved by elevating CPAP, but there is controversy as to the benefits of this form of titration. In some cases, raising CPAP to

eliminate all such evidence of elevated upper airway resistance results in further improvement of sleep structure and decrease in daytime sleepiness. By contrast, in other subjects, few if any symptoms occur when the patient has sustained elevated resistance, provided this occurs without causing repetitive arousal. In this latter setting, raising the CPAP is difficult to justify, although often done. Very limited studies attempting to justify this titration approach have not to date supported any benefit of one approach over another.

During CPAP titration, in addition to defining the events that should prompt raising the pressure, it is important to consider when the pressure may be too high (and thus needs to be lowered). Although it is generally assumed that the lowest effective pressure is most comfortable and excess pressure will disrupt sleep, this has not been shown by controlled trials. However, most titration studies should include periodic reductions in CPAP once breathing and sleep have been stabilized to test for the lowest pressure at which evidence of airway instability (see above) recurs. This pressure may be different at different times in the night, and it is almost always different in the supine position and during REM sleep. These observations challenge the concept of a single prescription of CPAP.

Auto-Titration

The above discussion has assumed that a single therapeutic pressure at which the upper airway is effectively splinted exists for a given patient, and that this pressure remains relatively unchanged over time (within each night and across nights). There is ample evidence that this is NOT true. Where it has been studied, it is strongly suggested that for many patients in the supine position, upper airway obstruction is more severe and/or takes more CPAP to treat (although these are not synonymous). There may also be differences in the CPAP needed during REM and non-REM sleep. Thus, when a single pressure is prescribed, most practitioners use the highest pressure needed during a prolonged period of titration (e.g., at night), knowingly over-treating during the rest of the time.

Beginning in about 1990, several investigators began to automate the titration algorithm for choosing a pressure. The concept evolved of a feedback loop that constantly adjusted the CPAP based on sensing either frank apnea, hypopnea, or indices of upper airway abnormality like snoring and/or the contour of the inspiratory airflow. These devices were called auto-titrating CPAP or Auto-CPAP. Two conflicting goals were suggested for optimizing their function—maximizing the efficacy of CPAP and improving patient compliance by reducing pressure to the minimal need at all times. The first of these was to respond to unexpected increases in need in order to prevent undertreatment. The latter was to prevent unnecessarily high values of pressure at a time they were not needed. As both the signal driving feedback and the time constants of the systems developed varied greatly, it is difficult to address the whole group of Auto-CPAP devices in a single study. In particular, the effectiveness of the decision process for raising and lowering the CPAP will dictate whether the final pressure profile is high or low compared with CPAP.

334 CONTINUOUS POSITIVE AIRWAY PRESSURE

This is not the logical target by itself, and only an outcome such as quality of sleep, improved hours of use by patients, and ultimately, improved daytime function and reduced sleepiness, can be used to evaluate the punitive value of Auto-CPAP over constant pressure. To date, however, limited data support this in large groups of patients. Some data suggest improved compliance with specific devices.

Having said that, no data suggest that the more reasonable of these devices is any LESS effective than CPAP, but some Auto-CPAP devices occasionally show changes in pressure that do not bear any logical relationship to the patient’s breathing (runaways), and there is every reason to assume these will impair sleep.

A logical approach to evaluating such devices needs to address several questions before beginning to ask whether long-term use is effective or better than traditional CPAP:

1.Which signal is driving the response of Auto-CPAP? Possible signals include the flow signal amplitude (apnea and hypopnea), shape (detection of starling resistor behavior in the form of ‘‘flow limitation shape’’ as described above), vibrations (e.g., snoring and airway instability), breathing pattern on a longer timescale, direct sound measurements, and direct measures of airway abnormality (e.g., measurement of impedance via forced oscillation technique). The existing devices are driven by different signals, and new devices appear frequently. When compared head-to-head these devices have different responses to breathing test-waveforms, and both bench and patient testing is not yet standardized.

2.What makes the pressure rise? Is a response sought to each abnormal breath or detection of abnormal impedance? Is the pressure adjusted after a ‘‘testing’’ protocol—e.g. a periodic deliberate lowering of the effective pressure to induce some endpoint of abnormality?

3.When is pressure lowered, and after how long? Is continuous testing possible (as with forced oscillations to measure impedance) to which pressure can be lowered when the control variable is low, or is ‘‘normal’’ a condition that, once achieved, provokes a prolonged period of constant pressure (e.g., what is the response to ‘‘normal breathing’’ when detected)? When pressure is lowered, is this a provocative test, or an attempt to detect over-treatment? How frequent are pressure decreases? The implications of these decreases and their endpoint are physiologically significant—‘‘testing’’ too frequently with a nonsubtle endpoint (e.g., an apnea or an arousal) will disrupt sleep. Testing too infrequently for decreasing pressure will produce ever increasing therapeutic pressure because there will be insufficient compensation for unavoidable errors in the algorithm’s detection of a need to raise pressure.

Furthermore, a constant tradeoff exists between the need to optimally set CPAP for a stable physiologic state (e.g., in stable stage 2 sleep in the supine position, a pressure of x cm H2O may be appropriate for long periods) and the need to respond with a rapid change in CPAP to

state changes affecting the airway (e.g., awakening, entering REM, or rolling from the lateral to the supine position). Each machine currently on the market and in development has made different decisions about the way to balance these needs, and the resultant behavior, although it can be described, is not clearly better or worse by simple criteria. Large numbers of patients are needed to show benefit in terms of daytime outcome or compliance with therapy, and these trials are not widely available, nor are the results from one machine easy to apply to another machine or even to a slightly modified algorithm.

This field is still in evolution, but there has been disappointment in the advantage of the approach as reflected in better therapy. Despite this, automation of titration may still have large benefits for patients, even if it is not ‘‘better’’ titration, or even ‘‘more comfortable’’ CPAP. This arises from a trusted algorithm being able to replace the costly CPAP titration study, which is currently usually done in an attended fully monitored laboratory setting. To date, only a few machines on the market have sufficiently reliable ‘‘auto-titration’’ that they can be left unattended and monitored on a first-time user of CPAP, with the resulting pressure behavior assumed to represent an accurate reflection of the patient’s need for CPAP. Even the best available machines still over-treat and under-treat some patients, and it seems advisable to recommend that evaluation of the results of a titration study be reviewed (at least off-line) by an expert with more than an assessment of the pressure profile the machine chose.

Our laboratory chooses to review all Auto-CPAP titrations by examining the flow profile and looking to see if overall we agree with the induced rises in pressure. We also review the pressure profile for rapid uncontrolled and unexpected rises in pressure that end with an arousal of the patient, and usually assume these are erroneous.

Finally, if Auto-CPAP is used for titrating a patient’s need with the intent of using a single pressure as a prescription, yet another ‘‘algorithm’’ must be invoked to translate a constantly fluctuating pressure into a single prescription. Review of the pressure and or flow tracings rarely results in a single pressure that is constant for much of the night. One must, on subjective grounds, discard excesses and ignore periods of inadequate therapy during the fluctuations. One proposal is to discard the highest pressures achieved during 5% to 15% of the night. There has been no testing of this approach by objective criteria of long-term benefit.

INDICATIONS FOR USE OF CPAP IN OSAS

Stated simply, CPAP is currently the first line of treatment and is indicated for reversal of sleep-induced abnormal upper airway behavior, provided it is severe and results in disruption of sleep with negative daytime consequences. When obstructive apneas and hypopneas occur very frequently and result in severe blood oxygen desaturations, it seems obvious that CPAP is needed. Formal trials of the benefits of CPAP have relatively conclusively shown benefit when more than 30 apnea/hypopneas occur per hour of sleep. This benefit is mostly in the form of reduced daytime sleepiness, although small studies have suggested

reductions in blood pressure, improvement in daytime cognitive performance, or reaction time after weeks to months of therapy. CPAP is now near universally accepted as the most effective therapy (better than surgery or oral appliances) but not always as the most acceptable therapy from the patient. This has resulted in compliance rates among moderate–severe apneics (see above definition), which range from 50% to 80%, leaving many patients suboptimally treated, or anxious to change to other treatments as they become available.

However, a more contentious issue is how mild can the physiological abnormality be before treatment with CPAP is either unnecessary or unacceptable to patients. Two relatively large clinical trials are currently underway to answer this question, but no definitive statement can be made at present. A therapeutic trial of CPAP may answer the question in individual patients who show abnormal respiratory events during sleep and have an overt complaint (such as excessive daytime sleepiness). The trial is considered successful provided that patients see a noticeable improvement in symptoms. Better documentation of the validity of this approach is urgently needed as recent studies have shown a very large number of subjects in the general population who have apnea–hypopnea indices ranging between 10 and 30 events per hour (up to 25% of the population), some of whom are asymptomatic, and others who have significant symptoms that might be due to this pathology. Anecdotally, many patients improve on CPAP, but many cannot adapt to chronic therapy. Some of these may benefit from alternative therapy, but CPAP may remain the most effective and definitive way to perform a therapeutic trial for all treatments for OSAHS.

ISSUES IN COMFORT/COMPLIANCE FOR OSAHS AND ANCILLARY TREATMENT ISSUES

Interfaces/Masks

As comfort is the most perceived issue affecting patient compliance, it is clear that the mask must be an important contributor to the patient’s willingness to use the device. Although this is accepted dogma, compliance rates over the years in which CPAP has been available are not clearly changing, and much of the willingness to use CPAP may also be affected by subjective patient perceptions of improvement (cost/benefit) and the reinforcement they get from the care provider. It is rare that a patient will use CPAP if the prescribing physician does not believe it works. Many types of nasal, oral, and full-face interfaces have been developed to maximize comfort. Nasal masks are currently most used, and details of material, shape, supporting extensions to relieve pressure points, and so on are beyond the scope of this discussion. Non-mask nasal interfaces also exist (‘‘pillows’’ or ‘‘prongs’’) and may help address issues of claustrophobia, variant facial anatomy preventing a good seal with a mask, and personal preference. Oral interfaces are less common, but they have a devout following by some patients. Finally, for those with large leakage out of the mouth when the nose is pressurized, full-face masks may provide an alternative. Chin straps are frequently used to reduce mouth leak. It is clear that the technologist who

knows the available masks and spends time trying multiple ones with a new patient will have greater success than one using the ‘‘one size fits all’’ approach.

Headgear

Like masks, a variety of headgear exist. These affect fit of the mask, pressure on the nasal bridge, tension of the straps, and even appearance. There is little published on the relative effect these have on compliance or patient preference, but it seems this is an important area.

Oxygen

Some patients (a minority) who use CPAP have a concomitant or related need for supplemental oxygen. As the oxygen is being delivered into a larger air stream, the rate of infusion (typically 2 to 10 L/min) may need to be different from that prescribed for a patient just breathing supplemental O2. Furthermore, simple examination of the circuit will show that the leak (intentional and unintentional at the mask) will have a large effect on the delivered concentration of the O2 bled into the air stream. A larger leak will change by a factor of 2–4 the final concentration of O2 at the patient’s nose. As CPAP masks are intrinsically leaky and variable, so it is predictable that the need for O2 will change. In patients without evidence of hypoventilation and central regulatory abnormalities (usually marked by daytime hypercapnea, or arterial PCO2 >45 mm Hg), giving too much is not a problem other than expense, so titration to the highest level needed to keep the oxygen saturation during all of sleep (including REM) >90% is the usual goal. However, in patients who tend to hypoventilate, excessive O2 will worsen CO2 retention and may lead to accumulation of serum bicarbonate, further depressing ventilatory drive even in the daytime. Thus, it is desirable to try to minimize O2 use.

Finally, it is not often appreciated that the location at which O2 is inserted into the CPAP circuit has a large effect. If the bleed is into the hose near the blower, the tubing promotes mixing and acts as a reservoir of a relatively fixed but lower O2-enriched gas. Pattern of breathing, i.e., time in inspiration and expiration and tidal volume, may have less effect, but the degree of leak will still play a large role. In contrast, if the O2 is bled directly into the mask, especially if this is beyond the leak in the circuit, the leak may have less effect. However, small changes in timing of breathing and mask size will have enormous effects on the inspired O2 concentration as buildup of a small volume of near pure O2 can accumulate during pauses and part of inspiration, whereas there is little volume to act as a reservoir and mixing chamber. This issue should be addressed by providing the location of O2 connection in any prescription so that it will at least match the titration technique.

Humidity

Although at first glance it is not clear why humidity should be needed if breathing occurs through the normal nasal mucosal humidifying mechanisms, drying of the nose and nasal reactive obstruction are common complaints in CPAP users. Recent literature suggests that humidifying the