Continuous Positive Airway Pressure (CPAP) Therapy

Introduction Ventilators are devices that periodically inflate the lungs with air, or other gas mixtures; allowing the lungs to deflate by their own elastic recoil after each inflation. There are many different types, the most common inflate the lungs with a pre-set volume of gas, but are limited in the pressure that can be applied and the rate of inflation. These devices operate in a sealed system, so that no gas can escape during inspiration. They are used to treat respiratory failure, when the lungs will not expand due to lung damage, or the lungs being filled with fluid due to heart failure, or due to failure of the respiratory muscles, or due to failure of the respiratory centre in the brain stem or spinal cord. Ventilators are also used during anaesthesia to deliver gases to the patient when the respiratory muscles are deliberately paralysed. Ventilators are also used in a semi-sealed system in which gas is allowed to by-pass the patients mouth if the flow is in excess of the patients requirements. In these applications the main aim of the therapy is to apply a pre-set pressure to the patients airways and lungs, either to splint the airways to prevent them collapsing, or to drive fluid out of the alveoli and back into the capillaries. This type of therapy is known as continuous positive airway pressure (CPAP) or bi-level positive airway pressure (BiPAP). The main difference between CPAP and BiPAP is that in BiPAP the air pressure is allowed to drop during expiration so as not to impede airflow out of the lungs. CPAP was originally devised by Professor Colin Sullivan of the University of Sydney to treat a condition called obstructive sleep apnoea in which the upper airway becomes completely blocked during sleep due to the tongue and soft palate falling onto the back of the pharynx. The mechanism is similar to snoring, during which the airway is only partially blocked. The technique is now used to treat a range of similar conditions. The first ventilators of this type were manufactured in the University of Sydney from vacuum cleaners that were run in reverse. The air flowed into one arm of a T-piece. The patient breathed from a second arm. Air flowed out of the third arm through a restriction that was used to control the pressure applied to the patient. Sleep Apnoea Sleep apnoea is defined as cessation of breathing of 10 s or more during sleep. There are three types: 1. obstructive apnoea in which airflow is absent despite ventilatory effort 2. central apnoea in which airflow is absent due to the lack of ventilatory effort 3. mixed apnoea which is usually due to a central apnoea followed by an obstructive apnoea. Normal sleep is broadly classified into rapid eye movement (REM) sleep and non-REM sleep. A person spends most of their sleeping time in non-REM sleep, during which heart rate, blood pressure and sympathetic nerve activity decrease. REM sleep occurs mostly during the second half of the night and occurs cyclically. During REM sleep there is increased electrical activity in the brain, increased sympathetic activity and abrupt

changes in heart rate and blood pressure. Paradoxically, in spite of the increased electrical activity in the brain there are periods of complete loss of muscle tone. The loss of tone in the pharyngeal muscles can cause the airways to collapse with the tongue and soft palate falling back onto the posterior (back) wall of the pharynx (Fig. 1). This blocks the airway. The blockage may be slight, with a small reduction in airflow. It may be moderate and result in snoring as the soft palate intermittently occludes the airway. It may be complete when both the tongue and the soft palate fall back onto the posterior pharyngeal wall and initiate an episode of apnoea. During sleep apnoea there is disruption of the normal cycling between REM and non-REM sleep. Obesity predisposes people to sleep apnoea, partly due to the increased fat in the neck narrowing the airways. The current increase in the incidence of obesity in Australia is also causing an increasing incidence of sleep apnoea.

Figure 1. During sleep, with the person lying on their back, the loss of muscle tone during REM sleep allows the tongue and/or the soft palate to fall back towards the posterior wall of the pharynx. If the movement is small, the upper airway is only slightly occluded, reducing airflow to a small extent. The soft palate alone may fall onto the posterior wall intermittently, causing snoring. If both the soft palate and tongue base fall onto the posterior pharyngeal wall, there may be complete occlusion of the upper airway, causing apnoea. (Bova R., Medicine Today, 6(4), April, 2005)

Physiological Response to Apnoea During Sleep When respiration is obstructed, there is a rise in the arterial partial pressure of carbon dioxide (paCO2) and a fall in the arterial partial pressure of oxygen (paO2) (Fig. 2). The rise in paCO2 is detected by the central chemoreceptors in the brain stem and causes an increase in respiratory effort. Similarly the fall in paO2 is detected by the carotid bodies of the internal carotid arteries in the neck and sends signals to the respiratory controller in the brain stem, leading to an increase in respiratory effort. These mechanisms tend to be less sensitive (have a lower gain) during sleep in normal people, but have a higher gain in people with sleep apnoea. This higher gain leads to a response even at normal arterial blood gas levels. Part of the response is to slow the heart and to increase the blood pressure by sympathetic constriction of the blood vessels, particularly in the muscles (Figs 3, 4). This constant response can result in a permanently raised blood pressure, putting strain on the heart due to the increased after load, i.e. increased pressure that the heart has to pump against. The increased effort by the respiratory muscles causes lower than normal pressures inside the chest. This results in an increased volume of blood being drawn into the ventricles before each heart beat (increased pre-load). This places further strain on the ventricles. Many patients with obstructive sleep apnoea eventually suffer from heart failure as a result of this loading.

Figure 2. Recordings from a sleeping patient with obstructive sleep apnoea. The left column is the control showing an episode of apnoea, the middle column shows the absence of sleep apnoea during CPAP therapy, and the right column shows the return of apnoea when CPAP ceases. The top trace shows the electroencephalogram (EEG) with arousal (large spikes) during an episode of apnoea in the control and recontrol columns. The second trace shows an EMG from a scalene muscle (an accessory respiration muscle) which demonstrates increased activity immediately following apnoea in the control and re-control columns. The next two traces are ribcage and abdominal respiratory movements, showing apnoea with little movement in the first half of the control and re-control columns followed by large

respiratory movements as the patient recovers; in contrast to the CPAP column which shows steady respiration throughout. The next two traces are tidal volume and oesophageal pressure, both of which reflect respiration and parallel the two traces above them. The trace second from the bottom is blood pressure and also show pulse rate, although difficult to see in this recording. Blood pressure and pulse rate both drop during apnoea, then rise during the recovery immediately afterwards. Overall the blood pressure in the control and re-control columns is above normal, whereas during CPAP therapy it is within the normal range. The bottom trace is arterial oxygen saturation and show the fall to 75% about 15 s after apnoea and a rise back to ~97% 15 s after recovery. 75% is dangerously low. (Courtesy of the University of Florida and Resmed) The constant heightened sympathetic response in people with obstructive sleep apnoea leads to insulin resistance, i.e. they require a higher level of insulin in their body to achieve the same lowering of blood glucose levels. These people frequently develop diabetes mellitus. The insulin resistance also leads to weight gain which worsens the airway obstruction. The combination of these effects also leads to an increased incidence of acute myocardial infarction (heart attack) and the development of arrhythmias.

Figure 3. Three separate records from the same patient with obstructive sleep apnoea, showing the patient awake in the top left panel, asleep with CPAP therapy in the top right panel and showing apnoeic episodes (OSA) while asleep with no therapy in the bottom panel. Each record has three traces: sympathetic nervous activity (SNA) showing normal low levels while asleep with CPAP therapy, but very high levels during apnoea. The second trace shows respiratory movements (RESP) showing a decrease during apnoea and large increase immediately following episodes of apnoea. The third trace is blood pressure (BP), also showing heart rate, and demonstrates the drop in blood pressure and heart rate during apnoea and the

large rise in both immediately after apnoea. (Courtesy of the University of Florida and Resmed)

Figure 4. Control of respiration during obstructive sleep apnoea. Decreasing O2 levels are detected by the peripheral chemoreceptors in the carotid bodies in the neck and the arch of the aorta. These signals are sent to the respiratory controller in the Nucleus tractus solitarius (NTS) in the medulla, which is also sensitive to increasing levels of CO2. There is increased activation of the respiratory muscles in an attempt to overcome the airway obstruction, of the vagus to slow the heart (diving reflex) and the sympathetic nervous system to constrict some blood vessels to conserve oxygen use. (Courtesy of the University of Florida and Resmed) The constant interruption of the sleep cycle means that these patients feel tired during the day and constantly drop off to sleep for short periods and do not perform well intellectually. They have a high incidence of motor vehicle accidents, which is believed to be of similar incidence to alcohol related motor vehicle accidents. They also have a higher than normal rate of industrial accidents. The use of CPAP on a regular basis appears to reverse some of these effects. It usually needs to be combined with a diet and exercise program. Diagnosis of the condition is made by analysing records, such as Figs 2 and 3 above, while the patient is sleeping overnight in a laboratory. This is known as a sleep study. Acute Pulmonary Oedema This is a condition that accompanies left heart failure. Left heart failure occurs when the left ventricle fails to pump blood adequately. The most common cause of left heart failure is high blood pressure. If the left ventricle fails to pump but the right ventricle continues to pump normally, there is a rise in pressure in the left atrium, the pulmonary veins and

the pulmonary capillaries. People with this condition become short of breath when they lie down, so spend most of their life, including sleep, being propped upright. The upright condition introduces a pressure gradient in the blood in the pulmonary veins due to gravity, so that the pressure is higher at the bottom of the lungs than the top. In turn the pulmonary capillary pressure is higher in the bottom of the lungs than the top. When the pressure becomes high enough, the capillaries leak fluid, and eventually cells, into the alveoli. This occurs initially at the bottom of the lungs, but at increasingly higher levels as the left heart failure worsens. The condition is fatal if untreated as the fluid in the alveoli interferes with gas exchange with the blood. Patients become increasingly short of breath and cough up pink, frothy sputum. The filling of the lungs with fluid is also known as congestion, so that this type of heart failure is also known as congestive heart failure. Treatment is by the use of rapidly acting diuretics, intravenous drugs that cause the kidneys to excrete fluid, and the use of CPAP, which increases the air pressure in the alveoli and drives the fluid back into the pulmonary capillaries. Cheyne-Stokes Respiration This is a breathing pattern associated with severe congestive heart failure characterised by rhythmic oscillation of tidal volume (depth of breathing) with regularly recurring periods of hyperpnoea (over-breathing), hypopnoea (under-breathing) and apnoea. It was first described by a physician, Cheyne in 1818, and then by Stokes in 1854 who described the condition as: The decline in the length and force of the respirations is as regular and remarkable as their progressive increase. The inspiration becomes each one less deep than the preceding, until they are all but imperceptible and then the state of apparent apnoea occurs. This is at last broken by the faintest possible inspiration; the next effort is a little stronger, until so to speak, the paroxysm of breathing is at its height, again to subside by a descending scale This symptom, as occurring in it highest degree, I have only seen during a few weeks previous to the death of the patient The exact cause of this phenomenon is unknown, although many medical practitioners have advanced theories that do not withstand analysis. Cheyne-Stokes breathing is associated with an increased sympathetic response, in the same way as sleep apnoea. It has been shown in experimental animals that a similar periodic breathing to CheyneStokes respiration can be induced by cutting the carotid arteries below the carotid bodies and rejoining them with a long length of tubing so that there is a large circulatory delay between the heart and the carotid bodies. This introduces a 180o phase change in the respiratory controller, and so makes it oscillate. However the circulatory delay required in experimental animals is far greater than seen in patients. It is possible that the gain of the controller, as shown by the increased sympathetic response to hypoxia, has also increased enough to cause positive feedback when combined with the small circulatory delay seen in patients. This has not been proven. It is also possible that Cheyne-Stokes respiration is not due to a classical linear control system oscillation with positive feedback, but is due to limit-cycle oscillation displayed by a non-linear controller. The respiratory controller is certainly a non-linear system. Cheyne-Stokes respiration is a bad prognostic sign for a patient with the survival rate over 2 years dropping to 56%, compared to a survival rate of 86% for severe congestive

heart failure without Cheyne-Stokes respiration. It has been shown that CPAP given over a prolonged period will improve the survival rate for some patients with Cheyne-Stokes respiration. As for sleep apnoea patients, the sympathetic response of Cheyne-Stokes respiration patients drops with prolonged CPAP therapy. Some research studies have found that BiPAP provides a better response than CPAP, but other studies have failed to find any difference between the responses to CPAP and BiPAP. CPAP Therapy The first device was simply a vacuum cleaner with the outlet pipe connected to the inlet of a mask at point 2 in Figure 5. An adjustable flow restriction at point 9 on the mask outlet caused the pressure to rise to about 6 cm of water above atmospheric pressure at the patients mouth. This increased pressure applied through the nostrils (not the mouth) was enough to hold the soft palate and the tongue forward and prevent obstruction (Fig. 1).

Figure 5. Mask for the first devices used for CPAP. The output of a vacuum cleaner blower was applied to the tube at point 2. A variable restriction at point 9 was adjusted to produce a pressure at point 3 of about 6 cm of water. There were problems with this device. The pressure was fixed, whereas the pressure needs by the patient depended on the stage of sleep and the position they were in and with variations in their nasal airflow resistance from day to day. The constant air flow into the patients nose tended to dry the mucous membranes and cause discomfort. The mask was uncomfortable. This led to a low compliance with the therapy. To improve therapy and compliance, pressure and flow sensors were placed at the blower outlet and there was an electronic control of the motor. To enable rapid changes in flow and pressure without having a noisy system, two centrifugal impellers were used in series (Fig. 6). The device was able to calculate the patients 95th percentile flow and regulate the pressure accordingly. It was also able to recognise moderate reductions in airflow, snoring and apnoea and adjust the pressure to treat these accordingly. Having treated a

given condition, the device would gradually return to the normal treatment pressure. Mask leaks could be detected and allowance made.

Figure 6. Two-stage blower with centrifugal impellers at either end of the motor. AutoSet Spirit CPAP System This is the most recent of the ResMed CPAP systems (Fig. 7). It contains a blower, pressure and flow sensors, and a control and data recording system. The air-flow generator control block diagram is shown in Figure 8.

Figure 7. AutoSet Spirit flow generator with tubing and mask. (Courtesy of ResMed)

Figure 8. Block diagram of the flow generator and feedback control system. (Courtesy of ResMed)

Figure 9. Airflow signal showing respiratory inflow (fine, solid line), long term average RMS inflow (dashed line), 50% long-term average RMS flow (dasheddotted line), 25% long-term average RMS flow (dotted line) and 2-second moving RMS flow (bold, solid line). The 50% and 25% lines are used as thresholds for the detection of hyponoea and apnoea respectively should the 2-second moving RMS fall below these thresholds for more than 10 seconds. (Courtesy of Resmed) Detection of Apnoea and Hypopnoea A long-term average root mean square (RMS) of respiratory in-flow is calculated (Fig. 9). This is used to set thresholds for the detection of abnormal events. A threshold for hypopnoea is defined as 50% of the long-term average RMS flow, and a threshold for apnoea is defined as 25% of the long-term average RMS flow. As well as the long-term average RMS flow, a 2-second moving RMS flow is calculated. Should the 2-second moving RMS flow drop below the hypopnoea or apnoea threshold for 10 s, then the condition of hypopnoea or apnoea is deemed to have occurred (Fig. 10).

Figure 10. Algorithm for the detection of apnoea and hypopnoea. (Courtesy of ResMed) Treatment of Apnoea and Hypopnoea Following detection of apnoea, the CPAP pressure is increased in amount proportion to the apnoea duration. If the current pressure is less than 10 cm of water, the pressure is increased at a maximum rate of 8 cm of water per minute duration of apnoea. The rate of increase diminishes as the pressure approaches 10 cm of water. The treatment is not designed to terminate a current apnoea, but to prevent further apnoeas occurring. The pressure increase is cumulative, so that if a new apnoea occurs, the increase is added to that from previous apnoeas. If no further apnoeas are detected, the pressure decays exponentially towards the preset minimum with a time constant of 20 minutes. Hypopnoea is not treated but is logged for review by the physician. Snoring Detection Snoring is detected by the pressure sensor. The pressure signal passes through a 30 Hz to 200 Hz band-pass filter. The output of the filter is rectified, then integrated. This signal is scaled by an empirically determined constant to a unit snore value, and an estimate of the motor noise (modeled as a second order polynomial) is subtracted. The resultant instantaneous snore signal is averaged over the inspiratory cycle, then compared with a standard snore threshold signal that excludes motor and airflow noise (Figure 11). If the threshold is exceeded the pressure is increased by 1.5 cm water per unit snore signal above the threshold. The maximum rate of increase of the pressure for snoring is 0.2 cm water per second. The response to snoring is added to any existing response to apnoea. Airflow Flattening Index A normal inspiratory flow-time curve is rounded, or quasi-sinusoidal in shape. The pressure gradient produced by contraction of the diaphragm is low at the beginning of inspiration, reaches a peak in the middle of inspiration and declines towards the end of inspiration. With adequate CPAP pressure, the airway behaves as a rigid tube and flow is a function of the pressure gradient.

Figure 11. Snoring detection algorithm. If the CPAP pressure is high enough to prevent apnoea and snoring, but not high enough to achieve full patency in the airway, the airway behaves like a floppy elastic tube. Increasing pressure gradient leads to narrowing of the tube so that above a certain gradient the airway behaves like a Starling resistor. When this happens the increased narrowing counteracts the increased pressure gradient so that the flow becomes constant and is independent of respiratory effort. When this happens the airflow looks more like a square wave than a sine wave (Fig. 12).

Figure 12. Two examples of flow limitation compared to normal breathing. If inspiratory effort only produces a pressure gradient above the threshold for a Starling resistor during the middle part of the breath, where effort is maximal, the flattening will be most noticeable over the middle half of inspiration (Fig. 12, middle panel). Conversely if inspiratory effort is very high, leading to a very high pressure gradient, inertial effects can lead to a very high initial air flow before the Starling effect limits the flow (Fig. 12, right panel). For this reason, the ResMed AotoSet Spirit CPAP system only looks at the middle half of inspiration in determining to what degree flow limitation may be present. A curvature or flattening index is derived, using the following steps (Fig. 13): 1. Any end-expiratory pause that has been included in the inspiratory half cycle is trimmed off as follows: the point where flow first reaches 75% of peak inspiratory flow is found, a backward search is then made to the point where flow last reached 25% of the peak. A linear extrapolation is made from the 75% point through the 25% point to the time where flow is zero. This is taken to be the earliest time at which inspiration could have started. 2. The approximate effect of the 0.1 Hz low pass filter on the flow-time curve is calculated and reversed as follows: a. Pass the wave through an identical 0.1 Hz low pass filter a second time b. Subtract the double filtered wave from the singly filtered wave to give a first-order approximation of the effect of the filter c. Add this difference back to the original wave. (For breaths of interest, inspiratory duration is about 2 seconds. The 0.1 Hz filter produces some distortion of the wave shape. For example, the trailing edge of a 2 s square wave sags by 20% of its amplitude, and the area under the curve is reduced by 10%. After correction, the trailing edge sags by only 2% of its amplitude, and the area under the curve is correct to 0.7%. The filter reduces the area under a sine wave by 10%, and after correction, the area is correct to 0.6%. This is more than adequate for current purposes.)

3.

4. 5.

6.

Figure 13. Trimming off any end-expiratory pause: (a) point where flow first reaches 75% of peak flow, (b) point where flow last reaches 25% of peak flow, prior to (a), (c) time at which line ab reaches x-axis, (d) if c is to left of origin, use origin. The inspiratory flow-time curve is scaled to unit length and unit mean height. (The curve is scaled as the shape, not the amplitude nor the duration, is required. Even with correct CPAP pressure, amplitude and duration are very variable, particularly in REM sleep, but also in stage 1 sleep and when the patient is awake and sighing.) Breaths with grossly abnormal shapes are detected by comparing with a stored template. Breaths that differ from the template by more than a threshold value at any point over the middle half are rejected. The default threshold is 1.0 units. It the breath is not rejected, the five-breath point wise moving average flow-time curve is updated. (The purpose of averaging the waveform over 5 breaths is to reduce the effect of cardiogenic airflow, the oscillation in the airflow caused by the heart beating against the left main bronchus. This could disguise an otherwise flattened flow-time curve. Swallows, coughs, talking etc can produce very abnormally shaped breaths. These are excluded in step 4 and do not influence the moving average.) The RMS deviation from unit scaled flow is calculated over the middle 50% of the normalized inspiratory time. This is the flattening, or curvature index. A mathematical square wave (total flow limitation) produces an index of zero. A normal index is approximately 0.2 units, and severe flow limitation typically produces an index of 0.1 units or less.

Pressure Increase Due to Detection of Flattening If the calculated flattening index is less than the pre-set threshold value, the mask pressure is increase. The default rate of increase in pressure is 3.0 cm water for each unit by which the flattening index is below the threshold, per breath. Moderate to severe flattening will cause the CPAP pressure to increase at 1.5 cm water per minute. The final applied pressure is the sum of the apnoea, snore and flow-flattening detection algorithms. Humidifier Air is humidified by having the output of the CPAP pass over a pool of heated water (Fig. 14).

Figure 14. Air humidifier. Patient Masks A variety of patient masks may be used, depending on the patients and clinicians preferences (Fig. 15).

Figure 15. Patient masks.

Mask Lead Detection The masks provide a fixed leak output port (mask vent) so that the patient does not rebreathe their expired carbon dioxide. The software can also measure any additional leak caused by poor mask fitting. Theoretically inspiratory and expiratory volumes (as measured by integrating the air-flow signal) should be equal. A net airflow (inspired expired) greater than the expected mask vent flow is equal to the leak. A leak flow of more than 0.4 l/s (24 l/min) is associated with patient discomfort, disturbed sleep and reduced efficacy of treatment. If a leak exceeds 0.7 l/s (42 l/min) for more than 20 s, the device will display, HIGH LEAK!!! Adjust mask, on the LCD screen. If enabled an audible alarm will also sound. Clinician Interface The clinician can connect the device to a computer to pre-set parameters and to download recorded information about the patients previous use of the device, incidences of apnoeas etc and recorded airflows and pressures. Patient Interface The device will automatically start and stop when the patient puts on or takes off the mask. For the first 3 minutes after starting, the device applies a fixed treatment pressure to give the patient a chance to adjust the mask to fit comfortably. Latest Model The ResMed S8 is the most recent model. It has the same functionality as the previous models but is more compact and quieter (Figure 16). It can be run from a variety of power sources including a 12 volt battery. Patient masks have also become more compact (Figure 17).

Figure 16 Escape S8 CPAP device for sleep disordered breathing.

Figure 17. Compact naso-pharangeal masks for use with CPAP devices. More Recent Developments The responses of these ventilators to apnoea detection, snoring and flow limitation are referred to as micro-control of sleep apnoea. Currently there is much research into developing macro-controllers of sleep apnoea which incorporate micro-control into a broader algorithm which can adapt to individual patient needs. Most machines can deliver BiPAP therapy as well as CPAP and there is an increasing tendency to use BiPAP (Figs 18, 19). Ventilators which automatically vary the airway pressure are now referred to as VPAP (variable positive airway pressure). This type of treatment is often referred to as pressure support (PS). This type of ventilation is also referred to as non-invasive ventilation (NIV). Invasive ventilation involves a sealed or semi-sealed system in which an endo-tracheal tube is inserted into the trachea with a cuff that is blown up to form an air-tight seal. The patients lungs are inflated by the ventilator and allowed to deflate through an exhaust valve. The patient has to be paralysed and heavily sedated or anaesthetized. The ventilator completely controls the patients respiration. This form of ventilation is only used for surgery requiring paralysis, severe respiratory failure or other serious illnesses requiring sedation and paralysis.

Figure 18 A macro-control system that adapts the micro-control (response to apnoea, snoring or airflow flattening) to changes in the patients state, such as airway impedance.

Currently the physician only needs to set the expiratory positive airway pressure (EPAP) and the inspired positive airway pressure (IPAP) can be set automatically and continually adjusted as the ventilator adapts to changes in the patients condition. The ventilator needs to recognize when the patient has sleep arousals. These can be due to pressure therapy being too aggressive, or, as in Figure 20, the patient suffering from an obstruction. Research is also underway to automatically set EPAP. These devices record the settings and patient variables and can thus start successive nights of therapy using the

previous nights final settings. Thus they can adapt over days as well as over short periods of time during the night. It has been proposed that in the future a trial period using an automated VPAP ventilator can be used instead of a sleep study for diagnosis and the settings learnt by the ventilator can then be applied from that time onwards if the trial shows that the patient does have OSA. The diagnosis is made by analysing the recorded instantaneous pressure and flow measurements as well as apnoea, hypopnoea, snoring and flow limitation detections. Two possible macro-control strategies are shown in Figures 21 and 22. The micro-control strategies can be expressed mathematically by the following equations:

Where P is the pressure change applied, K is a function of the detected sleep arousal, G is another function of the detected sleep arousal, Pmin is the minimum desired pressure S is an obstruction index (snoring or airflow flattening) ST is a threshold or tolerated obstruction index. Possible functions of K, G and ST are shown in Figures 23 to 27. A patients airway resistance can vary from day to day as well as during the night in different stages of sleep, so automatic changes in IPAP (or PS) may provide better treatment. Some patients find the pressure applied during inspiration to be disconcerting, so delaying sleep onset. With automatic settings, the pressure support can be kept low until the patient is asleep (as shown by regular respiration with a small reduction in tidal volume). Then higher pressure support can be given using an algorithm that slowly increases IPAP until the target ventilation (airflow into the lungs) is reached.

Figure 19 Micro-control response to apnoea, snoring or airflow flattening.

Figure 20 Arousal from sleep, as shown be the large 4th breath in this record, being a gasp or sigh, following obstruction, as shown by the previous three breaths exhibiting airflow flattening.

Figure 21 One possible macro-control strategy.

Figure 22 A second possible macro-control strategy.

Figure 23 K-gain pressure sensitivity reduction algorithm.

Figure 24 G-gain pressure sensitivity increase.

Figure 25 Obstruction threshold pressure sensitivity reduction algorithm.

Figure 26 Obstruction threshold pressure sensitivity increase algorithm.

Figure 27 Obstruction threshold as a function of treatment pressure.

VPAP ventilators have a triggered mode whereby a switch is made from EPAP to IPAP when the machines flow changes from expiration to inspiration. For this to happen a preset inspired airflow threshold must be exceeded. The threshold is a fixed fraction of the current peak flow, but constrained within certain limits. If apnoea is detected or the inspired airflow is too low to trigger the switch, after a pre-set period of time the ventilator switches to timed mode and applies its current IPAP. The time durations of inspiration and expiration are referred to as Ti and Te. Limits to these durations are TiMin, TiMax, TeMin and TeMax. TeMax is set by the physician. The others are set automatically by the ventilator to achieve the target airflow. Thus the ventilator has an automatically set variable backup respiration rate (number of breaths per minute) in case of apnoea. This can be set by the physician or is automatically set at the end of the ventilators learning period to the patients optimal breathing rate. The backup rate during spontaneous breathing (triggered breaths) is two-thirds of the apnoeic rate, which is set by the clinician. On transition from triggered to timed breaths, backup rate increases progressively, typically over 5 breaths, to the apnoeic rate. The rate of increase of the ventilated respiratory rate depends on how far ventilation (air flow rate into the lungs) is above or below the target ventilation. If the ventilation is below the target, the rate of increase of the respiratory rate is faster. Any triggered breath causes the backup rate to drop to two-thirds of the apnoeic rate and a return to triggered mode.

Maximum and minimum inspiration and expiration durations TiMin = apnoeic backup period * max. inspiratory fraction * 0.25 TiMax = maximum inspiratory fraction * current backup period, where current backup period = 60/current backup rate Current backup rate is measured in breaths per minute. Phase Phase of respiration is a useful concept in determining, during the cycling of respiration through inspiration and expiration, whether the patient is in flow triggered or timed mode. By definition phase = 0 at start of inspiration, phase = 0.5 at start of expiration, and phase approaches 1 at end of expiration. Rate of change of phase during inspiration is such that phase will reach 0.5 at TiMax, and rate of change of phase during expiration is such that phase will reach 1 at TeMax. If during inspiration phase increases smoothly and reaches 0.5 without a jump from a lower value, this means TiMax was reached, i.e. cycling occurred due to time when the phase reached 0.5 and no expiratory air flow was detected (Fig. 28)..

Figure 28 Timed inspiration during which no patient initiated airflow is detected.

If there is a jump in phase to 0.5, this signifies that cycling occurred due to patient initiated expiratory airflow triggering the end of inspiration and the start of expiration (Fig. 29).

Figure 29 Flow triggered breath during which the end of inspiration is triggered by detection of patient initiated expiratory airflow.

Similarly if during expiration the phase reaches 1 then drops from 1 to 0, this means that TeMax was reached, i.e. cycling occurred due to time and no inspiratory airflow was detected (Fig. 30).

Figure 30 Timed expiration during which no patient initiated inspiration is detected. Note that the inspiration has been terminated by patient initiated expiration triggering a switch from IPAP to EPAP.

If the phase did not reach 1 before dropping to 0, this means that triggering occurred due to detection of patient initiated inspiratory airflow (Fig. 31).

Figure 31 End of expiration triggered by inspiratory airflow detection. Note that inspiration has also been terminated by triggered detection of expiratory airflow.

An example of fully timed cycling (backup respiration) is shown in Fig. 32. Here there is no inspiratory nor expiratory airflow detected.

Figure 32 Backup respiration during which no patient initiated airflow is detected during inspiration or expiration.

Backup Respiratory Rate and Pressure Support The backup respiratory rate varies from breath to breath, so the slopes of the phase lines may vary from breath to breath. At each breath the backup period is calculated (= 60/backup rate, back up rate is measured in breaths per minute). Relative ventilation is defined as actual ventilation (litres/minute) divided by target ventilation. The error is the difference between the two. Relative ventilation error (RVE) = relative ventilation 1. RVE = 0 when ventilation is on target. RVE = -1 when there is no ventilation. RVE > 0 when ventilation is above target. Figure 33 shows how the ventilator changes its pressure support and backup rate when the patient stops breathing, probably as a result of going into rapid eye movement (REM) sleep. The 4th breath is a sigh and probably marks the onset of REM sleep. The patients muscles become paralysed and there is only expiratory effort from the 5th to 8th breath, then no respiratory from the 9th breath onwards. EPAP remains constant while IPAP and the backup rate are increased to try to meet the target ventilation.

Pressure

Flow

Phase

RVE
Figure 31 Probable onset of REM sleep with sigh at 4th breath. Ventilation changes from spontaneous to flow triggered expiration only from 5th to 8th breaths, then fully timed cycled breaths from 9th breath onwards.

Noninvasive Servoventilation Pressure support ( IPAP EPAP) is adjusted between a pre-set minimum (which may be 0, in which case the ventilator is essentially a CPAP machine) and a pre-set maximum, both being set by the clinician. The target ventilation is non-anatomical-deadspace ventilation, which is a crude approximation to alveolar ventilation. The aim is to prevent

high frequency small breaths being counted as effective ventilation. Anatomical deadspace is set manually and is an estimate, e.g. 100 ml for a 70 kg patient. Setting noninvasive servoventilation is a challenge because of leaks, and difficulty in estimating the appropriate target ventilation and speed of response. Leaks Constant leaks look like an inspiratory dc offset in the flow signal. They can be found by averaging over a few breaths. This can then be subtracted from the flow signal so that a stable base-line is achieved. This is necessary for sensitive triggering by patient initiated breaths. Large mouth leaks are more difficult to detect as the large compliance of the mouth (buccal cavity) makes it look like part of the lungs. Target Ventilation If the target is too high, the patient may fight the ventilator, which may lead to the ventilator fighting back and so disturbing the patients sleep. With VPAP the ventilator learns the correct setting over a one hour period while the patient is awake and resting with low support pressure. The target is set to 90% of the patients ventilation (litres per minute). The patients respiratory rate (breaths per minute) is also learnt at this time and the backup rate set (see later). The clinician can adjust any of these learnt values at any time. The dead space setting can also be adjusted at any time with an automatic correction of the target ventilation. Speed of Response If the speed of response is too slow to apnoea or hypopnea, unacceptable hypoxia may develop, especially at the onset of REM sleep. If the response is too fast, the patient may be aroused or have difficulty getting to sleep due to the response to the normal drop in respiratory rate and decrease in tidal volume with sleep onset. A patient being introduced to NIV may have their pCO2 well above the desired level initially. If the ventilation is set to a value that gives a normal pCO2 immediately, the patient may finish up with a respiratory alkalosis. This may decrease respiratory drive too much, decreasing upper airway muscle tone and exacerbate upper airway obstruction. There is a potential for active closure of the glottis. The solution is to set target ventilation initially to either the ventilation learnt from the awake resting period or that achieved by VPAP immediately after an acute exacerbation of OSA. The final target ventilation = initial target * (current pCO2/desired pCO2). The ramp up to the final target from the initial target is set to take place over several days to a few weeks. Clinician Setup Learn target ventilation and backup rate with patient awake Set target ventilation ramp if initial pCO2 above desired level Set EPAP Adjust inspiratory fraction of total cycle (default 40%) (in chronic obstructive pulmonary disease, COPD, decrease to 30 to 35%

Adjust cycling threshold (default of 25% of peak flow suits most patients), possibly increase in COPD or decrease in restrictive lung disease (e.g. pulmonary fibrosis)

Acknowledgements Material for these notes was provided by ResMed Pty Ltd, Bella Vista, NSW and the University of Florida.