Respiratory Physiology for the Anesthesiologist

Anesthesiologists take control of the respiratory function of millions of patients throughout the world each day. We learn to maintain gas exchange and use respiration to administer anesthetic gases through the completion of surgery, when we return this vital function to its legitimate owners, ideally with a seamless transition to a healthy postoperative course. With this being the nature of our trade, it is no surprise that a number of anesthesiologists have provided major contributions to the advancement of respiratory physiology, from early pioneers of the measurement of respiratory mechanics and gas exchange during anesthesia, to investigators of acute respiratory failure and extracorporeal gas exchange. Yet this article is for the clinical anesthesiologist: residents, educators, and those who fly solo, assisting them in perfecting their practice so that they can rouse every day healthy and happy individuals improved by their surgery. Aware that we cannot review every facet of this physiology, we focused our work on gas exchange and respiratory mechanics, developing four main themes (fig. 1): hypoxemia, hypercarbia, respiratory mechanics, and the mechanics of the lung and chest wall. To start, we recommend reading from classic textbooks of physiology.^1–3 

Hypoxemia deprives cells of their main source of energy—oxygen. When the Pao₂ decreases below normal levels, the gradient that drives oxygen from the blood into the interstitium and finally to the cell narrows, and may eventually compromise oxygen uptake. It is important in this context to differentiate the unique danger of a low Pao₂versus a decreased oxygen content; increasing oxygen content through, e.g., additional transfusions of red cells, will provide little help to the crucial lack of sufficient partial pressure. In situations of acute severe hypoxemia, swift treatment is often simple: administer oxygen! In less immediate but often complex situations such severe acute respiratory failure or intraoperative one-lung ventilation, understanding the physiology leading to hypoxemia is essential for choosing effective therapy.

Oxygen constitutes 21% of the air that we breathe and exerts 21% of the atmospheric pressure. At sea level (atmospheric pressure, 760 mmHg) once fully saturated by water vapor (47 mmHg), Po₂ in the inspired air (Pio₂) can then be calculated:

In the alveoli, oxygen is exchanged for carbon dioxide at a ratio of approximately 1.25:1, because a little more oxygen is used than carbon dioxide is produced. This phenomenon reflects the physiologic respiratory quotient (RQ), the ratio between carbon dioxide production (co₂) and oxygen consumption (o₂), normally approximately 0.8 when using a balanced diet. With a normal Paco₂ of 40 mmHg, Pao₂ at steady state will be

This is a simplified version of the alveolar air equation suitable for clinical use. Further development includes o₂ and co₂:

where co₂ is alveolar ventilation (_A) × its carbon dioxide concentration, with K = 1/760 mmHg; hence:

Under normal conditions, past the pulmonary capillaries, arterialized blood receives a small fraction of venous blood from minimal alveolar shunt (2 to 3%, increasing with age) and anatomic sources like bronchial, Thebesian, and some diaphragmatic veins, so that Pao₂ decreases just under 100 mmHg. Equation 2 also explains how hyperventilation may increase Pao₂ above 100 mmHg. The following paragraphs differentiate the main causes of hypoxemia based on the above schematic.

The most common cause of low Pio₂ is breathing at high altitude⁴ ; in La Paz, Bolivia, and in Lhasa, Tibet (both approximately 12,000 feet/3,600 meters), Pio₂ ≈ 100 mmHg, versus 150 mmHg at sea level (fig. 2). At high altitudes, the main concern for the anesthesiologist who treats native populations or transient workforces is hypoxemia; supplemental oxygen will be necessary for most patients and may be beneficial for the medical staff to maintain full physical and mental performance. Hypoxemia stimulates V_A mainly through carotid chemoreceptors⁴ ; hyperventilation may be profound, and the consequent decrease in Paco₂ will limit the reduction of Pao₂ according to equation 2. Near the summit of Mount Everest, at approximately 27,000 feet/8,200 meters and a barometric pressure of 284 mmHg, Pao₂ and Paco₂ measured in climbers breathing air were 25 and 13 mmHg, respectively.^5,6 

A low Pio₂ at sea level may be related to the accidental administration of a hypoxic mixture. The remarkable level of safety of anesthesia machines and portable field devices makes accidental delivery of a hypoxic mixture an exceptional event.

In addition to causes of low Po₂ upstream from the alveoli (low Pio₂), Pao₂ decreases mainly in relation to an increase in Paco₂. As per equation 2, Pao₂ falls when V_A decreases, at any given o₂ and any starting Paco₂. Table 1 applies the alveolar air equation to representative physiologic situations. Also, from equation 2, we appreciate the effect of fraction of inspired oxygen (Fio₂): a simple increase from 21 to 30% Fio₂ raises Pio₂ by 64 mmHg, which may restore Pao₂ to a value around 100 mmHg even in the presence of extreme hypoventilation.

In modern intensive care, we learn to take into account the possible contribution of extracorporeal carbon dioxide removal and carbon dioxide dialysis.^7,8  The lung of a patient undergoing low-flow, high-efficiency carbon dioxide removal for severe acute respiratory failure will maintain a comparatively high natural lung o₂ while achieving a very low V_A (low natural lung co₂) thanks to the high rate of carbon dioxide removal by the artificial lung. Very low respiratory quotient values, well below 0.5, are thus possible, requiring increased Fio₂ to maintain a constant Pao₂ despite an artificially normal or low Paco₂.

Oxygen transfer from the alveolus to the circulating hemoglobin is very rapid and seldom the cause of hypoxemia.⁹  The high partial pressure gradient and the thinness of the alveolo-capillary membrane allow ample time for oxygen transfer to reach steady state within the normal capillary blood transit time. Hence, transfer of oxygen is limited by pulmonary blood flow rather than diffusing capacity. However, regional capillary blood flow can vary substantially due to factors such as posture, alveolar pressure, and disease. Lung regions with short capillary transit time will then yield desaturated blood, which will not be compensated by areas with prolonged transit time, because hemoglobin will not saturate above 100%. Hence, any measurement of diffusion capacity, such as carbon monoxide diffusion, will be affected primarily by regional flow abnormalities rather than diffusion. Just like for oxygen, abnormal transfer of carbon monoxide is mostly due to ventilation/perfusion ratio (/) inequality, and carbon monoxide diffusion may be used as a generic index of abnormal pulmonary function, typically in the preoperative evaluation of patients scheduled to undergo pulmonary resections.¹⁰ 

The basic principle for alveoli to work as gas exchangers is to match ventilation (provides inflow and outflow for gases from and to the air) with perfusion (provides inflow and outflow of gases to and from the tissues). The two systems cross paths in the alveoli, where blood enters with different Po₂ and Pco₂ from the alveolar gas and exits with near equal concentrations. Normal V_A is approximately 4 l/min, and normal pulmonary blood flow (cardiac output) approximately 5 l/min, giving a / of 0.8. In reality, each individual alveolus’s / may differ substantially, from 0, that is shunt, to ∞, that is dead space and infinite combinations in between.

The most practical approach to / distribution is still the three compartments model proposed by Riley and Cournand^11,12  more than half a century ago: an ideal compartment, a nonventilated compartment, and a nonperfused compartment. In the nonventilated compartment, capillary blood reaches the pulmonary venous side (oxygenated blood) without participating at all in gas exchange. The proportion of nonoxygenated blood mixing into the arterialized side is defined as shunt or venous admixture and represents the amount of blood that would justify the deficit in arterial oxygenation if it were to cross the lung without participating to gas exchange. The degree of shunt is computed when mixed venous blood gases are available:

where _VA is the fraction of pulmonary blood flow exposed to ventilation, _TOT is the total pulmonary blood flow, Cco₂ is the oxygen content of ideal capillary blood computed from Pao₂ of the ideal compartment (equation 2), Cao₂ is the oxygen content of arterial blood, and Cvo₂ is the oxygen content of venous blood.

In reality, hypoxemia and hypercarbia in lung disease result from infinite combinations of / relationships that lie somewhere between shunt and dead space. The structure of the lung tends to predispose to / inequality, with both perfusion and ventilation being more efficient at the bases of the lungs, but the gradient of perfusion being steeper than ventilation, creating a situation prone to mismatch.^13,14  It is important to appreciate that although / inequality affects both oxygen and carbon dioxide exchange, the end result of is primarily hypoxemia, for two main reasons:

When low / develops in a portion of the lung, it decreases Po₂ and increases Pco₂ of arterialized blood. In spontaneously breathing patients, the higher Paco₂ stimulates ventilation, and Paco₂—but not Pao₂—will return toward baseline. If hypoxemia is severe, as at high altitudes, an additional stimulus to hyperventilation will further decrease Paco₂. In fully ventilated patients, who cannot increase their minute ventilation, Paco₂ will indeed increase, but the gradient between venous and arterial Pco₂ (46 to 40 = 6 mmHg) is very small, and the change is difficult to detect.

At first glance, it would seem that comparable but opposite changes of / would cancel each other out and yield normal gas exchange. This does not occur, and hypoxemia is the main result; this phenomenon is explained in terms of blood oxygen content and the hemoglobin–oxygen dissociation curve:

In an alveolo-capillary unit with normal /, Cao₂ is = 14 g/dl × 1.34 × 1 + 0.3 ≈ 19 ml of oxygen per dl, also expressed as 19 ml %. In a unit with low / and, e.g., arterial oxygen saturation (Sao₂) of 0.85, Cao₂ will be 16 ml %, close to a normal venous oxygen content (Cvo₂) of 14 ml %. In a unit with high / and a Pao₂ of, e.g., 400 mmHg, Cao₂ is almost 20 ml %, only slightly above normal. The high / unit cannot carry much more than 19 ml % of oxygen because once full saturation of hemoglobin is reached, a higher Pao₂ only increases dissolved oxygen, which contributes minimally to Cao₂.

Measuring regional / distribution is a complex task. The accepted standard has been the Multiple Inert Gas Elimination Technique, based on the infusion of a mixture of six dissolved inert gases with a known range of solubility, then measuring their concentrations in arterial blood and expired gas.^15,16  This technique has never been simplified enough for clinical use, nor does it provide spatial information of the areas of / inequality. Recently, the evolution of lung imaging with high-definition computerized tomography scan, positron emission tomography, and electrical impedance tomography¹⁷  is taking the visualization of / distribution closer to the clinical arena, although not yet to a quantitative evaluation of the precision provided by the Multiple Inert Gas Elimination Technique.

Pao₂ and Sao₂ are precise measurements of their respective physiologic entity, but alone do not provide adequate description of an ongoing respiratory impairment, nor allow accurate comparison among patients. It is important to know partial pressure (mmHg) or fraction (%) of oxygen upstream from the artery at which a certain Pao₂ is obtained¹⁸ : a Pao₂ of 100 mmHg has different physiologic implications whether obtained at a Pio₂ of 600 mmHg (Fio₂ 100%) versus a Pio₂ of 150 mmHg (Fio₂ 21%). The alveolo-arterial oxygen gradient is the most time-honored index of oxygenation, where Pao₂ is estimated from equation 2; the wider the alveolo-arterial oxygen gradient, the worse the performance of the lung as an exchanger. The most accurate application for this index is in patients intubated and breathing high Fio₂, where the linearity between this index and the level of venous admixture is best.¹⁸  More recently, the ratio between Pao₂and Fio₂ (Pao₂/Fio₂ or “P/F”) has largely substituted alveolo-arterial oxygen gradient because it does not require equation 2 and its relationship with venous admixture is less affected by Fio₂.^19,20 

Normally, transfer of oxygen from air to blood reaches full equilibrium within the alveolo-capillary unit, and Pao₂ ≈ Pao₂. Such degree of efficiency allows for even a low Pvo₂ to yield a normal Pao₂. However, in patients with intrapulmonary shunt, blood with low Pvo₂ enters the arterial stream without being saturated by alveolar oxygen and will decrease Pao₂ to a degree depending on the magnitude of the shunt fraction: the higher the shunt, the more the Pao₂ becomes dependent on the Pvo₂.

The next logical question is what causes a low Pvo_2. The following equation (Fick’s equation) illustrates the relationship between oxygen delivery to the tissues (Cao₂ × cardiac output) and oxygen utilization, oro₂:

Pvo₂ and mixed venous oxygen saturation (Svo₂) contribute to Cvo₂ (equation 4) and are used interchangeably as indices of tissue oxygen utilization.²¹  Unless indicated otherwise, Pvo₂ and Svo₂ refer to mixed venous blood, which is blood distal to the right atrium, where full mixing has occurred. Mixed venous blood can be sampled only via a pulmonary artery catheter. Central venous blood Po₂ obtained via a standard central venous catheter has been used as a surrogate, but it is not consistently accurate because is affected by separate currents of blood coming from venous districts with different Po₂.²² 

Pvo₂ may decrease for a defined number of factors: a low Pao₂, low hemoglobin concentration, low cardiac output, or a high o₂. With a low Pao₂, the lower Pvo₂ will further decrease Pao₂ until a new equilibrium of lower Pvo₂ and Pao₂ is reached. Normally, a Cao₂ of ~ 19 ml/100 ml of blood × cardiac output of 5 l/min yields a delivery of oxygen of ~ 1,000 ml/min; with a resting o₂ of ~ 250 ml/min, the (Cao₂ – Cvo₂) × cardiac output term is 750 ml/min, meaning that under normal circumstances, approximately 25% of the available oxygen is taken up by tissues. With a lower cardiac output or hemoglobin concentration and constant o₂, the term Cao₂ – Cvo₂ widens and Pvo₂ decreases, indicating an increased extraction of oxygen by the tissues. Note that increased extraction does not automatically indicate insufficient perfusion; hence, monitoring Pvo₂ or Svo₂²²  provides information on the status of oxygen tissue delivery but does not necessarily indicate a need to normalize oxygen delivery. It is important to note that the mathematical effect of the same change of each component of oxygen delivery on Pvo₂ may differ greatly from the clinical effect. This is shown in table 2, where halving any of the terms of the equation will equally half oxygen delivery and decrease Pvo₂. However, while halving hemoglobin concentration may be a relatively minor event in many critically ill patients, halving Sao₂ may be deadly. Furthermore, the physiologic relationship among the variables in discussion is complex and cannot be evaluated simply by the Fick’s equation. For example, accurate measurement of o₂ is elusive in the clinical setting,²³  and it is often referred to from tables; also, an increase in cardiac output in the presence of moderate shunt will generally increase shunt itself, negating the assumption that the other variables are constant.^24,25 

A succession of physiologic events decreases Pao₂ during general anesthesia.²⁶  Shortly after induction of anesthesia, the resting volume of the lung (functional residual capacity [FRC]), which is a reserve of oxygen, decreases by an average of 500 ml in adults; added to the volume lost by going from the erect to the supine position,²⁷  FRC is reduced from approximately of 3,000 ml to just more than 2,000, i.e., close to residual volume. The main cause is a loss of muscle tone with consequent cephalad displacement of the diaphragm; it is largely independent of the type of anesthesia and unrelated to neuromuscular blockade,²⁸  and is aggravated by pregnancy, obesity, and abdominal distention.²⁹  The decrease in resting lung volume causes a decrease in lung compliance, promotes cyclic airway closure at end-expiration and gas reabsorption, rapidly leading to atelectasis. Persistent blood flow during cyclic airway closure and through atelectasis causes areas of low / and shunt, contributing to hypoxemia. In the supine position, atelectases are prevalently located in the dorsal areas of the lung; normally limited to less than 10% of the lung volume, they may extend to 15 to 30% of it and be associated with persistent hypoxemia and respiratory complications.³⁰  Almost paradoxically, a high Fio₂ may add to lung collapse (absorption atelectasis) by substituting alveolar nitrogen with oxygen, the former not taking part in gas exchange. An Fio₂ of 80% may limit the degree of absorption atelectasis compared to100%. The combination of an Fio₂ not higher than 60%, and positive end-expiratory pressure (PEEP) during awakening from general anesthesia, may limit the degree of atelectasis and the incidence and duration of hypoxemia in the postoperative period.^31,32 

Different from hypoxemia, hypercarbia is generally well tolerated unless severe enough (above 80 to 100 mmHg) to cause obtundation and respiratory arrest. In susceptible patients, hypercarbia may be perilous through various mechanisms, by increasing intracranial pressure in patients with cerebral edema or exacerbating pulmonary hypertension in adults with right ventricular dysfunction and in children with congenital heart disease. On the other hand, moderate hypercarbia may be a favorable condition in a number of pathologic situations, such as in circulatory shock by improving local blood flow, and in acute respiratory distress syndrome (ARDS) as an accepted consequence of a ventilatory strategy of limited tidal volume.³³ 

Carbon dioxide and water are the end-products of aerobic metabolism; carbon dioxide is the main source of acid in the body and requires a highly efficient system for removal. In plasma, dissolved carbon dioxide (Paco₂ × 0.03) is in equilibrium with carbonic acid, hydrogen ions (buffered by hemoglobin), and bicarbonate ions, and is measured in mEq/l as “total CO₂” in a standard metabolic profile.³⁴  From the pulmonary capillary, carbon dioxide transfers to the alveolus by the same physical forces that transfer oxygen the other way. By the end of the capillary, carbon dioxide is in equilibrium in the two phases, and normally Paco₂ ≈ Paco₂. At steady state, Paco₂ results from the equilibrium of production by cellular metabolism (co₂) and elimination by V_A,

where K = 1/760 as in equation 2.

Elimination of carbon dioxide is the essential target of breathing. Paco₂ and V_A are tightly bound, and small increases in Paco₂ have immediate effect on the total minute ventilation: for each 1 mmHg Paco₂, minute ventilation increases by 1 to 2 l/min, retuning Paco₂ to its normal value (fig. 3). Based on the above schematic, we can differentiate the main causes of hypercarbia.

Increased aerobic metabolism occurs in hypermetabolic states such as exercise, fever, shivering, hyperthyroidism, excessive caloric/carbohydrate intake and, to its maximum extent, in neuroleptic malignant syndrome and malignant hyperthermia. Except for the latter two entities, increase of co₂ is either mild, as in high carbohydrate intake, or transient, as in shivering, and most of the time offset by the increase of V_A. When ventilation is limited by disease, sedation or residual anesthesia, an episode of high fever or shivering may overwhelm the capability of the respiratory system and result in levels of hypercarbia that require mechanical ventilatory support. co₂ may also change abruptly under non–steady state condition by changes in body carbon dioxide stores (bicarbonate and carbonic acid) or by exogenous infusion. For example, the rapid administration of sodium bicarbonate with the aim of neutralizing a metabolic acidemia (i.e., converting bicarbonate to carbon dioxide and water) results in a stoichiometric increase in carbon dioxide production: 100 mEq of sodium bicarbonate more than 20 min may cause up to 50% increase in co₂ of the same duration (5 mEq/min).

Hypoventilation causes hypercarbia and hypoxemia, particularly at low Fio₂. Looking at equation 2, at constant o₂, a decrease in V_A implies a decreased delivery of oxygen to the alveolus, while Paco₂ increases because a higher carbon dioxide concentration in the expired air is necessary to eliminate the same amount of carbon dioxide per unit time at constant co₂. Common causes of hypoventilation include a decreased central drive to breathe, as it occurs with hypnotics, volatile anesthetics, and opiates (fig. 3); respiratory muscle weakness, as in Guillain-Barré syndrome, myasthenia, and polyneuropathy of critical illness; and high resistive (severe asthma) or elastic (abdominal distention) ventilatory loads.

Ventilation is a to and fro process that starts in the airways, which are one-way conduits not involved in gas exchange. This fraction of tidal ventilation (150 to 200 ml in adults) is thus “dead volume” or anatomic dead space of the airways (V_D/V_{T AW}). In mechanically ventilated patients, V_D/V_{T AW} also includes any additional respiratory equipment where flow of gas is bidirectional, such as masks and corrugated tubing added distally to the Y-piece. In addition to V_D/V_{T AW}, where no carbon dioxide is ever exchanged, dead volume may also occur at the alveolar level or alveolar dead space (V_D/V_{T ALV}), where normally all carbon dioxide is exchanged, and represents a wasted fraction of V_A. In analogy with shunt in the three compartments model of Riley and Cournand,^11,12  V_D/V_{T ALV} is the proportion of alveolar ventilation that could justify the observed decrease in mixed Paco₂ if it were to be expired with a Pco₂ equal to the one of the inspired air. V_D/V_{T AW} + V_D/V_{T ALV} is physiologic dead space (V_D/V_{T PHYS}).

Measurementof V_D/V_{T PHYS} is commonly obtained from a modification of the Bohr equation³⁵  where Paco₂ substitutes Paco₂, the latter being difficult to measure:

Peco₂ is the mixed exhaled Pco₂, which includes both alveolar (≈ arterial) CO₂ and airways (≈ 0) CO₂, where a normal value of 30 mmHg gives a V_D/V_{T PHYS} = (40 to 30)/40 = 1/4 = 25%, i.e., 150 to 200 ml. The availability of continuous capnography allows estimation of Peco₂ from the midpoint of the initial rise of exhaled trace, or Phase I, and alveolar Pco₂ from the highest point (also end-tidal Pco₂, Petco₂); substituting Peco₂ with Petco₂ in equation 7, we measure V_D/V_{T ALV} continuously and noninvasively.³⁶  However, using Paco₂ as a surrogate of Paco₂ adds to V_D/V_{T PHYS} a source of carbon dioxide that does not contribute to Paco₂ because it is not dead space but shunt and / inequality.^35,37  Shunt widens the A - a gradient for carbon dioxide just like the A - a gradient for oxygen, but the effect of shunt on Paco₂ will be less visible than on Pao₂ because the difference in Pco₂ between venous and arterial blood is much smaller than the difference in Po₂. For example, a 30% shunt would increase Paco₂ by only 2 mmHg, a barely noticeable change; however, that would correspond to an increase in calculated V_D/V_{T PHYS} larger than 30% because of the increase in gradient, not in absolute Paco₂ value. Differentiating the mechanism leading to hypercarbia may be important; for example, the magnitude of V_D/V_{T PHYS} has been correlated with the rate of mortality in ARDS;³⁸  given the high shunt fraction present in severe ARDS, part of the increase in Paco₂ that contributes to the high calculated V_D/V_{T PHYS} may be due to decreased ventilation and not decreased blood flow. Similarly, in lung disease characterized by major alterations in / relationship, such as severe chronic obstructive pulmonary disease (COPD), a low Petco₂ may reflect the overall / inequality in addition to dead space.

Occasionally, alveolar units with altered mechanics may empty throughout expiration and add carbon dioxide, increasing Paco₂, and Petco₂ may exceed Paco₂, the limit being mixed venous carbon dioxide tension (Pvco₂). This is rare, and Petco₂ is normally equal to or lower than Paco₂; as long as the exhaled carbon dioxide trace on the capnogram shows a stable plateau, the Petco₂ can be accepted as representative of alveolar carbon dioxide.

The above discussion refers to steady state conditions, but we should also take into account the peculiarities of dynamic or transient carbon dioxide changes.³⁹  When ventilation increases, Pco₂ decreases rapidly, the rate of decrease being governed primarily by V_A. When ventilation decreases, the process is overall different, because the rate of increase of Pco₂ is governed only by co₂. When an individual holds breath for several seconds (apnea), an initial increase of Paco₂ toward the Pvco₂ value (an additional 6 mmHg or 21 ml of carbon dioxide with an FRC is 2.5 l) will take place rapidly—less than 6 s at constant cardiac output. However, past this equilibration time, any further increase of Pco₂ takes place at a much slower speed (3 to 6 mmHg/min) due to the high capacity of the body stores for carbon dioxide. Important dynamic changes in carbon dioxide occur also with changes in pulmonary blood flow. A fall in cardiac output at constant V_E causes a decrease in carbon dioxide transferred to the alveoli, reflected by lower Petco₂ and increased V_D/V_{T ALV}.

Air moves in and out of the lungs by the cyclic action of the respiratory muscles. Each breath starts with pressure applied by the respiratory muscles (negative pressure) or by a ventilator (positive pressure), generating a flow of gas through the airways that expands the volume of the lung and surrounding structures. This process is opposed each time by resistance to gas flow in the airways and elastance (stiffness) of the lung and surrounding structures. Note that the term elastance is seldom used in clinical lingo, in favor of its reciprocal, compliance; thus, an increased stiffness of the lung is a low compliance. These five variables—pressure, flow, volume, resistance, and compliance—are intimately intertwined in determining the mechanical process of breathing, whether it is during relaxed breathing, during arduous exercise, or supplied by a ventilator.⁴⁰ 

Normal airway resistance (R_aw) is 1 to 3 cm H₂O/l/s; normal compliance of the respiratory system (C_RS), i.e., lung plus chest wall, is 80 to 100 ml/cm H₂O.⁴¹  Separating PIP and P_PLAT is key for the correct measurement of R_aw and C_RS.⁴²  PIP is the highest pressure measured at the airway during a mechanical breath and is determined by size of the tidal volume (V_T), inspiratory flow rate, R_aw, and C_RS. P_PLAT is the pressure measured at the end of inspiration when flow is held, i.e., zero flow, and is determined solely by V_T and C_RS. Basic ventilatory settings do not separate PIP and P_PLAT, but most ventilators provide ways to measure and visualize them. An end-inspiratory pause closes the inspiratory valve without opening the expiratory valve, pausing flow for an adjustable period of 0.5 to 2 s, and displays the values of R_aw and C_RS. A typical inspiratory pause is executed with the ventilator on volume control, constant flow rate (“square flow waveform”); if possible, setting the flow rate at 60 l/min (1 l/s) simplifies R_aw calculation. We caution the practitioner to examine the actual traces on the ventilator screen, because inaccuracies do occur from unplanned changes in settings or artifacts due to spontaneous respiration, coughing, or leaks. Most critical care ventilators add the pause to a single breath without altering other parameters, thus allowing measurement on the ongoing ventilatory settings. Most anesthesia ventilators incorporate an end-inspiratory pause as a percentage of the set inspiratory time to all subsequent breaths; by doing so, the delivery of the breath changes, and with that the mechanics of the breath. Figure 4 shows such occurrence: by including the zero-flow pause in the set inspiratory time, the ventilator must deliver the set volume in a shorter time, which can only be done by increasing inspiratory flow, which in turn will result in higher PIP and sometimes in changes of P_PLAT and V_T, altering the measurement of R_aw and C_RS. Most anesthesia ventilators do not display the actual value of R_aw, but one can still get an approximate idea by assessing the difference between PIP and P_PLAT: when R_aw is increased by, e.g., bronchospasm, separation of PIP and P_PLAT is clearly noticeable, and may subsequently change with an intervention such as administering a bronchodilator. Under the appropriate circumstances, C_RS (but not R_aw) can be estimated without performing an inspiratory pause, provided that the pressure trace shows an identifiable plateau, as in figure 4C. This happens when flow is delivered with an initial peak followed by progressive decay throughout inspiration (“descending ramp” pattern), ideally reaching zero flow. This is the rule in pressure control ventilation and may be obtained also in volume control under appropriate settings.

As the mechanics of alveolar units vary, the time required to move air in and out may also vary both within and among areas of the lung. The time required by the respiratory system to change volume is described by the concept of time constant (τ):⁴³ 

Time constants are used to describe the exponential time change of a dynamic system to reach a new steady state, in our case filling or emptying of the lung or a lung unit. One τ is the time in seconds taken to achieve 63% of maximal volume change of the lung unit; every successive τ will change 63% of the previous volume; hence it takes 4 τ’s to reach about 95% of the new steady state. The clinical relevance of this concept is that it allows to distinguish different forms of respiratory failure from a mechanical standpoint. For example, asthma and COPD have slow τ because they have high R_aw and normal or high C_L; conversely, ARDS has fast τ because C_L is low and R_aw only moderately increased. The slow τ of asthma and COPD is responsible for lung hyperinflation and expiratory flow limitation. The fast τ of ARDS is responsible for the short inspiratory time (when not assisted by the ventilator) and tendency to de-recruit.

During expiration, the most relevant measurement is auto-PEEP—also “intrinsic PEEP” or “occult PEEP.”⁴⁴  Auto-PEEP is positive alveolar pressure (P_ALV) at end expiration, where normally we expect it to be equal to the pressure measured at the airway opening (P_PLAT), i.e., 0 cm H₂O (atmospheric pressure) or PEEP. Auto-PEEP occurs by two main mechanisms, shown in figure 5. First, there is insufficient time to exhale the full V_T because the relaxation time of the respiratory system is longer than the allocated expiratory time, be it spontaneous or set on the ventilator. Such is the situation of long τ in pure bronchospasm. Second is by the presence of expiratory flow limitation, characteristic of COPD due to loss of elastic recoil of the lung and supporting structure of the airways. As the volume of the lung decreases during exhalation, so does the diameter of the airways, possibly down to full collapse; this is often further amplified when the expiratory muscles are activated and generate a positive pressure around the airways.

Auto-PEEP has some of the effects of set PEEP: it may recruit the lung and may damage the lung. When generated during mechanical ventilation by prolonging inspiratory time, auto-PEEP may improve Pao₂ through alveolar recruitment⁴⁵ ; when generated by trapped gas in status asthmaticus, auto-PEEP may overstretch and damage the lung.⁴⁶  Just like set PEEP, auto-PEEP may decrease cardiac output and arterial blood pressure.⁴¹  Differently from set PEEP, auto-PEEP increases work of breathing: when breathing normally from FRC or from a set PEEP, a minimal negative pressure starts inspiratory flow and initiates a breath; with auto-PEEP, this negative pressure has to overcome the positive P_ALV before decreasing to less than 0 and starting inspiratory flow. This added pressure constitutes wasted effort because it does not generate any volume, and may ultimately cause fatigue in susceptible patients.⁴⁷ 

Measurement of auto-PEEP is carried out on a ventilator through an expiratory pause,⁴²  which closes the expiratory valve without opening the inspiratory valve, and equilibrates P_ALV across different lung units; the positive airway pressure trace will rise to a plateau above 0 or PEEP, the difference being auto-PEEP. Just like an inspiratory pause, this maneuver is optimal in the absence of spontaneous efforts and with a low respiratory frequency. Of note, even during a pause, the value of positive airway pressure will still be the lowest measured, and if the pause is not sufficiently long, there may be areas of the lung where P_ALV is still higher than the measured auto-PEEP.⁴⁸  When this maneuver is not available, as in most anesthesia ventilators, the presence of auto-PEEP can be inferred, but not quantified, by examining the expiratory flow trace, which will be truncated before reaching zero flow at end expiration, as shown in both panels of figure 5.

Bedside measurements of respiratory mechanics provide values that pertain to the respiratory system as a whole. Similar to /, minor mechanical inequalities develop during normal respiration, and their impact is trivial; however, regional changes in mechanics develop with the onset progression of respiratory disease, affect respiratory function, and may invalidate standard measurements.

Regional inhomogeneity in R_aw develops in asthma and COPD due to weakened elastic recoil, active bronchospasm, inflammatory changes, and trapping of gas. A common result is regional lung hyperinflation, described during spontaneous as well as general mechanical breathing,^48,49  and visually confirmed by advanced imaging techniques.⁵⁰  Dynamic reduction in airway caliber may progress to complete obstruction and exclude hyperinflated areas of the lung from standard measurements, underestimating the potential for lung damage, hypoventilation, and hemodynamic compromise.

Regional inhomogeneity of C_RS during general anesthesia and respiratory failure was demonstrated with the introduction of lung computerized tomography in the 1980s,^51,52  opening the way to the understanding of alveolar recruitment and damage during mechanical ventilation, and to the development of innovative strategies of lung protective ventilation.⁵³ Figure 6 shows computerized tomography scans of patients with acute respiratory failure and severe hypoxemia with similarly diminished C_RS by bedside measurement, whose regional mechanical characteristics are manifestly different. The “stiff” lungs will require high ventilating pressures that will distribute throughout severely diseased lung tissue, possibly causing diffuse damage. The “small” lungs will receive ventilating pressures intended to the entire lung but delivered mainly to a small portion of it, possibly causing damage to the small lung and doing very little to the consolidated lung. The last set of lungs seems impossible to ventilate, which was indeed the case: this patient was treated by extracorporeal membrane oxygenation for three weeks.

Further attention to ventilatory mechanics during anesthesia and intensive care requires understanding the interaction between lung and chest wall.⁵⁴  C_RS measured ordinarily at the bedside comprises the compliances in series of lung (C_L) and chest wall (C_CW). Like electrical capacitances in series, the reciprocal of each compliance adds up to the reciprocal of the total:

Normal values in ml/cm H₂O are: C_RS = 100, C_L = 200, and C_CW = 200. During mechanical ventilation, we are interested in the lung: it is the lung that we want to expand and we do not want to damage. By measuring C_RS, we extrapolate to the lung what we are actually measuring on the lung and chest wall, and for the most part, the information is valuable.

Under certain circumstances, understanding the mechanics of ventilation may require knowledge of the separate contributions of lung and chest wall. During normal respiration, at FRC the inward elastic force of the lung is equally balanced by the outward elastic force of the chest wall, resulting in a P_ALV equal to atmospheric (0) pressure, and a pressure in the pleural cavity (P_PL) slightly less than atmospheric pressure. Due in part to gravity, P_PL is more negative in ventral than dorsal areas of the lung in the supine position, reversed in the prone position,⁵⁵  and again more negative in the nondependent lung than in the dependent lung in the lateral position.⁴⁷  As FRC decreases after induction of general anesthesia, P_PL increases toward and above 0, promoting airway closure and atelectasis.²⁶  The key physiologic parameter here is the pressure generated across the lung, or transpulmonary pressure (P_TP):

A positive P_TP keeps the alveoli open: under normal circumstances at FRC, P_ALV is 0 and P_PL slightly negative, hence P_TP is positive; with PEEP, P_ALV is positive and P_PL less positive than P_ALV; with anesthesia, pulmonary edema, and abdominal distention, P_PL increases above P_ALV and P_TP turns negative, causing alveolar collapse and atelectasis.⁵⁶  Clinically, P_ALV is reliably estimated by P_PLAT, but P_PL requires indirect measurement by means of special thin-walled balloons positioned via the nares into the esophagus (esophageal pressure).⁵⁷  The value of esophageal pressure as a measurement of P_PL has been somewhat controversial, but recent studies have confirmed its reliability in selected situations with close attention to proper positioning and calibration.^58,59  Measuring P_TP as the target to PEEP titration in patients with ARDS has revealed the need for higher PEEP to provide effective recruitment of regional lung collapse and allow protective ventilation.^60,61  Under general anesthesia, the use of a lung protective strategy of higher PEEP with or without direct measurement of P_TP seems to ideally suit morbidly obese patients^62,63  and patients undergoing laparoscopic surgery, particularly in the steep Trendelenburg position, where P_PL increases further.

During a spontaneous inspiration, the chest wall expands by its elastic recoil and by the action of the inspiratory muscles, making P_PL, the pressure surrounding the lung, more negative than at end expiration. Since at the airway opening the pressure remains atmospheric, P_TP is positive and lung volume increases. During controlled mechanical ventilation, pressure is positive, P_ALV increases, P_PL increases a fraction of P_ALV, and P_TP is again positive. For the same pressure applied and constant respiratory mechanics, changes in P_TP will be the same, and V_T will be the same in spontaneous and mechanical breathing. During assisted breathing, where tidal ventilation is the result of a cooperative effort of patient and ventilator, the mechanical situation is more complex, with a few points worth noting. First, for the same V_T, the change in P_TP will be the same, independent of the mode of ventilation (fig. 4), and so will be the potential degree of lung damage. Second, when the lung volume increases during spontaneous ventilation, P_PL becomes more negative; when the lung volume increases during controlled tidal ventilation, P_PL becomes more positive; in both cases, P_ALV must be lower than the pressure in the airways for inspiratory flow to occur. Therefore, while during controlled mechanical ventilation P_ALV is higher than the set end expiratory pressure throughout the respiratory cycle, during spontaneous inspiration, P_ALV must become lower than the end expiratory pressure.

These differences have profound influences on the interaction between heart and lung function, and ultimately on hemodynamics and on the fluid balance of the lung interstitium. In particular, when spontaneous breathing generates significant decreases in intrathoracic and/or alveolar pressures, these may contribute to the generation of pulmonary edema, cardiogenic and not,^64,65  and may worsen lung injury.⁶⁶ 

Support for the preparation of this manuscript was provided solely from institutional and/or departmental sources.

The authors declare no competing interests.