Pulmonary complications are a major source of morbidity in surgical patients, second only to cardiovascular events as a cause of perioperative death. The overall incidence of pulmonary complications following all types of surgery is approximately 5 per cent. Several risk factors have been associated with a higher rate of pulmonary complications, including age, male gender, emergency surgery, American Society of Anesthesiology physical status classification, and the length of the surgical procedure.

While these conditions contribute to an overall increase of perioperative complications, two additional factors specifically predispose to the development of pulmonary complications: pre-existing respiratory disease and surgery of the chest or upper abdomen. Clinically important atelectasis, bronchospasm, retained secretions, and infectious complications may develop in 20 to 70 per cent of patients with pre-existing respiratory disease. This, in the most severe cases, can result in acute respiratory failure and the need for prolonged mechanical ventilation. Death from pulmonary complications occurs in 7 per cent of surgical patients with moderate to severe chronic lung disease. A recent prospective study confirmed that pulmonary complications occur in 33 per cent of patients with mild to moderate chronic lung disease who undergo upper abdominal surgery.

Evaluation for surgery must consider the effects of anaesthesia and surgery on respiratory function. In addition, special consideration has to be given to patients who are at high risk for perioperative pulmonary complications and to those who are scheduled to undergo lung surgery. Finally, optimization of their clinical status should be performed preoperatively, in an attempt to minimize morbidity.

Pulmonary function is altered in several ways during and after surgery. Important aspects include the breathing pattern, lung volumes, gas exchange, and defence mechanisms. These changes occur to some degree in every surgical patient, but have a greater impact in those with pre-existing respiratory disease.

Under normal circumstances, spontaneous minute ventilation (tidal volume × respiratory rate) increases in response to elevated arterial carbon dioxide tension and to a reduction of the arterial oxygen tension. Most drugs used during anaesthesia impair this normal ventilatory response to hypercapnia and hypoxaemia. Opioids can produce profound respiratory depression. Even small doses of morphine sulphate (for example, 7.5 mg subcutaneously) can blunt the ventilatory response to hypercapnia and hypoxia, and all the currently available opioid agonists seem to exert a similar effect at equipotent analgesic doses. The inhalational anaesthetics halothane, enflurane, and isoflurane also depress respiratory drive. Ventilatory response to hypoxaemia is virtually abolished at inspired concentrations of halothane as low as 0.1 to 0.2 per cent, which are likely to be present still during early recovery from general anaesthesia ( Fig. 1 65.). The potent inhalational agents decrease spontaneous minute ventilation, producing a decrease in tidal volume and resultant rapid, shallow breathing. This pattern increases the ratio of dead space to tidal volume (V&subD;/ V&subT;), leading to CO&sub2; retention, and may facilitate the development of alveolar collapse and atelectasis.

Changes in static lung volumes occur during surgery and anaesthesia. In 1933 Beecher reported a decrease in vital capacity of approximately 45 per cent and in functional residual capacity of approximately 20 per cent after laparotomy. These changes persisted from 1 to 2 weeks postoperatively. Subsequent studies confirmed and extended this original observation. The following characteristics have been established: functional residual capacity is reduced during general anaesthesia by about 20 per cent below the value measured in the awake, supine position. This change occurs early in the course of the surgical procedure, appears not to be influenced by use of muscle relaxants, and is more pronounced in the elderly. Upper abdominal incisions and thoracotomy are associated with the largest decrease in lung volumes postoperatively, followed by lower abdominal surgery; peripheral procedures do not cause persistent changes. The cause of the reduction in lung volumes was elucidated in 1974 by Froese and Bryan. Using lateral chest radiograms they found that the diaphragm ascended into the chest by about 2 cm during anaesthesia with or without paralysis, and this change accorded roughly with the decrease in functional residual capacity (Fig. 2) 66.

Breathing at low lung volumes may affect gas exchange. A 20 per cent reduction in functional residual capacity added to an equivalent decrease secondary to assumption of the supine position brings the resting lung volume (the functional residual capacity) quite close to residual volume. This circumstance may favour the development of atelectasis in the dependent zones of the lung following induction of general anaesthesia. Resistance to airflow also increases at lower lung volumes. Airway resistance is inversely proportional to the fourth power of the radius of the airway. At a reduced functional residual capacity, small decreases in volume result in a marked increase in resistance, which may cause hyperinflation and air-trapping in patients with asthma or chronic obstructive pulmonary disease.

Uncomplicated general anaesthesia produces abnormalities in gas exchange that may become clinically significant in patients with pre-existing respiratory disease. Adequate gas exchange is dependent upon homogeneous matching of ventilation and perfusion at the alveolar level. Inspired gas delivered to lung regions that have no pulmonary capillary blood flow cannot take part in gas exchange and, conversely, pulmonary blood flow distributed to regions without ventilation cannot be oxygenated. The term ventilation/perfusion ratio (V/ Q) is commonly used to refer to this relationship. Considering the lung as a whole, typical resting values may be 4 l/min for alveolar ventilation and 5 l/min for pulmonary blood flow. Thus, the overall V/ Q ratio would be 0.8, which happens to be close to the normal ratio between CO&sub2; production and O&sub2; consumption (respiratory quotient). In fact, ventilation and perfusion are not uniformly distributed and may range all the way from unventilated to unperfused alveoli, with infinite gradation in between.

A simplified approach to V/ Q physiology is shown in Fig. 3 67. The ventilated but unperfused alveoli comprise the dead space, the perfused but unventilated alveoli the shunt, and gas exchange is confined to the ‘ideal’ alveoli. In clinical practice, arterial hypoxaemia is frequently the result of low V/ Q ratios rather than of a true shunt and is more correctly referred to as venous admixture. Usually, an increased dead space may be offset by increasing minute ventilation. An abnormal venous admixture up to about 30 per cent may be corrected with higher inspired oxygen concentrations.

In the awake individual with normal lungs, the amount of inefficient ventilation causing venous admixture is minimal, resulting in an alveolar/arterial Po&sub2; gradient of about 10 mmHg. During uncomplicated general anaesthesia the alveolar/arterial Po&sub2; gradient may increase to 30 to 50 mmHg. The increase is secondary to the reduction in lung volumes and its effects on the development of atelectasis with alveolar and/or airway closure and resultant V/ Q mismatch. Furthermore, inhalational anaesthetics may impair gas exchange by inhibiting the physiological mechanism of hypoxic pulmonary vasoconstriction that tends to divert pulmonary blood flow away from areas of low ventilation to regions of increased ventilation, preserving the V/ Q relationship and arterial oxygenation.

Elimination of CO&sub2; is also affected during general anaesthesia, as a result of changes in the ratio of dead space to tidal volume (V&subD;/ V&subT;). Changes in V&subD;/ V&subT; as they occur during general anaesthesia are shown diagrammatically in Fig. 4 68. The anatomical dead space ((V&subD;/ V&subT;)&suba;; oral cavity, pharynx, and conducting airways) measures about 30 per cent of the tidal volume in spontaneously breathing individuals. During general anaesthesia, the (V&subD;/ V&subT;)&suba; also includes that segment of the anaesthetic circuit in which the gas flow is bidirectional (endotracheal tube or face-mask and tubing distal to the ‘Y-piece’ connector). When the trachea is intubated, the (V&subD;/ V&subT;)&suba; is slightly decreased because the upper airway is bypassed. When anaesthesia is administered by face-mask, on the other hand, the volume of the mask adds further dead space, and the (V&subD;/ V&subT;)&suba; increases to approximately 40 per cent.

The term ‘physiological dead space’, (V&subD;/ V&subT;)&subp;&subh;&suby;&subs;, defines areas of the lung parenchyma beyond the conducting airways where gas exchange does not occur. In the awake individual with normal lungs, no significant (V&subD;/ V&subT;)&subp;&subh;&suby;&subs; is detectable. During general anaesthesia, the (V&subD;/ V&subT;)&subp;&subh;&suby;&subs; increases to approximately 30 per cent of the tidal volume. Thus, a seemingly adequate minute ventilation of 5 to 6 l may result in an alveolar ventilation as low as 2 to 3 l per minute, which could cause CO&sub2; retention.

Studies carried out with the aid of the multiple inert gas washout technique allow summarization of the effects of anaesthesia on gas exchange as follows.

1.The alveolar/arterial Po&sub2; gradient is increased during anaesthesia, and this change is markedly affected by age.

2.The decrease in Po&sub2; is secondary to an increased distribution of flow to areas of decreased ventilation, most commonly the dependent areas.

3.The increase in V&subD;/ V&subT; seems to be secondary to increased distribution of ventilation to areas of lesser perfusion.

4.The major differences are between the awake and anaesthetized state; paralysis and controlled ventilation do not greatly alter overall gas exchange.

Inspired air is normally warmed and humidified in the nasopharynx. This process facilitates the clearance of airway secretions by optimizing ciliary function. Anaesthetic gases are essentially dry as they leave the standard anaesthesia machine; dryness tends to damage the respiratory epithelium. Endotracheal intubation exacerbates this problem by bypassing the upper airway. The cough mechanism is depressed during general anaesthesia as well as during spinal anaesthesia. The combination of these factors predisposes to inflammation of the respiratory mucosa and to retained secretions, which may favour alveolar collapse, bronchospasm, and infection.

Normal function of the immune system in humans is altered in the immediate postoperative period. Unfortunately, clinical testing in this area is confronted with the difficulty of separating the effects of multiple intraoperative factors that may impact on immunity. It appears that many of the immune changes seen in surgical patients are primarily the result of the surgical trauma and of endocrine responses, rather than of the anaesthetic exposure itself. In-vitro studies suggest that anaesthetic agents may have a direct effect on granulocyte and monocyte function and on the release of immunological mediators. The clinical significance of these observations, however, is not defined at the present time. A synopsis of the possible effects of anaesthetic agents on different components of the immune response is shown in Fig. 5 69.

Identification of patients at increased risk for perioperative respiratory complications is important since optimization of their clinical status can decrease morbidity. During clinical evaluation, one should give emphasis to the current respiratory status of the patient, to coexisting cardiovascular disease, to a prior history of treatment for pulmonary disease, to occupational exposures, and to the use of medications such as bronchodilators or corticosteroids. Among the risk factors possibly associated with an increased rate of perioperative pulmonary complications, the following have been reported consistently.

This is associated with an increased perioperative mortality because of the effects of smoking on both the cardiovascular and respiratory systems. A history of tobacco use is a sensitive predictor of lung disease and of postoperative respiratory complications: compared with non-smokers, cigarette-smokers have a two- to threefold greater incidence of perioperative pulmonary complications. Smoking also increases the concentration of carboxyhaemoglobin in the blood (3–15 per cent), thereby reducing the amount of haemoglobin available for oxygen transport. Cessation of smoking 12 to 18 h preoperatively significantly lowers carboxyhaemoglobin levels and may decrease heart rate, arterial blood pressure, and catecholamine levels. The impact of smoking cessation on the incidence of perioperative pulmonary complications, however, is not detectable unless smoking is discontinued for more than 8 weeks before surgery. This conclusion is in agreement with the observed improvement in respiratory symptoms and lung volumes that occurs over a period of months following the cessation of smoking.

Obesity reduces total lung volume and functional residual capacity, thus predisposing to hypoxaemia and atelectasis. Lower pre- and postoperative arterial oxygen saturation was found, compared with non-obese controls, in patients undergoing jejunoileal bypass for morbid obesity. The scanty data available from studies with a precise definition of obesity suggest that postoperative hypoxaemia and atelectasis are common in obese patients, but that the risk of severe respiratory morbidity is in general not increased.

A major concern when managing patients with bronchial asthma is the potential for exacerbation of bronchospasm secondary to stimulation of the airways. Severe bronchospasm may occur during endotracheal intubation, often as a result of ‘light anaesthesia’, bronchial secretions, and surgical manipulation of the viscera or the airway. Early studies suggested an increased risk of perioperative respiratory complications in patients with bronchial asthma when compared to non-asthmatics. However, even steroid-dependent asthmatic patients were shown to tolerate major surgery if proper perioperative care was given. Anaesthetic considerations in asthmatic patients include the use of volatile agents, which induce a direct bronchodilator effect independent of adrenergic stimulation. Ketamine exhibits bronchodilator properties. Opioids effectively blunt airway reflexes; synthetic opioids, such as fentanyl, do not induce histamine release and may decrease airway irritability during intubation and extubation of the trachea. When possible, regional anaesthesia or general anaesthesia administered by face-mask may be a valuable alternative to endotracheal intubation.

Patients with chronic obstructive pulmonary disease have been identified as a group at increased risk for perioperative respiratory complications, as opposed to those with restrictive disease in whom expiratory flow rate and cough mechanism are preserved. Many critical values of pulmonary function have been associated with an increased incidence of postoperative pulmonary complications. A summary of abnormal values of pulmonary function tests associated with increased perioperative respiratory complications is reported in Table 1 83. Although it is clear that patients with chronic obstructive pulmonary disease are at higher risk than is the normal population, no one best test or combination of tests has emerged as a predictor of morbidity. In fact, other studies fail to show a reproducible correlation between abnormal values of pulmonary junction tests and the rate of pulmonary complications. Thus, the predictive value of abnormal pulmonary function tests as indictors of surgical risk in patients with chronic obstructive pulmonary disease seems to be generic. Given that abnormal pulmonary function tests are usually associated with symptoms, they might be no more specific than the information gathered during a careful history and physical examination.

Virtually all patients with bronchogenic carcinoma of the lung have chronic obstructive pulmonary disease. While surgical therapy offers the best prospect of long-term survival, removal of lung tissue may reduce postoperative function, predisposing to long-term mechanical ventilatory support and possible death. When evaluating patients for lung surgery one should be able to predict postoperative function. The most frequently used method to predict lung function following lung resection is pulmonary scintigraphy. Ventilation scans with xenon-133, perfusion scans with technetium microaggregate, or combined V/ Q scans are equally effective in predicting postoperative forced vital capacity and forced expiratory volumes in 1s (FEV&sub1;). The selective function of the segment to be resected can be evaluated, and the predicted postoperative FEV&sub1; calculated. A predicted post-excision FEV&sub1; ≤ 800 to 1000 ml is generally considered of poor prognostic value.

Post-resection disability is related not only to limitations in ventilation but also to alterations in pulmonary blood flow and pressure; thus, criteria that stratify patients according to their ability to exercise have been studied. Measurement of maximum oxygen consumption, pulmonary artery pressures, and calculated pulmonary vascular resistances during bicycle ergometer testing have been advocated as possible criteria that could identify operative candidates who would otherwise not be considered for thoracotomy based on the results of pulmonary function tests. The predictive value of exercise testing needs confirmation in larger studies.

The efficacy of preoperative prophylactic measures in decreasing pulmonary complications has been clearly substantiated. Patients with abnormal pulmonary function tests who underwent a preoperative regimen of bronchodilators and physiotherapy had a postoperative rate of pulmonary complications of 21 per cent, as opposed to a 60 per cent incidence in patients with similar pulmonary function tests who did not receive preventive treatment.

Preoperative evaluation would be of limited value unless prophylactic measures are instituted in those patients identified as being at increased risk. High-risk categories of patients include those with chronic obstructive pulmonary disease as indicated by abnormal pulmonary function tests (particularly when scheduled for thoracic or upper abdominal surgery), smokers, and patients with acute respiratory infections or active bronchospasm. General measures that may be beneficial include the cessation of smoking, hydration, humidification of inspired gases, antibiotic treatment of infections, and treatment of associated cardiovascular and metabolic disease. More specifically, respiratory function may be improved over a short period of time before surgery by means of a programme of chest physical therapy and pharmacological treatment.

This includes breathing and coughing exercises, postural drainage, ambulation, and the use of mechanical aids to lung expansion. Physiotherapeutic treatment should be planned and carried out by specialized personnel in order to ensure maximal patient compliance. Preoperative education is likely to be more effective than attempting to teach principles of physiotherapy in the presence of postoperative pain and medications.

Several kinds of mechanical aids to physiotherapy are available. Intermittent positive pressure breathing (IPPB), popular in the 1960s and 1970s, is not always well tolerated and may transiently decrease pulmonary compliance. Forced expiratory exercise by means of blow bottles and similar devices is no longer recommended because of the potential for hyperinflation and air-trapping. Currently, manoeuvres designed to increase the functional residual capacity, such as incentive spirometry and continuous positive airway pressure by face-mask, are used more frequently. Preoperative IPPB, incentive spirometry, and deep-breathing exercises seems to be equally effective in reducing the incidence of pulmonary complications following upper abdominal surgery, as compared with no respiratory treatment. However, the use of IPPB is frequently associated with unpleasant side-effects such as bloating and abdominal distension.

Table 2 84 summarizes the characteristics of the most common agents used preoperatively in patients with pulmonary disease. Fluidification of bronchial secretions and resolution of bronchospasm are the main goals of this treatment. Bronchodilator therapy should be aggressively pursued even in those patients with chronic obstructive pulmonary disease who do not seem to respond to a single administration during pulmonary function testing: some of these patients will benefit from repeated treatments over the course of few days. End-points of preoperative pharmacological treatment should be the resolution of acute symptoms such as bronchospasm and dyspnoea, and improvement in the patient's level of activity, as well as in the ability to expectorate and perform physiotherapeutic manoeuvres. Repeating a set of pulmonary function tests may offer an objective documentation of the effects of treatment.

Pulse oximetry provides continuous measurement of arterial oxygen saturation by spectrophotometry. Oxygenated and reduced haemoglobin have different spectra of light absorption: the relative concentration of oxygenated haemoglobin (expressed as percentage saturation) in arterial blood is derived by a microprocessor from the ratio of absorption of two different wavelengths specific for the two forms of haemoglobin. The sensing unit of the pulse oximeter (oxysensor) can be applied to areas of the body (fingertips, toes, earlobes, nose, tongue) where a light beam is shone through and recorded by an emitting and a receiving diode. Adequate arterial blood flow is necessary in order to provide an adequate signal: motion, vasoconstriction, hypothermia, and very low blood pressure interfere with the detection of the signal. The intraoperative use of continuous oximetry identifies episodes of arterial oxygen desaturation earlier and more frequently than does clinical observation. Postoperatively, pulse oximetry is widely employed during immediate recovery from general anaesthesia and in the intensive care unit. Continuous monitoring of arterial oxygen saturation in these settings may reveal episodes of hypoxaemia that could otherwise go undetected. During weaning from mechanical ventilation, recording of arterial oxygen saturation may allow for changes in ventilatory parameters with less frequent arterial blood sampling.

This provides measurement of end-tidal CO&sub2; and display of exhaled CO&sub2; waveforms (capnography). The most commonly used capnometers continuously withdraw a small (150 ml/min) sample of gas distally to the ‘Y-piece’ connector of the breathing circuit. Modifications to face-masks, nasal airways, and nasal cannulas have been designed to facilitate capnography in the awake patient, enabling the capnograph to serve as an apnoea monitor. The CO&sub2; tension recorded at end-expiration (end-tidal CO&sub2;) reflects the Pco&sub2; in the alveolar gas, which in normal circumstances is slightly lower (2–4 mmHg) than the arterial Pco&sub2;. A higher arterial to end-tidal Pco&sub2; gradient reflects an increase in V&subD;/ V&subT;. A sudden fall in the end-tidal Pco&sub2; may be due to an acute decrease in cardiac output, to pulmonary embolism or to air embolism. Useful applications of capnometry in the operating room include: detection of accidental oesophageal intubation by absence of a CO&sub2; waveform, inadequate ventilation, disconnection of a component of the breathing system, CO&sub2; rebreathing from an exhausted CO&sub2; absorber or a malfunctioning valve. Capnometry may allow early detection of air embolism during sitting craniotomy, spinal fusion or hip surgery, by a sudden decrease of end-tidal CO&sub2; secondary to a decreased cardiac output or to sampling of gas lower in CO&sub2;, which diffuses into the alveoli from the pulmonary capillaries. Capnography may be useful in the intensive care unit in mechanically ventilated patients, where the adequacy of ventilation in response to physiological or mechanical changes may be assessed immediately.

No evidence exists that the type of anaesthetic technique affects the outcome of patients with chronic respiratory disease. Potential advantages of a regional anaesthetic (spinal, epidural, extremity blocks) are avoidance of the reduction in respiratory volumes and consequent alterations in gas exchanges associated with general anaesthesia, as well as the potential benefit of not intubating the trachea. On the other hand, a regional anaesthetic may weaken the function of respiratory muscles and blunt the proprioceptive reflexes from the diaphragm, causing hypoventilation; oversedation of an awake patient may also lead to hypoventilation, particularly in the elderly. Few studies have looked at the outcome of patients with respiratory disease in respect to the use of regional versus general anaesthesia, and the results of these are contradictory: although two large studies report better outcomes following spinal or epidural anaesthesia, the finding was not reproduced when comparable groups of patients were studied. Not surprisingly a universally applicable answer is not available. When suitable, extremity blocks (brachial, lumbosacral, ankle block) are intuitively a safe alternative, since they do not interfere with respiratory mechanics. Spinal or epidural anaesthesia may be very effective in a co-operative patient for procedures of the lower extremities or lower abdomen, while they may turn out an unwise choice in an elderly patients with dyspnoea at rest, scheduled for long and traumatizing surgery. Most instances are not so clear-cut, and the decision must ultimately be made by an experienced anaesthesiologist. The statement often forwarded, that a patient with severe chronic obstructive pulmonary disease ‘needs a spinal’ because he or she ‘may never be extubated’ is inappropriate.

Another approach now commonly employed for thoracic and upper abdominal surgery utilizes epidural analgesia in combination with ‘light’ general endotracheal anaesthesia. A mixture of a narcotic analgesic and local anaesthetic (for example, fentanyl 10 &mgr;g/ml in bupivacaine 0.075–0.125 per cent at 3–6 ml/h) is infused into the epidural space via a catheter inserted into the thoracic or lumbar spine, while general anaesthesia is maintained with nitrous oxide in oxygen and, if necessary, supplemented with a low concentration of a volatile agent. The epidural infusion can be continued beyond the surgery for analgesia. The adjunct of epidural analgesia greatly reduces the amount of general anaesthetic, facilitates extubation, and provides excellent pain relief postoperatively without significant ventilatory depression. The possible impact of this combined technique on the development of postoperative pulmonary complications has been assessed in several studies. While a few studies indicate a beneficial effect when compared to general anaesthesia alone, others found no difference; none, however, found higher rates of complication in the epidural groups. These studies are difficult to compare because of the different definitions of pulmonary complications, and the lack of uniformity with anaesthetic techniques used. However, the satisfaction of patients was uniformly superior when epidural analgesia was continued postoperatively, compared with intermittent intramuscular injection of opioids.

Many of the mechanisms that affect respiratory function during anaesthesia and surgery persist for a variable period of time postoperatively. In most instances, pulmonary complications may be viewed as a continuation of physiological derangements initiated on the operating table. Characteristically, postoperative patients breath rapidly, with shallow tidal volumes; vital capacity is reduced for several days postoperatively. Prolonged immobilization, tight bandages, splinting from incisional pain, and diaphragmatic dysfunction, all contribute to the persistence of low lung volumes.

The most important changes occur after thoracic and upper abdominal surgery. However, any type of surgery, when associated with immobilization, malnutrition, advanced age, and a decreased respiratory reserve, may create grounds for pulmonary complications. Patients with long-standing respiratory failure are particularly vulnerable in the perioperative period: pulmonary or systemic complications may easily result in acute respiratory failure requiring mechanical ventilation.

Causes of inability to sustain spontaneous ventilation at the end of surgery include excessive administration of volatile anaesthetics, narcotic analgesics, or muscle relaxants; prolonged and traumatizing procedures with large fluid requirement; and intraoperative complications such as lung collapse, pneumothorax, pulmonary oedema. In these circumstances, treatment of the intercurrent complication is generally sufficient to allow rapid recovery of respiratory function. Inability of sustaining adequate spontaneous ventilation later in the postoperative course is most frequent in patients with severe chronic obstructive pulmonary disease. Frequent causes of acute respiratory failure at this time are: pneumonia, tracheobronchitis, aspiration pneumonitis and cardiac failure. Pneumonia impairs gas exchange and decreases lung compliance: excessive bronchial secretions increase airway resistance and worsen bronchospasm. Patients with limited respiratory reserve tolerate poorly the consequent increased work of breathing and will require prolonged mechanical ventilation.

Hypoxaemia may initially be treated with enriched inspired oxygen concentrations provided by a non-rebreathing face-mask, high-flow oxygen delivery systems, or by continuous positive airway pressure delivered by mask. Adjustable positive pressure ventilation triggered by the patient's inspiratory effort can also be delivered mechanically without the use of an eudotracheal tube. These systems may be effective in the short-term management of acute respiratory failure secondary to exacerbations of chronic pulmonary disease or to congestive heart failure. No data are available, however, in surgical patients. The potential for gastric distention associated with positive airway pressure without endotracheal intubation suggests the need for cautious use of these techniques in patients recovering from gastrointestinal surgery.

Ultimately, the surgical patient with acute respiratory failure may require endotracheal intubation and mechanical ventilation to avoid exhaustion, acidosis, and hypoxaemia. Intubation should be carried out by an experienced physician. Oral intubation under direct laryngoscopy is generally the easier route, and the one of choice in unstable patients when rapid re-establishment of the airway is mandatory. Nasal intubation is better tolerated during prolonged mechanical ventilation; the endotracheal tube is more effectively secured and accidental extubation less likely. Potential drawbacks include the possibility of mucosal damage and bleeding, which can make subsequent laryngoscopy very difficult, the limited internal diameter and length of the endotracheal tube, and the possibility of subsequent sinusitis. Naso-tracheal intubation should be avoided in patients with coagulopathies, and in those with major fractures of the facial bones or of the base of the skull. Table 3 85 summarizes issues concerning the oral and nasal approaches to intubation.

Patients with acute or chronic respiratory failure are often chronically ill and malnourished, and the added stress of surgery and infection further depletes their metabolic reserves. When tolerated, enteral nutrition may be used instead of total parenteral nutrition. This route seems to be more physiological, preserves the function of the intestinal mucosal barrier, and avoids the possible complications related to the use of parenteral nutrition, including morbidity resulting from venous access.

As outlined earlier, a programme of physical therapy, including postural drainage, deep-breathing exercises, and early mobilization should be undertaken by specialized personnel. Getting the patient out of bed is a most effective lung expansion manoeuvre, which may generate a 10 to 20 per cent increase in functional residual capacity. Ambulation with the help of a nurse and a respiratory therapist is possible in co-operative intubated patients.

Adequate analgesia is essential postoperatively since pain can result in tachycardia and hypertension and limit respiratory and general activity. Intermittent intramuscular injections do not allow reliable pain control. In patients in the intensive care unit, where respiratory function can be closely monitored, opioids should be given intravenously.

In recent years, neuroaxial administration of opioids has become common practice in many surgical intensive care units. Preservative-free morphine injected intrathecally (1 mg or less) or epidurally (2–4 mg), can give excellent analgesia for 24 to 36 h. The epidural route is more frequently used, since it carries less risk of postdural puncture headache and infection, and thus is safer when prolonged drug delivery is desirable. Continuous epidural infusion through an indwelling catheter allows the use of short-acting, highly lipid-soluble opioids, such as fentanyl, thereby limiting the potential for ventilatory depression. Since opioids and local anaesthetics block different nociceptors in the spinal cord, mixtures of the two can be used to improve analgesia and minimize side-effects.

Self-administered analgesia (patient-controlled analgesia) allows standard drugs such as morphine or meperidine to be given from bedside devices operated directly by the patient. These devices typically employ infusion pumps with safety limits on infusion rates. Patient-controlled analgesia provides more stable blood levels of opioids when compared to intermittent pro re nata administration, and consequently better pain control with smaller overall amounts of the drug.

Regional nerve blocks may also be considered: multiple intercostal nerves blockaded with a local anaesthetic (for example, 0.5 per cent bupivacaine with adrenaline 1:200000) provide 4 to 8 h of analgesia following thoracotomies and subcostal and flank incisions. Non-steroidal anti-inflammatory agents may be an effective alternative or adjunct to opioids; ketorolac is a new non-steroidal anti-inflammatory agent with analgesic properties comparable to morphine but with no detectable respiratory depression.

Sedation and adequate sleep is also an important aspect of the care of these patients; the practice of withdrawing or minimizing sedation in intubated patients to facilitate weaning is, in many cases, inappropriate. Benzodiazepines and neuroleptic agents are frequently used for sedation. In patients requiring large doses of sedatives because of excessive agitation, a continuous infusion may be appropriate. Midazolam (a short acting benzodiazepine) or propofol (an ultra-short-acting potent hypnotic) administered in continuous infusion can provide satisfactory sedation without untoward haemodynamic and respiratory effects; recovery after discontinuation of the infusion was almost immediate in the majority of patients.

The extent of acute lung disease is a major determinant of the patient's ability to sustain spontaneous ventilation: extubation will not be possible while major parenchymal damage from pneumonia or trauma persists, causing hypoxaemia, increased dead space, and impaired lung compliance. Appropriate antibiotic therapy must be guided by serial cultures. Fluid balance should be evaluated daily, and diuretic therapy administered when clinical evidence of fluid retention develops. Bronchospasm mandates aggressive treatment, as outlined earlier; factors capable of precipitating bronchospasm include pulmonary oedema, direct stimulation of the airways, pain, agitation, and (although probably not commonly) &bgr;-blockade therapy. Chest physical therapy should be accompanied by adequate analgesia. Excessive amounts of bronchial secretions may hinder extubation: treatment of tracheobronchitis with antibiotics, inhalation of mucolytics, and chest physical therapy enhance the patient's ability of clearing secretions.

Breathing through a narrow, long tube increases the patient's work and may limit weaning. A major increase in resistance occurs in adults if endotracheal tubes below size 7.0 (internal diameter in mm) are used, in conjunction with an increase in minute ventilation. Above this size, resistance does not increase appreciably, and the discomfort and risk of changing the tube is not justified in most patients.

The timing of an elective tracheostomy is not defined. Airway damage by the endotracheal tube has become less common since the introduction of high-volume, low-pressure cuffs. It is generally accepted that patients can be safely ventilated through an oro- or nasotracheal tube for 7 to 14 days. Elective tracheostomy has a low incidence of complications and is often unexpectedly welcomed by the patient, who appreciates the lower resistance to breathing and the better comfort as compared with an oro- or nasotracheal tube. Modifications of the tracheostomy tube allow the patient to talk when he is not mechanically ventilated (fenestrated tracheostomy) and during positive pressure ventilation (‘talking’ tracheostomy).

Modern ventilators for use with patients in intensive care units are capable of delivering different modes of ventilation, allowing greater versatility in the choice of the modality that seems most appropriate for each patient. Regardless of the mode selected, the first decisions to be made when starting a patient on ventilatory support are about tidal volume, respiratory rate, and inspired oxygen concentration. Tidal volumes 50 to 80 per cent larger than spontaneous are often necessary to compensate for the circuit compressible volume and the increased physiological dead space, (V&subD;/ V&subT;)&subp;&subh;&suby;&subs;. Since ventilator circuits include humidifiers and long, large-bore corrugated tubing, a considerable percentage of the tidal volume is compressed in the system and never reaches the patient. Most systems have compressible volume loss factors of 3 to 5 ml/cmH&sub2;O airway pressure. Thus, with a peak airway pressure of 50 cmH&sub2;O, 150 to 250 ml of volume is lost. (V&subD;/ V&subT;)&subp;&subh;&suby;&subs; may increase because of dilation of the airways and decreased cardiac output during positive pressure ventilation, and because of the lung disease itself.

Slow respiratory rates and large tidal volumes are generally employed in patients with chronic obstructive pulmonary disease, in whom a short expiratory time impairs CO&sub2; elimination and favours air-trapping. Rapid rates may be employed safely in patients with restrictive disease.

The duration of inspiration may be controlled in three different ways, depending upon the design of the ventilator. In volume-cycled modes, such as continuous mandatory ventilation, intermittent mandatory ventilation, and assist/controlled ventilation, inspiration terminates when the preset volume is delivered. In the absence of a leak, this method guarantees the tidal volume, independent of changes in compliance and resistance. In pressure-cycled modes, such as pressure-limited ventilation, inspiration terminates when a preset mouth pressure is reached. In this mode, pressure but not volume has to be dialled to determine tidal volume, which, consequently, is not guaranteed and is affected by changes in respiratory compliance and resistance. In time-cycled ventilators, inspiration terminates after a preset time, and tidal volume depends on the inspiratory flow rate; these ventilators are still used in the operating room, but are not practical for prolonged ventilation in the intensive care unit.

Inspired oxygen concentration (Fio&sub2;) is generally regulated to the lowest possible level to maintain adequate arterial oxygen tension (60 to 80 mmHg) in order to avoid oxygen toxicity. The ‘safe’ level of Fio&sub2; in humans is not known: oxygen toxicity is directly related to Fio&sub2; and to the length of exposure. Inspired concentrations of O&sub2; below 50 per cent should be safe even for prolonged periods of time. Arterial oxygenation may be improved, and Fio&sub2; decreased, by applying positive end-expiratory pressure.

Positive end-expiratory pressure (PEEP) improves oxygenation in patients with acute respiratory failure by increasing functional residual capacity, by redistributing oedema fluid to the interstitial compartment, and possibly by decreasing cardiac output and shunting. Although there is a progressive increase in arterial P&subO;&sub2; with an increase in functional residual capacity, increasing levels of positive end-expiratory pressure are associated with the potential for hyperinflation, increased intrathoracic pressure, decreased preload and increased afterload to the right ventricle, shifting of the intraventricular septum with decreased filling of the left ventricle, and barotrauma. Since benefits and complications of PEEP must be balanced, there is no agreement on what is the optimal level.

We will illustrate the salient features of a few among the most common modalities of mechanical ventilation, emphasizing that none has been proved to be universally superior and that the choice is dictated by the characteristics of each individual patient, the time course of the respiratory failure, and the preference of experienced physicians.

This implies that the machine is fully sustaining the task of breathing. Controlled mandatory ventilation necessitates suppression of the patient's spontaneous efforts, accomplished either with overventilation or with heavy sedation and occasionally with paralysis. This mode is reserved for patients with absent respiratory drive or with the most severe degrees of failure and is obviously not ideal during the process of withdrawal of mechanical support.

This allows spontaneous breathing complemented by a variable input from the ventilator. The amount of ventilation provided can be progressively reduced as the patient's clinical condition improves. Intermittent mandatory ventilation is a simple and effective mode of weaning. A potential problem of this method is the inability of some patients to couple their spontaneous activity with the mechanical breaths. In the synchronized modification of intermittent mandatory ventilation the patient's breath inhibits the mechanical cycle, which is delivered only after a preset pause.

This allows the patient to initiate every breath spontaneously; the machine responds to the negative pressure exerted by the patient's inspiratory effort by delivering a preset tidal volume. By progressively decreasing the size of the assisted breath, the patient is allowed more work until the input from the ventilator becomes negligible. Although very appealing in principle, assist/controlled ventilation has several drawbacks. The less than perfect design of many ‘demand valves’ may cause either an excessive work to trigger the mechanical breath or, on the other hand, the delivery of breaths in response to stimuli such as coughing or moving, which may cause periods of distinct discomfort to the patient. Also, since the size of the assisted breath is the only variable that can be set, this does not always match the patient's pattern of breathing: when the patients peak-inspiratory flow is higher than the one supported by the ventilator the mechanical breath may reach the patient while exhalation has started, causing discomfort and increased work of breathing.

This is a relatively new mode of assist ventilation. A preset amount of pressure is delivered in response to the patient's inspiratory effort; the tidal breath is delivered at an inspiratory flow (80–100 l/min) which is always higher than the patient's, thus avoiding dysynchrony. The only preset parameter is the inspiratory pressure; the patient controls the entire breathing cycle, receiving a smooth assistance from the ventilator. Potential advantages of this method are a more physiological pattern of breathing, decreased work of breathing, and slower respiratory rate, as compared with intermittent mandatory ventilation, and the potential for less barotrauma. The success of inspiratory pressure support ventilation is also due to the much improved overall design of the ventilators that can supply it; these ventilators, however, are also extremely expensive, and they should be reserved for the few patients who might really benefit from advanced modes of ventilatory support.

Once the patient who has been subjected to prolonged mechanical ventilation finally gets close to being separated from the machine, some objective criteria to predict the success of extubation, are appropriate. Bedside measurements of respiratory system mechanics are easily obtainable: suggested values include a forced vital capacity approaching 1 litre, a tidal volume of 300 to 400 ml, a negative inspiratory force lower than 40 cmH&sub2;O, and a respiratory rate lower than 30 breaths/min.

Common sense should warn against predetermined criteria: if they are too strict, few patients will fail, but many others will be subjected to unnecessarily prolonged mechanical support. If liberal criteria are chosen, many patients inevitably fail. The answer is that no index or combination of variables can be an accurate predictor. Only the daily observation of the patient's progress, coupled with the observation of many previous states and with continuous learning about the underlying pathophysiology can improve the success rate in the treatment of respiratory failure.

Aitkenhead AR, Pepperman ML, Willatts SM, et al. Comparison of propofol and midazolam for sedation in critically ill patients. Lancet 1989; 704–8.

Appelberg M, Gordon L, Fatti LP. Preoperative pulmonary evaluation of surgical patients using the vitalograph. Br J Surg 1974; 61: 57–9.

Beecher HK. Effect of laparotomy on lung volume: demonstration of a new type of pulmonary collapse. J. Clin Invest 1933; 12: 651.

Bersten A, et al. Treatment of severe cardiogenic pulmonary edema with continuous positive airway pressure delivered by face-mask. N Engl J Med 1991; 325: 1825–30.

Brochard L, et al. Reversal of acute exacerbations of chronic obstructive lung disease by inspiratory assistance with a face-mask. N Engl J Med 1990; 323: 1523–30.

Celli BR, Rodriguez KS, Snider GL. A controlled trial of intermittent positive pressure breathing, incentive spirometry and deep breathing exercises in preventing pulmonary complications after abdominal surgery. Am Rev Respir Dis 1984; 130: 12–15.

Froese AB, Bryan AC. Effects of anesthesia and paralysis on diaphragmatic mechanics in man. Anesthesiologgy 1974; 41: 242–55.

Gaensler EA. Analysis of ventilatory defects by timed capacity measurements. Am Rev Tuberc 1951; 65: 256.

Gracey DR, Divertie MB, Didier EP. Preoperative pulmonary preparation of patients with chronic obstructive pulmonary disease. Chest 1979; 76: 123–9.

Keagy BA, et al. Elective pulmonary lobectomy: factors associated with morbidity and operative mortality. Ann Thorac Surg 1985; 40: 349–52.

Latimer RG, et al. Ventilatory patterns and pulmonary complications after upper abdominal surgery determined by preoperative and postoperative computerized spirometry and blood gas analysis. Am J Surg 1971; 122: 622–32.

Oh SH, Patterson R. Surgery in corticosteroid-dependent asthmatics. J Allergy Clin Immunol 1974; 53: 345–51.

Pedersen T, Eliasen K, Henriksen H. A prospective study of risk factors and cardiopulmonary complications associated with anesthesia and surgery: risk indicators of pulmonary morbidity. Acta Anesthes Scand 1990; 34: 144–55.

Pontoppidan H. Mechanical aids to lung expansion in non-intubated patients. Am Rev Respir Dis 1980; 122: 109–19.

Stein M, Cassara EL. Preoperative pulmonary evaluation and therapy for surgical patients. JAMA 1970; 211: 787–90.

Tarhan S, et al. Risk of anesthesia and surgery in patients with chronic bronchitis and chronic obstructive pulmonary disease. Surgery 1973; 74: 720–6.

Tisi GM. Preoperative evaluation of pulmonary function. Am Rev Respir Dis 1979; 119: 293–346.

Vaughan RW, Engelhart RC, Wise L. Postoperative hypoxemia in obese patients. Ann Surg 1974; 180: 877–82.

Veterans Affairs TPN cooperative study group. Perioperative total parenteral nutrition in surgical patients. N Engl J Med 1991; 325: 525–32.

Warner MA, Divertie MB, Tinker JH. Preoperative cessation of smoking and pulmonary complications in coronary artery bypass patients. Anesthesiology 1984; 60: 380–3.