35 Postoperative Respiratory Care
Patients undergoing cardiac surgery experience physiologic stresses from anesthesia, thoracotomy, surgical manipulation, and cardiopulmonary bypass (CPB). Each of these interventions can create transient deleterious effects on pulmonary function even with normal lungs; the effects may be exaggerated in the presence of preexisting pulmonary pathology. Important pulmonary changes after cardiac surgery include diminished functional residual capacity after general anesthesia and muscle relaxants,1 transient reduction in vital capacity (VC) after median sternotomy and intrathoracic manipulation, atelectasis, and increased intravascular lung water.2 Acute functional residual capacity reduction creates arterial hypoxemia because of a mismatch between ventilation and perfusion, and diminishes lung compliance with increased work of breathing. This additional work of breathing, which increases oxygen consumption by up to 20% in spontaneously breathing patients,3 also increases myocardial work at a time when myocardial reserves may be limited. Changes in spirometric measurements and respiratory muscle strength can last up to 8 weeks after surgery.4
Thus, a sizeable proportion of cardiac surgical patients can be expected to have respiratory complications. In our experience, about a fourth of cardiac surgical patients were extubated in the operating room or within 4 hours of intensive care unit (ICU) arrival, and about half within 8 hours of ICU arrival. Median postprocedure intubation time was 7.6 hours. About 8% of patients experienced prolonged mechanical ventilation (defined as ≥ 72 hours after ICU arrival), and about 7% required reintubation of the trachea either shortly after initial extubation or because of delayed respiratory failure. Acute lung injury (ALI), sometimes progressing to acute respiratory distress syndrome (ARDS), can occur in up to 12% of postoperative cardiac patients.5 Tracheostomy was performed in 1.4% of post-CPB patients to facilitate recovery and weaning from ventilatory support. Although these figures represent the experience of one referral center, others have reported similar results (Table 35-1). A more recent study at this same institution showed a trend toward less ventilator dependency but little change in the rate of tracheostomy.6 In this study, 5.5% experienced ventilator dependency, whereas 1.45% required tracheostomy.
Risk factors for respiratory insufficiency
The lung is especially vulnerable because disturbances may affect it directly (atelectasis, effusions, pneumonia) or indirectly (via fluid overload in heart failure, as the result of mediator release from CPB, shock states, or infection, or via changes in respiratory pump function as with phrenic nerve injury). Postoperative status will be determined, in part, by the patient’s preoperative pulmonary reserve, as well as by the level of stress imposed by the procedure. Thus, a patient with reduced VC caused by restrictive lung disease undergoing minimally invasive surgery may have fewer postoperative pulmonary issues than a relatively healthy patient undergoing simultaneous CABG and valve replacement with its longer accompanying operative/anesthetic and CPB times. Respiratory muscle weakness contributes to postoperative pulmonary dysfunction, and prophylactic inspiratory muscle training has been shown to improve respiratory muscle function, pulmonary function tests, and gas exchange. Training reduces the percentage of patients requiring more than 24 hours of postoperative ventilation support from 26% to 5%.7
Assessing Risk Based on Preoperative Status
A number of robust models are available to stratify mortality outcome by preoperative risk factors in patients undergoing cardiac surgery.8 The independent (predictive) variables and their weighting vary somewhat from model to model and vary between models predicting mortality versus those predicting morbidity or length-of-stay outcomes,9 but the commonalities are greater than the differences. The Society of Thoracic Surgeons National Adult Cardiac Surgery Database is widely used in the United States and offers, in addition to a mortality prediction, a model customized to predict prolonged ventilation.10,11 The EuroSCORE is commonly used in Europe.12 Factors common to outcome risk adjustment models include age, sex, body surface area, presence of diabetes or renal failure, chronic lung disease, peripheral vascular disease, cerebrovascular disease, prior cardiac surgery, and emergency or unstable status.9–11 Chronic obstructive pulmonary disease (COPD) might be expected to be a major risk for postoperative respiratory morbidity and mortality and appears as a factor in many models. However, hospital mortality with mild-to-moderate COPD is not especially high; it is the minority of patients with severe COPD, especially those older than 75 years and receiving steroids, who are at greatest risk.13 Patients with preexisting COPD have greater rates of pulmonary complications (12%), atrial fibrillation (27%), and death (7%).13 Obesity, defined by increased body mass index, does not appear to increase the risk for postoperative respiratory failure.14 In contrast, even modest increases of serum creatinine concentration (> 1.5 mg/dL) are independently associated with greater morbidity and mortality.9,15
At least four studies have used multivariate regression techniques to elucidate factors specifically associated with postoperative respiratory failure (Table 35-2). The studies differ in their end points for outcome and in their choice of preoperative versus operative versus postoperative variables. Spivack et al16 examined 513 consecutive patients undergoing CABG and identified reduced left ventricular ejection fraction, preexisting congestive heart failure, angina, current smoking, and diabetes mellitus as predictors of mechanical ventilation support beyond 48 hours. In this study, pulmonary diagnosis, lung mechanics, and blood gas parameters were not independently useful in predicting outcome. Branca et al17 found that the mortality rate predicted by the Society of Thoracic Surgeons model10 was the single best predictor of mechanical ventilation support for longer than 72 hours, but also identified mitral valvular disease, age, vasopressor and inotrope use, renal failure, operative urgency, type of operation, preoperative ventilation, prior cardiac surgery, female sex, myocardial infarction within 30 days, and previous stroke as contributors.17 Rady et al18 examined both preoperative and intraoperative factors and noted that transfusion of more than 10 units of blood products or total CPB time in excess of 120 minutes were important operative events in addition to the usual preoperative predictors of extubation failure. Canver and Chandra19 looked only at operative and postoperative predictors versus the end point of mechanical ventilation for more than 72 hours; they found that prolonged CPB time, sepsis and endocarditis, gastrointestinal bleeding, renal failure, deep sternal wound infection, new stroke, and bleeding requiring reoperation were important predictors of prolonged ventilatory support. None of these models, general or specific for respiratory complications, is sufficiently sensitive or specific to prohibit consideration of surgery for an individual patient, but they do provide the clinician with early warning for patients at high risk.
Operating Room Events
Identification of the patient who is difficult to intubate is important for planning extubation for a time when sufficient personnel and equipment are available to deal with a potentially difficult reintubation. Opioids and neuromuscular blocking agents with long half-lives might be expected to influence extubation time. It is not the specific duration of action of these drugs but rather the skill of the anesthesiologist in knowing how to use them well that influences extubation time. Reoperative patients are at risk18–20 partly because of longer CPB time, increased blood transfusion, and the additional likelihood of bleeding in this population. CPB time is repeatedly identified as a risk,18–20 and a correlation between CPB time and inflammatory cytokine release has been demonstrated.21 However, levels of C-reactive protein, an inflammatory marker, do not correlate with outcomes such as time on mechanical ventilation.22 Genetic polymorphisms are associated with respiratory complications,23 suggesting that risk prediction may require more sophisticated understanding of individual patient variables. Recent observations of dose-dependent reductions in adverse events after CABG in patients receiving statins are also intriguing.24,25
Low cardiac output states may be important predictors of prolonged ventilation because prolonged periods of inadequate perfusion result in additional mediator release. Patients maintained on an intra-aortic balloon pump (IABP) or a ventricular assist device may have borderline or insufficient cardiac output; it makes little sense to impose the additional work of breathing3 until their cardiac issues have resolved. Cardiovascular collapse occasionally occurs at the time of chest closure secondary to severe distension or edema of the lungs. Physiologically, this acts much like cardiac tamponade, and the solution is to leave the chest open for 24 to 48 hours. An open chest delays early extubation and also has a potential to produce long-term ventilator dependency should infection or sternal osteomyelitis develop.
The prognostic and therapeutic implications of an IABP depend on the reasons for which the device was inserted (Figure 35-1). Not surprisingly, mortality and ventilation-dependency rates are lowest in those not requiring any mechanical support. In patients in whom the IABP was placed before surgery for unstable angina, definitive surgery should correct the problem and removal of the IABP and extubation need not be delayed. In all other scenarios, intubation and ventilatory support may be required beyond the time of removal of the IABP because of residual cardiac dysfunction, fluid overload, or associated organ injury. Patients whose IABP was placed for preoperative cardiogenic shock, as an assist to separating from CPB, or for low output states in the postoperative period have a high mortality risk and frequently need prolonged ventilatory support.
Positive end-expiratory pressure (PEEP) while on CPB has been advocated as one method to prevent atelectasis. This turns out to be impractical in patients with COPD because air trapping interferes with surgical exposure. Recruitment maneuvers after CPB have variable impact on intubation time; most studies show it to be ineffective in reducing the need for long-term ventilatory support. Alveolar recruitment maneuvers can be performed without deleterious effects, even in morbidity obese patients, as long as intravascular volume is adequate.26 There are no compelling data that fluid management choices or the use of steroids before CPB have substantial effects on intubation time or respiratory failure. A number of studies27,28 suggest that infusion of small volumes of hypertonic saline prepared in a hydroxyethyl starch solution may reduce total fluid needs, improve cardiac index, and lessen the pulmonary gas exchange compromise, but studies have not examined whether these differences in the operating room translate to improved outcome or shorter length of stay.
Postoperative Events
The expected postoperative course is a short period of ventilatory support while the patient is warmed, allowed to awaken, and observed for bleeding or hemodynamic instability. Preoperative risks, issues with difficult intubation, and operating room events should be communicated from the operating room team to the ICU team at the time of ICU admission. Box 35-1 outlines criteria to be met before routine extubation. Reduced cuff leak volume reliably identifies patients at risk for laryngeal edema. Intravenous methylprednisolone can reduce the incidence of postextubation stridor.29 Prophylactic nasal continuous positive airway pressure (CPAP) at 10 cm H2O for a minimum of 6 hours has been shown to reduce hypoxemia, pneumonia, and reintubation rates after elective cardiac surgery.30
BOX 35-1 Criteria to be met Before Early Postoperative Extubation
Although postoperative care of low-risk cardiac surgical patients has come to resemble a recovery room model, high-risk patients benefit from postoperative involvement of anesthesiologists, cardiologists, and critical care specialists.31 Numerous individual studies and systematic reviews32 have confirmed the value of full-time intensive care specialists in a variety of settings, although there is at least one dissenting opinion.33 Wide variation continues in adherence to the Leapfrog group physician staffing standard,34 despite demonstrated financial return on investment.35 It is likely that a robust organizational environment, rather than the mere presence of intensivists, is necessary to achieve the best results.36
Hospital-acquired infections are an important cause of postoperative morbidity and nosocomial pneumonia is common in patients receiving continuous mechanical ventilation. The historic risk for ventilator-associated pneumonia (VAP) appears to be around 1% per day when diagnosed using protected specimen brush and quantitative culture techniques.37 More recent data suggest that VAP rates can be decreased by an order of magnitude with careful attention to patient management.38,39 Strategies believed to be effective at reducing the incidence of VAP include early removal of nasogastric or endotracheal tubes, formal infection control programs, hand washing, semirecumbent positioning of the patient,40 daily sedation “vacation,”41 avoiding unnecessary reintubation, providing adequate nutritional support, avoiding gastric overdistention, use of the oral rather than the nasal route for intubation, scheduled drainage of condensate from ventilator circuits,42 and maintenance of adequate endotracheal tube cuff pressure.43 Strategies that are not considered effective include routine changes of the ventilator circuit, dedicated use of disposable suction catheters, routine changes of in-line suction catheters, daily replacement of heat and moisture exchangers, and chest physiotherapy.44 The literature supports both continuous aspiration of subglottic secretions and use of silver-coated endotracheal tubes to reduce the incidence of VAP.45–47
Diagnosis of acute lung injury and acute respiratory distress syndrome
Intravascular and intra-alveolar fibrin deposition are common. Procoagulant activity becomes enhanced in ARDS, and bronchoalveolar lavage will reveal increased tissue factor levels.48 The clinical presentation is typically an acute onset of severe arterial hypoxemia refractory to oxygen therapy, with a PaO2 to FiO2 (P/F ratio) of less than 200 mm Hg. ARDS is classically diagnosed only in the absence of left ventricular failure, which complicates the diagnosis in the postoperative cardiac patient who may also be in heart failure. Other findings in ARDS include decreased lung compliance (< 80 mL/cm H2O) and bilateral infiltrates on chest radiograph.49 Murray et al50 created a Lung Injury Score that awards points for affected quadrants on chest radiograph, P/F ratio, amount of PEEP applied, and the static compliance of the lung. Scores above zero, but less than 2.5, are considered ALI, and scores greater than 2.5 meet the threshold for ARDS.
The proliferative phase of ARDS occurs on days 3 to 7 as inflammatory cells accumulate as a result of chemoattractants released by the neutrophils. At this stage, the normal repair process would remove debris and begin repair, but a disordered repair process may result in exuberant fibrosis, stiff lungs, and inefficient gas exchange. Evidence suggests that careful fluid and ventilator management may affect this process.51,52 Conventional ventilator support after cardiac surgery is to maintain large tidal volumes (VT; typically 10 mL/kg) to reopen atelectatic but potentially functional alveolae. The problem is that the compromised lung is no longer homogenous, and high pressures can further damage the remaining normal lung. Direct mechanical injury may occur as a result of overdistention (volutrauma), high pressures (barotrauma), or shear injury from repetitive opening and closing. “Biotrauma” also may occur as a result of inflammatory mediator release and impaired antibacterial barriers. Nahum et al53 showed that dissemination of Escherichia coli via bacterial translocation from the lung was highest in dogs ventilated with a high-VT strategy. Thus, current clinical practice with known or suspected lung injury is to limit inflation pressures. The maximal “safe” inflation pressure is not known, but evidence favors keeping peak inspiratory pressures less than 35 cm H2O and restricting VT to ≤6 mL/kg of ideal body weight in patients at risk for ALI.54 The landmark ARDSNet trial randomized patients to 6 versus 12 mL/kg of ideal body weight and demonstrated a significant difference in 28-day survival with the low-VT group.55 The same study showed significant decreases in interleukin-6 (IL-6) release when the low-VT strategy was used. Most recently, ventilation with lower VT has been shown to be beneficial in critically ill patients even without ALI, as measured by plasma IL-6 levels and progression to lung injury,56 but this issue has not yet been studied in the cardiac surgical population. A conservative strategy of fluid administration has been shown to improve oxygenation and shorten the duration of mechanical ventilation.51
Therapy with acute lung injury/acute respiratory distress syndrome
Maintaining a lung protective ventilatory strategy can involve permissive hypercapnia,57 if normal Pco2 levels cannot be achieved with low VT. The acid-base changes must be monitored carefully, especially in patients with reactive pulmonary vasculature. Prone positioning can be useful in achieving oxygenation.58 A short daily turn to the prone position does not appear to improve outcome in ARDS, although one post hoc analysis found lower mortality in the sickest patients.59 Lower VT with increasing amounts of PEEP may increase alveolar recruitment and thus improve oxygenation.60 Taken to an extreme, patients with ALI may be ventilated with high-frequency oscillation, which is essentially high PEEP with tiny (smaller than dead space), frequently delivered VT. Other techniques for patients who did not respond successfully to conventional therapy include extracorporeal CO2 removal,61 extracorporeal membrane oxygenation,62 partial liquid ventilation,63 inhaled nitric oxide,64,65 and inhaled prostacyclin.66 Although clearly beneficial to some individuals in extreme circumstances, prospective controlled trials are lacking. High-dose corticosteroids have been in and then out of fashion for treatment of ARDS. More recently, Miduri et al gave prolonged lower dose methylprednisolone therapy for unresolving ARDS and were able to document improvements in P/F ratio, better ICU survival, and shortened duration of mechanical ventilation.67
In the healthy cardiac surgical population, the use of PEEP usually is not necessary.68,69 Increased levels of PEEP may decrease cardiac output, unless volume loading is used to stabilize preload by maintaining transmural filling pressures.70 The effects of PEEP are most marked in the presence of abnormal right ventricular function, particularly if the right coronary artery is compromised.71 PEEP neither protects against the development of ARDS72 nor reduces the amount of mediastinal bleeding after cardiac surgical procedures involving CPB.73 Most clinicians will routinely use 5 cmH2O of PEEP in ventilated patients. However, greater levels of PEEP (often 8 to 15+ cmH2O) are usually necessary to maintain adequate oxygenation with ALI or developing ARDS; application of PEEP in the postoperative patient usually involves a trade-off between cardiac and pulmonary goals.
Lung Recruitment
In laboratory and clinical models, an important component of a lung protective ventilatory strategy is recruitment of the lung. This is the closed-chest analogy to the open-chest recruitment maneuver typically done at the end of CPB to re-expand the collapsed lung. The goal of opening the lung is to allow ventilation to occur at a point on the pressure-volume curve that avoids repetitive atelectasis (by staying above a critical closing pressure) and at the same time avoids overinflation.74 ARDSNet data suggest that the short-term effects of recruitment maneuvers are highly variable and that further study is necessary to determine the role of recruitment maneuvers in the management of ALI/ARDS.75 With anesthetic techniques geared to early extubation, suboptimal oxygenation in the early postoperative period appears to be more common today. In most instances, impaired oxygenation is due to atelectasis and responds quickly to brief recruitment maneuvers. These should be performed with caution because of the adverse impact of increased airway pressure on venous return and cardiac output if the patient is intravascularly “empty.” Although the benefits of early application of the lung protective ventilatory strategy in the high-risk cardiac surgical patient have not been studied formally, the benefits of lung protective ventilatory strategy in other populations suggest that this strategy should be considered as soon as ALI is identified, and perhaps even prophylactically in high-risk patients.56
Permissive Hypercapnia
Conventional management is to maintain Paco2 within a “normal” or eucapnic range, classically between 35 and 45 mm Hg. A patient who chronically retains CO2 would be considered eucapnic at his or her higher baseline Paco2. The traditional reason for maintaining eucapnia is primarily that acute deviation from a normal or acclimatized Paco2 will result in alkalemia or acidemia to which the kidneys will respond by retaining or excreting bicarbonate ion. Normal kidneys can compensate for a Pco2-induced pH change in 12 to 36 hours.76 If high airway pressures would be required to maintain a “normal” Paco2, then Paco2 values up to 60 mm Hg are acceptable as long as cardiovascular stability is present and the pH remains greater than 7.30. It has been hypothesized that increased Pco2 levels might even be protective and that low levels of Pco2 could play a role in organ injury.77 Permissive hypercapnia should be used judiciously in patients with pulmonary hypertension because acidosis can exacerbate pulmonary vasoconstriction and further impair right ventricular function and cardiac output.78 (See Chapter 24.)
Cardiopulmonary Interactions
An understanding of cardiopulmonary interactions associated with mechanical ventilation is critical to the cardiothoracic intensivist. Hemodynamic changes may occur secondary to changes in lung volume and intrathoracic pressure even when VT remains constant.79 Pulmonary vascular resistance and mechanical heart-lung interactions play prominent roles in determining the hemodynamic response to mechanical ventilation. Because lung inflation alters pulmonary vascular resistance and right ventricular wall tension, there are limits to intrathoracic pressure that a damaged heart will tolerate. High lung volumes also may mechanically limit cardiac volumes. In patients with airflow obstruction, occult PEEP (auto-PEEP) may also contribute to hypotension and low cardiac output.80 Auto-PEEP can be detected by respiratory waveform monitoring or by pressure monitoring with the ventilator’s expiratory port held closed at end-exhalation. Auto-PEEP may respond to bronchodilators and/or increased expiratory time to permit more complete exhalation.
General Support Issues
Patients requiring long-term ventilatory support are prone to a number of complications including venous thromboembolism, central venous catheter–related bloodstream infections, surgical site infections, VAP, pressure ulcers, nutritional depletion, delirium, and gastrointestinal bleeding. The Agency for Healthcare Research and Quality has identified a number of patient safety practices applicable to the ICU patient; high on the list are appropriate venothromboembolism prophylaxis, use of perioperative β-blockers, use of maximum sterile barriers during catheter insertion, and appropriate use of antibiotic prophylaxis.81 Standard practice in long-term ventilator patients includes prophylaxis against gastrointestinal bleeding with histamine blockers or proton pump inhibitors (unless the patient is receiving continuous gastric feedings), head of bed elevation to 30 degrees or more in hemodynamically stable patients,40 a brief daily wake up from sedation,82 use of in-line suction catheters, glucose control,83 and appropriate venothromboembolism prophylaxis. Box 35-2 summarizes these risk-reduction efforts. Ensuring that each of these goals is met on each patient every day requires extra work but can be accomplished with a daily goals form84 or with information technology.85
Impediments to weaning and extubation
Factors that limit the removal of mechanical ventilatory support include delirium, neurologic dysfunction, unstable hemodynamics, respiratory muscle dysfunction, renal failure with fluid overload, and sepsis. Figure 35-2 outlines one approach to identifying readiness to wean and possible alternative approaches to weaning.
Neurological Complications
Delirium is common in long-stay ICU patients and is associated with greater costs in the ICU and for the entire hospital stay.86 Delirium after cardiac surgery is a common complication in cardiovascular ICUs; estimated incidence rates are approximately 30% in the general cardiac surgical population to 83% in mechanically ventilated patients.87,88 Delirium resolves spontaneously or with pharmacologic intervention in almost all patients by postoperative day 6. Evidence from a large trial of mostly medical critical care patients suggests dexmedetomidine is associated with less delirium than midazolam.89 Dexmedetomidine has been shown to be safe and effective in the post-CABG population.90 Alcohol or benzodiazepine withdrawal should be considered in the differential diagnosis of delirium. Recent evidence suggests that ketamine may attenuate delirium in CPB patients, possibly because of anti-inflammatory effects.91 Initial postoperative management of agitation consists of reassurance and orientation of the patient, as well as control of pain with opioids. Agitation accompanied by disorientation may be worsened by benzodiazepines, which should be restricted to treatment of oriented but anxious patients or for prophylaxis of alcohol and benzodiazepine withdrawal. If the patient remains agitated and disoriented, haloperidol is useful.92
