While intraoperative mortality has diminished greatly over the last several decades, the risk of death within 30 days of surgery remains stubbornly high and is ultimately related to perioperative organ failure. Perioperative strokes, while rare (<2% in noncardiac surgery), are associated with a more than 10-fold increase in mortality. Rapid identification and treatment are key to maximizing long-term outcomes. Postoperative delirium (POD) and postoperative cognitive dysfunction (POCD) are separate but related perioperative neurological disorders, both of which are associated with poor long-term outcomes. To date, there are few known interventions that can ameliorate the risk of perioperative central nervous system dysfunction. Major adverse cardiac events (MACE) are a major contributor to adverse clinical outcomes following surgical procedures. Recently, advances in diagnostic strategies (eg, high-sensitivity cardiac troponin [hs-cTn] assays) have improved our understanding of MACE. Recently, the dabigatran in patients with myocardial injury after noncardiac surgery (MINS; Management of myocardial injury After NoncArdiac surGEry) trial demonstrated that a direct thrombin inhibitor could improve outcomes following MINS. While the risk of acute respiratory distress syndrome (ARDS) after surgery is approximately 0.2%, other less severe complications (eg, pneumonia, reintubation) are closer to 2%. While intensive care unit (ICU) concepts related to ARDS have migrated into the operating room, whether or not adverse pulmonary outcomes impact long-term outcomes in surgical patients remains a matter of debate. The standardization of acute kidney injury (AKI) definition has improved the ability of clinicians to measure and study the incidence of this important source of perioperative morbidity. AKI is associated with increased mortality as well as nonrenal morbidity (eg, myocardial infarction) after major surgery. Gastrointestinal complications after surgery range from ileus (common in abdominal procedures and associated with an increased length of stay) to less common complications such as mesenteric ischemia and gastrointestinal bleeding, both of which are associated with very high mortality. Outside of cardiothoracic surgery, the incidence of perioperative hepatic injury is not well described but, in this population, is associated with worsened long-term outcomes. Hyperglycemia is a common perioperative complication and occurs in patients undergoing both cardiac and noncardiac surgery. Both hyper- and hypoglycemia are associated with worsened long-term outcomes in cardiac and noncardiac surgery. Better diagnosis and increased understanding of perioperative organ injury has led to an increased appreciation for the specific role that particular organ systems play in poor long-term outcomes and has set the stage for targeted therapeutic interventions.
Due to advances in surgical techniques, anesthesiology, and monitoring equipment, the risk of intraoperative mortality in patients undergoing surgical procedures requiring anesthesia has diminished over the last several decades.1 Despite these improvements in intraoperative outcomes, the risk of mortality, either in the hospital or within 30 days, for adult patients undergoing noncardiac surgery remains approximately 1.4%.2,3 The purpose of this review is to describe the impact of individual organ system dysfunction on perioperative morbidity and mortality. Importantly, many studies of organ failure focus on cardiac surgery or noncardiac surgery, but not both, presumably because cardiac surgery patients are exposed to very specific risks (eg, large vessel cannulation, aortic manipulation and cross clamping, cardiopulmonary bypass) that uniquely expose them to perioperative organ dysfunction. In instances in which these patient populations have been studied separately, we will address organ dysfunction in cardiac and noncardiac surgery as distinct groups to maintain consistency.
The incidence of perioperative stoke in noncardiac and nonmajor vascular surgery ranges from 0.1% to 1.9%,4 although the incidence of “covert stroke” (identified on magnetic resonance imaging [MRI] but not necessarily associated with physical manifestations) in patients >65 and undergoing noncardiac surgery may be as high as 10%.5 Strokes are significantly more common in cardiac and major vascular surgery with a reported incidence of 1.9% to 9.7%.6
In the nonsurgical setting, stroke mortality is 12.6%.7 However, in the perioperative setting, the impact of stroke is much higher. After general surgery, for instance, perioperative stroke can have a mortality of 26%.8,9 Data analyzed from the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQUIP) database showed perioperative stroke was associated with an 8-fold increase in all-cause mortality within 30 days of surgery.10 Further analysis of the ACS-NSQUIP database found that perioperative stroke in noncarotid major vascular surgery was associated with 3-fold increase in 30-day all-cause mortality and an increase hospital length of stay.11
Perioperative stroke not only imparts a higher mortality but also increases morbidity. The Perioperative Ischemic Evaluation (POISE) trial showed that patients with nonfatal perioperative strokes required help with activities of daily living. Some patients were incapacitated, and more than 50% were transferred to a long-term care facility.12 Unsurprisingly, perioperative stroke also increases the risk of transfer to a skilled facility as oppose to home after surgery. The Veterans Affairs National Surgery Quality Improvement Project reported a 7-fold increase in mortality from perioperative stroke in patients undergoing noncarotid vascular surgery. In addition, the mean length of hospital stay was increased by 5 days.13
The NeuroVISION trial found that perioperative stroke is associated with increased risk of cognitive decline at 12 months. There was also an increased risk of perioperative delirium and transient ischemic attacks at 1-year follow-up.14
The speed at which perioperative strokes are treated is essential to maximizing functional outcomes. Postoperatively, 5% to 15% of perioperative strokes occur in immediate settings such as the postoperative care unit (PACU), with most strokes occurring in the 24 hours period after surgery.11 The MR. CLEAN (Multicenter Randomized Clinical Trial of Endovascular Treatment for Acute Ischemic Stroke in the Netherlands) trial showed that longer time from symptom to artery puncture was linked to worse functional outcomes; each hour carried a 5.3% absolute decrease in likelihood of functional independence and an absolute increase in mortality of 2.2%.15 When successful reperfusion was the final end point, time had an even more robust impact. Each hour imparted a 7.7% decrease in functional outcome.15
Postoperative delirium (POD) and postoperative cognitive dysfunction (POCD) are 2 distinct entities that are commonly conflated. Delirium manifests as disturbances in attention and changes in cognition, both of which all develop over a short period of time usually hours to days. Fluctuations in these changes are commonly seen throughout the day. POCD, by contrast, is a deterioration in cognitive function temporally associated with surgery.16
Overall mortality in patients who experience delirium is high. The estimated 1- and 6-month mortality in elderly (nonsurgical) patients is 14% and 22%, respectively, which is twice as high compared to elderly patients without delirium. At 6 months after hospital admission, 43.2% of elderly patients with delirium remained in some form of institutional care.17 In intensive care unit (ICU) patients specifically, the duration of delirium is correlated with 1-year18 mortality.
The Safety and Efficacy of Dexmedetomidine Compared with Midazolam (SEDCOM) trial investigated the number of days of ICU delirium and its relationship to mortality, ventilator time, and ICU length of stay. Compared to 0 days of ICU delirium, an independent dose-response increase in mortality was seen from day 1 of delirium (hazard ratio, 1.70; 95% confidence interval [CI], 1.27–2.29; P < .001), day 2 of delirium (hazard ratio, 2.69; CI, 1.58–4.57; P < .001), and beyond day 3 of delirium (hazard ratio, 3.37; CI, 1.92–7.23; P < .001). Duration of delirium was the strongest independent factor in both ventilator time and ICU length of stay.19
The morbidity associated with delirium may extend to surgical populations. Propensity matching in 112 patients followed for 30 months suggests that in hip fracture patients, the relative risk (RR) of mortality is 1.6 in those who experience POD as compared to controls.20 In this study, POD was also associated with an increased risk of dementia or mild cognitive impairment (RR 1.9) as well as institutionalization (RR 1.8).20 That said, it is important to understand whether or not delirium is dangerous in and of itself or simply a marker for other processes associated with aging, both of which ultimately contribute to poor outcomes. Ziman et al21 examined 5-year mortality in elderly patients undergoing noncardiac surgery and found that, after adjustments for covariates, delirium was not predictive of mortality. This raises the possibility that prior work on delirium has not been properly controlled and/or utilized insufficient covariates. This concept is further supported by the fact that there are no prospective data suggesting that interventions targeting perioperative delirium can improve outcomes.
As described above, POCD is a subtle disorder that primarily effects thought processes. A wide variety of cognitive domains such as executive function, memory, attention, and concentration are affected.22 POCD is more persistent problem involving change in cognitive function diagnosed by neuropsychological testing.23 It is distinct from POD, although the 2 are likely related.
For instance, in nonsurgical patients, POD is associated with decline in cognitive functioning and functional dependence.24,25 In cardiac surgical patients, those who experience POD are more likely to have a persistent drop from baseline in Mini-Mental Examination scores at 6 months compared to those who do not.26 Conversely, patients who experience POCD are also at increased risk for POD.27
Poor postoperative outcomes in major elective surgery are related to preoperative cognitive impairment in elderly patients. A prospective study involving 186 patients undergoing major elective surgery at the Veterans Affairs Medical Center found that preoperative cognitive impairment (based on the Mini-Cog test) was associated with one or more of postoperative complications, higher 6-month mortality, increased length of stay, higher 30-day readmission rate, and increased rate of discharge to long-term care facility.27 Similar data were demonstrated using the Montreal Cognitive Assessment (MoCA).28
The development of cognitive dysfunction after surgery is also associated with poor outcomes. This is best studied in cardiac surgery patients, with data suggesting that POCD after cardiac surgery leads to an increase risk of mortality22 as well as permanent cognitive decline.29 That said, noncardiac surgical patients also demonstrate an association between POCD and mortality.30
Interestingly, a randomized controlled trial of general anesthesia to regional anesthesia in patients >60 years old found no significant difference in POCD at 3 months,31 suggesting that advance age, comorbidities, and perioperative stressors are the cause of POCD and not the anesthetic itself. This finding is supported by Mason et al’s32 meta-analysis suggesting that mode of anesthesia (regional versus general) plays no role in either POD or POCD. A retrospective study examining neurocognitive testing in patients with CAD who underwent surgery or medical management over a period of 12 months found no difference between groups at any timepoint, suggesting that underlying medical comorbidities were the cause of any changes in neurological function.33
Major adverse cardiac events (MACE) are a major contributor to adverse clinical outcomes following surgical procedures. In particular, myocardial ischemia is the most common cause of 30-day mortality after noncardiac surgery34,35 and has led to the development of the concept of myocardial injury after noncardiac surgery (MINS).36 Unfortunately, MACE have no standard definition but typically include myocardial ischemia, heart failure, recurrent angina, need for cardiovascular intervention (eg, percutaneous intervention [PCI])37 and, in some cases, arrhythmias.38 Some authors have also included stroke, pulmonary edema, pulmonary embolism, and all-cause mortality,37,38 but for the purposes of this review, we will only include adverse events directly related to the myocardium.
Traditionally, clinical diagnoses (eg, for acute myocardial infarction, heart failure, or arrhythmias) were utilized to identify MACE. Recently, advances in biomarkers have made it easier for clinicians to identify cardiovascular abnormalities that might be missed by clinical signs and symptoms alone. For instance, the increasing use of high-sensitivity cardiac troponin (hs-cTn) assays has made it possible for clinicians and researchers to detect myocardial injury that might have been missed by clinical diagnoses alone, making it possible to better stratify patients based on their projected outcome. Even among the cardiologists, the definition of myocardial infarction is constantly evolving, with the fourth, most current Universal Definition of Myocardial Infarction Expert Consensus Document reflecting the use of hs-cTn.39
Similarly, heart failure diagnosis has been revolutionized by the use of both plasma brain natriuretic peptide (BNP) and plasma N-terminal pro-BNP (NT-proBNP). Pro b-type natriuretic peptide (pBNP) is produced by the ventricles40 and is formed when the prohormone “proBNP” is cleaved into BNP and NT-proBNP.41 The landmark “Breathing Not Properly” study demonstrated that BNP outperformed history, physical examination findings, and other laboratory values in discriminating cardiac and noncardiac causes of dyspnea and in emergency medicine setting.42 BNP has also facilitated the diagnosis of heart failure with preserved ejection fraction (HFpEF) in a group of patients who experience wall stress and myocardial injury before the development of reduced systolic function.43
Database studies that rely on clinical classification software suggest that the Incidence of MACE is 3% after major noncardiac surgery.3 Importantly, presence of MACE increases the risk of both in-hospital death (adjusted odds ratio [aOR] = 1.22) and ischemic stroke (aOR = 1.05).3
The higher the peak troponin value after noncardiac surgery, the higher the risk of death (for values <0.02, adjusted hazard ratio [aHR] = 2.41; for 0.03–0.29, aHR = 5.00; for >0.3, aHR = 10.48)2 (Figure 1). When peak fourth-generation troponin T (TnT) is used to identify myocardial injury (cutoff of <0.02 ng/mL), the incidence of myocardial injury in the perioperative period is approximately 1.9%.2 Newer “highly sensitive” troponin assays offer even greater ability to improve on the diagnosis of MINS as well as risk-stratify patients. These increases in troponin—some which might be subclinical—are also associated with an increased risk of death, with projected mortality ranging from 0.1% (<5 ng/L) to 29.6% (>1000 ng/L)35 (Figure 2).
The development of MINS is not only of prognostic significance. It is thought that >80% of MINS events are due to myocardial ischemia,36 and the dabigatran in patients with MINS MANAGE (Management of myocardial injury After NoncArdiac surGEry) trial was developed to test whether or not dabigatran, a direct thrombin inhibitor, could improve outcomes following MINS. This international, multicenter, randomized, placebo controlled trial included 1754 patients who had experienced MINS. The primary outcome of major vascular complications was significantly reduced in the intervention group (hazard ratio 0.73, 95% CI, 0.55–0.93) with no difference in complications44 (Figure 3). This study was not powered to detect changes in mortality but a trend toward improved mortality (hazard ratio 0.90, 95% CI, 0.69–1.18) was noted, hopefully setting the stage for a larger trial to include this important outcome.
Several studies have suggested that preoperative BNP and NT-proBNP may be useful in predicting cardiovascular events that occur following noncardiac surgery.45–49 In patients who undergo cardiac surgery, elevated BNP is associated with a 2.44 hazard ratio for nonfatal MACE50 as well as postsurgical heart failure (hazard ratio 1.93)51 and performs similarly to TnT as a predictive tool.50 Less is known about the utility of BNP in predicting outcomes after noncardiac surgery, but a study suggests that BNP elevations are correlated with prolonged length of stay.52
Postoperative atrial fibrillation (POAF) is a common adverse event following cardiac surgery but is now increasingly recognized in noncardiac surgical patients as well. In patients undergoing cardiac surgery, the incidence of new onset POAF ranges from approximately 18% to 32% depending on the operation.53–56 In the cardiac surgical population, POAF is associated with an increased risk of stroke that persists for years after the operation55,57,58 as well as increased hospital length of stay, in-hospital mortality, and postdischarge mortality (at 6 months) as well as an increased risk of complications in almost every organ system.56,58–60
In a broader surgical population, the incidence of POAF is closer to 1% to 2%57 although this varies by subspecialty. Colorectal surgical patients, for instance have a 6.6% risk of POAF, and those who experience POAF have higher in-hospital and 1-year mortality.61 Similarly, POAF is associated with increased risk of in-hospital mortality in patients who have undergone lobectomy (POAF = 10%)62 as well as esophagectomy (POAF = 21%).63 Importantly, patients who experience POAF are older than those who do not.63 In noncardiac surgical patients, POAF nearly triples the risk of stroke at 5 years57 (Figure 4).
While not often classified as a MACE, a retrospective analysis of approximately 3400 patients undergoing cardiac surgery suggested that the need for inotropes or vasopressors for >48 hours postoperatively was associated with a significant increase in early mortality after cardiac surgery.64 More recently, application of the vasoactive-inotropic score (VIS, composite measure of hemodynamic drugs administered within the first 24 hours of cardiac surgery) has been shown to be predictive of postsurgical complications as well as long-term survival.65
Two decades ago, the Acute Respiratory Distress Syndrome (ARDS) Network (ARDSNet) demonstrated that lung injury in critically ill patients was a modifiable risk factor for adverse outcomes in critically ill patients, by demonstrating that low tidal volume ventilation in the setting of ARDS improved mortality.66 A decade later, “lung-protective” strategies were still being inconsistently adopted in ICUs, with some centers reporting <50% compliance.67 The definition of ARDS continues to evolve, with the Berlin Definition being reported in 2012.68 A core feature of all ARDS definitions, however, has been ineffective gas exchange. The Berlin definition of ARDS focuses on 4 elements—timing, chest imaging, origin of the injury, and severity (measured with blood oxygenation indexed to fractional inspired oxygen concentration [Fio2]).
Thankfully, accurately reporting a correlation between ARDS and clinical outcomes in the perioperative setting will be a challenge because the incidence of ARDS in surgical patients is quite low (approximately 0.2% in 2 separate analyses, 1 including over 50,000 patients).69,70 Interestingly, even in the pre-ARDSNet era, it was acknowledged that poor gas exchange after surgery was predictive of poor clinical outcomes. Rady et al71 examined clinical outcomes in 1461 cardiothoracic surgery patients who had normal preoperative lung function. They found that 12.3% of them developed lung dysfunction defined as a Pao2/Fio2 (P/F) ratio <150. Those who experienced postoperative pulmonary dysfunction exhibited higher serum creatinine, higher incidence of low cardiac output syndrome, higher lactate values, longer mechanical ventilation, higher infection rate, and higher hospital mortality, all of which were statistically significant.71
That said, other (non-ARDS) forms of lung injury occur in the perioperative period, and they may impact clinical outcomes. An analysis of approximately 1200 patients undergoing abdominal, neurosurgical, and orthopedic procedures found that 33% experienced some form of postoperative pulmonary complication (including prolonged need for supplemental oxygen [20.6%], atelectasis [17.1%], pleural effusion [9.7%], pneumonia [1.8%], and reintubation [1.7%]).70 These authors also reported that presence of any primary pulmonary complication significantly increased the risk of mortality (RR 2.3, P < .05).70
While the majority of research in the relationship of tidal volumes and lung injury has focused on critically ill patients, in 2010, Lellouche et al64 tested whether or not intraoperative tidal volumes had an impact on perioperative outcomes in patients undergoing cardiac surgery and made 2 important findings. First, cardiac surgical patients typically received tidal volumes far in excess of what would be considered lung protective (11.5 mL/kg predicted body weight, almost twice the recommended volume by ARDSNet) and second, that the probability of organ failure after surgery was influenced by tidal volume64 (Figure 5). Unfortunately, this study did not measure lung injury itself, making it impossible to directly attribute organ dysfunction with a decrease in pulmonary function. Still, the authors’ observation of a strong relationship between prolonged intubation (>24 hours) and the risk of mortality after cardiac surgery64 (Figure 6) is at least suggestive that lung injury may contribute to mortality after surgery, with the acknowledgement that there are several nonpulmonary (eg, bleeding) causes for prolonged intubation in this patient population.
The strongest evidence supporting the hypothesis that lung injury plays a substantial role in the development of adverse perioperative outcomes comes from Futier et al’s72 group, who randomized 400 adults undergoing major abdominal surgery to lung protective versus a more traditional ventilation strategy. They found a substantial reduction in the primary outcome (composite of both pulmonary and extrapulmonary complications within 7 days of the procedure) in patients who received lung-protective ventilation.72 A similar, but small (58 participants) study showed improvements in postoperative lung function (ie, less lung injury) when lung-protective ventilation was used intraoperatively.73
The role of intraoperative lung function and injury on postoperative outcomes should not, however, be unquestioned. A multicenter PROVHILO (PROtective Ventilation using HIgh versus LOw positive end-expiratory pressure) trial randomizing 900 surgical patients to low positive end-expiratory pressure (PEEP, 2 cm H2O) without recruitment maneuvers versus high PEEP (12 cm H2O) with recruitment maneuvers found no difference in lung function postoperatively, or in either pulmonary or extrapulmonary complications.74 Furthermore, a recent, large (961 subjects) trial of low tidal volume ventilation in critically ill patients without ARDS demonstrated no difference in ventilator-free days or any other clinical outcome measured.75 While ICU studies cannot be directly extrapolated to the OR environment, the size of this study as well as the PROVHILO data are compelling and at least raises the possibility that lung-protective ventilation may be useful primarily in patients experiencing ARDS. Additionally, it is increasingly realized that critical care trials including several hundred patients may not be adequately powered, despite the report of a P value <.05, and it is worth noting that the ARDSNet trial was stopped after the fourth interim analysis (because of efficacy) after only 861 patients were enrolled in the study.
Until recently, there were no standardized definitions for acute kidney injury (AKI), which made it difficult to measure the impact of AKI on outcomes as well as test potential therapeutic strategies. In 2004, the Risk, Injury, Failure, Loss of kidney function, and End-stage kidney disease (RIFLE) criteria was developed, followed by the Acute Kidney Injury Network (AKIN) criteria in 2007 and the Kidney Disease: Improving Global Outcomes (KDIGO).76 These definitions have made it easier for researchers and clinicians to study this significant contributor to perioperative morbidity.
Retrospective analyses suggest that AKI is present in 10% to 12% of hospital admissions in the United States, and in the general medical population is associated with increased length of stay, high mortality (10%–17%), and increased hospital costs.77,78 In those who require dialysis and survive at least 90 days, the risk of developing end-stage renal disease at 180 days after ICU admission is 8.5%.79 A smaller study comparing the intensity of renal replacement therapy in ICU patients needing dialysis found that 5% to 6% had developed end-staged renal disease (ESRD) by 90 days.80
In large, noncardiac surgical populations, the risk of AKI was reported to be 0.8% to 1%.81,82 These earlier studies were limited by their nonstandard definitions for AKI (eg, decrease in creatinine clearance of 50 mL/min or 40% from baseline81 or ACS-NSQUIP renal morbidity outcome)82 and should be interpreted with caution when comparing them to more modern studies which use RIFLE, AKIN, and KDIGO criteria.
In cardiac surgical patients specifically, there appears to be a substantial increased risk of mortality that is associated with the development of AKI. One study of almost 3000 cardiac surgical patients found that survival decreased in proportion to the grade of AKI as measured by the RIFLE criteria.83 Two similar studies using KDIGO have found nearly identical results.84,85
In patients undergoing major surgical procedures (defined as surgical procedures requiring admission to an ICU), the incidence of AKI (based on RIFLE criteria) is 31.6%.86 In this large single-center analysis, the aHR for death was 1.18 for RIFLE-Risk and 1.57 for RIFLE-Injury. Interestingly, even in patients who experienced complete recovery of renal function, the risk of death was elevated (aHR 1.20)86 (Figure 7). Using the KDIGO criteria for AKI, another group reported an AKI incidence of 57% in surgical patients admitted to the ICU.87 Similarly, these authors reported a positive relationship between AKI severity and mortality (odds ratio 2.19, 3.88, and 7.18 for KDIGO 1, 2, and 3, respectively).87
An increase in creatinine of 50 μM after cardiac surgery (when compared to baseline values) was associated with a significant increase in mortality in 3434 adult cardiac surgical patients64 (Figure 8). Interestingly, AKI after cardiac surgery also increases the risk of nonrenal morbidity, such as myocardial infarction.88–90
Gastrointestinal (GI) dysfunction, in particular ileus, is a widely appreciated contributor to adverse outcomes in patients undergoing bowel surgery. In colorectal surgical patients, for instance, postoperative ileus increases length of stay by 4.9 to 8.4 days91,92 and is associated with increased hospital costs.91
Even in patients undergoing nongastrointestinal surgery, GI complications, though rarer, can be a major contributor to outcomes. While causation is difficult to establish, it is clear that surgical patients who experience GI dysfunction are also at increased risk of other complications. For instance, in patients undergoing spinal fusion surgery, a database study examining over 250,000 patients from the Healthcare Cost and Utilization Project National Inpatient Sample found that the presence of ileus was associated with a 5.7 day increase in length of stay, as well as an increased incidence of deep venous thrombosis, myocardial infarction, aspiration pneumonia, sepsis, and death, all of which were highly significant.93
Similar results have been reported in other surgical populations. GI complications occur in 0.58% to 2.9% of cardiac surgical patients94–96 and are associated with a significant increase in mortality.95,97 In lung transplantation, major GI complications occur in 7.4% of patients and prolong both ICU and hospital stay, but do not appear to impact overall survival.98
Mesenteric ischemia, GI bleeding, and bowel obstruction, while rare, are particularly dangerous. In cardiac surgery patients, mesenteric ischemia occurs in 0.06% to 0.31% of cases but is associated with a 37% to 77% mortality.97,99 In lung transplantation, the incidence is 8%.98 GI hemorrhage and bowel obstruction occur in 0.52% and 0.25% of cardiac surgical patients and are associated with 18% and 13% mortality, respectively.97
Unlike injury to the cardiovascular, pulmonary, renal, GI, and endocrine systems, there are no mechanical or pharmacologic treatments for hepatic injury. After cardiac surgery requiring cardiopulmonary bypass, 10% of patients will develop hyperbilirubinemia.100 In this patient population, hyperbilirubinemia at 3.5 days after surgery had a profoundly negative impact on long-term survival.100 A similar, but smaller study also identified an association between postoperative hyperbilirubinemia and increased risk of mortality, as well as duration of inotropic support, mechanical ventilation, and both ICU and hospital length of stay.101 Postoperative hyperbilirubinemia has also been associated with increased complications following esophagectomy.102 Compared to other organ systems (eg, cardiovascular, renal), substantially less is known about impact of perioperative hepatic dysfunction on outcomes, presenting interested clinicians with an important research opportunity.
The role of the endocrine system, in particular glucose control, began to receive increased focus in 2001 after publication of a randomized controlled trial showing improved outcomes in critically ill patients (many of whom had undergone cardiac surgery) in whom blood glucose was kept between 80 and 110 mg/dL.103 Three subsequent randomized controlled trials, one of which included over 6000 patients, failed to show a difference between “intensive” and more liberal blood glucose targets in critically ill patients.104–106 None of these studies focused on surgical patients specifically.
In patients who have undergone cardiac surgery, perturbations in blood glucose do appear to be correlated with worse long-term outcomes. For instance, cardiac surgical patients who experience an episode of blood glucose <70 mg/dL are at increased risk of all-cause mortality approximately 5 years after surgery.107 Other study suggests that perioperative hyperglycemia is associated with worsened neurocognitive dysfunction and lower 5-year survival after cardiac surgery108 as well as an increased risk of infection,109,110 length of stay, and respiratory complications.109 It is estimated that an episode of hyperglycemia (>180 mg/dL) is associated with approximately $3192 in additional cost in the cardiac surgery population.109
The relationship between blood glucose control and outcomes is not limited to cardiac surgery. An analysis of over 3000 patients undergoing general, vascular, and urologic surgery found that patients with intraoperative hyperglycemia were at increased risk of infectious complications including surgical site infections, pneumonia, urinary tract infections, and sepsis.111 Perioperative hyperglycemia has also been identified as a risk factor for surgical site infection as well as periprosthetic infection in patients undergoing orthopedic surgery (Table).112
Table. – Key Relationships Between Organ Dysfunction and Perioperative Outcomes
|Organ system||Insult||Outcome effect||Ref|
|Neurologic||Stroke||Perioperative mortality 26% (noncardiac surgery)||6,7|
|Delirium||Relative risk of mortality 1.6 (hip surgery)||20|
|Cognitive dysfunction||Hazard ratio of mortality 1.63 (cardiac surgery)||22|
|Cardiac||MINS||Strong relationship with 30-d mortality||2,35|
|POAF||Increased mortality (cardiac surgery)||56,58–60|
|Prolonged vasopressors||Increased mortality||64,65|
|Pulmonary||Extreme tidal volumes||Increased organ failure (cardiac surgery)||64|
|Renal||AKI||Increased mortality (cardiac surgery, SICU admission)||83–86|
|GI||Ileus||Increased length of stay (spinal fusion)||93|
|Mesenteric ischemia||37%–77% mortality (cardiac surgery)||97,99|
|GI bleed||18% mortality (cardiac surgery)||97|
|Endocrine||Hyperglycemia||Increased infectious complications (noncardiac surgery)||111|
While there is clearly an association between hyperglycemia and poor outcomes in a variety of surgical populations, given the weight of evidence in nonsurgical patients suggesting that strict glucose control does not improve outcomes, it would be premature to suggest intensive glucose control in the perioperative period. Two small, randomized controlled trials of intensive versus conservative glucose control in cardiac surgery patients found no outcome difference between groups113,114 but both studies may have been underpowered. Another study randomizing patients to intensive versus conventional glucose therapy intraoperatively (with identical regimens postoperative) found increased risk of death or stroke in the intensive group, further complicating analysis of the associative data.113
While intraoperative mortality during surgery requiring anesthesia is almost imperceptibly low, 30-day mortality after noncardiac anesthesia remains stubbornly high at up to 2%. Perioperative organ dysfunction plays a substantial role on outcomes following surgery. Advances in diagnostics (eg, MRI, highly-sensitive troponin) have allowed clinicians to detect organ injuries that may not be clinically evident, and which will likely improve the ability of clinicians to predict outcomes in this vulnerable patient population as well as better understand the underlying causes and test potential treatment strategies.