Evaluation of Fever and Infections in Cardiac Surgery Patients

Nosocomial Infections

In the acute period following the first 48 to 72 hours, the differential diagnosis expands considerably and a detailed search for infection should be undertaken (Table 1). In general, surgical site infections are the most common nosocomial infection in postoperative patients. ¹² Pneumonia is also common, occurring in more than 4% of cardiac surgery patients. Risk factors include need for reintubation, mechanical ventilation for more than 48 hours, neurologic dysfunction, hypotension, and multiple blood product transfusions. ¹³ Catheter-related bloodstream infections are of particular concern given the potential for seeding of endovascular prosthetic material. Urethral catheterization is routine in this population, and prolonged use increases the risk of urinary tract infections. ¹⁴ Clostridium difficile infection (CDI) usually develops in the subacute period but can present almost immediately after antibiotic receipt. ¹⁵ Furthermore, the risk of CDI may be substantial even when antibiotic exposure is minimal. For example, one recent study found that 1.5% of postsurgical patients whose only antibiotic exposure was perioperative antibacterial prophylaxis developed CDI. ¹⁶ Nosocomial sinusitis is less common but should be considered in those with nasoenteric tubes, and total incidence rates may be as high as 8% in intensive care unit patients. ¹⁷ It is also important to recognize that infection can be present without fever, especially in immunocompromised patients or those receiving nonsteroidal anti-inflammatory drugs or corticosteroids.

OPEN IN VIEWER

Table 1.

Major Causes of Fever in Postoperative Cardiac Surgery Patients.

Noninfectious	Infectious
Before 48-72 hours
Surgical trauma inflammation a	Rare: community-acquired infection present before surgery, nosocomial infections in patients hospitalized prior to surgery, or infections that prompt surgery (e.g., endocarditis)
Cardiopulmonary bypass a
Transfusion reaction
Bleeding/hematoma
Drug reaction
Malignant hyperthermia
Preexisting traumatic injuries
Myocardial infarction
Stroke or intracranial bleed
Drug or alcohol withdrawal
Hyperacute transplant rejection
After 48-72 hours
Transfusion reaction a	Sternal wound infection and mediastinitis a (usually presents at ~1 week)
Bleeding/hematoma a	Pneumonia or aspiration a
Drug reaction a	Urinary tract infection a
Postpericardiotomy syndrome a (usually several weeks postoperatively)	Intravascular catheter-related infection
Myocardial infarction	Bloodstream infection secondary to any of the above ±prosthetic endovascular infection
Stroke or intracranial bleed	Clostridium difficile–associated diarrhea
Drug or alcohol withdrawal	Leg wound cellulitis (at site of vein harvest)
Deep vein thrombosis	Sinusitis
Acute transplant rejection	Empyema
Pancreatitis	Thoracic aortic graft infection
Acalculous cholecystitis	Left ventricular assist device pocket infection (usually within 1 month) or driveline infection (usually after 1 month)
Hyperthyroidism
Acute adrenal insufficiency
Gout or pseudogout

a

Refers to the most common etiologies in each respective category.

Deep Sternal Wound Infection and Mediastinitis

Mediastinitis is one of the most important complications in cardiac surgery patients, occurring in 0.8% to 5% of cases and carrying mortality rates ranging from 12% to 50%.^18-22 Mediastinitis usually results from intraoperative wound contamination. The wound can also be easily disrupted in the early postoperative phase by coughing or become contaminated when the chest is left open. Staphylococcus aureus is the most common culprit, but other common causes include Gram-negative bacilli, coagulase-negative staphylococci, and streptococci. ²¹ Candida and Aspergillus have also been reported as pathogens, as well as a range of more unusual organisms.^23,24 Numerous risk factors have been identified, including diabetes, obesity, and prolonged operation (Table 2).^22,25-36

OPEN IN VIEWER

Table 2.

Risk Factors for Sternal Wound Infection and Mediastinitis.

Patient Specific	Surgery or Situation Specific
Diabetes a	Prolonged operation a
Obesity a	Use of internal mammary artery grafts
Peripheral vascular disease	Open sternum postoperatively
Prior cardiac surgery	Heart transplant
Tobacco use	Need for reoperation in early postoperative period
Osteoporosis	Postoperative hyperglycemia
Preoperative chest deformities	Length of stay in hospital >5 days before surgery
Malnutrition	Prolonged postoperative intensive care unit course
Immunocompromised status	Improper timing of antibiotic prophylaxis

a

Marks the conditions that are most consistently identified as major risk factors in the literature.

The median time to onset is 7 days, but presentation may occasionally be delayed for months.^19,37 Patients present with a combination of fever, leukocytosis, chest pain, and sternal cellulitis with or without drainage. Systemic signs of infection are common and help distinguish mediastinitis from superficial sternal wound infection. Indeed, bacteremia may occur in 57% of cases and can sometimes precede signs of sternal inflammation. ³⁸

The diagnosis is straightforward in the presence of classic symptoms and is confirmed when pus is discovered in the mediastinum on reoperation. However, diagnosis can be more challenging when patients present only with systemic symptoms. Computed tomography (CT) scan is a sensitive test and may show the hallmark signs of mediastinal fluid collections and pneumomediastinum. However, CT specificity depends highly on the timing relative to surgery, with reported rates of 33% to 39% when performed before 2 to 3 weeks compared with 85% to 100% when performed later.^39,40 One group reported that sternal puncture with aspiration for gram stain and culture had excellent positive and negative predictive value and decreased time to diagnosis. ⁴¹ Mediastinitis missed by CT but diagnosed by gallium scintigraphy has been reported, ⁴² but widespread use of nuclear imaging for this purpose is not common.

Treatment requires both antimicrobial therapy and surgical debridement. Empiric therapy should include broad spectrum antibiotics directed at S.aureus (including methicillin-resistant S.aureus) and Gram-negative bacilli, with adjustment based on culture results. Vancomycin with cefepime or piperacillin-tazobactam are usually appropriate choices. We generally favor empiric antifungal therapy aimed at Candida in severely ill patients with suspected mediastinitis; echinocandins or fluconazole are reasonable choices in this scenario, depending on local fungal susceptibilities. Duration of therapy is typically 2 to 3 weeks after sternal resection and pectoral flap, except with residual osteomyelitis where up to 6 weeks may be required. ⁴³

Postpericardiotomy Syndrome

Postpericardiotomy syndrome is an inflammatory condition that affects 15% to 20% of patients after surgical trauma to the pericardium, pleura, or both. ⁵⁰ The diagnosis should be suspected when there is fever, pleuritic chest pain, leukocytosis, and elevated inflammatory markers, with no other obvious infectious source. ⁵¹ Electrocardiography may show diffuse ST-segment elevation consistent with pericarditis, and pleural and pericardial effusions are common. Symptoms usually develop weeks after surgery, but can also occur in the immediate postoperative period.^50,52 Nonsteroidal anti-inflammatory drugs are the cornerstone of supportive therapy. Colcichine prophylaxis reduces the incidence of postpericardiotomy syndrome and may also be beneficial for treatment.^53,54 A short course of corticosteroids should be considered for refractory or relapsing cases. Prognosis is generally good, but 10% to 15% will recur, and a small fraction will develop constrictive pericarditis. ⁵¹

History and Intraoperative Course

Particular attention should be paid to comorbidities known to predispose to infection (particularly diabetes and immunocompromised status) and drug allergies. Substance or alcohol abuse should raise suspicion for withdrawal, as well as HIV or endocarditis for intravenous drug users. In noncommunicative patients, nursing staff providing direct care to the patient should be questioned for presence of cough, sputum amount and quality, diarrhea, and occult areas of skin breakdown. Medications should be carefully reviewed, including potentially important omissions from outpatient regimens (eg, inadvertent cessation of glucocorticoids leading to adrenal insufficiency). Dates of catheter placement should be reviewed. Cardiovascular surgeries done on an elective basis screen carefully for active infection preoperatively, but patients and family members should be questioned for evidence of signs of infection that may have since developed. For patients hospitalized prior to surgery, records should also be carefully reviewed for fevers, antibiotic exposure, and laboratory and radiology test results.

Important details of the intraoperative course that influence development of early postoperative fever include the type of surgery, duration of cardiopulmonary bypass, induction of hypothermia, intraoperative complications, and number and type of blood products. For surgeries related to endocarditis, the extent of infection and adequacy of debridement should be ascertained, as well as quality of prosthetic material used.

Physical Exam and Vital Signs

The physical exam should focus on the surgical site for excessive or new pain, tenderness, drainage, erythema, or dehiscence. Recipients of bypass grafting should also have the site of vein harvest examined. Other signs of infection may include erythema or thrombophlebitis around catheter sites and purulent output from drains and thoracostomy tubes. Cardiac exam may reveal a pericardial friction rub indicative of postpericardiotomy syndrome, and a new murmur can indicate prosthetic valve endocarditis. Abnormal lung sounds may be suggestive of pneumonia. Abdominal tenderness may be a sign of pancreatitis, acalculous cholecystitis, ischemic bowel, or severe CDI. The skin should be examined for rashes and any pressure sores or ulcers. Any focal neurological deficit mandates urgent neuroimaging to assess for stroke or intracranial bleed. Extremities should be evaluated for signs of ischemia, or new unilateral swelling that might indicate deep vein thrombosis.

Extreme hyperthermia (exceeding 106°F) is typically only seen with a select few, rare noninfectious etiologies such as malignant hyperthermia, neuroleptic malignant syndrome, and thyroid storm. Bradycardia and/or conduction abnormalities can be a manifestation of prosthetic valve endocarditis. Hypotension in the presence of fevers should raise concern for sepsis.

Laboratory Tests

The leukocyte count peaks in the first 3 days after cardiac surgery. ⁵⁵ Thus, leukocytosis that persists or worsens after three day, particularly with an increased proportion of band forms and neutrophils, should raise concern for infection. C-reactive protein levels appear to have a similar trajectory but are nonspecific and not routinely followed in postoperative patients. Additional laboratory tests should be focused by the clinical picture; for example, abdominal pain should prompt liver function tests, pancreatic enzymes, or appropriate imaging. The utility of procalcitonin in surgical patients is discussed later in this review.

Microbiological Evaluation

For fevers occurring after 48 to 72 hours, 2 sets of blood cultures (with at least 1 peripheral specimen) are generally indicated. Negative cultures must be interpreted in the context of any preceding antibiotic use. Although commensal organisms such as coagulase-negative staphylococcus are often contaminants, they also have high pathogenic potential in cardiac surgery patients with endovascular prostheses. Isolation of the same organism from multiple blood culture sets increases the likelihood of a true bloodstream infection.

If local signs of surgical site infection are present, fluid from the incision site should be sampled in an aseptic manner and cultured. Culturing fluid from existing drains can be misleading as the organisms that are found usually represent drain colonization, and should generally be avoided. ⁵⁶ If pneumonia is suspected, a tracheal aspirate (in intubated patients) should be obtained for gram stain and culture. Bronchoalveolar lavage may be considered, particularly in immunocompromised patients who are at higher risk for unusual pathogens. Urinalysis and urine culture should be obtained if symptoms of dysuria are present, or with unexplained fevers in a noncommunicative patient. However, distinguishing infection from colonization can be difficult for patients with urethral catheters. New onset of diarrhea in febrile postsurgical patients should prompt testing for C.difficile.

Imaging Studies

Chest radiographs are commonly obtained but can be difficult to interpret given the frequency of atelectasis and effusions in cardiac surgery patients. Pleural effusions should rarely prompt thoracentesis unless there is a high clinical index of suspicion of parapneumonic effusion or empyema. Chest CT scan should be considered if a pulmonary infection or embolism is suspected, as chest radiographs are frequently inconclusive. CT is also useful in cases of suspected mediastinitis, but clinicians must recognize its poor specificity in the first 2 to 3 weeks after surgery.

Empiric Antimicrobial Therapy for Suspected Sepsis

Empiric antibiotics should be reserved for cases with a high index of suspicion for infection or who have clinical signs of severe sepsis. Distinguishing sepsis from the vasoplegic shock that can follow cardiopulmonary bypass is difficult, but a reasonable approach is to administer broad spectrum antibiotics following appropriate cultures, with discontinuation after 48 hours if infectious workup is negative. The choice of antibiotics should be guided by local antibiogram susceptibility patterns, but vancomycin with an antipseudomonal β-lactam such as cefepime or piperacillin-tazobactam are usually appropriate starting options. We generally reserve empiric antifungal therapy for patients with risk factors for invasive Candida infections, such as immunosuppression (including post–cardiac transplant patients), presence of a central venous catheter or vascular device for more than 5 to 7 days (or shorter if a femoral site was used), recent receipt of total parenteral nutrition, or a significant amount of body surface colonization with Candida. In addition, antifungal therapy is appropriate in severely ill patients with suspected mediastinitis and those with suspected left ventricular device infections. If used, we recommend an echinocandin as first-line therapy; fluconazole can be used as well, though clinicians should be aware of the higher risk of resistance with several Candida strains.

Prosthetic Valve Endocarditis

Prosthetic valve endocarditis (PVE) is a serious complication with highest risk in the first 3 months after surgery. ⁵⁷ Early-onset PVE (within the first 2 months of surgery) is typically nosocomial in origin and results from direct intraoperative contamination or bacteremia arising from other sites. S.aureus and coagulase-negative staphylococci are the most common pathogens, followed by Gram-negative bacilli, enterococci, and Candida.^58,59 The pathophysiology of late-onset PVE resembles native valve endocarditis (NVE) and usually results from transient bacteremia in ambulatory patients; streptococci, staphylococci, and enterococci are the most common causes. ⁵⁸ Although bioprosthetic valves have a higher long-term risk of endocarditis compared to mechanical valves, ⁶⁰ studies have reached differing conclusions regarding which carries higher risk for early endocarditis.^57,61-63

Any positive blood culture in a postoperative patient with a prosthetic valve should raise concern for PVE. Compared with NVE, PVE more commonly leads to conduction abnormalities, new or changing murmurs, heart failure, and arterial emboli.^64-66 The standard for diagnosis remains the Duke Criteria, ⁶ , for which the major criteria are multiple blood cultures positive for organisms typical for infective endocarditis, and echocardiographic evidence of vegetations or valvular destruction. Transesophageal echocardiogram (TEE) is significantly superior to transthoracic echocardiogram for prosthetic valves (>90% vs <40% sensitivity) and can better detect invasive complications such as abscesses, fistulae, perforation, and paraprosthetic leaks.^68-70 TEE can still miss cases (particularly with early disease), but a repeat TEE 1 week later when there is a high degree of suspicion can improve diagnostic yield. ⁷¹ Recently, several investigators have explored real-time 3-dimensional TEE for PVE and reported improved anatomic and functional assessment of the prosthetic valve, which may improve surgical planning and better predict embolism risk.^72,73

A minimum of 6 weeks of antimicrobial therapy is standard for PVE, and surgery is required more often than in NVE due the higher rate of invasive infections and difficulty eradicating infection. Surgery is indicated for PVE associated with abscess, dehiscence, or valvular obstruction or regurgitation, as well as persistent/recurrent bacteremia or emboli despite appropriate antibiotic therapy. ⁷⁴ In addition, certain pathogens are much more likely to require surgery because of their aggressive nature (particularly fungi and S.aureus) or lack of efficacious antibiotic therapy (ie, multidrug-resistant Gram-negative bacilli and enterococci). However, identifying patients who will benefit from surgery is nuanced and must take into account multiple factors.

Left Ventricular Assist Device Infections

Infections are a major complication associated with left ventricular assist devices (LVADs) and are associated with mortality rates as high as 20%. ⁷⁵ The smaller second-generation LVADs have much lower infection rates compared with first-generation devices, but overall rates still range from 30% to 50%.^75-77 Percutaneous driveline infection is the most common LVAD infection and usually results from migration of skin flora along the driveline, presenting as exit site cellulitis. Diagnosis can be challenging as signs can be mimicked by local site irritation and superficial swab cultures are difficult to interpret. Eighty-five percent of driveline infections occur more than 30 days after device implantation with an average time of onset of 6 months.^78,79

In contrast, the majority of pocket infections present within 30 days as they arise from seeding of the pocket space at the time of or shortly after operation. ⁸⁰ Later infections usually result from spread of organisms via the driveline exit site. Pocket infections can manifest as abscesses or wound drainage with overlying cellulitis and systemic symptoms. Distinguishing superficial driveline infections from pocket infections is critical. The former can often be managed with 2 to 4 weeks of antimicrobial therapy with or without surgical debridement, while pocket infections require surgical debridement as well as chronic suppressive antimicrobial therapy. ⁸¹ Diagnosis is made by visualization of the abscess in the pocket space by ultrasound or CT scan, and aspiration with culture should be done to identify the causative organism.

Left ventricular assist device recipients are at highest risk for bloodstream infections within the first 30 days, which can result in endovascular involvement of the valves or blood-contacting surfaces of the pump and cannula. These are the most serious of the LVAD-related infections, and most commonly are due to staphylococci, enterococcci, Candida, and Gram-negative bacilli. Patients may present with signs of sepsis, and less commonly with new dysfunction of the pump valves or septic emboli to distant sites. ⁸² The diagnosis of pump/cannula infection is usually presumptive when multiple positive blood cultures are obtained with no other focus for infection identified. TEE may identify an intracardiac vegetation adjacent to the cannula, but image quality usually suffers from the reflective internal metal surfaces. From an infectious perspective, optimal management involves complete device removal and a prolonged course of antibiotics; however, since most patients are completely dependent on their device, chronic suppression (either indefinitely, or until heart transplantation) is generally used with LVAD retention. ⁸¹

Vascular Graft Infections

Thoracic aortic prosthetic grafts are frequently used in cardiovascular surgery. Overall, prosthetic grafts are complicated by infection in 1% to 6% of cases. ⁸³ Direct inoculation of the surgical site or hematogenous spread may be responsible, and infections usually manifest within the first 2 months of surgery. ⁸⁴ Early infection often presents with overt signs such as fever, chills, pain, and discharge from the surgical site, whereas late infections usually present more subtly. Gram-positive, Gram-negative, and polymicrobial infections are all possible, with staphylococci being the most common pathogens.^84,85

There is no consensus approach to the diagnosis of suspected prosthetic graft infections, but evaluation should at least begin with 2 sets of blood cultures. Positive cultures combined with other clinical or radiographic signs of infection are often enough to make a presumptive diagnosis. CT scan with contrast has high sensitivity and specificity in acute infection, with characteristic findings that include air bubbles, periprosthetic fluid collections, and false aneurysms. ⁸⁶ However, CT performs poorly with chronic infection, and distinguishing from postsurgical changes can be difficult in the immediate postoperative period. ⁸⁵

Nuclear medicine studies can often confirm vascular graft infection. Tagged leukocyte scans have excellent performance for late and subacute infections, with sensitivity approaching 100% and specificity of 94% in one series. ⁸⁷ Gallium scintigraphy may be more specific but less sensitive. ⁸⁸ However, nuclear studies are of limited utility in the early postoperative period due to surgical inflammation, suffer from poor insight into vascular anatomy, and are impractical in emergent situations.

If suspicious radiologic findings are discovered and the microbiologic etiology remains unknown, percutaneous sampling of suspected abscesses (or exploratory surgery in some cases) can provide a definitive diagnosis. Unless severe sepsis is present, every attempt should be made to withhold antimicrobial therapy until blood and/or fluid cultures are obtained. Microbiologic diagnosis is critical because of the wide spectrum of pathogens that can cause infection and the prolonged duration of therapy that will be required (including possible lifelong antibiotic suppression). Reoperation is sometimes necessary as well.

Procalcitonin

Procalcitonin (PCT) is a biomarker for bacterial infection that has been extensively studied in respiratory infections and sepsis. PCT has also been explored in postoperative fever, although interpretation is complicated by differences in degree of PCT elevation after different types of surgery. ⁸⁹ Fortunately, several studies have specifically examined PCT in cardiac surgery patients. Aouifi et al ⁹⁰ found that PCT levels peaked on postoperative day 1 and returned to baseline by day 3 in a control group of noninfected cardiac surgery patients, but in patients with suspected infection, PCT sensitivity was 85% and specificity 95% using a cutoff of 1 ng/mL. The investigators also found that higher PCT levels correlated with severity of infection and overall had superior performance compared to C-reactive protein, although sensitivity was decreased by recent antibiotic use. Other studies have also demonstrated utility in children after cardiac surgery.^91,92

Thus, PCT appears to be a useful marker for differentiating bacterial infection from other causes of postoperative fever in cardiac surgery patients, particularly with dramatically elevated levels and/or increasing levels with serial tests. However, false negatives and positives are not uncommon, and clinicians must interpret results in the context of other clinical findings.

Positron Emission Tomography/Computed Tomography (PET/CT)

While traditionally used for detection and surveillance of malignancy, PET/CT is emerging as a valuable tool for detection of infectious conditions. With regards to cardiovascular surgery patients, PET/CT appears to be useful for diagnosing prosthetic valve endocarditis as well as cardiac implantable electronic device infections, despite concerns that the physiologic uptake of the tracer in myocardium might limit its use in this setting. ⁹³ PET/CT also has the advantage of potentially detecting metastatic foci in endocarditis. ⁹⁴ A recent prospective study of 72 patients with suspected PVE found that abnormal FDG uptake around the prosthetic valve on PET/CT had a sensitivity of 73% and specificity of 80% for diagnosing PVE (determined using the modified Duke criteria during a 3-month follow-up period as the gold standard). ⁹⁵ When abnormal FDG uptake was incorporated as a potential new major criterion of the modified Duke criteria, the sensitivity of the combined criteria was 97%, suggesting that PET/CT can have substantial diagnostic impact when used as part of a global assessment that includes clinical, microbiological, and echocardiographic evaluation.

Positron emission tomography/computed tomography also appears to be superior in diagnosing vascular prosthetic graft infections compared to CT alone, with one study reporting sensitivity of 93% and specificity of 91%.^96-98 Compared with other nuclear imaging modalities, PET/CT has the added advantage of delineating precise anatomic information about the extent of infection, which can help guide the strategy for reoperation. However, performance depends on criterion for interpreting positivity, and false positives have been reported. ⁹⁹

The routine use of PET/CT is limited by its high cost as well as lack of availability at many hospitals. In addition, the major limiting factor for PET/CT from a clinical standpoint is the potential for false positive FDG uptake in the early postoperative period from surgical inflammation. The precise timeframe after surgery at which PET/CT may be useful requires further investigation, but a minimum of 4 weeks has been suggested. ¹⁰⁰ Thus, while the exact role of PET/CT in this setting is still an area of research, it seems appropriate to consider in patients more than 4 weeks removed from surgery where infection of prosthetic material is strongly suspected but conventional imaging studies are negative.