Temperatures in the febrile range are a frequent occurrence in patients with total knee arthroplasties (TKA), but the source and clinical importance of postoperative fever are unclear. Common sources investigated include pneumonia, atelectasis, pulmonary embolus, wound infection, and urinary tract infection, but postoperatively screening patients with a barrage of tests to determine the source of postoperative fever seems to have a low clinical yield. 11–13,27 Work outside the field of orthopaedics has shown that the presence of atelectasis does not correlate with fever. 4,7,26 In addition, fevers after total joint arthroplasty do not seem to be predictive of wound infection or urinary tract infection. 13,27 Blood culture does not seem to be a valuable screening tool for postoperative fever, 8,27 and fevers do not seem to be predictive for the development of deep vein thrombosis or pulmonary embolus. 6 However, in small, isolated reports, the presence of a hematoma at a wound site has been implicated as a possible source of fever. 3,10
For patients with knee arthroplasties, most of the pathologic entities considered as possible sources for postoperative fever have been ruled out, and several investigators have theorized that such fevers are part of the normal physiologic response to the trauma of surgery. 11,13,27 It seems likely that the inflammatory response to total joint arthroplasty could produce a transient release of endogenous pyrogens, resulting in fever, but this theory has not been substantiated with laboratory or clinical tests. The cytokines interleukin-1β (IL-1β), interleukin-6 (IL-6), and tumor necrosis factor-α (TNF-α) are involved intimately in the body’s inflammatory response to injury, and elective surgery has been shown to cause a temporary elevation in serum IL-6 levels. 9,17,24,25 This systemic response seems to be even more robust in the presence of postoperative fevers. 9,17,24 One theory for such serum IL-6 elevation is that its production is stimulated by the local inflammatory response at the surgical site, but to our knowledge, no one has investigated the local and systemic cytokine profiles and correlated them with the febrile response. The role of IL-1 and TNF-α in this process also has not been clarified, although animal studies have shown that subcutaneous injections of pyogenic stimuli result in a local increase in IL-1 and TNF-α levels, which is followed by a systemic increase in IL-6 and fever. 5,19,22,23 In addition, pretreating such animals with antibodies to IL-1 or TNF-α has attenuated the febrile response. 5,19 Fever associated with such subcutaneous injections also was limited in IL-1 deficient knockout mice. 14
The studies described above have interesting implications for the orthopaedic surgeon. These data suggest that the postoperative fever routinely seen after total joint arthroplasties may be the natural adaptive response to local injury to bone, marrow, and muscle. Determining the local and systemic cytokine changes after a hip arthroplasty and correlating such changes to the febrile response would elucidate this relationship. If such fevers are part of the natural adaptive inflammatory response after joint arthroplasty, a decreased use of antipyretics and expensive laboratory tests would be justified. Based on the hypothesis that postoperative fevers after TKA are the result of the local and systemic release of inflammatory cytokines, the goals of the current investigation were to determine: (1) whether IL-1β, TNF-α, and IL-6 concentrations increase in the wound drainage fluid and serum after TKA; and (2) whether these cytokines are more abundant in patients who become febrile.
MATERIALS AND METHODS
Twenty consecutive patients admitted to our institution between April and July 2001 for TKA were enrolled in the study. The operative indication for each patient was osteoarthritis of the knee. No patient had inflammatory arthritis, a systemic inflammatory or autoimmune disorder, a history of any type of cancer or chronic illness, or bilateral TKAs.
Each patient had TKA by the senior author. Anesthesia was administered according to patient and anesthesiologist preference based on preoperative comorbidities. The anesthesia team was not involved in the study, and the choice of anesthesia type was not influenced by the study. Patients were given 1 g intravenous cefazolin before placement of the tourniquet, and then 1 g every 8 hours for 24 hours after surgery. At surgery, each patient had an indwelling urinary catheter placed, which was removed on the second postoperative day. The surgery was done via a medial parapatellar approach; the Howmedica Duracon prosthesis (Rutherford, NJ) with all cemented components were used. Before wound closure, a 1/8-inch silastic wound drain with vacuum suction was placed, which was removed on the second postoperative day.
Patient temperatures were obtained preoperatively and then every 4 hours after surgery for 72 hours. Drain output was recorded every 3 hours until drain removal 48 hours after surgery. Systemic leukocyte counts with differential, hematocrit, and platelet counts were measured immediately preoperatively and then at 1, 24, and 48 hours after surgery. Blood cultures, urine analysis with microscopic evaluation and culture, and a radiograph of the chest were obtained on the second postoperative day for each patient with a temperature greater than 38.5° C. For data analysis, patients were divided into two groups: patients who experienced a temperature greater than 38.5° C at any point during the 72-hour postoperative period (febrile group) and patients whose temperature readings remained less than 38.5° C (afebrile group). A fever workup, including a radiograph of the chest, blood cultures, and urinary cultures, was done for each patient who became febrile and who showed no evidence of pneumonia, atelectasis, or positive blood culture (Table 1).
Peripheral blood and drain fluid samples were obtained immediately preoperatively (except synovial fluid, which was obtained intraoperatively at the time of arthrotomy) and at 1, 6, 24, and 48 hours after surgery. Samples could not be collected at all points for all patients because the wound drains sometimes clotted and stopped draining before the final point. Therefore, the number of samples analyzed per patient varied. Drain reservoirs were emptied 30 minutes before sample collection to ensure that all fluid collected had drained from the wound at the specified point. The blood and drain fluid samples were spun in a centrifuge for 5 minutes at 3000 rpm to separate the serum. The supernatant was removed and stored at −70° C until sample analysis was done. Concentrations of IL-1, IL-6, and TNF-α in the samples then were determined using commercially available enzyme-linked immunosorbent assays (ELISA) (Biosource International, Camarillo, CA). All of the ELISAs were done according to the manufacturer’s instructions. Various dilutions of serum and drainage fluid were tested using these ELISA kits, and control samples with known cytokine concentrations were tested with every assay. Samples were tested in duplicate, and the results are expressed as picograms per milliliter.
Each patient had a 1-year followup to rule out the development of a wound infection or any other source of postoperative fever.
Comparisons of mean cytokine levels, leukocyte counts, and oral temperatures at the five different times were done using one-way analysis of variance (ANOVA) with the post hoc Tukey’s test. Mean demographic and outcome variables in the febrile and afebrile groups were compared using Student’s t tests. Spearman rank correlation coefficients were used to test for a correlation between drain IL-6, serum IL-6, and serum leukocyte counts. Statistical significance was set at p less than 0.05 for all tests. All statistical analyses were done with SPSS 8.0 for Windows (SPSS Science, Chicago, IL).
Of the 20 patients in the study group, 10 (50%) had temperatures in the febrile range (temperature ≥38.5° C) during the 3 postoperative study days. The mean temperature curve for all patients shows that the peak temperature was seen on the first postoperative day, with a gradual decrease during the next 2 days (Fig 1). There was little difference between the patients who were febrile and the patients who were afebrile in terms of possible confounding factors, including age, gender, anesthesia type, operative time, operative blood loss, drain output, hematocrit, and transfusion requirement (Table 1). Twelve patients (60%) required transfusion. Of these, five were in the afebrile group and seven were in the febrile group. Nine patients received autologous blood, one received homologous blood, and two received both. No cell saver or reperfusion drain blood was given. Escherichia coli was grown in the urine culture of one patient who was afebrile. This patient was treated with a 5-day course of ciprofloxacin. As of the 1-year followup, no patient in either group had a wound infection develop.
Mean local and serum cytokine levels for all patients measured preoperatively and postoperatively at the four points are shown in Figures 2 and 3. There was a slight increase in TNF-α in the drain fluid at 1 hour, but this change was not statistically significant. There also were no significant changes seen in serum TNF-α levels. Mean IL-1β levels in the drain fluid peaked at 24 hours, with concentrations at that time being significantly higher than those preoperatively. In no patient’s serum was the IL-1β concentration high enough to be detected by the current ELISA (minimum detectable concentration = 8 pg/mL). There was no significant difference between patients who were febrile and patients who were afebrile in terms of drain or serum concentrations of IL-1β or TNF-α.
As seen in Figures 2 and 3, drain and serum IL-6 concentrations increased significantly during the postoperative period. The patients who were febrile had significantly higher concentrations of IL-6 in the serum and drain samples at 24 and 48 hours than did the patients who were afebrile (Fig 4). Using Spearman rank correlation coefficients, a significant correlation was found between the maximum concentration of IL-6 seen in the serum and that of the drain fluid (Table 2).
There also was a significant increase in the mean serum leukocyte count seen during the postoperative period, with a peak on the second postoperative day (Fig 5A). The elevated leukocyte count was attributable to an acute inflammatory response, as shown by a similar increase in the percentage of neutrophils seen on the manual differentiation of the leukocyte count (Fig 5B). Patients who were febrile had higher mean leukocyte counts, but this difference was significant only on the third postoperative day (Fig 5C). With the use of Spearman rank correlation coefficients, a significant correlation was found between maximum postoperative leukocyte count and serum and between maximum postoperative leukocyte count and drain IL-6 levels (Table 2).
The development of postoperative fever is common after orthopaedic procedures, but the source and clinical importance of these fevers is understood poorly. 11–13,27 The results of the current study support this finding: 50% (10 of 20) of the patients had temperatures of 38.5° C or greater develop during the 72 hours after TKA. The current data also concured with those in other studies of fever after joint arthroplasty in showing that wound infection, urinary tract infection, pneumonia, atelectasis, and deep venous thrombosis usually are not the source of such fevers. 11,13,27 In the current study, there was one urinary tract infection in a patient who was febrile and none in patients who were afebrile. Another potential cause of fever is the transfusion of blood products. Kennedy et al 13 reported an increased risk of febrile-range temperature (>39° C) with a decrease in hematocrit and the transfusion of blood, but this finding has not been substantiated by other studies of postoperative fever in this population. 11,27 We found no association between the development of fever and the hematocrit level or the use of transfusions.
The current results not only suggest that the febrile response is not attributable to such pathologic processes, but they also support the idea that the postoperative inflammatory response may be a primary cause of these fevers. We found elevated levels of IL-1β, IL-6, and TNF-α at the surgical site and elevated IL-6 levels in the serum after total joint arthroplasty. In addition, there were significantly higher levels of drain and serum IL-6 in patients who were febrile than in patients who were afebrile at 24 and 48 hours after TKA. A positive correlation (Table 2) was found between serum IL-6, drain IL-6, and serum leukocyte levels, suggesting that TKA is followed by an increased local production of IL-1β, IL-6, and TNF-α at the surgical site that results in the release of IL-6 into the local circulation, leading to leukocytosis and fever.
Several authors have theorized that fevers seen in the early postoperative period are part of the normal inflammatory response to the trauma of total joint arthroplasty, 11,13,27 but there is little experimental evidence to support this theory. Although some reports have indicated high levels of inflammatory cytokines in the wound drainage blood, 1,15,28 the clinical importance of these high cytokine levels was not clear; nevertheless, the authors advised caution with the reperfusion of unwashed blood. Kristiansson et al 16 found high levels of IL-1β, IL-6, and TNF-α in wound drainage blood after total hip arthroplasty, levels that were significantly higher than those found in the systemic circulation at the same points. Given that wound drainage after total joint arthroplasty has shown increased levels of IL-1β, IL-6, and TNF-α and that high circulating levels of IL-6 also have been observed in such patients, it seems likely that these cytokines play a role in postoperative fever. Two studies of patients who did not have orthopaedic issues have shown an association between circulating levels of IL-6 and postoperative fever. 9,24 Both studies concluded that IL-6 may be involved in postoperative fevers, but neither study analyzed local cytokine levels as a possible source of elevated serum IL-6. To our knowledge, no previous study has investigated the association between postoperative fevers and the local and systemic cytokine profiles.
Several animal studies have suggested that cytokines IL-1β, IL-6, and TNF-α are involved in the febrile response. 2,14,18–22 Cartmell et al 2 and Miller et al 23 helped elucidate the roles of these cytokines by creating a pyogenic inflammatory stimulus in rats by injecting the bacterial cell wall product lipopolysaccharide into a subcutaneous air pouch. The rats had a predictable fever develop after lipopolysaccharide injection. Samples from the fluid in the pouch first showed increased levels of IL-1β and TNF-α and then increased levels of IL-6. Serum samples showed increased levels of IL-6, but no elevation IL-1β or TNF-α. Injection of IL-1β receptor antagonist or antisera to TNF-α blocked the elevation of circulating IL-6 and the fever seen with local lipopolysaccharide injection. This series of experiments suggested that IL-1β and TNF-α interact locally in response to an infectious or inflammatory stimulus, resulting in the production of IL-6 that then is released into the circulation, leading to fever.
It is not clear why some patients have a more robust local and systemic cytokine response than others. Some authors have suggested that the amount of tissue trauma and surgical time may play a role in the magnitude of the inflammatory response. 9,24 This theory was not substantiated by the current study: all patients had the same procedure with the same surgeon, and there was little difference in operative times or blood loss between patients who became febrile and those who did not. Another possible reason for a more robust local inflammatory response in some patients may be bacterial contamination. In theory, patients who have higher local and systemic IL-6 levels develop may have a subclinical amount of bacterial contamination at the wound site, which is addressed successfully by the immune system and therefore does not progress to a frank infection. Most likely, however, there is simply a genetic difference between patients in the magnitude of the inflammatory response to the trauma of TKA.
Febrile-range temperatures commonly are seen in patients after TKA. The results of the current study suggest that such fevers usually are not the result of a pathologic process but, rather, part of the normal inflammatory response to the trauma of surgery that results in local and systemic elevations in the endogenous pyrogen IL-6. Therefore, costly tests aimed at determining the cause of fevers in the 72 hours after TKA are not warranted unless findings on physical examination suggest a specific source.
1. Andersson I, Tylman M, Bengtson JP, Bengtsson A: Complement split products and pro-inflammatory cytokines in salvaged blood after hip and knee arthroplasty. Can J Anaesth 48:251–255, 2001.
2. Cartmell T, Poole S, Turnbull AV, Rothwell NJ, Luheshi GN: Circulating interleukin-6 mediates the febrile response to localised inflammation in rats. J Physiol 526:653–661, 2000.
3. Chmel H, Palmer JA, Eikman EA: Soft tissue hematoma as a cause of fever in the adult. Diagn Microbiol Infect Dis 11:215–219, 1988.
4. Clarke DE, Kimelman J, Raffin TA: The evaluation of fever in the intensive care unit. Chest 100:213–220, 1991.
5. Cooper AL, Brouwer S, Turnbull AV, et al: Tumor necrosis factor-alpha and fever after peripheral inflammation in the rat. Am J Physiol 267:R1431–R1436, 1994.
6. Diamond PT, Macciocchi SN: Predictive power of clinical symptoms in patients with presumptive deep venous thrombosis. Am J Phys Med Rehabil 76:49–51, 1997.
7. Engoren M: Lack of association between atelectasis and fever. Chest 107:81–84, 1995.
8. Fanning J, Neuhoff RA, Brewer JE, et al: Frequency and yield of postoperative fever evaluation. Infect Dis Obstet Gynecol 6:252–255, 1998.
9. Frank SM, Kluger MJ, Kunkel SL: Elevated thermostatic setpoint in postoperative patients. Anesthesiology 93:1426–1431, 2000.
10. Gemer O, Shenhav S, Segal S, et al: Sonographically diagnosed pelvic hematomas and postcesarean febrile morbidity. Int J Gynaecol Obstet 65:7–9, 1999.
11. Guinn S, Castro Jr FP, Garcia R, Barrack RL: Fever following total knee arthroplasty. Am J Knee Surg 12:161–164, 1999.
12. Kenan S, Liebergall M, Simchen E, Porat S: Fever following orthopedic operations in children. J Pediatr Orthop 6:139–142, 1986.
13. Kennedy JG, Rodgers WB, Zurakowski D, et al: Pyrexia after total knee replacement: A cause for concern? Am J Orthop 26:549–552, 554, 1997.
14. Kozak W, Kluger MJ, Soszynski D, et al: IL-6 and IL-1 beta in fever: Studies using cytokine-deficient (knockout) mice. Ann N Y Acad Sci 856:33–47, 1998.
15. Kristiansson M, Soop M, Saraste L, et al: Cytokine and coagulation characteristics of retrieved blood after arthroplasty. Intensive Care Med 21:989–995, 1995.
16. Kristiansson M, Soop M, Sundqvist KG, et al: Local vs. systemic immune and haemostatic response to hip arthroplasty. Eur J Anaesthesiol 15:260–270, 1998.
17. Lahat N, Zlotnick AY, Shtiller R, Bar I, Merin G: Serum levels of IL-1, IL-6 and tumour necrosis factors in patients undergoing coronary artery bypass grafts or cholecystectomy. Clin Exp Immunol 89:255–260, 1992.
18. Louthan FB, Meduri GU: Differential diagnosis of fever and pulmonary densities in mechanically ventilated patients. Semin Respir Infect 11:77–95, 1996.
19. Luheshi G, Miller AJ, Brouwer S, et al: Interleukin-1 receptor antagonist inhibits endotoxin fever and systemic interleukin-6 induction in the rat. Am J Physiol 270:E91–E95, 1996.
20. Luheshi G, Rothwell N: Cytokines and fever. Int Arch Allergy Immunol 109:301–307, 1996.
21. Luheshi GN: Cytokines and fever: Mechanisms and sites of action. Ann N Y Acad Sci 856:83–89, 1998.
22. Luheshi GN, Stefferl A, Turnbull AV, et al: Febrile response to tissue inflammation involves both peripheral and brain IL-1 and TNF-alpha in the rat. Am J Physiol 272:R862–R868, 1997.
23. Miller AJ, Luheshi GN, Rothwell NJ, Hopkins SJ: Local cytokine induction by LPS in the rat air pouch and its relationship to the febrile response. Am J Physiol 272:R857–R861, 1997.
24. Miyawaki T, Maeda S, Koyama Y, Fukuoka R, Shimada M: Elevation of plasma interleukin-6 level is involved in postoperative fever following major oral and maxillofacial surgery. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 85:146–152, 1998.
25. Pullicino EA, Carli F, Poole S, et al: The relationship between the circulating concentrations of interleukin 6 (IL-6), tumor necrosis factor (TNF) and the acute phase response to elective surgery and accidental injury. Lymphokine Res 9:231–238, 1990.
26. Roberts J, Barnes W, Pennock M, Browne G: Diagnostic accuracy of fever as a measure of postoperative pulmonary complications. Heart Lung 17:166–170, 1988.
27. Shaw JA, Chung R: Febrile response after knee and hip arthroplasty. Clin Orthop 367:181–189, 1999.
© 2003 Lippincott Williams & Wilkins, Inc.
28. Southern EP, Huo MH, Mehta JR, Keggi KJ: Unwashed wound drainage blood: What are we giving our patients? Clin Orthop 320:235–246, 1995.