Analysis of the Causes of Elevated C-Reactive Protein Level in the Early Postoperative Period After Primary Total Knee Arthroplasty
Tae Won Kim, , Dong Hwan Kim, , Won Seuk Oh, , Jae Ang Sim, , Yong Seuk Lee, , Beom Koo Lee,
The Journal of Arthroplasty. Volume 31, Issue 9, September 2016, Pages 1990–1996
Measurement of C-reactive protein (CRP) levels as a screening test for acute periprosthetic joint infection has high sensitivity and low specificity. We performed the present study to analyze the causes of elevated CRP levels in the early postoperative period after primary total knee arthroplasty (TKA). This study is intended to help the postoperative care of patients through understanding the factors associated with postoperative elevation of CRP.
The records for 627 patients who underwent primary TKA between January 2005 and May 2013 were examined. We excluded 50 patients for whom TKA with inflammatory arthritis or revision TKA was performed. We measured serial CRP levels during the 4-week early postoperative period in all included cases to find the cases that showed a CRP pattern of elevation–depression–elevation (a bimodal pattern). We analyzed the causes of re-elevated CRP levels in patients with a bimodal pattern of CRP change.
Of the 577 included patients, 76 showed bimodal CRP elevation patterns. Eighteen elevations were caused by postoperative infections (periprosthetic infection), 10 by cardiovascular problems, 11 by gastrointestinal problems, 12 by urologic problems, 10 by respiratory problems, and 15 had unknown origins.
Our study shows that elevated CRP levels after TKA can have various causes. Although there may be other causes for an elevated CRP, it is essential to perform a work-up for prosthetic joint infections. In addition, there seems to be a need to evaluate noninfectious causes and infection of other sites, in addition to periprosthetic infection.
- total knee arthroplasty;
- C-reactive protein;
- periprosthetic infection;
- early postoperative period;
- bimodal pattern
Total knee arthroplasty (TKA) is a reliable method for treating severe arthritis accompanied by pain and deformity. The frequency of its use has been increasing along with the aging population. However, various postoperative complications may lead to undesirable outcomes, such as periprosthetic infection, which occurs in about 1%-3% of all TKAs. The most important issues with this problem are rapid diagnosis and early treatment . However, no diagnostic methods have both a high sensitivity and a high specificity for the detection of postoperative periprosthetic infection. The requirements for a screening test to rule out the possibility of periprosthetic infection are a high sensitivity, a high negative predictive value, and cost-effectiveness .
Serum C-reactive protein (CRP) level is generally used as a screening test for infection because it is simple, cost-effective, and highly sensitive . However, the main problem with the CRP level test is that it has low specificity. For example, CRP levels may increase under several conditions in addition to infection, such as trauma, surgery, burns, tissue necrosis, immunologically mediated inflammatory diseases, and advanced cancer. Moreover, other clinical situations such as vigorous exercise, and even some psychiatric diseases, are associated with mild CRP changes . In addition, a definite threshold level of CRP required for diagnosing infection is clearly not defined, and the reported results vary 5, 6 and 7. Moreover, the sensitivity and specificity for diagnosing infections depend on different CRP threshold levels under different conditions 6, 7 and 8. There is a report that a relatively high false negative rate is possible using absolute CRP level alone in cases where postoperative infection is diagnosed early . CRP levels after TKA usually peak at day 2 or 3 postoperatively and diminish abruptly if there are no postoperative complications 10 and 11. However, there was a report that a bimodal pattern of CRP level change has been used to diagnose postoperative infections. In this case, an increase in CRP levels after an initial decrease was observed on postoperative day 13 in the case of deep infection, and on postoperative day 10, in the case of superficial infection .
For the previously mentioned reasons, CRP level has several disadvantages when used as an exclusive screening test. Recently, new diagnostic criteria for periprosthetic infections have been proposed which include joint fluid analysis, frozen biopsy, microbiological culture, and molecular biological tests, and clinical features, in addition to serological tests such as CRP level 13 and 14.
Although various methods for interpreting CRP have been reported previously, there is no definitive guideline on how to interpret CRP clinically 5, 6, 7, 8 and 12. As CRP has a high sensitivity but a low specificity, its elevation due to different etiologies can be clinically observed. In addition, the use of CRP may confuse diagnoses in elderly populations where a concurrent illness may elevate CRP and where TKA is often indicated.
The authors experienced a number of incidents involving abnormal increases in CRP during the treatment of patients after TKA and made many efforts to differentiate the infection of the surgical site after arthroplasty from other causes. However, as mentioned in a number of previous studies, because the variation in CRP results is relatively wide and the absolute standard level of CRP is not clearly established to diagnose infection, the authors started trying to understand why the normally elevated CRP level decreases and suddenly increases again during the treatment of patients after TKA. We uncovered other causative cases besides infection, which was the main concern, and determined that using the bimodal pattern would be more reasonable than using the absolute value of CRP in designing this retrospective study.
For the reasons previously mentioned, we analyzed the causes of elevated CRP level within a 4-week postoperative period after primary TKA. This analysis is intended to help in the postoperative care of patients by reporting several causes of elevated CRP that can be confused with periprosthetic infection. In addition, this study is intended to help the postoperative care of patients through understanding the factors associated with postoperative elevation of CRP.
The hypothesis for this study is that early postoperative CRP level after TKAs may be elevated by various causes because elderly patients are more susceptible to postoperative complications that may elevate CRP levels.
Materials and Methods
From the 627 patients who had undergone 755 TKAs at our hospital between January 2005 and May 2015, we examined 577 patients who had undergone 705 arthroplasties. We excluded 50 patients (50 TKAs) for whom TKA with inflammatory arthritis or revision TKA was performed.
Bilateral TKAs were performed simultaneously or serially. When TKA was performed serially, the second arthroplasty was used for CRP analysis. Because the surgery on the opposite side is conducted 1-2 weeks after the first surgery, the second surgery would become a confounding variable if the first surgery is used as the standard due to the design of this study, in which the variation in CRP over 4 weeks is analyzed. This was determined inappropriate and thus the second surgery was set as a standard. The arthroplasties were performed by 2 experienced orthopedic surgeons, and each patient had his or her CRP measured postoperatively.
Postoperative CRP tests were conducted daily in all patients for the first 2 weeks of hospitalization as the absence or presence of early infection was considered significant. The test was also performed in the third and fourth postoperative week through the outpatient facility after the patients were discharged. In particular, the hospitalization period was longer than 2 weeks for patients with bimodal early CRP patterns. For these patients, the test was performed daily until discharge.
The study was performed by reviewing the electronic medical records (EMRs) of the patients and investigating the graphs for CRP level during the first 4 weeks of the postoperative period. A method for defining the CRP threshold on a specific postoperative day was possible but unusable because the day representing re-elevated CRP varied depending on the cause. Instead of a threshold level, the causes of elevation were analyzed for patients who showed a bimodal pattern (where CRP was initially elevated, declined, and then re-elevated) in the first 4 weeks after surgery.
Among the 577 patients, a total of 76 patients (14 men and 62 women) met the requirements. Of those, 23 patients had undergone bilateral arthroplasty. The average age was 71.3 years. Most patients had normal preoperative CRP levels. As an exception, when both knees were operated on in series, surgery on the opposite side was performed 1-2 weeks after the first surgery even if the CRP level was not normalized after the first surgery. However, these patients were excluded from the CRP pattern analysis for the normal group.
This study was approved by the Institutional Review Board of our hospital.
All medical records including progress notes, consultation notes, blood tests, urinalysis results, and radiologic images for the patients at the time of re-elevated CRP level were used to determine the cause of the CRP elevation. Causes of CRP elevation were divided into surgical infection and nonsurgical infection, then divided into respiratory, gastrointestinal, urological, or cardiovascular etiology. All causes that could not be traced with EMR were marked separately. We statistically analyzed the CRP bimodal pattern for each cause using SPSS 22.0. We compared the start date of re-elevation for the mean CRP level and the mean second CRP peak level in the bimodal curve using 1-way analysis of variance or Kruskal–Wallis test.
Postoperative Infection (Surgical Site Infection and Periprosthetic Infection)
Infections were classified as superficial or deep. Deep infection was diagnosed in cases with widespread fever, obvious pus, leukocyte count greater than 50,000 in a joint fluid analysis, or a positive bacterial culture. Superficial infection was diagnosed when warmth, erythema, or pain occurred at the surgical site with no abnormal findings in the joint cavity. These were treated solely with additional intravenous antibiotics.
Simple upper respiratory infection including acute sinusitis was considered to be the cause when symptoms were clearly described in the internal medicine or ear-nose-and-throat department clinical notes. Pneumonia was considered to be the cause if diagnosis was confirmed by the presence of symptoms, chest x-ray, chest computed tomography (CT), or a clinical opinion made by a respiratory physician.
Transient paralytic ileus with constipation was considered to be the cause when CRP re-elevation was accompanied by increased gas content in abdominal CT and a 3-day history of constipation followed by improvement and CRP normalization after an enema or medical treatment. In addition, in patients with recurrent diarrhea, nausea, vomiting, anorexia, or abdominal pain accompanied by fever, we found and classified the causes using abdominal CT, sonography, or endoscopy performed by gastrointestinal physicians.
Urinary tract infection and acute renal failure were considered to be the causes by confirming clinical symptoms and urinalysis results and when diagnoses were made by urologists or internal medicine physicians.
Cardiovascular inflammation or ischemic heart disease was considered a potential cause. If an ultrasound scan or chest CT confirmed a diagnosis of deep vein thrombosis (DVT) with no other causes, DVT was considered to be the cardiovascular cause.
If causes were not obvious from EMR or investigations, those cases were classified to be idiopathic.
Among the 577 patients, a total of 76 patients showed a bimodal pattern of CRP change. Postoperative infections were identified in 18 patients (24%). In the remaining 58 patients (76%), CRP changes were because of other causes or were idiopathic (Fig. 1).
Of the 18 patients with postoperative infections, 12 had deep infections and 6 had superficial infections. The rate of deep infection was 1.7% (12 cases of 707 cases). Superficial infections were treated with intravenous antibiotics while acute deep infections were treated with debridement, washing, and intravenous antibiotics in all cases (Table 1).
Of the 10 patients with respiratory infections, 5 had simple upper respiratory infections and all improved with conservative treatment. Four patients were diagnosed with pneumonia, and one of them had aspiration pneumonia. One patient was diagnosed of acute sinusitis and treated with antibiotics (Table 2).
A total of 11 patients were diagnosed with abdominal CT. Among them, 7 had paralytic ileus or constipation accompanying CRP elevation. Gastrointestinal problems were improved, and CRP normalized by enema or medical treatment. Of the remaining 4 cases, there was one diagnosis each of ischemic colitis, pseudomembranous colitis, acute cholecystitis, and aggravation of chronic cholecystitis (Table 3).
Of the 12 patients with urological causes of CRP elevation, 11 had a urinary tract infection and 1 had acute renal failure. Of those with urinary tract infections, 3 were diagnosed with cystitis, which was treated with antibiotics. One patient received treatment for acute renal failure (Table 4).
Ten patients were diagnosed with DVT, and all improved with conservative treatment (Table 5).
In 15 patients, etiologies could not be traced. In these cases, the second peak value for CRP was significantly lower compared to the cases with known causes. The CRP level in the follow-up lab decreased again without any specific treatment in all 15 cases. However, the limitation of this study is that there was no further effort to find the cause after the decrease in these cases.
Based on the overall statistical analysis, there was a statistically significant difference in the start date of re-elevation from the mean CRP level between cases of periprosthetic infection and those with cardiovascular causes (DVT; P = .001), that is, the reincreasing tendency appeared significantly later in cases of periprosthetic infection than with DVT. In addition, there was a statistically significant difference in the second peak CRP value between the cases of surgical site infection and those with cardiovascular causes (DVT; P = .004). In other words, the second peak CRP value was significantly higher in the cases of surgical site infection compared with the DVT cases. Moreover, the second peak CRP value was statistically significantly low in the cases with unknown causes compared to the remaining 5 groups with various known systematic causes (P = .000, .001, .008, .000, and .004; Table 6).
In summary, among the 577 patients, 76 showed a bimodal pattern of CRP changes. The most common cause was postoperative infection. In addition, respiratory, gastrointestinal, urological, cardiovascular, and idiopathic causes were detected (Fig. 1). All 76 patients showed a common bimodal pattern of CRP changes, with slight daily differences according to the cause, over 4 weeks in the postoperative period (Table 6).
Postoperative infection after TKA is a serious complication for orthopedic surgeons. However, it is generally difficult to diagnose and treat as there are few therapeutic options. For this reason, various diagnostic methods and treatment plans for postoperative infection after TKA are being introduced 5 and 6.
Diagnosis of postoperative infection after TKA can be made based on a clinical history of delayed wound healing, surgical wound fistula, and persistent postoperative pain. Other diagnostic methods recently introduced include hematologic tests, joint fluid analysis, imaging, pathologic tests, microbiological culture, and molecular biological techniques. Of these methods, hematologic tests are most widely used as a primary screening test to predict infection 5, 6, 13 and 14.
CRP, a serological marker synthesized in the liver whose level is elevated in response to acute-phase reactions such as infection, trauma, or malignancy, is used in serological tests. CRP is used more frequently as a screening test for infection after arthroplasty than other serological markers because of its sensitivity, rapidity, and cost-effectiveness .
CRP hits a peak on days 2 or 3 after surgery and diminishes to a normal value of less than 1 mg/dL by 3 weeks after surgery. The highest peak on days 2-3 after surgery reflects tissue damage from the operation. Unless other inflammation occurs, CRP levels return to preoperative levels within 3 weeks . In this study, CRP peaked (10.96 mg/dL) on the second postoperative day on average for normal patients who did not show a bimodal pattern, and then, consistently declined. The value reached a minimum of less than 1 mg/dL (0.87 mg/dL) for the first time on the 10th postoperative day and then decreased to 0.38 mg/dL in the third postoperative week, eventually returning back to a normal level of less than 0.5 mg/dL, which was the reference level, within 3 weeks (Fig. 2). When CRP level is abnormally high after 3 weeks postoperatively or shows a bimodal pattern (elevation–depression–elevation), infection is suspected . However, it is reported that CRP levels are able to increase under several conditions such as trauma, surgery, burns, tissue necrosis, immunologically mediated inflammatory diseases, and advanced cancer in addition to infection. Even some psychiatric diseases and vigorous exercise are associated with mild CRP changes .
Another problem is that both the sensitivity and specificity for diagnosing infection can be differentially effected by the threshold level of CRP or the time when the level is checked 6, 7, 8 and 9. Therefore, instead of using a threshold CRP level in this study, the causes of CRP change were analyzed for patients who showed a bimodal pattern in the first 4 weeks after surgery. Suh et al  reported using the bimodal pattern of CRP level change to diagnose postoperative infection, where re-elevation of CRP levels after a continuous initial depression was observed on postoperative day 13 in cases of deep infection and on postoperative day 10 in case of superficial infection.
The present study was retrospectively performed to investigate the causes of abnormal CRP elevation by reviewing the EMR of patients. Results showed that only 24% of the elevated CRP cases were caused by postoperative infection, whereas 56% had other identifiable causes and 20% were of unknown causes. In particular, we were able to confirm gastrointestinal causes including transient paralytic ileus with severe constipation, urologic causes including urinary tract infection, respiratory causes including upper respiratory infection, and cardiovascular causes including DVT, as relatively common causes of CRP level elevation. Particularly, the patients with elevated CRP caused by transient paralytic ileus with severe constipation were those who did not excrete for at least 3 days and did not pass gas without complaints of abdominal distension and discomfort. Plain abdominal X-ray and abdominal CT were performed to confirm fecal impaction and transient paralytic ileus, followed immediately by an enema. One day after a considerable amount of gas was passed with excretion on enema, the symptoms improved and CRP declined (Fig. 3). Another noteworthy result was that many gastrointestinal causes and DVT were not related to bacterial or viral infections
It is known that thrombophlebitis occurs in acute-phase DVT, thereby elevating serum inflammatory markers such as interleukin 6, interleukin 8, or CRP 16, 17 and 18. In this study, all patients classified as DVT showed improvements in clinical patterns with declining CRP as treatment began after the diagnosis of DVT. Thus, DVT was considered the cause of CRP increase in these patients. As a result, it was concluded that DVT can be considered a possible cause for postoperative CRP elevation after TKA.
In addition, tissue necrosis is a potent acute-phase stimulus, and after ischemic heart disease, there is a major CRP response, the magnitude of which reflects the extent of myocardial necrosis. Myocardial necrosis due to abrupt closure of coronary artery, in case of acute myocardial infarction, leads to a systemic and regional humoral and cellular inflammatory response aiming to promote the local myocardial healing process and scar formation. In the early phase of ischemic heart disease, cytokines play an important cytoprotective role, mainly by reducing cell apoptosis. Plasma CRP concentration increases after the cytokines activation in the initial hours of ischemic heart disease 19 and 20. In addition, recent report suggests that the atherosclerotic process is characterized by a low-grade inflammation altering the endothelium of the coronary arteries and is associated with an increased level in markers of inflammation . In this study, there was no patient who had diagnosed as ischemic heart disease such as myocardial infarction accompanying CRP elevation.
Parvizi et al  reported that the incidence of postoperative ileus after lower extremity reconstruction ranges from 0.3% to 4.0% and if not recognized or improperly addressed, may result in more severe complications such as bowel perforation, peritonitis, sepsis, multiorgan failure, and even death. Bederman et al  reported that postoperative ileus menifests as abdominal pain, abdominal distension, anorexia, nausea or vomiting, and failure to pass stool or gas and that bilateral TKA is one of the risk factors leading to postoperative ileus.
In this study, there were 7 patients who had paralytic ileus or constipation accompanying CRP elevation. Gastrointestinal problems were improved, and CRP normalization occurred after an enema or medical treatment. We classified paralytic ileus with severe constipation as the cause of CRP elevation for those 7 patients. In addition, ischemic colitis, pseudomembranous colitis, acute cholecystitis, and aggravation of chronic cholecystitis were diagnosed by abdominal CT, and all were resolved with appropriate treatment with no infection of the joint. Therefore, it is thought that gastrointestinal problems are significant causes of postoperative CRP elevation and that abdominal CT is important for differential diagnosis and appropriate treatment.
In this study, there was one patient who was diagnosed with an acute renal failure accompanying CRP elevation in the early postoperative period after primary TKA. Acute renal failure was classified as the cause based on a previous article reporting that CRP could increase in acute renal failure .
After classifying patients according to the cause of their CRP elevation, we statistically analyzed the differences in the graphs of daily mean CRP levels for the 4-week postoperative period (Table 6). Two significant characteristics were found. In other words, a reincreasing pattern appeared significantly later in cases of surgical site infection compared with the DVT cases. In addition, the second peak CRP value was significantly high in cases of surgical site infection compared with the DVT cases. The second peak CRP value was significantly lower in case with unknown causes compared to the remaining 5 groups with various known systematic causes.
Several limitations exist in this study. First, this is a retrospective study using past medical records so we cannot know information that was not recorded. Second, it is possible that causes were not explored if the CRP returned to normal after the peak in the follow-up period. Third, there may be multiple causes, which can have complex effects on CRP level. Although we used a bimodal curve pattern, false positive or false negative results are possible. Finally, it is possible that CRP variation is dependent on the duration of antibiotic treatment. It is thought that most patients classified as having an unknown cause may have had a false positive for the bimodal curve pattern. In addition, it is possible that patients excluded from analysis may have had several complications during the acute postoperative period.
The clinical importance of this study is that the most important cause of postoperative CRP elevation is periprosthetic infection, and immediate treatment after accurate diagnosis is necessary. Although there may be other causes for an elevated CRP, it is essential to perform the work-up for prosthetic joint infections. In addition, when making a diagnosis, it is important to bear in mind that various diseases may lead to CRP elevation. By doing this, we were able to find various causes other than postoperative infection and provide appropriate treatments according to the cause.
In addition, the authors cannot suggest an efficient and cost-effective algorithm to discern the cause in patients showing a bimodal pattern during the follow-up of CRP after TKA based on this study alone. However, considering the results of this study, in which the time of reincrease was faster in DVT cases than in cases of infection, it is believed that ultrasonography or lower extremity venography is required to confirm the diagnosis if DVT is suspected after physical examination (calf tenderness or edema) to distinguish DVT in addition to early infection in the early stage—within the first postoperative week. Moreover, we suggest that the postoperative CRP test may be conducted for the first 2 weeks of hospitalization once on postoperative 2 or 3 days for detecting the highest peak, once on postoperative 4 or 5 days for detecting normally decreasing pattern, and additionally, once or twice during postoperative 1-2 weeks for detecting abnormally re-elevating pattern. The CRP test according to this schedule may be also performed through the outpatient facility after the patients would be discharged before postoperative 2 weeks.
Therefore, we must keep in mind that during the early postoperative period after TKAs, CRP levels may be elevated by various causes besides postoperative infection because elderly patients with comorbid illness frequently have TKAs. Therefore, if CRP levels show abnormal patterns after TKAs, using a multimodal approach and determining an appropriate treatment after a sudden diagnosis in patients is critical to improving surgical outcomes.
It is important to know that elevated CRP level after TKA can have various causes. Although there may be other causes for elevated CRP, it is essential to perform the work-up for prosthetic joint infections. In addition, there seems to be a need to evaluate noninfectious causes and other site infection, in addition to periprosthetic infection.