- Email: [email protected]
Serum White Blood Cell Count and Differential Do Not Have a Role in the Diagnosis of Periprosthetic Joint Infection
The Journal of Arthroplasty Vol. 27 No. 8 Suppl. 1 2012
Serum White Blood Cell Count and Differential Do Not Have a Role in the Diagnosis of Periprosthetic Joint Infection Nader Toossi, MD, Bahar Adeli, BA, Mohammad R. Rasouli, MD, Ronald Huang, BS, and Javad Parvizi, MD, FRCS
Abstract: Serum white blood cell (WBC) count and neutrophil differential are frequently ordered during preoperative workup of suspected cases of periprosthetic joint infection (PJI). However, their roles in diagnosis of PJI have remained unclear despite previous studies. In this study, preoperative serum WBC and neutrophil percentages were retrieved from hospital charts. The diagnostic cutoff point determined by receiver operating characteristic curve analysis was 7800 cells/μL with 55% sensitivity and 66% specificity for WBC count, whereas the cutoff value for neutrophil percentage was 68% with 52% sensitivity and 75% specificity. Our study confirms the long-held belief that serum WBC count and differential has minimal role in routine workup of patients with suspected PJI. Keywords: periprosthetic joint infection, total joint arthroplasty, serum WBC count, serum neutrophil percentage. © 2012 Published by Elsevier Inc.
inflammatory markers such as C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), interleukin 6, procalcitonin, and tumor necrosis factor α may also play a role in diagnosis of PJI [5,8-11]. Elevation of the serum white blood cell (WBC) count (WCC) and neutrophil differential has been the hallmark for diagnosis of many infections [12]. Thus, it is reasonable to assume that the level of WBC and neutrophil differential in serum may also be elevated with PJI. Using the latter reasoning, clinicians have routinely ordered WBC and differential in patients suspected of PJI. A number of previous studies have cast doubt on the value of blood WCC and differential for diagnosis of PJI. A recent meta-analysis by Berberi et al [13] detected a pooled sensitivity of 45% and specificity of 87% for WCC in diagnosis of PJI. Part of the problem with interpretation of the literature relates to the fact that different definition of PJI was used by each study, which influences the accuracy of a diagnostic test. Furthermore, none of the studies evaluated the accuracy of WCC and differential in isolation. This study, using a single institutional database in a consecutive series of patients, was designed to address many of the shortcomings of the current literature to determine the role of serum WCC and differential in diagnosis of PJI. A standard definition for PJI was applied to all patients, and receiver operating characteristic (ROC) curves were used to determine the statistical parameters for the test.
Many decades have passed since Charnley [1] reported an infection rate of more than 9% after total hip arthroplasty. Because of the use of prophylactic antibiotics, clean air operating rooms, and a number of other strategies, the incidence of periprosthetic joint infection (PJI) has declined remarkably to be as low as 1% [2,3]. Despite this trend, PJI has remained one of the most challenging complications of total joint arthroplasty. Treatment of PJI is costly because of the subsequent need for reoperation, and the lengthy hospitalization often required eradicating the infection [4]. Presently, there is no universally accepted diagnostic test that is absolute or reliable for detection of PJI [5]. Hence, diagnosis of PJI is made based on clinical suspicion, serologic tests, imaging studies, and isolation of organisms from joint culture samples [3,5-7]. In recent years, studies have demonstrated that other
From the Department of Orthopedics, Rothman Institute of Orthopedics at Thomas Jefferson University Hospital, Philadelphia, Pennsylvania. Supplementary material available at www.arthroplastyjournal.org. Submitted August 9, 2011; accepted March 13, 2012 The Conflict of Interest statement associated with this article can be found at doi:10.1016/j.arth.2012.03.021. Reprint requests: Javad Parvizi, MD, FRCS, Department of Orthopedics, Rothman Institute of Orthopaedics, Thomas Jefferson University Hospitals, 925 Chestnut St, 2nd 26 Floor, Philadelphia, PA, 19107. © 2012 Published by Elsevier Inc. 0883-5403/2708-0012$36.00/0 doi:10.1016/j.arth.2012.03.021
51
52 The Journal of Arthroplasty Vol. 27 No. 8 Suppl. 1 September 2012
Methods and Material After obtaining approval of the institutional review board, the computerized and prospective database on all revision total joint arthroplasties was reviewed to identify all patients who underwent surgery for failure of either total knee arthroplasty or total hip arthroplasty. The registry consisted of 1856 revision surgeries performed on 1543 patients (817 females) at our institution between January 2003 and August 2010. The mean age of the patients at the time of surgery was 66 years with an SD of 12 years (range, 25-96 years). Using strict institutional criteria [14], 751 joints were found to have PJI. There were 463 knees and 288 hips in our septic revision group, and female patients comprised 49% of this group. In the aseptic group, there were 687 hips and 418 knees, and 54% of the patients were female. Patients were diagnosed to have PJI if they fulfilled one of the following criteria: 1. Positive culture: a More than 5 colonies on 1 plate b Light growth and greater in a single culture c Very light growth on 2 cultures if pathogen has the same resistance profile d 3 positive cultures in broth 2. Intraoperative purulence 3. Draining sinus tract or 3 of the following 4: a. ESR greater than 30 mm/h b. CRP greater than 10 mg/L c. Synovial WCC greater than 1760 cells/μL or 10 700 cells/μL acute postoperative (from synovial fluid) d. Synovial neutrophils percentages greater than 73% or 89% acute postoperative (from synovial fluid) The database was explored to extract all relevant data including the level of serum WCC and neutrophil differential. All patients had a record of their preoperative WCCs, but there were 572 patients for whom differential had not been ordered preoperatively, leaving 930 patients in whom both the serum WCC and neutrophil differential were available. The mean time interval from preoperative blood sampling to revision surgery was 12 days with an SD of 8 days. Forty-three percent of patients in aseptic group and 33% in septic group did not have any differentials. Statistical Analysis All statistical analyses were performed using Medcalc software version 11.6.1 (MedCalc Software, Mariakerke, Belgium). The mean and 95% confidence interval (95% CI) of WCCs and neutrophil percentages of the infected and noninfected cases were determined, and independent-samples Student t test was used to compare mean WCC and neutrophil percentage. P b .05 (2-sided) was considered to be statistically significant.
We constructed ROC curves to determine the area under the ROC curve (AUC) and cutoff value for WCC and neutrophil percentage. The AUC, which implies the diagnostic accuracy of the test, was calculated. An AUC of 1 demonstrates an ideal test with a 100% sensitivity and specificity, whereas an AUC of less than 0.5 indicates that the diagnostic test is not useful. The ROC curve constructed correlates the true-positive and falsepositive rates for a series of data points. Finding a point on the curve nearest to the top left corner of the diagram meant the most optimum for determining cutoff point. We determined cutoff points for WCC and neutrophil percentage as such. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the corresponding cutoff value were calculated. In addition, PPV and NPV were calculated for combination of WCC and neutrophil percentage results. For this purpose, a binary variable was defined. If both WCC and neutrophil percentage had predicted infection based on obtained cutoff points, the new variable was coded as “septic.” If both of them refuted infection, the variable value was coded as “aseptic.” Because of the high prevalence of infected joints, we calculated the Bayesian positive and negative predictive values for each test in addition to conventional predictive values using the following formulae [15]: Bayesian PPV = ðSensitivity × PrevalenceÞ =fðSensitivity × PrevalenceÞ þ½ð1−SpecificityÞ × ð1−PrevalenceÞg
Bayesian NPV = ½ðSpecificity × ð1−PrevalenceÞ =f½ð1−SensitivityÞ × Prevalence + ½Specificity × ð1−PrevalenceÞg
Results The prevalence of PJI in this cohort was found to be 40%. The mean of serum WCC in infected joints was 9236 cells/μL (95% CI, 8896-9575 cells/μL), whereas the mean for the noninfected group was 7331 cells/μL (95% CI, 7204-7458 cells/μL). There was statistically significant difference between 2 groups regarding mean WCC (P b .001). The mean of neutrophil percentage for the noninfected and infected joints were 63% (95% CI, 62.2%-63.6%) and 69% (95% CI, 67.5%-70%) respectively. There was statistical difference between the 2 means (P b .001). The AUC for WCC was 0.637 (95% CI, 0.614-0.659), and that for neutrophil percentage was 0.652 (95% CI, 0.623- 0.679). The cutoff points were found to be 7800 cells/μL for WCC and 69% for neutrophil percentage (Fig.; available online at www.arthroplastyjournal.org). The sensitivity and specificity for WCC were 55% (95%
Serum WCC and Differential in the Diagnosis of PJI Toossi et al
CI, 51.8%-59.0%) and 66% (95% CI, 63.3%-68.9%) respectively, whereas the corresponding values for neutrophil percentage were 53% (95% CI, 47.6%57.7%) and 75% (95% CI, 71.9%-78.2%) successively. Table (available online at www.arthroplastyjournal.org) demonstrates other calculated diagnostic characteristics of serum WBC and neutrophil percentage. Bayesian PPV was 32% for both WCC and neutrophil percentage. Bayesian NPV was 84% for WCC and 88% for neutrophil percentage. When both tests had values lower than or equal to obtained cutoff points (aseptic), the calculated NPV was 55% (95% CI, 50.68%58.82%). Positive predictive value was calculated for septic group, when both tests had values higher than obtained cutoff points, which was 60% (95% CI, 53.93%-66.30%).
Discussion There remains a dire need for a test that can accurately and predictably diagnose PJI. In the meantime, clinicians resort to a number of serum and synovial tests in evaluation of patients with suspected PJI. White blood cell count and neutrophil differential is one such test that has been in common use. There are conflicting reports regarding the diagnostic value of serum WCC and neutrophil percentage [16-19]. A recent systematic review by Berberi et al [13] evaluated the role of inflammatory blood laboratory levels as markers of PJI. In their study, among 4 assessed serologic markers including interleukin 6, CRP, ESR, and serum WCC, the latter test had the lowest diagnostic value as determined by odds ratio. There are other studies that have found no statistically significant difference between preoperative WCC in infected and aseptic revision arthroplasty cases [16,17]. In 1 study, the sensitivity of serum WCC in diagnosing infection in total hip arthroplasty was reported to be as low as 14%. On the other hand, some authors have concluded that serum WCC is a useful test, as its mean is significantly higher in patients with infection compared with aseptic failures [19]. Our study also detected a statistical significance in the mean value of serum WCC and neutrophil percentage between infected and noninfected cases. The latter, however, is irrelevant when assessing the diagnostic value of a test. Despite the difference in the mean value, there were many patients with PJI in whom the WCC and neutrophil percentage was within normal limits, both in our study and those prior [19]. One of the challenges in interpreting the data related to serum WCC and neutrophil differential relates to the cutoff point used by various studies. According to the study by Spangehl et al [9], a cutoff value of 11 000 cells/ μL for WBC provided a sensitivity of 20%, specificity of 96%, PPV of 54%, and NPV of 85% (79%-90%). In the same study, a neutrophil percentage more than 75% was considered to be suggestive of infection. Using
53
that threshold, neutrophil percentage had a sensitivity of 24%, specificity of 89%, PPV of 30%, and NPV of 85% [9]. To overcome the problem of choosing a random and arbitrary cutoff value, we used ROC curves to statistically determine the appropriate cutoff value for both WCC and neutrophil percentage. Using the values identified by ROC analyses, we demonstrated a higher sensitivity and lower specificity for serum WCC and neutrophil percentage. Another strength of our study relates to the use of a standard definition for PJI for the entire cohort. The latter overcomes the problem with current literature when different definition for PJI is used between various studies. Our study consists of a large consecutive cohort of patients with both aseptic failure and PJI, which makes statistical comparisons more relevant. Because of the relatively higher prevalence of PJI cases that were referred to our center for treatment, we used Bayesian predictive values that have been shown to be more accurate in such situation [15]. Our study has some limitations. Although details of medical comorbidities were collected, we cannot, with certainty, claim that the reason for elevation of serum WCC in some patients may not have occurred as a result of an underlying inflammatory condition or infection in another site. There were a fair number of patients (572/1543 or 37%) in our cohort in whom serum neutrophil percentage had not been ordered. Most of these patients were in the aseptic group in whom ordering differential of WCC would be considered unnecessary. Finally, the standard definition for PJI that is used at our institution may be criticized by some authorities, as it lacks data on frozen section. The latter is not done routinely at our institution. Although we can agree that the criteria chosen to determine PJI may not be acceptable to all, it carries the benefit of applying one standard to all patients and streamlining the data for analyses. Despite the aforementioned limitations, this study using appropriate statistical analyses in a relatively large cohort of patients demonstrates that serum WBC and neutrophil percentage has minimal to no role in diagnosis of PJI. We believe that this test should be abandoned for routine use and only used in patients with systemic infections.
References 1. Charnley J. A clean-air operating enclosure. Br J Surg 1964;51:202. 2. Fitzgerald Jr RH. Total hip arthroplasty sepsis. Prevention and diagnosis. Orthop Clin North Am 1992;23:259. 3. Peersman G, Laskin R, Davis J, et al. Infection in total knee replacement: a retrospective review of 6489 total knee replacements. Clin Orthop Relat Res 2001;15. 4. Persson U, Montgomery F, Carlsson A, et al. How far does prophylaxis against infection in total joint replacement offset its cost? Br Med J (Clin Res Ed) 1988;296:99.
54 The Journal of Arthroplasty Vol. 27 No. 8 Suppl. 1 September 2012 5. Bauer TW, Parvizi J, Kobayashi N, et al. Diagnosis of periprosthetic infection. J Bone Joint Surg Am 2006;88:869. 6. Toms AD, Davidson D, Masri BA, et al. The management of peri-prosthetic infection in total joint arthroplasty. J Bone Joint Surg Br 2006;88:149. 7. Pandey R, Berendt AR, Athanasou NA. Histological and microbiological findings in non-infected and infected revision arthroplasty tissues. The OSIRIS Collaborative Study Group. Oxford Skeletal Infection Research and Intervention Service. Arch Orthop Trauma Surg 2000;120:570. 8. Di Cesare PE, Chang E, Preston CF, et al. Serum interleukin-6 as a marker of periprosthetic infection following total hip and knee arthroplasty. J Bone Joint Surg Am 2005;87:1921. 9. Spangehl MJ, Masri BA, O'Connell JX, et al. Prospective analysis of preoperative and intraoperative investigations for the diagnosis of infection at the sites of two hundred and two revision total hip arthroplasties. J Bone Joint Surg Am 1999;81:672. 10. Della Valle CJ, Sporer SM, Jacobs JJ, et al. Preoperative testing for sepsis before revision total knee arthroplasty. J Arthroplasty 2007;22:90. 11. Bottner F, Wegner A, Winkelmann W, et al. Interleukin-6, procalcitonin and TNF-alpha: markers of peri-prosthetic infection following total joint replacement. J Bone Joint Surg Br 2007;89:94. 12. Dale DC. A new look at an old laboratory test: the WBC count. J Gen Intern Med 1991;6:264.
13. Berbari E, Mabry T, Tsaras G, et al. Inflammatory blood laboratory levels as markers of prosthetic joint infection: a systematic review and meta-analysis. J Bone Joint Surg Am 2010;92:2102. 14. Ghanem E, Parvizi J, Burnett RSJ, et al. Cell count and differential of aspirated fluid in the diagnosis of infection at the site of total knee arthroplasty. J Bone Joint Surg Am 2008;90:1637. 15. Streiner DLNG. PDQ epidemiology. 2nd ed. London: B.C. Decker; 1996. 16. Nilsdotter-Augustinsson A, Briheim G, Herder A, et al. Inflammatory response in 85 patients with loosened hip prostheses: a prospective study comparing inflammatory markers in patients with aseptic and septic prosthetic loosening. Acta Orthop 2007;78:629. 17. Savarino L, Tigani D, Baldini N, et al. Pre-operative diagnosis of infection in total knee arthroplasty: an algorithm. Knee Surg Sports Traumatol Arthrosc 2009; 17:667. 18. Muller M, Morawietz L, Hasart O, et al. Diagnosis of periprosthetic infection following total hip arthroplasty— evaluation of the diagnostic values of pre- and intraoperative parameters and the associated strategy to preoperatively select patients with a high probability of joint infection. J Orthop Surg Res 2008;3:31. 19. Virolainen P, Lahteenmaki H, Hiltunen A, et al. The reliability of diagnosis of infection during revision arthroplasties. Scand J Surg 2002;91:178.
Serum WCC and Differential in the Diagnosis of PJI Toossi et al
54.e1
WBC count 100
Sensitivity
80 60 40 20 0
A 0
20
40 60 80 100-Specificity
100
PMN% 100
Sensitivity
80 60 40 20 0
B 0
20
40 60 80 100-Specificity
100
Fig. Receiver operating characteristic curves for serum WCC (A) and neutrophil percentage (B), respectively. The abscissa (100-specificity) represents the false-positive rate, and the ordinate (sensitivity), the true-positive rate.
Table. Diagnostic Test Characteristics for Serum WCC and Neutrophil Percentage Sensitivity (95% CI)
Specificity (95% CI)
PPV (95% CI)
NPV (95% CI)
Serum WCC 55.39% (51.76%-58.99%) 66.15% (63.3%-68.9%) 52.59% (49.04%-56.12%) 68.54 (65.66%-71.33%) Serum neutrophil percentage 52.67% (47.6%-57.7%) 75.17% (71.91%-78.22% 52.55% (47.48%-57.59% 75.07% (71.81%-78.13%
Serum White Blood Cell Count and Differential Do Not Have a Role in the Diagnosis of Periprosthetic Joint Infection Nader Toossi, MD, Bahar Adeli, BA, Mohammad R. Rasouli, MD, Ronald Huang, BS, and Javad Parvizi, MD, FRCS
Abstract: Serum white blood cell (WBC) count and neutrophil differential are frequently ordered during preoperative workup of suspected cases of periprosthetic joint infection (PJI). However, their roles in diagnosis of PJI have remained unclear despite previous studies. In this study, preoperative serum WBC and neutrophil percentages were retrieved from hospital charts. The diagnostic cutoff point determined by receiver operating characteristic curve analysis was 7800 cells/μL with 55% sensitivity and 66% specificity for WBC count, whereas the cutoff value for neutrophil percentage was 68% with 52% sensitivity and 75% specificity. Our study confirms the long-held belief that serum WBC count and differential has minimal role in routine workup of patients with suspected PJI. Keywords: periprosthetic joint infection, total joint arthroplasty, serum WBC count, serum neutrophil percentage. © 2012 Published by Elsevier Inc.
inflammatory markers such as C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), interleukin 6, procalcitonin, and tumor necrosis factor α may also play a role in diagnosis of PJI [5,8-11]. Elevation of the serum white blood cell (WBC) count (WCC) and neutrophil differential has been the hallmark for diagnosis of many infections [12]. Thus, it is reasonable to assume that the level of WBC and neutrophil differential in serum may also be elevated with PJI. Using the latter reasoning, clinicians have routinely ordered WBC and differential in patients suspected of PJI. A number of previous studies have cast doubt on the value of blood WCC and differential for diagnosis of PJI. A recent meta-analysis by Berberi et al [13] detected a pooled sensitivity of 45% and specificity of 87% for WCC in diagnosis of PJI. Part of the problem with interpretation of the literature relates to the fact that different definition of PJI was used by each study, which influences the accuracy of a diagnostic test. Furthermore, none of the studies evaluated the accuracy of WCC and differential in isolation. This study, using a single institutional database in a consecutive series of patients, was designed to address many of the shortcomings of the current literature to determine the role of serum WCC and differential in diagnosis of PJI. A standard definition for PJI was applied to all patients, and receiver operating characteristic (ROC) curves were used to determine the statistical parameters for the test.
Many decades have passed since Charnley [1] reported an infection rate of more than 9% after total hip arthroplasty. Because of the use of prophylactic antibiotics, clean air operating rooms, and a number of other strategies, the incidence of periprosthetic joint infection (PJI) has declined remarkably to be as low as 1% [2,3]. Despite this trend, PJI has remained one of the most challenging complications of total joint arthroplasty. Treatment of PJI is costly because of the subsequent need for reoperation, and the lengthy hospitalization often required eradicating the infection [4]. Presently, there is no universally accepted diagnostic test that is absolute or reliable for detection of PJI [5]. Hence, diagnosis of PJI is made based on clinical suspicion, serologic tests, imaging studies, and isolation of organisms from joint culture samples [3,5-7]. In recent years, studies have demonstrated that other
From the Department of Orthopedics, Rothman Institute of Orthopedics at Thomas Jefferson University Hospital, Philadelphia, Pennsylvania. Supplementary material available at www.arthroplastyjournal.org. Submitted August 9, 2011; accepted March 13, 2012 The Conflict of Interest statement associated with this article can be found at doi:10.1016/j.arth.2012.03.021. Reprint requests: Javad Parvizi, MD, FRCS, Department of Orthopedics, Rothman Institute of Orthopaedics, Thomas Jefferson University Hospitals, 925 Chestnut St, 2nd 26 Floor, Philadelphia, PA, 19107. © 2012 Published by Elsevier Inc. 0883-5403/2708-0012$36.00/0 doi:10.1016/j.arth.2012.03.021
51
52 The Journal of Arthroplasty Vol. 27 No. 8 Suppl. 1 September 2012
Methods and Material After obtaining approval of the institutional review board, the computerized and prospective database on all revision total joint arthroplasties was reviewed to identify all patients who underwent surgery for failure of either total knee arthroplasty or total hip arthroplasty. The registry consisted of 1856 revision surgeries performed on 1543 patients (817 females) at our institution between January 2003 and August 2010. The mean age of the patients at the time of surgery was 66 years with an SD of 12 years (range, 25-96 years). Using strict institutional criteria [14], 751 joints were found to have PJI. There were 463 knees and 288 hips in our septic revision group, and female patients comprised 49% of this group. In the aseptic group, there were 687 hips and 418 knees, and 54% of the patients were female. Patients were diagnosed to have PJI if they fulfilled one of the following criteria: 1. Positive culture: a More than 5 colonies on 1 plate b Light growth and greater in a single culture c Very light growth on 2 cultures if pathogen has the same resistance profile d 3 positive cultures in broth 2. Intraoperative purulence 3. Draining sinus tract or 3 of the following 4: a. ESR greater than 30 mm/h b. CRP greater than 10 mg/L c. Synovial WCC greater than 1760 cells/μL or 10 700 cells/μL acute postoperative (from synovial fluid) d. Synovial neutrophils percentages greater than 73% or 89% acute postoperative (from synovial fluid) The database was explored to extract all relevant data including the level of serum WCC and neutrophil differential. All patients had a record of their preoperative WCCs, but there were 572 patients for whom differential had not been ordered preoperatively, leaving 930 patients in whom both the serum WCC and neutrophil differential were available. The mean time interval from preoperative blood sampling to revision surgery was 12 days with an SD of 8 days. Forty-three percent of patients in aseptic group and 33% in septic group did not have any differentials. Statistical Analysis All statistical analyses were performed using Medcalc software version 11.6.1 (MedCalc Software, Mariakerke, Belgium). The mean and 95% confidence interval (95% CI) of WCCs and neutrophil percentages of the infected and noninfected cases were determined, and independent-samples Student t test was used to compare mean WCC and neutrophil percentage. P b .05 (2-sided) was considered to be statistically significant.
We constructed ROC curves to determine the area under the ROC curve (AUC) and cutoff value for WCC and neutrophil percentage. The AUC, which implies the diagnostic accuracy of the test, was calculated. An AUC of 1 demonstrates an ideal test with a 100% sensitivity and specificity, whereas an AUC of less than 0.5 indicates that the diagnostic test is not useful. The ROC curve constructed correlates the true-positive and falsepositive rates for a series of data points. Finding a point on the curve nearest to the top left corner of the diagram meant the most optimum for determining cutoff point. We determined cutoff points for WCC and neutrophil percentage as such. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for the corresponding cutoff value were calculated. In addition, PPV and NPV were calculated for combination of WCC and neutrophil percentage results. For this purpose, a binary variable was defined. If both WCC and neutrophil percentage had predicted infection based on obtained cutoff points, the new variable was coded as “septic.” If both of them refuted infection, the variable value was coded as “aseptic.” Because of the high prevalence of infected joints, we calculated the Bayesian positive and negative predictive values for each test in addition to conventional predictive values using the following formulae [15]: Bayesian PPV = ðSensitivity × PrevalenceÞ =fðSensitivity × PrevalenceÞ þ½ð1−SpecificityÞ × ð1−PrevalenceÞg
Bayesian NPV = ½ðSpecificity × ð1−PrevalenceÞ =f½ð1−SensitivityÞ × Prevalence + ½Specificity × ð1−PrevalenceÞg
Results The prevalence of PJI in this cohort was found to be 40%. The mean of serum WCC in infected joints was 9236 cells/μL (95% CI, 8896-9575 cells/μL), whereas the mean for the noninfected group was 7331 cells/μL (95% CI, 7204-7458 cells/μL). There was statistically significant difference between 2 groups regarding mean WCC (P b .001). The mean of neutrophil percentage for the noninfected and infected joints were 63% (95% CI, 62.2%-63.6%) and 69% (95% CI, 67.5%-70%) respectively. There was statistical difference between the 2 means (P b .001). The AUC for WCC was 0.637 (95% CI, 0.614-0.659), and that for neutrophil percentage was 0.652 (95% CI, 0.623- 0.679). The cutoff points were found to be 7800 cells/μL for WCC and 69% for neutrophil percentage (Fig.; available online at www.arthroplastyjournal.org). The sensitivity and specificity for WCC were 55% (95%
Serum WCC and Differential in the Diagnosis of PJI Toossi et al
CI, 51.8%-59.0%) and 66% (95% CI, 63.3%-68.9%) respectively, whereas the corresponding values for neutrophil percentage were 53% (95% CI, 47.6%57.7%) and 75% (95% CI, 71.9%-78.2%) successively. Table (available online at www.arthroplastyjournal.org) demonstrates other calculated diagnostic characteristics of serum WBC and neutrophil percentage. Bayesian PPV was 32% for both WCC and neutrophil percentage. Bayesian NPV was 84% for WCC and 88% for neutrophil percentage. When both tests had values lower than or equal to obtained cutoff points (aseptic), the calculated NPV was 55% (95% CI, 50.68%58.82%). Positive predictive value was calculated for septic group, when both tests had values higher than obtained cutoff points, which was 60% (95% CI, 53.93%-66.30%).
Discussion There remains a dire need for a test that can accurately and predictably diagnose PJI. In the meantime, clinicians resort to a number of serum and synovial tests in evaluation of patients with suspected PJI. White blood cell count and neutrophil differential is one such test that has been in common use. There are conflicting reports regarding the diagnostic value of serum WCC and neutrophil percentage [16-19]. A recent systematic review by Berberi et al [13] evaluated the role of inflammatory blood laboratory levels as markers of PJI. In their study, among 4 assessed serologic markers including interleukin 6, CRP, ESR, and serum WCC, the latter test had the lowest diagnostic value as determined by odds ratio. There are other studies that have found no statistically significant difference between preoperative WCC in infected and aseptic revision arthroplasty cases [16,17]. In 1 study, the sensitivity of serum WCC in diagnosing infection in total hip arthroplasty was reported to be as low as 14%. On the other hand, some authors have concluded that serum WCC is a useful test, as its mean is significantly higher in patients with infection compared with aseptic failures [19]. Our study also detected a statistical significance in the mean value of serum WCC and neutrophil percentage between infected and noninfected cases. The latter, however, is irrelevant when assessing the diagnostic value of a test. Despite the difference in the mean value, there were many patients with PJI in whom the WCC and neutrophil percentage was within normal limits, both in our study and those prior [19]. One of the challenges in interpreting the data related to serum WCC and neutrophil differential relates to the cutoff point used by various studies. According to the study by Spangehl et al [9], a cutoff value of 11 000 cells/ μL for WBC provided a sensitivity of 20%, specificity of 96%, PPV of 54%, and NPV of 85% (79%-90%). In the same study, a neutrophil percentage more than 75% was considered to be suggestive of infection. Using
53
that threshold, neutrophil percentage had a sensitivity of 24%, specificity of 89%, PPV of 30%, and NPV of 85% [9]. To overcome the problem of choosing a random and arbitrary cutoff value, we used ROC curves to statistically determine the appropriate cutoff value for both WCC and neutrophil percentage. Using the values identified by ROC analyses, we demonstrated a higher sensitivity and lower specificity for serum WCC and neutrophil percentage. Another strength of our study relates to the use of a standard definition for PJI for the entire cohort. The latter overcomes the problem with current literature when different definition for PJI is used between various studies. Our study consists of a large consecutive cohort of patients with both aseptic failure and PJI, which makes statistical comparisons more relevant. Because of the relatively higher prevalence of PJI cases that were referred to our center for treatment, we used Bayesian predictive values that have been shown to be more accurate in such situation [15]. Our study has some limitations. Although details of medical comorbidities were collected, we cannot, with certainty, claim that the reason for elevation of serum WCC in some patients may not have occurred as a result of an underlying inflammatory condition or infection in another site. There were a fair number of patients (572/1543 or 37%) in our cohort in whom serum neutrophil percentage had not been ordered. Most of these patients were in the aseptic group in whom ordering differential of WCC would be considered unnecessary. Finally, the standard definition for PJI that is used at our institution may be criticized by some authorities, as it lacks data on frozen section. The latter is not done routinely at our institution. Although we can agree that the criteria chosen to determine PJI may not be acceptable to all, it carries the benefit of applying one standard to all patients and streamlining the data for analyses. Despite the aforementioned limitations, this study using appropriate statistical analyses in a relatively large cohort of patients demonstrates that serum WBC and neutrophil percentage has minimal to no role in diagnosis of PJI. We believe that this test should be abandoned for routine use and only used in patients with systemic infections.
References 1. Charnley J. A clean-air operating enclosure. Br J Surg 1964;51:202. 2. Fitzgerald Jr RH. Total hip arthroplasty sepsis. Prevention and diagnosis. Orthop Clin North Am 1992;23:259. 3. Peersman G, Laskin R, Davis J, et al. Infection in total knee replacement: a retrospective review of 6489 total knee replacements. Clin Orthop Relat Res 2001;15. 4. Persson U, Montgomery F, Carlsson A, et al. How far does prophylaxis against infection in total joint replacement offset its cost? Br Med J (Clin Res Ed) 1988;296:99.
54 The Journal of Arthroplasty Vol. 27 No. 8 Suppl. 1 September 2012 5. Bauer TW, Parvizi J, Kobayashi N, et al. Diagnosis of periprosthetic infection. J Bone Joint Surg Am 2006;88:869. 6. Toms AD, Davidson D, Masri BA, et al. The management of peri-prosthetic infection in total joint arthroplasty. J Bone Joint Surg Br 2006;88:149. 7. Pandey R, Berendt AR, Athanasou NA. Histological and microbiological findings in non-infected and infected revision arthroplasty tissues. The OSIRIS Collaborative Study Group. Oxford Skeletal Infection Research and Intervention Service. Arch Orthop Trauma Surg 2000;120:570. 8. Di Cesare PE, Chang E, Preston CF, et al. Serum interleukin-6 as a marker of periprosthetic infection following total hip and knee arthroplasty. J Bone Joint Surg Am 2005;87:1921. 9. Spangehl MJ, Masri BA, O'Connell JX, et al. Prospective analysis of preoperative and intraoperative investigations for the diagnosis of infection at the sites of two hundred and two revision total hip arthroplasties. J Bone Joint Surg Am 1999;81:672. 10. Della Valle CJ, Sporer SM, Jacobs JJ, et al. Preoperative testing for sepsis before revision total knee arthroplasty. J Arthroplasty 2007;22:90. 11. Bottner F, Wegner A, Winkelmann W, et al. Interleukin-6, procalcitonin and TNF-alpha: markers of peri-prosthetic infection following total joint replacement. J Bone Joint Surg Br 2007;89:94. 12. Dale DC. A new look at an old laboratory test: the WBC count. J Gen Intern Med 1991;6:264.
13. Berbari E, Mabry T, Tsaras G, et al. Inflammatory blood laboratory levels as markers of prosthetic joint infection: a systematic review and meta-analysis. J Bone Joint Surg Am 2010;92:2102. 14. Ghanem E, Parvizi J, Burnett RSJ, et al. Cell count and differential of aspirated fluid in the diagnosis of infection at the site of total knee arthroplasty. J Bone Joint Surg Am 2008;90:1637. 15. Streiner DLNG. PDQ epidemiology. 2nd ed. London: B.C. Decker; 1996. 16. Nilsdotter-Augustinsson A, Briheim G, Herder A, et al. Inflammatory response in 85 patients with loosened hip prostheses: a prospective study comparing inflammatory markers in patients with aseptic and septic prosthetic loosening. Acta Orthop 2007;78:629. 17. Savarino L, Tigani D, Baldini N, et al. Pre-operative diagnosis of infection in total knee arthroplasty: an algorithm. Knee Surg Sports Traumatol Arthrosc 2009; 17:667. 18. Muller M, Morawietz L, Hasart O, et al. Diagnosis of periprosthetic infection following total hip arthroplasty— evaluation of the diagnostic values of pre- and intraoperative parameters and the associated strategy to preoperatively select patients with a high probability of joint infection. J Orthop Surg Res 2008;3:31. 19. Virolainen P, Lahteenmaki H, Hiltunen A, et al. The reliability of diagnosis of infection during revision arthroplasties. Scand J Surg 2002;91:178.
Serum WCC and Differential in the Diagnosis of PJI Toossi et al
54.e1
WBC count 100
Sensitivity
80 60 40 20 0
A 0
20
40 60 80 100-Specificity
100
PMN% 100
Sensitivity
80 60 40 20 0
B 0
20
40 60 80 100-Specificity
100
Fig. Receiver operating characteristic curves for serum WCC (A) and neutrophil percentage (B), respectively. The abscissa (100-specificity) represents the false-positive rate, and the ordinate (sensitivity), the true-positive rate.
Table. Diagnostic Test Characteristics for Serum WCC and Neutrophil Percentage Sensitivity (95% CI)
Specificity (95% CI)
PPV (95% CI)
NPV (95% CI)
Serum WCC 55.39% (51.76%-58.99%) 66.15% (63.3%-68.9%) 52.59% (49.04%-56.12%) 68.54 (65.66%-71.33%) Serum neutrophil percentage 52.67% (47.6%-57.7%) 75.17% (71.91%-78.22% 52.55% (47.48%-57.59% 75.07% (71.81%-78.13%