The Journal of Bone & Joint Surgery

The Journal of Bone & Joint Surgery, Volume 89, Issue 7

Scientific Articles | July 01, 2007

Use of Erythrocyte Sedimentation Rate and C-Reactive Protein Level to Diagnose Infection Before Revision Total Knee Arthroplasty: A Prospective Evaluation

Nelson V. Greidanus, MD, MPH, FRCSC¹; Bassam A. Masri, MD, FRCSC¹; Donald S. Garbuz, MD, MHSc, FRCSC¹; S. Darrin Wilson, MBBCh, MD, FRCS¹; M. Gavan McAlinden, MBBCh, MPH, MD, FRCS¹; Min Xu, MSc¹; Clive P. Duncan, MD, MSc, FRCSC¹

¹ Department of Orthopaedics, University of British Columbia, Laurel Pavilion, 910 West 10th Avenue, Vancouver, BC V5Z 4E3, Canada. E-mail address for N.V. Greidanus: nelson.greidanus@vch.ca

View Disclosures and Other Information

Disclosure: The authors did not receive any outside funding or grants in support of their research for or preparation of this work. Neither they nor a member of their immediate families received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. A commercial entity (Zimmer, Warsaw, Indiana) paid or directed in any one year, or agreed to pay or direct, benefits in excess of $10,000 to a research fund, foundation, division, center, clinical practice, or other charitable or nonprofit organization with which one or more of the authors, or a member of his or her immediate family, is affiliated or associated.

A commentary is available with the electronic versions of this article, on our web site () and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM).

Investigation performed at the Division of Lower Limb Reconstruction and Oncology, Department of Orthopaedics, University of British Columbia, Vancouver, British Columbia, Canada

The Journal of Bone and Joint Surgery, Incorporated

J Bone Joint Surg Am, 2007 Jul 01;89(7):1409-1416. doi: 10.2106/JBJS.D.02602

5 Recommendations (Recommend) | 3 Comments | Saved by 3 Users Save Case

Article

Comments (2)

text A A A

Abstract

Background: Despite the widespread use of several diagnostic tests, there is still no perfect test for the diagnosis of infection at the site of a total knee arthroplasty. The purpose of this study was to evaluate the diagnostic test characteristics of the erythrocyte sedimentation rate and C-reactive protein level for the assessment of infection in patients presenting for revision total knee arthroplasty.

Methods: One hundred and fifty-one knees in 145 patients presenting for revision total knee arthroplasty were evaluated prospectively for the presence of infection with measurement of the erythrocyte sedimentation rate and the C-reactive protein level. The characteristics of these tests were assessed with use of two different techniques: first, receiver-operating-characteristic curve analysis was performed to determine the optimal positivity criterion for the diagnostic test, and, second, previously accepted criteria for establishing positivity of the tests were used.

Results: A diagnosis of infection was established for forty-five of the 151 knees that underwent revision total knee arthroplasty. The receiver-operating-characteristic curves indicated that the optimal positivity criterion was 22.5 mm/hr for the erythrocyte sedimentation rate and 13.5 mg/L for the C-reactive protein level. Both the erythrocyte sedimentation rate (sensitivity, 0.93; specificity, 0.83; positive likelihood ratio, 5.81; accuracy, 0.86) and the C-reactive protein level (sensitivity, 0.91; specificity, 0.86; positive likelihood ratio, 6.89; accuracy, 0.88) have excellent diagnostic test performance.

Conclusions: The erythrocyte sedimentation rate and the C-reactive protein level provide excellent diagnostic test information for establishing the presence or absence of infection prior to surgical intervention in patients with pain at the site of a knee arthroplasty.

Level of Evidence: Diagnostic Level I. See Instructions to Authors for a complete description of levels of evidence.

Figures in this Article

Infection, a potentially devastating and expensive complication^1,2, has been shown to develop after up to 2% of uncomplicated knee arthroplasties (sixty-one of 3930^3,4). While patients with infection at the site of a knee replacement may present with systemic symptoms and signs compatible with sepsis, many present with features that are indistinguishable from those of failure for aseptic reasons. The misdiagnosis of infection as aseptic loosening, arthrofibrosis, or unexplained pain can easily lead to a delay in appropriate treatment or to inappropriate surgery. Therefore, when evaluating pain at the site of a knee arthroplasty, and prior to embarking on revision knee arthroplasty, the surgeon must distinguish knees with infection from those without infection. This presents a diagnostic challenge, as none of the tests commonly used to diagnose infection appear to be 100% sensitive or specific when applied to patients who have had an arthroplasty⁵. Furthermore, there is a paucity of published data on preoperative diagnostic test characteristics prior to revision total knee arthroplasty⁶. Previous studies have often been performed in a retrospective manner, with consequent difficulty in the control of variables such as antibiotic therapy at the time of specimen collection and systemic inflammatory conditions that may cause elevation of the inflammatory markers being measured⁵. Additionally, some authors have drawn conclusions regarding test utility on the basis of small numbers of patients in whom infection was the cause of the failure^7,8. The purpose of our study was to prospectively evaluate the diagnostic test characteristics of the erythrocyte sedimentation rate and C-reactive protein level performed to establish the diagnosis of infection prior to revision knee surgery.

Materials and Methods

Materials and Methods | Results | Discussion | Appendix | References

We performed a prospective analysis of diagnostic tests used to detect infection prior to revision knee replacement. The original cohort consisted of 201 consecutive patients treated with a total of 207 revision knee replacements at a single tertiary referral center from 1997 to 2001. The mean age of the patients was sixty-seven years (range, forty-one to ninety years) at the time of the revision operation. Twenty-six patients (twenty-six knees) were excluded from the study because they had a known systemic disease or a condition that could result in an abnormal erythrocyte sedimentation rate and/or C-reactive protein level, such as rheumatoid arthritis or other inflammatory arthritides. Nineteen patients (nineteen knees) who presented for a second-stage revision arthroplasty following insertion of an antibiotic spacer for treatment of infection were also excluded. Eleven patients (eleven knees) who were already receiving antibiotics at the time of clinical assessment, and before a definitive diagnosis of infection was made at our institution, were excluded from our primary data set⁹. Therefore, the final cohort for analysis in the primary data set consisted of 151 knees in 145 patients. There were seventy-four women and seventy-one men. The primary diagnosis was osteoarthritis in all patients.

The eleven patients (eleven knees) who had been excluded from the primary data set because they were already receiving antibiotics prior to a definitive diagnosis of infection were included in a secondary data set, together with the patients from the primary data set. This group (156 patients [162 knees]) was analyzed to determine if antibiotic use at the time of the measurements of the erythrocyte sedimentation rate and C-reactive protein level would alter the performance of those tests.

All subjects gave informed consent to participate in the study, and the protocol was approved by the research ethics boards of our hospital and university. All consenting patients underwent testing consisting of a measurement of the preoperative erythrocyte sedimentation rate, a measurement of the C-reactive protein level, knee aspiration and culture of the aspirate, and culture of intraoperative specimens obtained at the time of the revision arthroplasty.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 1: Receiver-operating-characteristic (ROC) curve, which consists of data points with connecting ties as the diagonal segments to create a continuous curve, for patients not receiving antibiotics. Note that the upper left corner of the curve designates the optimal cut-point for the erythrocyte sedimentation rate (ESR) (22.5 mm/hr) and the C-reactive protein level (CRP) (13.5 mg/L).

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 2: Receiver-operating-characteristic (ROC) curve, which consists of data points with connecting ties as the diagonal segments to create a continuous curve, for all patients, including the eleven taking antibiotics at the time of the measurements of the erythrocyte sedimentation rate (ESR) and C-reactive protein level (CRP).

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 3: Receiver-operating-characteristic (ROC) curve, which consists of data points with connecting ties as the diagonal segments to create a continuous curve, demonstrating erythrocyte sedimentation rate (ESR) cut-points of 22.5 and 30 mm/hr.

Diagnostic Tests

Definition of Infection

A knee was considered to be infected if two or all three of the samples of either the aspirate or the tissue were positive for growth on bacterial culture at any time in a period of up to five days.

Erythrocyte Sedimentation Rate and C-Reactive Protein Level

The serological tests were evaluated with use of two separate methods: (1) developing new receiver-operating-characteristic curves and cut-points for the optimal performance of these tests, and (2) using predetermined cut-points obtained from the published literature.

The receiver-operating-characteristic curves for the erythrocyte sedimentation rate and C-reactive protein level were established with use of SPSS software (version 10.0; SPSS, Chicago, Illinois) (Figs. 1, 2, 3, 4). The receiver-operating-characteristic curve (plot of sensitivity [y axis] versus 1 — specificity [x axis]) illustrates diagnostic test performance and determines the optimal positivity criterion (cut-point) for the definition of a positive test¹⁰. The graph that is constructed correlates true-positive and false-positive rates for a series of cut-points for any test. As the criteria for a positive test are made more stringent, the cut-point on the curve moves to the left and down. Conversely, as the criteria for a positive test are made more liberal, the cut-point on the curve moves to the right and up (greater sensitivity, lower specificity). The optimal cut-point (threshold for positivity) for a diagnostic test is often chosen as that in the upper left corner of the receiver-operating-characteristic curve, which represents the point at which the greatest sensitivity and specificity are obtained¹¹.

For the analysis of the operating test with use of predetermined abnormal levels, we assumed that the cut-off values for a diagnosis of infection after total knee arthroplasty were =30 mm/hr for the erythrocyte sedimentation rate and =10 mg/L for the C-reactive protein level. These levels are based on previously published criteria for the evaluation of infection at the site of a total hip arthroplasty^1,12,13.

Aspiration of the Knee Joint

The aspiration was performed, with use of aseptic technique, by the introduction of an 18-gauge needle into the knee joint anterolaterally at the level of the joint space. The joint fluid that was obtained was divided into three aliquots and then processed, and the results were reported separately for the three specimens. This protocol was designed to minimize the risk of laboratory contamination and the occurrence of false-positive results as well as to reduce the risk of false-negative results. An aspiration specimen was considered positive for infection if there was growth on two or all three of the cultures, including late bacterial growth on subculture¹, at any time up to five days after incubation. The aspiration was repeated when the findings were inconsistent, in terms of demonstrating either different organisms or different sensitivity profiles, among the various cultures.

View Large | Download Slide(.ppt)
Add to My JBJS

Fig. 4: Receiver-operating-characteristic (ROC) curve, which consists of data points with connecting ties as the diagonal segments to create a continuous curve, demonstrating C-reactive protein level (CRP) cut-points of 10 and 13.5 mg/L.

Cultures of Intraoperative Specimens

Intraoperative antibiotics were withheld until knee tissues were obtained. A patient was considered to have an infection if there was growth, including late bacterial growth on subculture¹, on two or all three of the cultures of the intraoperative specimens after five days of incubation.

Data Analysis

Each test result was defined as true-positive, false-positive, true-negative, or false-negative in relation to the final diagnosis of infection, and a summary 2 × 2 table was created for each test (Table I). Sensitivity, specificity, positive and negative predictive values, accuracy, and positive and negative likelihood ratios were determined for each test (Table II and Appendix). The accuracy of the diagnostic test is equivalent to the area under the curve of the respective receiver-operating-characteristic curve^11,14. Microsoft Excel (Microsoft, Redmond, Washington) and SPSS software (version 10.0; SPSS) were used to process all data.

View Larger

TABLE I Summary 2 × 2 Table

	Infection
Test	Positive	Negative
Positive	a	b
Negative	c	d

Add to My JBJS

TABLE I Summary 2 × 2 Table

	Infection
Test	Positive	Negative
Positive	a	b
Negative	c	d

View Larger

TABLE II Diagnostic Test Characteristics*

Sensitivity	a/(a + c)
Specificity	d/(b + d)
Pos. likelihood ratio	a/(a + c)/[b/(b + d)]
Neg. likelihood ratio	c/(a + c)/[d/(b + d)]
Probability of infection	(a + c)/(a + b + c + d)
Pos. predictive value	a/(a + b)
Neg. predictive value	d/(c + d)
P (test negative)	(c + d)/(a + b + c + d)
Overall accuracy	(a + d)/(a + b + c + d)

See Table I for definitions of a, b, c, and d.

Add to My JBJS

TABLE II Diagnostic Test Characteristics*

Sensitivity	a/(a + c)
Specificity	d/(b + d)
Pos. likelihood ratio	a/(a + c)/[b/(b + d)]
Neg. likelihood ratio	c/(a + c)/[d/(b + d)]
Probability of infection	(a + c)/(a + b + c + d)
Pos. predictive value	a/(a + b)
Neg. predictive value	d/(c + d)
P (test negative)	(c + d)/(a + b + c + d)
Overall accuracy	(a + d)/(a + b + c + d)

See Table I for definitions of a, b, c, and d.

Results

Materials and Methods | Results | Discussion | Appendix | References

An infection was diagnosed in forty-five (29.8%) of the 151 knees. This relatively high prevalence was due to the fact that our institution functions as a major referral center for all patients with an infection at the site of a prosthetic joint in a catchment area of four million people.

The infecting organisms in the forty-five knees are listed in the Appendix. The most common organism was Staphylococcus (thirty-seven knees), which was sensitive to methicillin and oxacillin in the majority of cases (twenty-six of thirty-seven knees). More than one organism grew on culture of the specimens from three knees.

The receiver-operating-characteristic curve analysis demonstrated that the erythrocyte sedimentation rate and the C-reactive protein level are both excellent diagnostic tests (Figs. 1, 2, 3, 4). If a test has good diagnostic characteristics its apex should appear in the upper left corner of the plot of sensitivity versus (1 — specificity). This is the case for both the erythrocyte-sedimentation-rate test and the C-reactive-protein test. A useless diagnostic test would be illustrated by a straight line at 45° to the axes of sensitivity and (1 — specificity).

Our receiver-operating-characteristic curve analysis suggests that the optimal cut-points for the erythrocyte sedimentation rate and C-reactive protein level are 22.5 mm/hr and 13.5 mg/L, respectively. This means that any erythrocyte sedimentation rate of =22.5 mm/hr should be considered positive for infection. Likewise, a C-reactive protein level of =13.5 mg/L should be considered positive. These cut-points correspond to the numerical threshold at which both the sensitivity and the specificity of the test are optimized given the distribution of the test results in the population under study (maximum test accuracy). Application of these criteria results in excellent performance of the erythrocyte-sedimentation-rate and C-reactive-protein tests to rule out infection (Table III). The erythrocyte sedimentation rate had a sensitivity of 0.93, a specificity of 0.83, a positive likelihood ratio of 5.81, a negative predictive value of 0.96, and an overall accuracy of 0.86. The C-reactive protein level had a sensitivity of 0.91, a specificity of 0.86, a positive likelihood ratio of 6.89, a negative predictive value of 0.95, and an overall accuracy of 0.88. In our cohort, the pretest probability of infection can be considered to be the prevalence of infection in patients with symptoms presenting for revision total knee arthroplasty, which was 0.298. If a patient is tested with measurement of the erythrocyte sedimentation rate alone and the result is positive (above the cut-point), then the corresponding posttest probability of infection is the positive predictive value (Table III), which was 0.71. If the result is negative (below the cut-point), then the posttest probability of infection is 1 minus the negative predictive value (1.0 — 0.96), which was 0.04. Likewise, if a patient is tested with measurement of the C-reactive protein level alone and the result is positive, then the corresponding posttest probability of infection would be 0.74; if the result is negative, then the corresponding posttest probability of infection would be 1.0 — 0.95, or 0.05.

View Larger

TABLE III Diagnostic Test Characteristics for Patients Not Receiving Antibiotics (N = 151 Knees)

	Erythrocyte Sedimentation Rate*	C-Reactive Protein Level*	Erythrocyte Sedimentation Rate or C-Reactive Protein Level Positive*	Erythrocyte Sedimentation Rate and C-Reactive Protein Level Positive*
Use of cut-point suggested by receiver-operating-characteristic curves (erythrocyte sedimentation rate, 22.5 mm/hr; C-reactive protein level, 13.5 mg/L)
Sensitivity	0.93 (0.86-1.00)	0.91 (0.82-0.99)	0.95 (0.89-1.01)	0.88 (0.79-0.98)
Specificity	0.83 (0.76-0.9)	0.86 (0.8-0.93)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	5.81 (3.73-9.05)	6.89 (4.19-11.33)	4.22 (2.95-6.03)	13.54 (6.56-27.94)
Neg. likelihood ratio	0.07 (0.02-0.23)	0.10 (0.04-0.26)	0.05 (0.01-0.22)	0.12 (0.05-0.27)
Pos. predictive value	0.71 (0.59-0.82)	0.74 (0.63-0.86)	0.64 (0.52-0.75)	0.84 (0.74-0.95)
Neg. predictive value	0.96 (0.93-1.00)	0.95 (0.91-0.99)	0.97 (0.94-1.00)	0.95 (0.91-0.99)
Overall accuracy	0.86 (0.81-0.92)	0.88 (0.82-0.93)	0.82 (0.76-0.88)	0.92 (0.87-0.96)
Use of cut-point suggested by literature (erythrocyte sedimentation rate, 30 mm/hr; C-reactive protein level, 10 mg/L)
Sensitivity	0.82 (0.71-0.93)	0.93 (0.86-1.00)	0.95 (0.89-1.00)	0.79 (0.67-0.91)
Specificity	0.87 (0.81-0.93)	0.83 (0.75-0.90)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	6.7 (3.95-11.35)	5.49 (3.58-8.43)	4.22 (2.95-6.03)	12.15 (5.84-25.27)
Neg. likelihood ratio	0.20 (0.1-0.38)	0.08 (0.02-0.24)	0.05 (0.01-0.22)	0.21 (0.12-0.39)
Pos. predictive value	0.74 (0.61-0.86)	0.7 (0.58-0.81)	0.64 (0.52-0.75)	0.83 (0.72-0.94)
Neg. predictive value	0.92 (0.86-0.97)	0.96 (0.93-1.00)	0.97 (0.94-1.00)	0.91 (0.86-0.96)
Overall accuracy	0.86 (0.8-0.91)	0.86 (0.8-0.91)	0.82 (0.76-0.88)	0.89 (0.84-0.94)

The 95% confidence intervals are given in parentheses.

Add to My JBJS

TABLE III Diagnostic Test Characteristics for Patients Not Receiving Antibiotics (N = 151 Knees)

	Erythrocyte Sedimentation Rate*	C-Reactive Protein Level*	Erythrocyte Sedimentation Rate or C-Reactive Protein Level Positive*	Erythrocyte Sedimentation Rate and C-Reactive Protein Level Positive*
Use of cut-point suggested by receiver-operating-characteristic curves (erythrocyte sedimentation rate, 22.5 mm/hr; C-reactive protein level, 13.5 mg/L)
Sensitivity	0.93 (0.86-1.00)	0.91 (0.82-0.99)	0.95 (0.89-1.01)	0.88 (0.79-0.98)
Specificity	0.83 (0.76-0.9)	0.86 (0.8-0.93)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	5.81 (3.73-9.05)	6.89 (4.19-11.33)	4.22 (2.95-6.03)	13.54 (6.56-27.94)
Neg. likelihood ratio	0.07 (0.02-0.23)	0.10 (0.04-0.26)	0.05 (0.01-0.22)	0.12 (0.05-0.27)
Pos. predictive value	0.71 (0.59-0.82)	0.74 (0.63-0.86)	0.64 (0.52-0.75)	0.84 (0.74-0.95)
Neg. predictive value	0.96 (0.93-1.00)	0.95 (0.91-0.99)	0.97 (0.94-1.00)	0.95 (0.91-0.99)
Overall accuracy	0.86 (0.81-0.92)	0.88 (0.82-0.93)	0.82 (0.76-0.88)	0.92 (0.87-0.96)
Use of cut-point suggested by literature (erythrocyte sedimentation rate, 30 mm/hr; C-reactive protein level, 10 mg/L)
Sensitivity	0.82 (0.71-0.93)	0.93 (0.86-1.00)	0.95 (0.89-1.00)	0.79 (0.67-0.91)
Specificity	0.87 (0.81-0.93)	0.83 (0.75-0.90)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	6.7 (3.95-11.35)	5.49 (3.58-8.43)	4.22 (2.95-6.03)	12.15 (5.84-25.27)
Neg. likelihood ratio	0.20 (0.1-0.38)	0.08 (0.02-0.24)	0.05 (0.01-0.22)	0.21 (0.12-0.39)
Pos. predictive value	0.74 (0.61-0.86)	0.7 (0.58-0.81)	0.64 (0.52-0.75)	0.83 (0.72-0.94)
Neg. predictive value	0.92 (0.86-0.97)	0.96 (0.93-1.00)	0.97 (0.94-1.00)	0.91 (0.86-0.96)
Overall accuracy	0.86 (0.8-0.91)	0.86 (0.8-0.91)	0.82 (0.76-0.88)	0.89 (0.84-0.94)

The 95% confidence intervals are given in parentheses.

Combination testing of both the erythrocyte sedimentation rate and the C-reactive protein level, with the result considered positive if at least one of the two tests was positive, demonstrated increased sensitivity (0.95) in comparison with the sensitivity of either single test alone, while specificity decreased to 0.77. Combination testing of the erythrocyte sedimentation rate and C-reactive protein level increased the specificity of the test battery if a result was considered to be positive only when both the erythrocyte sedimentation rate and the C-reactive protein level were positive; the specificity increased to 0.93, while the sensitivity decreased to 0.88. When both the erythrocyte sedimentation rate and the C-reactive protein level were positive in our cohort, in which the prevalence of infection was 0.298, the posttest probability of infection (positive predictive value) was 0.84. If both the erythrocyte sedimentation rate and the C-reactive protein level are negative, then the posttest probability of infection is 1 minus the negative predictive value when the erythrocyte sedimentation rate or the C-reactive protein level is positive (0.97), or 0.03. Therefore, combination testing can provide the clinician with posttest probabilities of infection that are greater in magnitude than that obtained from any single test alone (Table III).

Analysis with use of the positivity criteria for the erythrocyte sedimentation rate and C-reactive protein level derived from the published literature^1,12,13 demonstrated similar results. The published literature suggests that an erythrocyte sedimentation rate of =30 mm/hr be considered positive and a C-reactive protein level of =10 mg/L be considered positive. Many consider these cut-points to be the standard for interpreting erythrocyte sedimentation rates and C-reactive protein levels, although they were derived from cohorts consisting largely of patients who had undergone a hip arthroplasty. We performed analyses and constructed receiver-operating-characteristic curves with use of these traditional cut-points for comparative purposes and to demonstrate the improved diagnostic test performance with our proposed cut-points (Table III).

Eleven knees in eleven patients were receiving antibiotics for the management of an established infection in the knee (as demonstrated by previous bacterial growth in cultures of aspiration or tissue specimens performed at another institution) at the time of transfer to our institution. These eleven knees were excluded from our primary data set because of the ongoing use of antibiotics, which may have affected the reliability of the cultures of the aspiration and tissue specimens performed at our institution. All eleven patients had negative results of the cultures performed at our institution, and we attributed those findings to the prior commencement of antibiotics resulting in false-negative cultures. However, we included these eleven knees in a secondary data set to evaluate whether ongoing antibiotic use altered the performance of the tests of the erythrocyte sedimentation rate and C-reactive protein level (Table IV). We recognize that these eleven patients represent an important subgroup that may be representative of many patients referred for assessment and management after antibiotic therapy has already started. Newly performed cultures of specimens from such patients may be falsely negative as a result of the commencement of antibiotic therapy. In all of these eleven patients with an infection (as indicated by positive cultures performed at the referring institution), the erythrocyte sedimentation rate and C-reactive protein level were elevated above our thresholds for a positive test (22.5 mm/hr and 13.5 mg/L for the erythrocyte sedimentation rate and C-reactive protein level, respectively). Including the eleven knees in the eleven patients who were taking antibiotics at the time of the assessment for infection in the data set (162 knees in 156 patients) did not appear to affect the diagnostic test characteristics of the erythrocyte sedimentation rate and C-reactive protein level (Fig. 2). The receiver-operating-characteristic curves, individual diagnostic test characteristics, and combined test characteristics and their confidence intervals (Fig. 2, Table IV) were essentially unchanged in comparison with those illustrated in Figure 1 and Table III. Therefore, although eleven is a small number of patients, and this may be a limitation of the study, it appears that, in the clinical situation in which a patient with symptoms related to a possible infection at the site of a knee arthroplasty was started on empiric antibiotic therapy prior to referral, the diagnostic test characteristics of the erythrocyte sedimentation rate and C-reactive protein level remain reliable and accurate.

View Larger

TABLE IV Diagnostic Test Characteristics for All Patients, Including Eleven Receiving Antibiotics (N = 162 Knees)

	Erythrocyte Sedimentation Rate*	C-Reactive Protein Level*	Erythrocyte Sedimentation Rate or C-Reactive Protein Level Positive*	Erythrocyte Sedimentation Rate and C-Reactive Protein Level Positive*
Use of cut-point suggested by receiver-operating-characteristic curves (erythrocyte sedimentation rate, 22.5 mm/hr; C-reactive protein level, 13.5 mg/L)
Sensitivity	0.92 (0.85-0.99)	0.89 (0.8-0.97)	0.94 (0.88-1.00)	0.87 (0.78-0.95)
Specificity	0.84 (0.77-0.91)	0.86 (0.8-0.93)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	5.83 (3.75-9.08)	6.8 (4.14-11.19)	4.21 (2.94-6.02)	13.42 (6.51-27.69)
Neg. likelihood ratio	0.08 (0.03-0.22)	0.12 (0.05-0.26)	0.07 (0.02-0.21)	0.13 (0.06-0.27)
Pos. predictive value	0.75 (0.64-0.85)	0.77 (0.67-0.88)	0.68 (0.57-0.78)	0.87 (0.78-0.95)
Neg. predictive value	0.95 (0.91-0.99)	0.93 (0.89-0.98)	0.96 (0.92-1.00)	0.93 (0.88-0.98)
Overall accuracy	0.87 (0.81-0.92)	0.87 (0.82-0.92)	0.83 (0.77-0.89)	0.91 (0.87-0.95)
Use of cut-point suggested by literature (erythrocyte sedimentation rate, 30 mm/hr; C-reactive protein level, 10 mg/L)
Sensitivity	0.81 (0.71-0.92)	0.9 (0.83-0.98)	0.94 (0.88-1.00)	0.77 (0.66-0.88)
Specificity	0.87 (0.81-0.94)	0.83 (0.76-0.9)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	6.73 (3.98-11.37)	5.40 (3.51-8.3)	4.21 (2.94-6.02)	12.00 (5.78-24.91)
Neg. likelihood ratio	0.20 (0.11-0.36)	0.10 (0.04-0.25)	0.07 (0.02-0.21)	0.23 (0.14-0.39)
Pos. predictive value	0.77 (0.66-0.88)	0.73 (0.63-0.84)	0.68 (0.57-0.78)	0.85 (0.75-0.95)
Neg. predictive value	0.90 (0.84-0.96)	0.94 (0.9-0.99)	0.96 (0.92-1.00)	0.89 (0.83-0.95)
Overall accuracy	0.85 (0.8-0.91)	0.85 (0.8-0.91)	0.83 (0.77-0.89)	0.88 (0.83-0.93)

The 95% confidence intervals are given in parentheses.

Add to My JBJS

TABLE IV Diagnostic Test Characteristics for All Patients, Including Eleven Receiving Antibiotics (N = 162 Knees)

	Erythrocyte Sedimentation Rate*	C-Reactive Protein Level*	Erythrocyte Sedimentation Rate or C-Reactive Protein Level Positive*	Erythrocyte Sedimentation Rate and C-Reactive Protein Level Positive*
Use of cut-point suggested by receiver-operating-characteristic curves (erythrocyte sedimentation rate, 22.5 mm/hr; C-reactive protein level, 13.5 mg/L)
Sensitivity	0.92 (0.85-0.99)	0.89 (0.8-0.97)	0.94 (0.88-1.00)	0.87 (0.78-0.95)
Specificity	0.84 (0.77-0.91)	0.86 (0.8-0.93)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	5.83 (3.75-9.08)	6.8 (4.14-11.19)	4.21 (2.94-6.02)	13.42 (6.51-27.69)
Neg. likelihood ratio	0.08 (0.03-0.22)	0.12 (0.05-0.26)	0.07 (0.02-0.21)	0.13 (0.06-0.27)
Pos. predictive value	0.75 (0.64-0.85)	0.77 (0.67-0.88)	0.68 (0.57-0.78)	0.87 (0.78-0.95)
Neg. predictive value	0.95 (0.91-0.99)	0.93 (0.89-0.98)	0.96 (0.92-1.00)	0.93 (0.88-0.98)
Overall accuracy	0.87 (0.81-0.92)	0.87 (0.82-0.92)	0.83 (0.77-0.89)	0.91 (0.87-0.95)
Use of cut-point suggested by literature (erythrocyte sedimentation rate, 30 mm/hr; C-reactive protein level, 10 mg/L)
Sensitivity	0.81 (0.71-0.92)	0.9 (0.83-0.98)	0.94 (0.88-1.00)	0.77 (0.66-0.88)
Specificity	0.87 (0.81-0.94)	0.83 (0.76-0.9)	0.77 (0.69-0.85)	0.93 (0.88-0.98)
Pos. likelihood ratio	6.73 (3.98-11.37)	5.40 (3.51-8.3)	4.21 (2.94-6.02)	12.00 (5.78-24.91)
Neg. likelihood ratio	0.20 (0.11-0.36)	0.10 (0.04-0.25)	0.07 (0.02-0.21)	0.23 (0.14-0.39)
Pos. predictive value	0.77 (0.66-0.88)	0.73 (0.63-0.84)	0.68 (0.57-0.78)	0.85 (0.75-0.95)
Neg. predictive value	0.90 (0.84-0.96)	0.94 (0.9-0.99)	0.96 (0.92-1.00)	0.89 (0.83-0.95)
Overall accuracy	0.85 (0.8-0.91)	0.85 (0.8-0.91)	0.83 (0.77-0.89)	0.88 (0.83-0.93)

The 95% confidence intervals are given in parentheses.

Discussion

Materials and Methods | Results | Discussion | Appendix | References

While many different diagnostic tests may be used to assist in establishing the presence or absence of infection, many of these tests (culture of intraoperative specimens, histopathological analysis of frozen sections, and final histopathological analyses) cannot aid the surgeon with preoperative decision-making or preoperative planning as the specimens are not obtained until the knee prosthesis is exposed at the time of the operation^15-19. Additional tests, such as nuclear medicine imaging, may be highly sensitive but not specific to the diagnosis of infection^4,20. The ideal investigation for the diagnosis of infection should be able to be carried out prior to the operation and should involve a minimum number of tests, which in combination should be accurate, convenient to the patient, cause minimal morbidity, and be cost-effective.

The erythrocyte sedimentation rate and C-reactive protein level are nonspecific markers of acute inflammation, which may be elevated in infective, neoplastic, or inflammatory conditions^21,22. Recent reports suggest that the C-reactive protein level and erythrocyte sedimentation rate rise promptly following hip and knee arthroplasty and that the C-reactive protein level and erythrocyte sedimentation rate return to normal at approximately three weeks and two months postoperatively, respectively^13,23,24. Elevated erythrocyte sedimentation rates (means of 53 to 63 mm/hr) have been reported in several series of patients with an infection at the site of a total knee arthroplasty, and as a result the erythrocyte sedimentation rate has been widely used in an effort to detect infection prior to operations^20,25. While measurement of the erythrocyte sedimentation rate is commonly used as a diagnostic test, concern about its inadequate sensitivity has been expressed^5,20,25,26. This concern may be due to a number of deficiencies in previous studies on erythrocyte sedimentation rates in patients with infection at the site of a joint arthroplasty, as the majority have been retrospective, have included a small number of knees, or have used inconsistent criteria for the definition of the so-called gold-standard test for infection. Reports on the use of the C-reactive protein level for the diagnosis of infection suggest that it may be more sensitive, specific, and responsive than the erythrocyte sedimentation rate^11,27-29.

In order to adequately assess the diagnostic characteristics of the erythrocyte sedimentation rate and C-reactive protein level, it is necessary to determine the appropriate positivity criteria through receiver-operating-characteristic curve analysis. Furthermore, strict criteria should be adhered to with regard to inclusion and exclusion of patients (with exclusion of those with systemic illness, inflammatory arthritis, and other comorbid diseases that may result in abnormal elevations of the erythrocyte sedimentation rate and C-reactive protein level) and with regard to the definition of the so-called gold standard test for infection in the study.

To the best of our knowledge, this is the first study to demonstrate clinically relevant positivity criteria for preoperative serological tests for the diagnosis of infection in patients presenting for revision total knee arthroplasty. It appears that a threshold of 22.5 mm/hr for the erythrocyte sedimentation rate and 13.5 mg/L for the C-reactive protein level provide the clinician with the greatest amount of diagnostic information from these two serological tests. These cut-points result in maximum test accuracy. With the use of these thresholds, the erythrocyte sedimentation rate and the C-reactive protein level are highly sensitive (0.93 and 0.91, respectively) for the diagnosis of infection in patients with symptoms related to knee arthroplasty. In addition, they retain good specificity (0.83 and 0.86, respectively).

Because relatively few tests are both highly sensitive and highly specific, two or more tests are often used to evaluate a diagnostic possibility and therefore potentially increase the test sensitivity or specificity. It is therefore important to understand the combined operating characteristics of the erythrocyte-sedimentation-rate and C-reactive-protein tests¹¹. Our results suggest that, if the clinician administers both tests simultaneously and considers a positive result of at least one test as an indication of infection, then the sensitivity increases to 0.95, which is higher than the sensitivity of either test performed alone (Table III). However, combining the two tests decreases specificity (to 0.77). The apparent increase in sensitivity suggests that this combination may be desired to optimize screening for infection (i.e., it is good for ruling out disease), as previous authors have suggested that the ideal screening test should be highly sensitive¹¹. The results in our cohort suggest that when the erythrocyte sedimentation rate and C-reactive protein level are measured together as a so-called screening battery (with a positive erythrocyte sedimentation rate or C-reactive protein level indicative of a positive test battery) then the sensitivity is 0.95, the negative predictive value is 0.97, and the corresponding posttest probability of infection is 0.03 (1 minus the negative predictive value). This means that if both tests are negative, then there is only a 3% probability that infection is present in a patient with symptoms related to a knee arthroplasty. Alternatively, if the clinician desires greater specificity from the diagnostic test battery, he or she should consider the combination positive only if both the erythrocyte sedimentation rate and the C-reactive protein level are positive (specificity, 0.93; sensitivity, 0.88; and positive predictive value, 0.84). This specificity is higher than the specificity of either single test alone. High specificity is a desired feature for tests that are being used to confirm the presence of disease¹¹. In our cohort of patients with symptoms related to a knee arthroplasty, if both the erythrocyte sedimentation rate and C-reactive protein level were positive then the posttest probability of infection was 0.84, which is an 84% probability of infection being present.

Despite the lack of published data to substantiate the traditional positivity criteria for the erythrocyte sedimentation rate and C-reactive protein level (30 mm/hr and 10 mg/L, respectively), our data demonstrate that these cut-points result in reasonable test performance (Figs. 3 and 4). However, as these cut-points do not occur at the apex of the receiver-operating-characteristic curve, they do not represent the ideal threshold at which both the sensitivity and the specificity of the tests are optimized.

It is important to acknowledge potential limitations of the presented data. Error and bias may have arisen as a result of the absence of a perfect "gold standard" test and our need to use an imperfect "gold standard" test. We used the growth of bacteria on culture of fluid or tissue from the joint as our uniform gold standard. Although not perfect, the rigor of this standard should minimize the risk of error from a so-called fuzzy gold standard, and so-called verification bias^30,31.

The current study validates measurements of the erythrocyte sedimentation rate and C-reactive protein level as excellent diagnostic tests with which to detect infection before revision knee arthroplasty is performed. Careful selection and administration of these tests, in conjunction with an astute history and physical examination, may help to ensure that each patient with pain at the site of a total knee arthroplasty presenting for consideration of revision surgery is managed with the most appropriate therapeutic regimen.

Appendix

Materials and Methods | Results | Discussion | Appendix | References

A table showing the infecting organisms grown on culture and additional information pertaining to terminology, definitions, and utility of the diagnostic tests are available with the electronic versions of this article, on our web site at (go to the article citation and click on "Supplementary Material") and on our quarterly CD-ROM (call our subscription department, at 781-449-9780, to order the CD-ROM). ?

References

Materials and Methods | Results | Discussion | Appendix | References

Spangehl MJ, Masri BA, O'Connell JX, Duncan CP. 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: 672-83.81672 1999 [PubMed]

Hanssen AD, Rand JA. Evaluation and treatment of infection at the site of a total hip or knee arthroplasty. Instr Course Lect.1999;48: 111-22.48111 1999 [PubMed]

Sculco TP. The economic impact of infected total joint arthroplasty. Instr Course Lect.1993;42: 349-51.42349 1993 [PubMed]

Grogan TJ, Dorey F, Rollins J, Amstutz HC. Deep sepsis following total knee arthroplasty. Ten-year experience at the University of California at Los Angeles Medical Center. J Bone Joint Surg Am.1986;68: 226-34.68226 1986

Levitsky KA, Hozack WJ, Balderson RA, Rothman RH, Gluckman SJ, Maslack MM, Booth RE Jr. Evaluation of the painful prosthetic joint. Relative value of bone scan, sedimentation rate, and joint aspiration. J Arthroplasty.1991;6: 237-44.6237 1991 [PubMed][CrossRef]

Hanssen AD, Rand JA, Osmon DR. Treatment of the infected total knee arthroplasty with insertion of another prosthesis. The effect of antibiotic-impregnated bone cement. Clin Orthop Relat Res.1994;309: 44-54.30944 1994

Datz FL, Anderson CH, Ahluwalia R, Whiting JH, Gabor FV, Morton KA, Christian PE, Crebs K, Neptune M, Rauh DA. The efficacy of indium-111-polyclonal IgG for the detection of infection and inflammation. J Nucl Med.1994;35: 74-83.3574 1994 [PubMed]

Becker W, Palestro CJ, Winship J, Feld T, Pinsky CM, Wolf F, Goldenberg DM. Rapid imaging of infections with a monoclonal antibody fragment (LeukoScan). Clin Orthop Relat Res.1996;329: 263-72.329263 1996 [CrossRef]

Laiho K, Maenpaa H, Kautiainen H, Kauppi M, Kaarela K, Lehto M, Belt E. Rise in serum C reactive protein after hip and knee arthroplasties in patients with rheumatoid arthritis. Ann Rheum Dis.2001;60: 275-7.60275 2001 [CrossRef]

Swets JA, Pickett RM, Whitehead SF, Getty DJ, Schnur JA, Swets JB, Freeman BA. Assessment of diagnostic technologies. Science.1979;205: 753-9.205753 1979 [PubMed][CrossRef]

Black ER, Panzer RJ, Mayewski RJ, Griner PF. Characteristics of diagnostic tests and principles for their use in quantitative decision making. In: Black ER, Bordley DR, Tape TG, Panzer RJ. Diagnostic strategies of common medical problems. Philadelphia: American College of Physicians; 1999. p 8-10.8 1999

Niskanen RO, Korkala O, Pammo H. Serum C-reactive protein levels after total hip and knee arthroplasty. J Bone Joint Surg Br.1996;78: 431-3.78431 1996

White J, Kelly M, Dunsmuir R. C-reactive protein level after total hip and total knee replacement. J Bone Joint Surg Br.1998;80: 909-11.80909 1998 [CrossRef]

Jaescheke R, Guyatt G, Lijmer J. Diagnostic tests. In: Guyatt G, Rennie D, editors. User's guide to the medical literature: essentials of evidence-based clinical practice. Chicago: AMA Press; 2002. p 127-28.127 2002

Mariani BD, Martin DS, Levine MJ, Booth RE Jr, Tuan RS. Polymerase chain reaction detection of bacterial infection in total knee arthroplasty. Clin Orthop Relat Res.1996;331: 11-22.33111 1996 [PubMed][CrossRef]

Della Valle CJ, Scher DM, Kim YH, Oxley CM, Desai P, Zuckerman JD, Di Cesare PE. The role of intraoperative Gram stain in revision total joint arthroplasty. J Arthroplasty.1999;14: 500-4.14500 1999 [PubMed][CrossRef]

Mirra JM, Amstutz HC, Matos M, Gold R. The pathology of the joint tissues and its clinical relevance in prosthesis failure. Clin Orthop Relat Res.1976;117: 221-40.117221 1976 [PubMed]

Feldman DS, Lonner JH, Desai P, Zuckerman JD. The role of intraoperative frozen sections in revision total joint arthroplasty. J Bone Joint Surg Am.1995;77: 1807-13.771807 1995 [PubMed]

Lonner JH, Desai P, Dicesare PE, Steiner G, Zuckerman JD. The reliability of analysis of intraoperative frozen sections for identifying active infection during revision hip or knee arthroplasty. J Bone Joint Surg Am.1996;78: 1553-8.781553 1996 [PubMed]

Windsor RE, Insall JN, Urs WK, Miller DV, Brause BD. Two-stage reimplantation for the salvage of total knee arthroplasty complicated by infection. Further follow-up and refinement of indications. J Bone Joint Surg Am.1990;72: 272-8.72272 1990 [PubMed]

Covey DC, Albright JA. Clinical significance of the erythrocyte sedimentation rate in orthopaedic surgery. J Bone Joint Surg Am.1987;69: 148-51.69148 1987 [PubMed]

Shih LY, Wu JJ, Yang DJ. Erythrocyte sedimentation rate and C-reactive protein values in patients with total hip arthroplasty. Clin Orthop Relat Res.1987;225: 238-46.225238 1987 [PubMed]

Moreschini O, Greggi G, Giordano MC, Nocente M, Margheritini F. Postoperative physiopathological analysis of inflammatory parameters in patients undergoing hip or knee arthroplasty. Int J Tissue React.2001;23: 151-4.23151 2001 [PubMed]

Bilgen O, Atici T, Durak K, Karaeminogullari O, Bilgen MS. C-reactive protein values and erythrocyte sedimentation rates after total hip and total knee arthroplasty. J Int Med Res.2001;29: 7-12.297 2001

Barrack RL, Jennings RW, Wolfe MW, Bertot AJ. The value of preoperative aspiration before total knee revision. Clin Orthop Relat Res.1997;345: 8-16.3458 1997 [PubMed]

Rand JA, Chao EY, Stauffer RN. Kinematic rotating-hinge total knee arthroplasty. J Bone Joint Surg Am.1987;69: 489-97.69489 1987 [PubMed]

Choudhry RR, Rice RP, Triffitt PD, Harper WM, Gregg PJ. Plasma viscosity and C-reactive protein after total hip and knee arthroplasty. J Bone Joint Surg Br.1992;74: 523-4.74523 1992 [PubMed]

Sanzen L, Carlsson AS, Josefsson G, Lindberg LT. Revision operations on infected total hip arthroplasties. Two- to nine-year follow-up study. Clin Orthop Relat Res.1988;229: 165-72.229165 1988 [PubMed]

Aalto K, Osterman K, Peltola H, Rasanen J. Changes in erythrocyte sedimentation rate and C-reactive protein after total hip arthroplasty. Clin Orthop Relat Res.1984;184: 118-20.184118 1984 [PubMed]

Begg CB. Biases in the assessment of diagnostic tests. Stat Med.1987;6: 411-23.6411 1987 [PubMed][CrossRef]

Phelps CE, Hutson A. Estimating diagnostic test accuracy using a "fuzzy gold standard". Med Decis Making.1995;15: 44-57.1544 1995 [CrossRef]

Merkel KD, Brown ML, Dewanjee MK, Fitzgerald RH. Comparison of indium-labeled-leukocyte imaging with sequential technetium-gallium scanning in the diagnosis of low-grade musculoskeletal sepsis. A prospective study. J Bone Joint Surg Am.1985;67: 465-76.67465 1985 [PubMed]

Palestro CJ, Swyer AJ, Kim CK, Goldsmith SJ. Infected knee prosthesis: diagnosis with In-111 leukocyte, Tc-99m sulfur colloid, and Tc-99m MDP imaging. Radiology.1991;179: 645-8.179645 1991 [PubMed]

Topics

erythrocyte sedimentation rate ; infection ; c-reactive protein measurement ; knee replacement arthroplasty ; receiver operating characteristic

Username

Password

Access for Patients

Forgot Password?

Have a subscription to the print edition?

Not a Subscriber?

Buy this Article

Get online access for 30 days for $35

New to JBJS?

Subscribe to the online edition for 30 days for $75

Register for a FREE limited account to get full access to all CME activities, to comment on public articles, or to sign up for alerts.

Have a subscription to the print edition?

Current subscribers to The Journal of Bone & Joint Surgery in either the print or quarterly DVD formats receive free online access to JBJS.org.

Activate your online account now

Forgot your password?

Enter your username and email address. We'll send you a reminder to the email address on record.

Username:*

Email Address:*

Forgot your username or need assistance? Please contact customer service at subs@jbjs.org. If your access is provided
by your institution, please contact you librarian or administrator for username and password information. Institutional
administrators, to reset your institution's master username or password, please contact subs@jbjs.org

References

Accreditation Statement

These activities have been planned and implemented in accordance with the Essential Areas and policies of the Accreditation Council for Continuing Medical Education (ACCME) through the joint sponsorship of the American Academy of Orthopaedic Surgeons and The Journal of Bone and Joint Surgery, Inc. The American Academy of Orthopaedic Surgeons is accredited by the ACCME to provide continuing medical education for physicians.

CME Activities Associated with This Article

Submit a Comment

Please read the other comments before you post yours. Contributors must reveal any conflict of interest.
Comments are moderated and will appear on the site at the discretion of JBJS editorial staff.

* = Required Field

Comment Author(s) * (if multiple authors, separate names by comma)

Example: John Doe

Affiliation & Institution*

Comment Title*

Comment*

Cancel

Nelson V. Greidanus

Posted on October 18, 2007

Optimal Operating Points Determination for ESR and CRP

University of British Columbia

The stated purpose of our published manuscript was â€œto prospectively evaluate the diagnostic test characteristics of the erythrocyte sedimentation rate(ESR) and C-reactive protein(CRP) level for the diagnosis of infection prior to revision total knee arthroplastyâ€ (1). We collected data prospectively and evaluated the diagnostic performance of the ESR and CRP against the so-called â€˜gold standardâ€™ diagnosis of infection---positive growth of bacteria on culture of intra-articular fluid or tissue specimen. Microsoft Excel(Redmond, Washington) and SPSS(version 10.0; SPSS) software were employed to tabulate, graphically illustrate, and analyze the data including the development of receiver-operator-characteristic (ROC) curves for the respective ESR and CRP tests. Our goals were to plot the ROC curves, identify optimal cutpoints for the ESR and CRP, and to objectively illustrate the performance of these tests using standardized nomenclature with 95% confidence intervals. We selected to illustrate the â€˜ideal cutpointsâ€™ alongside the so-called â€˜traditional cutpointsâ€™ for ESR and CRP so that the readership may understand the differences in test performance should they choose to use cutpoints from the historical literature or those recommended from the hip arthroplasty literature(2).

Dr. Lin et al. identify a number of methods for ROC curve evaluation and â€˜cut-pointâ€™ determination. While numerous methods are available for determining the optimal cut-point there is no evidence that one method is superior or is the so-called â€˜gold standardâ€™(3). One method that is commonly utilized is to use the apex of the ROC curve (upper left corner of the curve) as the optimal cut-point as this represents the point at which sensitivity and specificity are optimized. This is the method that we used, which corresponds with the Youden index(4). While we recommend the selection of cutpoints of 22.5 mm/hr for ESR and 13.5mg/L for CRP on the basis of this method we do not claim statistical superiority (p <_.05 of="of" this="this" cutpoint="cutpoint" over="over" the="the" traditional="traditional" cutpoints="cutpoints" _30mm="_30mm" hr="hr" and="and" _10mg="_10mg" l="l" for="for" esr="esr" crp="crp" respectively.="respectively." is="is" evident="evident" from="from" overlapping="overlapping" _95="_95" confidence="confidence" intervals="intervals" in="in" table="table" _3.="_3." variance="variance" seen="seen" point="point" estimates="estimates" our="our" modest="modest" sample="sample" size="size" do="do" not="not" permit="permit" us="us" to="to" demonstrate="demonstrate" statistical="statistical" superiority="superiority" proposed="proposed" reported="reported" crp.="crp." however="however" knowledge="knowledge" report="report" first="first" an="an" evidence-based="evidence-based" determination="determination" optimal="optimal" diagnose="diagnose" infection="infection" prior="prior" revision="revision" knee="knee" arthroplasty.="arthroplasty." data="data" analytic="analytic" methods="methods" suggest="suggest" that="that" these="these" are="are" _22.5mm="_22.5mm" _13.5mg="_13.5mg" p="p" /> When performing â€˜area under curveâ€™(AUC) analyses we calculated the AUC (with 95% confidence intervals) for ESR ROC as 0.925 (0.884-0.967) and for CRP ROC as 0.908 (0.857-0.960). This AUC data does not prove statistical superiority of one test over the other but rather demonstrates that both the ESR and CRP are â€˜highly accurateâ€™ diagnostic tests as defined by criteria of(5). Our calculations and subsequent recommendations for using ESR and CRP together in â€˜combination testingâ€™ (for improved sensitivity and specificity) are based on Bayesion methods as recommended by Black et al.(6). We did not develop ROC curves and AUC analyses for the 'combination testing' of ESR and CRP and therefore are unable to comment on AUC differences which might occur in such scenarios.

In summary, we have demonstrated objective â€˜evidence-basedâ€™ cutpoints for ESR and CRP in the population of patients assessed for revision total knee arthroplasty. Furthermore, we have developed and demonstrated ROC curves which prove that the ESR and CRP tests are highly accurate when evaluating for possible infection. Finally, we have suggested â€˜combination testingâ€™ of the ESR and CRP as a method to increase the diagnostic information from these tests whether screening patients for the possibility of infection or seeking confirmation of a presumptive diagnosis of infection.

References:

1. Greidanus NV, Masri BA, Garbuz DS, Wilson SD, McAlinden MG., Xu M, Duncan CP. Use of erythrocyte sedimentation rate and C-reactive protein level to diagnose infection before revision total knee arthroplasty. A prospective evaluation. J Bone Joint Surg Am 2007;89:1409-1416.

2. Spangehl MJU, Masri BA, O'Connell JX, Duncan CP. Prospective analysis of preoperative and intraoperative investigations for the diagnosis of injection at the sites of two hundred and two evision total hip arthroplasties. J bone Joint Surg Am. 1999;81:672-83.

3. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000;45:23-41.

4. Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32-35.

5. Swetts JA. Measuring the accuracy of diagnostic systems. Science 1988; 240: 1285-1293.

6. Black ER, Panzer RJ, Mayewski RJ, Griner PF. Characteristics of diagnostic tests and principles for their use in quantitative decision making. In: Black ER, Bordley DR, Tape TG, Panzer RJ. Diagnostic strategies of common medical problems. Philadelphia: American College of Physicians; 1999. p 8-10

Zong-I Lin, M.D.

Posted on August 22, 2007

Optimal Operating Points Determination

Chung Shan Medical University Hospital, TAIWAN

To The Editor:

In their recent article, "Use of Erythrocyte Sedimentation Rate and C-Reactive Protein Level to Diagnose Infection Before Revision Total Knee Arthroplasty. A Prospective Evaluation" Greidanus et al.(1) did not publish the data of the area under the curve (AUC) of the respective receiver-operating characteristic (ROC) curve of erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, ESR or CRP positive, and ESR and CRP level positive. They also did not compare the AUC between ESR or CRP positive and ESR and CRP level positive by statistical tests.

We found, with or without receiving antibiotics, there were no statistically significant difference between the overall accuracy of 22.5 mm/hr for ESR and of 30 mm/hr for ESR, of 13.5 mg/L for CRP and of 10 mg/L for CRP, of 22.5 mm/hr for ESR or 13.5 mg/L for CRP and of 30 mm/hr for ESR or 10 mg/L for CRP, and of 22.5 mm/hr for ESR and 13.5 mg/L for CRP and of 30 mm/hr for ESR and 10 mg/L for CRP. The combination of ESR and CRP tests provides better diagnostic information statistically when selecting 22.5 mm/hr for ESR and 13.5 mg/L for CRP, not 30 mm/hr for ESR and 10 mg/L for CRP.

In addition, Greidanus et al.(1) also did not mention clearly which method they used to determine the optimal operating point of ESR and CRP, probably by choosing cut-points at the apex of the ROC. We rechecked the cut-off points of ESR and CRP by the Youden index (J), maximum {Sensitivity + Specificity -1}(2), and we reached the same conclusion. We could not, however, perform the t test to compare the difference because of lacking of the complete data set, and we wonder about the statistical significance. The criteria for the optimal cut-off determination include a plot of sensitivity and specificity as a function of the cut-off value, Youden index, odds ratio, likelihood ratio, efficiency, kappa, misclassification-cost term, optimized likelihood ratio for a given prevalence, slope approach, and many more(3). We wish the authors will provide details about their method for determining the optimal cut-off points and the statistical significance.

The authors did not receive any outside funding or grants in support of their research for or preparation of this work. Neither they nor a member of their immediate families received payments or other benefits or a commitment or agreement to provide such benefits from a commercial entity. No commercial entity paid or directed, or agreed to pay or direct, any benefits to any research fund, foundation, division, center, clinical practice, or other charitable or nonprofit organization with which the authors, or a member of their immediate families, are affiliated or associated.

References:

2. Youden WJ. Index for rating diagnostic tests. Cancer 1950;3:32-35.

3. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000;45:23-41.

Print
PDF
Email

Recipient(s) will receive an email with a link (good for 5 days) to 'Use of Erythrocyte Sedimentation Rate and C-Reactive Protein Level to Diagnose Infection Before Revision Total Knee Arthroplasty: A Prospective Evaluation'.
Your Name:*
Example: John Doe

Email Address:* CC Me:
Enter your valid email address. Example: jdoe@example.com

Recipient's Email Address:*

Separate multiple email address with semi-colons (up to 5).

Subject:* 's The Journal of Bone & Joint Surgery: 'Use of Erythrocyte Sedimentation Rate and C-Reactive Protein Level to Diagnose Infection Before Revision Total Knee Arthroplasty: A Prospective Evaluation'
Subject for your email.

Message:

(Optional, message will truncate at 1000 characters)

Processing your request... Please Wait...

Copyright © in the material you requested is held by The Journal of Bone & Joint Surgery, Inc. (unless otherwise noted). This email ability is provided as a courtesy, and by using it you agree that that you are requesting the material solely for personal, non-commercial use, and that it is subject to Journal of Bone & Joint Surgery, Inc.'s Terms of Use. The information provided in order to email this topic will not be used to send unsolicited email, nor will it be furnished to third parties. Please refer to The Journal of Bone & Joint Surgery Inc.'s Privacy Policy for further information.

Copyright © The Journal of Bone & Joint Surgery Inc. All rights reserved.
Share
Get Citation

Nelson V. Greidanus, Bassam A. Masri, Donald S. Garbuz, S. Darrin Wilson, M. Gavan McAlinden, Min Xu, Clive P. Duncan; Use of Erythrocyte Sedimentation Rate and C-Reactive Protein Level to Diagnose Infection Before Revision Total Knee ArthroplastyA Prospective Evaluation. The Journal of Bone & Joint Surgery. 2007 Jul;89(7):1409-1416.

Download citation file:

RIS (Zotero)

EndNote

BibTex

Medlars

ProCite

RefWorks

Reference Manager

Copyright © The Journal of Bone and Joint Surgery, Inc.
Rights & Permissions
Article Alerts

Processing your request... Please Wait.
Submit a Comment
Data Supplements