The Journal of Bone & Joint Surgery

The Journal of Bone & Joint Surgery, Volume 93, Issue 18

Scientific Articles | September 21, 2011

Differentiating Between Methicillin-Resistant and Methicillin-Sensitive Staphylococcus aureus Osteomyelitis in Children: An Evidence-Based Clinical Prediction Algorithm

Kevin L. Ju, MD¹; David Zurakowski, PhD¹; Mininder S. Kocher, MD, MPH¹

¹ Department of Orthopaedic Surgery, Children's Hospital Boston, 300 Longwood Avenue, Boston, MA 02115. E-mail address for M.S. Kocher: mininder.kocher@childrens.harvard.edu

View Disclosures and Other Information

Disclosure: None of the authors received payments or services, either directly or indirectly (i.e., via his or her institution), from a third party in support of any aspect of this work. One or more of the authors, or his or her institution, has had a financial relationship, in the thirty-six months prior to submission of this work, with an entity in the biomedical arena that could be perceived to influence or have the potential to influence what is written in this work. No author has had any other relationships, or has engaged in any other activities, that could be perceived to influence or have the potential to influence what is written in this work. The complete Disclosures of Potential Conflicts of Interest submitted by authors are always provided with the online version of the article.

Disclosure statement for author(s): PDF

Investigation performed at Children's Hospital Boston, Boston, Massachusetts

A commentary by David J. Zaleske, MD, is linked to the online version of this article at jbjs.org.

J Bone Joint Surg Am, 2011 Sep 21;93(18):1693-1701. doi: 10.2106/JBJS.J.01154

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

Article

Comments (0)

text A A A

Abstract

Background:

Although osteomyelitis was once commonly due to methicillin-sensitive Staphylococcus aureus (MSSA), methicillin-resistant Staphylococcus aureus (MRSA)—which causes more virulent and invasive infections—has emerged as an increasingly important cause. Differentiating clinically between MRSA and MSSA can be challenging, but is necessary in order to promptly administer appropriate antibiotics and maintain vigilance against possible sequelae of MRSA osteomyelitis. The purpose of our study was to develop a clinical prediction algorithm to distinguish between MRSA and MSSA osteomyelitis in children.

Methods:

A retrospective review of 129 children presenting with culture-proven Staphylococcus aureus osteomyelitis between 2000 and 2009 was performed. The demographics, symptoms, vital signs, and laboratory test values in the MSSA group (n = 118) and the MRSA group (n = 11) were compared with use of univariate analysis. Multivariate logistic regression with backward stepwise selection was then used to identify independent multivariate predictors of MRSA osteomyelitis, and each of these predictors was subjected to receiver operating characteristic curve analysis to determine the optimal cutoff value. Finally, a prediction algorithm for differentiating between MRSA and MSSA osteomyelitis on the basis of these independent predictors was constructed.

Results:

Patients with MRSA osteomyelitis differed significantly from those with MSSA osteomyelitis with regard to non-weight-bearing status, antibiotic use at presentation, body temperature, hematocrit value, heart rate, white blood-cell count, platelet count, C-reactive protein level, and erythrocyte sedimentation rate. Four significant independent multivariate predictors were identified: a temperature of >38°C, a hematocrit value of <34%, a white blood-cell count of >12,000 cells/μL, and a C-reactive protein level of >13 mg/L. The predicted probability of MRSA osteomyelitis, determined on the basis of the number of these predictors that a child satisfied, was 92% for all four predictors, 45% for three, 10% for two, 1% for one, and 0% for zero predictors. Receiver operating characteristic curve analysis was used to evaluate the predictive accuracy of the number of multivariate predictors, and this analysis revealed a steep shoulder and an area under the curve of 0.94 (95% confidence interval, 0.88 to 1.00).

Conclusions:

Our proposed set of four predictors provided excellent diagnostic performance in differentiating between MRSA and MSSA osteomyelitis in children, and thus would be able to guide patient management and facilitate timely antibiotic selection.

Level of Evidence:

Diagnostic Level IV. See Instructions to Authors for a complete description of levels of evidence.

Figures in this Article

Introduction

Osteomyelitis is one of the most common invasive bacterial infections in children and has the potential to result in a poor musculoskeletal outcome and in systemic illness. Staphylococcus aureus is the most prevalent etiologic agent in osteomyelitis that occurs during childhood^1,2. Although methicillin-sensitive Staphylococcus aureus (MSSA) was once commonly the cause of such bone infections, methicillin-resistant Staphylococcus aureus (MRSA), either hospital-acquired or community-acquired, has emerged as an increasingly important cause over the past decade^3-5.

A rising concern regarding the use of broad-spectrum antibiotics has accompanied the appearance of resistant bacterial strains^6,7. Vancomycin has long been the primary choice for treating serious MRSA infections, but vancomycin-resistant strains of Staphylococcus aureus have recently appeared⁸. Increased use of vancomycin also appears to be responsible for the dramatic increase in colonization and infection by vancomycin-resistant enterococci in American hospitals^9,10. Furthermore, widespread use of macrolides to treat MRSA has added selective pressure for strains to evolve constitutive or inducible resistance to clindamycin and other macrolides^11,12. The emergence of resistant Staphylococcus aureus strains lowers the probability that our once-reliable antibiotics will be effective against future infections, including osteomyelitis⁸. Therefore, prudent and appropriate antibiotic selection is important.

Evidence has been accumulating that infections produced by MRSA are more virulent and more deeply invasive than those produced by MSSA. Musculoskeletal infections caused by MRSA are associated with a longer hospital stay, a greater number of surgical interventions, and higher complication and mortality rates^13-15 than those caused by MSSA, as well as with higher rates of deep venous thrombosis, septic pulmonary emboli, and abscess formation^3,16,17. However, differentiating between MRSA and MSSA osteomyelitis when a patient first presents can be challenging. Both can result in fever, pain, and loss of function of the involved extremity. Furthermore, once cultures have been acquired, it may require several days before the strain of the pathogen is identified and antibiotic susceptibility data become available. The purpose of our study was to develop an evidence-based clinical prediction algorithm to distinguish between MRSA and MSSA osteomyelitis on the basis of clinical data available at the time the patient presents for treatment.

Materials and Methods

Study Design

Cases of MSSA and MRSA osteomyelitis treated at a tertiary-care children's hospital between January 1, 2000, and May 31, 2009, were retrospectively evaluated after approval from our hospital's institutional review board. Patients were identified through the hospital's microbiology database on the basis of cultures from bone and surrounding soft tissue that were positive for MSSA or MRSA. The complete medical records of the patient were then obtained. We analyzed only patients with a positive Staphylococcus aureus culture who had been diagnosed with osteomyelitis on the basis of the overall clinical impression, laboratory test results, culture results, and radiographs.

One hundred and sixty patients with a positive culture for MSSA and thirteen patients with a positive culture for MRSA were identified. Thirty-seven of the patients with MSSA and one of the patients with MRSA were excluded because the medical records were incomplete or because peripheral blood had not been obtained for laboratory testing, in order to avoid bias associated with incomplete data. In addition, one patient with MSSA and one patient with MRSA were excluded because they were immunocompromised. Finally, four patients with MSSA were excluded because they were younger than one year of age (since the ranges of normal laboratory values for such children are substantially different from those of older children). Data analyzed for each patient included age, sex, duration of symptoms, weight-bearing status, antibiotic use at the time of presentation, history of hospitalization in the past six months, place of residence, past medical history, vital signs at the time of presentation (tympanic membrane temperature, heart rate, and blood pressure), serum white blood-cell count (WBC) and differential, hematocrit, platelet count, erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP) level. Our hospital measures the C-reactive protein level with use of a particle-enhanced immunoturbidimetric assay performed on a Roche Cobas C analyzer (Roche Diagnostics, Indianapolis, Indiana), which has been shown to provide sensitive and accurate measurement of the C-reactive protein level^18-20. This assay was standardized against the BCR470/CRM470 standard (Reference Preparation for Proteins in Human Serum) from the Institute for Reference Materials and Measurement (Geel, Belgium)²¹. Children were classified as having hospital-associated MRSA if they had been hospitalized within six months prior to admission, were a resident in a long-term-care facility, were receiving dialysis, or had a permanent indwelling catheter^22,23. Children who did not possess any of these risk factors were classified as having community-associated MRSA.

Statistical Methods

Univariate analysis was used to compare the MSSA and the MRSA osteomyelitis group with respect to sixteen variables: age, sex, duration of symptoms, weight-bearing status, antibiotic use at the time of presentation, history of prior hospitalization, body temperature, hematocrit, heart rate, systolic blood pressure, diastolic blood pressure, platelet count, absolute neutrophil count, erythrocyte sedimentation rate, white blood-cell count, C-reactive protein level, and location of the osteomyelitis. Continuous variables were analyzed with use of a two-sample Student t test and reported as the mean and the standard deviation, with the exception of the duration of symptoms and the C-reactive protein level; these two variables did not follow a normal distribution and were therefore analyzed with use of the Mann-Whitney U test and reported as the median and the interquartile range. The Pearson chi-square test was used to compare the groups with respect to the location of the osteomyelitis (i.e., upper extremity, lower extremity, or spine). A p value of 0.05 was considered significant; a Bonferroni adjustment for multiple comparisons was not performed since we were comparing only two study groups. Multivariate logistic regression was then applied to identify the significant independent predictors of MRSA infection by considering candidate variables with p values of <0.05 in the univariate analysis²⁴. A backward stepwise selection procedure was used to establish the final multivariate model, and the likelihood ratio test was used to assess the significance of the results²⁵. Receiver operating characteristic curve analysis was performed for each multivariate predictor to determine the area under the curve. The optimal cutoff value (i.e., threshold) for each predictor was chosen with use of the Youden index, which identifies the point on a receiver operating characteristic curve that is farthest from the line of nondiscrimination²⁶. Thus, the cutoff value for each predictor variable was optimized to maximize sensitivity (i.e., the probability of correctly classifying patients with MRSA) while minimizing 1–specificity (i.e., the probability of incorrectly classifying patients with MSSA), as described in detail by Pepe²⁷. Next, multivariate logistic regression was used to calculate the probability of MRSA infection (and the associated 95% confidence interval [CI]) for each of the sixteen possible combinations of positive and negative predictors²⁸. A simplified model for estimating the probability of MRSA infection, in which the logistic regression was based on the number of positive predictors rather than on each possible combination, was also evaluated. The Pearson chi-square test for trend was calculated to assess the strength of the relationship between an increasing number of predictors and an increasing percentage of patients who had MRSA osteomyelitis. Statistical analysis was performed with use of SPSS software (version 18.0; SPSS/IBM, Chicago, Illinois). A two-tailed p value of <0.05 was considered significant.

Source of Funding

No external funding was received for this study.

Results

Eleven (9%) of the 129 patients with osteomyelitis had an MRSA infection, and 118 (91%) had an MSSA infection. The prevalence of MRSA infection among patients with osteomyelitis at our institution increased from 2000 to 2009, as was the trend at many other institutions. During the first half of the study period (2000 through 2004), 5.5% of the osteomyelitis cases were due to MRSA, but this percentage increased to 12.1% during the period from 2005 through 2009. Demographic characteristics, including age (p = 0.85) and sex (p = 0.16), were similar in the two groups. The average age of the patients was eleven years (range, one to eighteen years), and ninety-five (74%) of the patients were male. MRSA osteomyelitis was located in the upper extremity in one patient, in the lower extremity in eight, and in the spine in two. MSSA osteomyelitis was located in the upper extremity in sixteen patients, in the lower extremity in ninety-five, and in the spine in seven. The location of the osteomyelitis was not correlated with the organism responsible (chi-square = 2.39 [two degrees of freedom], p = 0.30). The percentage of patients who were taking antibiotics at the time of presentation was higher in the MRSA group than in the MSSA group (45% compared with 17%, p = 0.04), and the percentage of patients who were not able to bear any weight across the affected bone was also higher in the MRSA group (82% compared with 31%, p < 0.01) (Table I). Comparison of the two study groups with use of univariate analysis revealed that patients with MRSA had a significantly lower hematocrit value (p < 0.01) and higher values for seven other variables: body temperature (p < 0.001), heart rate (p < 0.01), serum white blood-cell count (p < 0.001), absolute neutrophil count (p <0.01), platelet count (p < 0.001), erythrocyte sedimentation rate (p < 0.001), and C-reactive protein level (p < 0.001). The duration of symptoms (p = 0.98) and the systolic (p = 0.90) and diastolic (p = 0.24) blood pressure did not differ significantly between the groups.

View Larger

TABLE I Univariate Comparison Between the MRSA and MSSA Osteomyelitis Groups*

Variable	MSSA Group (N = 118)	MRSA Group (N = 11)	P Value
Age†(yr)	11.2 ± 3.9	11.5 ± 5.9	0.85
Sex			0.16
Male	89 (75%)	6 (55%)
Female	29 (25%)	5 (45%)
Duration of symptoms‡(d)	7 (4 to 28)	14 (3 to 28)	0.98
Prior hospitalization	13 (11%)	3 (27%)	0.14
Antibiotic use at presentation	20 (17%)	5 (45%)	0.04
Non-weight-bearing	36 (31%)	9 (82%)	<0.01
Temperature†(°C)	37.2 ± 0.9	38.8 ± 1.3	<0.001
Heart rate†(bpm)	92 ± 22	114 ± 20	<0.01
Systolic blood pressure†(mm Hg)	115 ± 13	116 ± 18	0.90
Diastolic blood pressure†(mm Hg)	65 ± 11	69 ± 17	0.24
Serum white blood-cell count†(× 103 cells/μL)	9.31 ± 3.68	16.92 ± 5.87	<0.001
Absolute neutrophil count†(× 103 cells/μL)	7.3 ± 6.9	13.1 ± 5.9	<0.01
Hematocrit†(%)	36.0 ± 3.5	32.5 ± 3.3	<0.01
Platelet count†(× 103 cells/μL)	328 ± 134	514 ± 248	<0.001
Erythrocyte sedimentation rate†(mm/hr)	44.9 ± 30.6	93.2 ± 32.1	<0.001
C-reactive protein‡(mg/L)	3.8 (0.8 to 8.9)	19.5 (13.8 to 34.1)	<0.001

*MRSA = methicillin-resistant Staphylococcus aureus, and MSSA = methicillin-sensitive Staphylococcus aureus.
†Values are given as the mean and the standard deviation.
‡Values are given as the median, with the interquartile range in parentheses.

Add to My JBJS

TABLE I Univariate Comparison Between the MRSA and MSSA Osteomyelitis Groups*

Variable	MSSA Group (N = 118)	MRSA Group (N = 11)	P Value
Age†(yr)	11.2 ± 3.9	11.5 ± 5.9	0.85
Sex			0.16
Male	89 (75%)	6 (55%)
Female	29 (25%)	5 (45%)
Duration of symptoms‡(d)	7 (4 to 28)	14 (3 to 28)	0.98
Prior hospitalization	13 (11%)	3 (27%)	0.14
Antibiotic use at presentation	20 (17%)	5 (45%)	0.04
Non-weight-bearing	36 (31%)	9 (82%)	<0.01
Temperature†(°C)	37.2 ± 0.9	38.8 ± 1.3	<0.001
Heart rate†(bpm)	92 ± 22	114 ± 20	<0.01
Systolic blood pressure†(mm Hg)	115 ± 13	116 ± 18	0.90
Diastolic blood pressure†(mm Hg)	65 ± 11	69 ± 17	0.24
Serum white blood-cell count†(× 103 cells/μL)	9.31 ± 3.68	16.92 ± 5.87	<0.001
Absolute neutrophil count†(× 103 cells/μL)	7.3 ± 6.9	13.1 ± 5.9	<0.01
Hematocrit†(%)	36.0 ± 3.5	32.5 ± 3.3	<0.01
Platelet count†(× 103 cells/μL)	328 ± 134	514 ± 248	<0.001
Erythrocyte sedimentation rate†(mm/hr)	44.9 ± 30.6	93.2 ± 32.1	<0.001
C-reactive protein‡(mg/L)	3.8 (0.8 to 8.9)	19.5 (13.8 to 34.1)	<0.001

Multivariate logistic regression analysis confirmed that four of these variables were independent predictors of MRSA infection, including temperature (p < 0.001), hematocrit value (p = 0.003), white blood-cell count (p < 0.001), and C-reactive protein level (p < 0.001). Three of the eleven MRSA infections were hospital-associated, but these cases did not differ significantly from the eight community-associated MRSA cases with respect to these four predictors (all p values > 0.05). Receiver operating characteristic curve analyses indicated that the area under the curve ranged from 0.770 to 0.902 for these four predictors, indicating that all four possessed excellent diagnostic accuracy in differentiating MRSA from MSSA osteomyelitis (Table II). The optimal cutoff values for maximizing sensitivity and specificity were a temperature of >38°C, a hematocrit value of <34%, a serum white blood-cell count of >12,000 cells/μL, and a C-reactive protein level of >13 mg/L. The two groups differed markedly in the percentage of patients above the cutoff values for temperature, white blood-cell count, and C-reactive protein and in the percentage of patients below the cutoff value for hematocrit (Fig. 1). Nine (82%) of the eleven patients with MRSA osteomyelitis were above the optimal cutoff values for temperature, for the white blood-cell count, and for the C-reactive protein level, whereas only 19% of the patients with MSSA osteomyelitis were above the cutoff value for temperature, 16% were above the cutoff for the white blood-cell count, and 11% were above the cutoff for the C-reactive protein level (all p < 0.001). Eight (73%) of the patients with MRSA osteomyelitis were below the cutoff for hematocrit compared with thirty (25%) of the patients with MSSA osteomyelitis (p = 0.003).

View Larger

TABLE II Area Under the Curve and Cutoff Value for Each Multivariate Predictor of MRSA Osteomyelitis*

Predictor	Area Under the Curve	95% CI	P Value	Selected Cutoff Value
Temperature	0.824	0.641 to 0.999	<0.001	>38°C
Hematocrit	0.770	0.627 to 0.913	0.003	<34%
White blood-cell count	0.902	0.839 to 0.965	<0.001	>12,000 cells/μL
C-reactive protein	0.830	0.672 to 0.988	<0.001	>13 mg/L

*Based on receiver operating characteristic curve analysis. MRSA = methicillin-resistant Staphylococcus aureus, and CI = confidence interval.

Add to My JBJS

TABLE II Area Under the Curve and Cutoff Value for Each Multivariate Predictor of MRSA Osteomyelitis*

Predictor	Area Under the Curve	95% CI	P Value	Selected Cutoff Value
Temperature	0.824	0.641 to 0.999	<0.001	>38°C
Hematocrit	0.770	0.627 to 0.913	0.003	<34%
White blood-cell count	0.902	0.839 to 0.965	<0.001	>12,000 cells/μL
C-reactive protein	0.830	0.672 to 0.988	<0.001	>13 mg/L

*Based on receiver operating characteristic curve analysis. MRSA = methicillin-resistant Staphylococcus aureus, and CI = confidence interval.

Fig. 1

Histogram showing the percentage of patients in the MSSA (methicillin-sensitive Staphylococcus aureus) and MRSA (methicillin-resistant Staphylococcus aureus) groups with positive clinical predictors (body temperature >38°C, hematocrit <34%, white blood-cell count [WBC] >12,000 cells/μL, and C-reactive protein [CRP] level >13 mg/L, representing the optimal cutoff value for each predictor as determined by multivariate analysis).

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

We then evaluated a model that considered only whether the value of a particular predictor was above or below the appropriate cutoff value. Again, multivariate logistic regression analysis showed that each of the four previously identified predictors remained a significant independent predictor of MRSA infection. The likelihood of MRSA infection was 6.32 times higher if the temperature was >38°C than if it was not (p = 0.01), the likelihood was 4.96 times higher if the hematocrit value was <34% (p = 0.02), the likelihood was 3.85 times higher if the white blood-cell count was >12,000 cells/μL (p = 0.04), and the likelihood was 3.90 times higher if the C-reactive protein level was >13 mg/L (p = 0.04).

This final multivariate model containing the four dichotomous predictors was then used to develop a prediction algorithm for the MRSA probability based on the sixteen (2⁴) possible combinations of positive and negative predictor values (see Appendix). The predicted probability of MRSA infection ranged from 91% (95% CI, 70% to 98%) for patients who had all four predictors to 0.1% (95% CI, 0% to 2%) for patients who had none of the four predictors. Finally, regression modeling was used to develop a simplified algorithm that considered only the number of multivariate predictors that were positive; this algorithm was very similar to the previous one but was based on only five possible outcomes rather than sixteen (Table III).

View Larger

TABLE III Derivation of the Simplified Algorithm for the Predicted Probability of MRSA Infection on the Basis of the Number of Multivariate Predictors*

No. of Predictors†	Patients with MSSA (N = 118)	Patients with MRSA (N = 11)	Probability of MRSA (%)	95% CI (%)
0	63 (53%)	0 (0%)	0	0 to 3
1	30 (25%)	1 (9%)	1	0 to 20
2	20 (17%)	2 (18%)	10	2 to 40
3	5 (4%)	2 (18%)	45	16 to 78
4	0 (0%)	6 (55%)	92	70 to 98

*MRSA = methicillin-resistant Staphylococcus aureus, MSSA = methicillin-sensitive Staphylococcus aureus, and CI = confidence interval.
†The four predictors are body temperature >38°C, hematocrit < 34%, serum white blood-cell count > 12.0 × 10³ cells/μL, and C-reactive protein >13 mg/L.

Add to My JBJS

TABLE III Derivation of the Simplified Algorithm for the Predicted Probability of MRSA Infection on the Basis of the Number of Multivariate Predictors*

No. of Predictors†	Patients with MSSA (N = 118)	Patients with MRSA (N = 11)	Probability of MRSA (%)	95% CI (%)
0	63 (53%)	0 (0%)	0	0 to 3
1	30 (25%)	1 (9%)	1	0 to 20
2	20 (17%)	2 (18%)	10	2 to 40
3	5 (4%)	2 (18%)	45	16 to 78
4	0 (0%)	6 (55%)	92	70 to 98

Six (55%) of the eleven patients with an MRSA infection had all four predictors, two (18%) had three predictors, two (18%) had two predictors, one (9%) had one predictor, and zero had none of the four predictors. On the other hand, none of the 118 patients with an MSSA infection had all four predictors, five (4%) had three predictors, twenty (17%) had two predictors, thirty (25%) had one predictor, and sixty-three (53%) had none of the four predictors (Table III, Fig. 2). The Pearson chi-square test for trend indicated a very strong relationship between the number of predictors and the percentage of patients who had an MRSA infection (chi-square = 76.63 [four degrees of freedom], p < 0.0001).

Fig. 2

Histogram showing the percentage of patients in the MSSA (methicillin-sensitive Staphylococcus aureus) and MRSA (methicillin-resistant Staphylococcus aureus) groups with zero, one, two, three, or all four positive clinical predictors. The Pearson chi-square test indicated a strong relationship between the number of positive predictors and the percentage of patients who had an MRSA infection (chi-square = 76.63, p < 0.0001).

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

Receiver operating characteristic curve analysis was used to evaluate the predictive accuracy of the number of multivariate clinical predictors, and the area under the curve (0.944; 95% CI, 0.880 to 1.000) demonstrated outstanding diagnostic performance (Fig. 3). The curve showed a steep shoulder close to the upper left corner, indicative of a very useful prediction rule with an excellent tradeoff between the true-positive rate (signal) and the false-positive rate (noise). Our algorithm did an excellent job in distinguishing MRSA from MSSA osteomyelitis. For example, the operating points along the receiver operating characteristic curve indicated that the presence of any two of the four multivariate predictors had a sensitivity of 0.91 (i.e., a true-positive rate of 91%) and a specificity of 0.79 (i.e., a false-positive rate of 21%) for predicting the presence of an MRSA infection.

Fig. 3

Receiver operating characteristic curve for the simplified MRSA (methicillin-resistant Staphylococcus aureus) prediction algorithm. The area under the curve was 0.944, indicating excellent diagnostic performance for the set of four clinical predictors (temperature, hematocrit, serum white blood-cell count, and C-reactive protein). The 45° dashed line represents the line of nondiscrimination, corresponding to an uninformative algorithm.

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

Discussion

Osteomyelitis is among the most common invasive bacterial infections in the pediatric population. Although historically most cases have been caused by MSSA, over the past decade an increasing number of cases have been caused by MRSA^3-5, which is associated with greater morbidity and mortality. Both pathogens can produce a similar clinical picture initially, with fever, pain, and loss of function of the affected extremity. Developing a way to differentiate between the two bacterial strains is important in order to select appropriate antibiotics in a timely manner and maintain vigilance against the possible complications associated with MRSA infection.

Univariate analysis revealed several significant differences between patients with MRSA osteomyelitis and those with MSSA osteomyelitis. In general, children with MRSA osteomyelitis appeared more seriously ill, with a higher temperature, greater tachycardia, more common inability to bear weight, greater leukocytosis and thrombocytosis, greater elevation in the erythrocyte sedimentation rate and the C-reactive protein level, and a lower hematocrit value. Children with MRSA were also more likely to be taking an antibiotic at the time of presentation. However, the antibiotic regimen may not have been active against MRSA. Finally, the mean duration of symptoms prior to presentation appeared to be longer in the MRSA group than in the MSSA group. However, given the p value of 0.98, it is extremely unlikely that this longer duration represented an actual difference. Furthermore, the interquartile ranges of the two groups were equally wide (four to twenty-eight days in the MSSA group and three to twenty-eight days in the MRSA group), demonstrating the variability in the duration of symptoms reported by the patient's parents. Thus, the observed difference in the mean duration of symptoms between the two groups is most likely due to random chance.

Some of the clinical parameters that were found to be significantly associated with MRSA infection in the univariate analysis (Table I) could conceivably respond similarly to an infection (i.e., the phenomenon of covariance), which would obviate the need to use all of them in a prediction algorithm. Therefore, we subsequently performed multivariate logistic regression to identify independent predictors of MRSA infection. We were able to identify four powerful independent predictors that, as a group, afforded excellent diagnostic performance in differentiating between MRSA and MSSA osteomyelitis in children. We then determined an optimal cutoff value for each of the four predictors, with the use of receiver operating characteristic curve analysis, in order to make the algorithm more user-friendly by translating four continuous variables into four dichotomous variables while maintaining good discrimination between the two groups. Over 50% of the patients with an MRSA infection had all four of the predictors, whereas none had zero predictors. Likewise, over 50% of the patients with an MSSA infection had zero of the four predictors, and none had all four (Fig. 2).

We then used these dichotomous predictors to construct two algorithms for predicting the probability of MRSA osteomyelitis, one using all sixteen possible combinations of the four predictors (see Appendix) and a simplified one using just the number of predictors that a patient possessed (Table III). The simplified algorithm indicated a very high probability of MRSA infection (92%) if a patient possessed all four predictors, an intermediate probability if two or three predictors were present (10% and 45%, respectively), and a very low probability if zero or one predictors were present (0% and 1%, respectively). It should be noted that these predicted probabilities of MRSA osteomyelitis do not depend on the prevalence of MRSA osteomyelitis. We are not stating positive predictive values, which do depend on the estimated prevalence and on the Bayes theorem, because we cannot assume that our study population represented a random sample of the general population.

The management of patients on the basis of the predicted risk of MRSA osteomyelitis needs to take into account the consequences of false-positive and false-negative diagnoses. Thus, the following proposed management scheme may be appropriate. Before any antibiotics are administered, we advise obtaining blood cultures and, whenever possible, cultures from the site of the infection; these are used to determine antibiotic sensitivity, and the antibiotic selection can be adjusted once these data become available. Patients who have a very high probability of MRSA osteomyelitis (i.e., four predictors) should be started immediately on antibiotics active against MRSA, and they should be monitored for sequelae associated with MRSA infection. The same management strategy may also be appropriate for patients who have an intermediate probability of MRSA osteomyelitis (i.e., two or three predictors), because of the morbidity associated with not appropriately treating MRSA osteomyelitis. However, clinical suspicion and the local prevalence of MRSA will likely inform this decision. Finally, for patients with a very low probability of MRSA osteomyelitis (i.e., zero or one predictor), it would likely be appropriate to use antibiotics that target MSSA. In fact, on the basis of this algorithm, 74% of the patients in our study could have initially received an antibiotic that was not as broad-spectrum, since they only possessed zero or one of the predictors. Regardless of the initial antibiotic regimen chosen, following the patient closely for signs of clinical improvement, or the lack thereof, remains crucial.

In the past, MRSA was predominantly a nosocomial pathogen seen in patients with certain health care exposures²⁹. However, in recent years there has been a surge in reports from around the world documenting the appearance of MRSA infection in previously healthy children in the community^30-34, and these children often lacked the classic MRSA risk factors^4,35,36. The majority of MRSA infections in children involve the skin and soft tissues, with cases of osteomyelitis representing only about 6% of all MRSA infections³³. Therefore, it is not surprising that published reports comparing the clinical characteristics and risk factors associated with MRSA and MSSA infections involving the skin and soft tissues^35,37-40 are much more common than reports involving osteomyelitis^14,23,41. Our study, in conjunction with previous work involving MRSA osteomyelitis, helps to shed more light on this increasingly important clinical entity.

Our findings are generally similar to those of previous studies involving MRSA and MSSA osteomyelitis. Hawkshead et al.¹⁴ retrospectively reviewed ninety-seven children with osteomyelitis to determine whether those with an infection caused by MRSA (n = 11) differed clinically from those with an infection caused by MSSA (n = 14), a non-Staphylococcus aureus pathogen (n = 15), or an organism that could not be cultured (n = 34). As in our study, MRSA osteomyelitis was associated with a significantly higher white blood-cell count, C-reactive protein level, and erythrocyte sedimentation rate at admission compared with osteomyelitis due to MSSA or other pathogens. Saavedra-Lozano et al.⁴¹ retrospectively reviewed 290 children with culture-positive or culture-negative osteomyelitis. As part of their study, they compared patients with MSSA (n = 72) and MRSA (n = 36) osteomyelitis with respect to differences at presentation. In accordance with our results, they found no significant difference between the two groups with respect to age, sex, duration of symptoms before admission, or antibiotic use. In addition, they reported that a significantly greater percentage of patients with MRSA osteomyelitis were anemic, had a C-reactive protein level of >4 mg/L, and had an erythrocyte sedimentation rate of >40 mm/hr at presentation compared with patients with MSSA osteomyelitis, which also accords with our findings. Finally, Erdem et al. recently published a retrospective study comparing the characteristics and outcomes of sixty-two Hawaiian children with culture-proven osteomyelitis, which was caused by MSSA in forty-seven children and by MRSA in fifteen²³. The groups did not differ significantly with respect to age, sex, or the duration of illness prior to admission. While Erdem et al. did not report the erythrocyte sedimentation rate and C-reactive protein level at the time of admission, the peak value of each of these parameters was significantly higher in the MRSA group than in the MSSA group. In addition, both the white blood-cell count and the absolute neutrophil count showed a trend toward a higher level at admission and a higher peak level in the MRSA group, although the differences were not significant. While the general findings of Erdem et al. support our results, it should be noted that 59% of their subjects were of Native Hawaiian or Pacific Islander descent and thus may not be demographically comparable with our population. Overall, the results of our study are consistent with previous reports regarding MRSA and MSSA osteomyelitis, and we have developed a unique algorithm to accurately predict the probability of MRSA osteomyelitis.

The limitations of our study include its retrospective case-control design, which is theoretically subject to bias. However, we avoided information bias associated with analysis of incomplete data by excluding patients who had incomplete medical records or whose peripheral blood had not undergone laboratory testing. Furthermore, we avoided selection bias associated with the inclusion of patients with presumptive and inconsistent diagnoses by excluding patients who did not have an explicit, culture-proven diagnosis of Staphylococcus aureus osteomyelitis. However, we were not able to determine the genotypes of the individual MRSA and MSSA isolates because of the retrospective nature of the study, and we cannot therefore be certain that the genotypes of these two organisms did not change during our study period. Another limitation of our study is the relatively low number of MRSA osteomyelitis cases that were included in the analysis. However, multivariate analysis identified four overwhelmingly significant independent predictors of MRSA osteomyelitis (Table II), confirming the associations determined by our univariate analysis and minimizing the likelihood that the associations were due to confounders or were false positives.

The results of our statistical analyses, combined with the clear differences between the MRSA and MSSA groups with respect to the four predictors, make it unlikely that these findings are spurious. Nevertheless, as with any new clinical prediction rule, follow-up studies should be conducted to evaluate the accuracy and utility of the prediction algorithm before it is put into clinical practice. Therefore, we will prospectively validate the ability of our algorithm to identify cases of MRSA osteomyelitis, both at our own institution and in new clinical settings, and we will then investigate whether the resulting information leads to more timely selection of an appropriate antibiotic and improved clinical outcomes.

In an era in which greater emphasis is being placed on evidence-based medicine, clinical prediction rules are one of the tools in a physician's armamentarium for making more informed and accurate clinical decisions. We previously created a clinical prediction rule for distinguishing between septic arthritis and transient synovitis of the hip in children⁴², then validated it in a new population⁴³, and finally developed a clinical practice guideline for the treatment of septic arthritis in children⁴⁴. We have now taken the initial step in differentiating between MRSA and MSSA osteomyelitis in children by developing our algorithm to predict the probability of MRSA infection, which will ultimately help to facilitate the timely selection of appropriate antibiotics and to maintain vigilance against the complications associated with MRSA infection.

Appendix

A table showing the probability of MRSA osteomyelitis associated with each possible combination of the four clinical predictors is available with the online version of this article as a data supplement at jbjs.org.

References

Blyth MJ; Kincaid R; Craigen MA; Bennet GC. The changing epidemiology of acute and subacute haematogenous osteomyelitis in children. J Bone Joint Surg Br. 2001;83:99-102.[CrossRef][PubMed]

Craigen MA; Watters J; Hackett JS. The changing epidemiology of osteomyelitis in children. J Bone Joint Surg Br. 1992;74:541-5.[PubMed]

Arnold SR; Elias D; Buckingham SC; Thomas ED; Novais E; Arkader A; Howard C. Changing patterns of acute hematogenous osteomyelitis and septic arthritis: emergence of community-associated methicillin-resistant Staphylococcus aureus. J Pediatr Orthop. 2006;26:703-8.[CrossRef][PubMed]

Herold BC; Immergluck LC; Maranan MC; Lauderdale DS; Gaskin RE; Boyle-Vavra S; Leitch CD; Daum RS. Community-acquired methicillin-resistant Staphylococcus aureus in children with no identified predisposing risk. JAMA. 199825;279:593-8.[CrossRef]

Jarvis WR; Thornsberry C; Boyce J; Hughes JM. Methicillin-resistant Staphylococcus aureus at children's hospitals in the United States. Pediatr Infect Dis. 1985;4:651-5.[CrossRef][PubMed]

Aronoff SC. Antimicrobials in children and the problem of drug resistance. Am Fam Physician. 1996;54:44-6, 54, 6.[PubMed]

Altemeier WA 3rd. A pediatrician's view. More rifle, less shotgun. Pediatr Ann. 1996;25:474, 87, 500.[PubMed]

Appelbaum PC. MRSA—the tip of the iceberg. Clin Microbiol Infect. 2006;12 Suppl 2:3-10.[CrossRef][PubMed]

Carmeli Y; Samore MH; Huskins C. The association between antecedent vancomycin treatment and hospital-acquired vancomycin-resistant enterococci: a meta-analysis. Arch Intern Med. 1999;159:2461-8.[CrossRef][PubMed]

de Bruin MA; Riley LW. Does vancomycin prescribing intervention affect vancomycin-resistant enterococcus infection and colonization in hospitals? A systematic review. BMC Infect Dis. 2007;7:24.[CrossRef][PubMed]

Fokas S; Fokas S; Tsironi M; Kalkani M; Dionysopouloy M. Prevalence of inducible clindamycin resistance in macrolide-resistant Staphylococcus spp. Clin Microbiol Infect. 2005;11:337-40.[CrossRef][PubMed]

Azap OK; Arslan H; Timurkaynak F; Yapar G; Oruç E; Gagir U. Incidence of inducible clindamycin resistance in staphylococci: first results from Turkey. Clin Microbiol Infect. 2005;11:582-4.[CrossRef][PubMed]

Martínez-Aguilar G; Avalos-Mishaan A; Hulten K; Hammerman W; Mason EO Jr; Kaplan SL. Community-acquired, methicillin-resistant and methicillin-susceptible Staphylococcus aureus musculoskeletal infections in children. Pediatr Infect Dis J. 2004;23:701-6.[CrossRef][PubMed]

Hawkshead JJ 3rd; Patel NB; Steele RW; Heinrich SD. Comparative severity of pediatric osteomyelitis attributable to methicillin-resistant versus methicillin-sensitive Staphylococcus aureus. J Pediatr Orthop. 2009;29:85-90.[CrossRef][PubMed]

Cosgrove SE; Sakoulas G; Perencevich EN; Schwaber MJ; Karchmer AW; Carmeli Y. Comparison of mortality associated with methicillin-resistant and methicillin-susceptible Staphylococcus aureus bacteremia: a meta-analysis. Clin Infect Dis. 2003;36:53-9.[CrossRef][PubMed]

Gonzalez BE; Teruya J; Mahoney DH Jr; Hulten KG; Edwards R; Lamberth LB; Hammerman WA; Mason EO Jr; Kaplan SL. Venous thrombosis associated with staphylococcal osteomyelitis in children. Pediatrics. 2006;117:1673-9.[CrossRef][PubMed]

Hollmig ST; Copley LA; Browne RH; Grande LM; Wilson PL. Deep venous thrombosis associated with osteomyelitis in children. J Bone Joint Surg Am. 2007;89:1517-23.[CrossRef][PubMed]

Senju O; Takagi Y; Uzawa R; Iwasaki Y; Suzuki T; Gomi K; Ishii T. A new immuno quantitative method by latex agglutination—application for the determination of serum C-reactive protein (CRP) and its clinical significance. J Clin Lab Immunol. 1986;19:99-103.[PubMed]

Price CP; Trull AK; Berry D; Gorman EG. Development and validation of a particle-enhanced turbidimetric immunoassay for C-reactive protein. J Immunol Methods. 1987;99:205-11.[CrossRef][PubMed]

Eda S; Kaufmann J; Roos W; Pohl S. Development of a new microparticle-enhanced turbidimetric assay for C-reactive protein with superior features in analytical sensitivity and dynamic range. J Clin Lab Anal. 1998;12:137-44.[CrossRef][PubMed]

Baudner S; Bienvenu J; Blirup-Jensen S; Carlstrom A; Johnson A; Ward-Milford A. The Certification of a Matrix Reference Material for Immunochemical Measurement of 14 Human Serum Proteins. CRM470, Report EUR 15243 EN. 1993:1-186.

Mongkolrattanothai K; Aldag JC; Mankin P; Gray BM. Epidemiology of community-onset Staphylococcus aureus infections in pediatric patients: an experience at a Children's Hospital in central Illinois. BMC Infect Dis. 2009;9:112.[CrossRef][PubMed]

Erdem G; Salazar R; Kimata C; Simasathien T; Len KA; Bergert L; Melish M. Staphylococcus aureus osteomyelitis in Hawaii. Clin Pediatr (Phila). 2010;49:477-84.[CrossRef][PubMed]

Katz MH. Multivariable analysis: a practical guide for clinicians. 2nd ed. New York: Cambridge University Press; 2006.

Vittinghoff E; McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol. 2007;165:710-8.[CrossRef][PubMed]

Perkins NJ; Schisterman EF. The Youden Index and the optimal cut-point corrected for measurement error. Biom J. 2005;47:428-41.[CrossRef][PubMed]

Pepe MS. The statistical evaluation of medical tests for classification and prediction. New York: Oxford University Press; 2004.

Harrell FE. Regression modeling strategies: with applications to linear models, logistic regression, and survival analysis. New York: Springer; 2001.

Doebbeling BN. The epidemiology of methicillin-resistant Staphylococcus aureus colonisation and infection. J Chemother. 1995;7 Suppl 3:99-103.[PubMed]

Bukhari EE; Al-Otaibi FE. Severe community-acquired infection caused by methicillin-resistant Staphylococcus aureus in Saudi Arabian children. Saudi Med J. 2009;30:1595-600.[PubMed]

Kuhfahl KJ; Fasano C; Deitch K. Scapular Abscess, Septic Emboli,and Deep Vein Thrombosis in a Healthy Child Due to Community-Acquired Methicillin-Resistant Staphylococcus aureus: Case Report. Pediatr Emerg Care. 2009;25:677-80.[CrossRef][PubMed]

Rozenbaum R; Sampaio MG; Batista GS; Garibaldi AM; Terra GM; Souza MJ; Vieira EN; Silva-Carvalho MC; Teixeira LA; Figueiredo AM. The first report in Brazil of severe infection caused by community-acquired methicillin-resistant Staphylococcus aureus (CA-MRSA). Braz J Med Biol Res. 2009;42:756-60.[CrossRef][PubMed]

Gerber JS; Coffin SE; Smathers SA; Zaoutis TE. Trends in the incidence of methicillin-resistant Staphylococcus aureus infection in children's hospitals in the United States. Clin Infect Dis. 2009;49:65-71.[CrossRef][PubMed]

Yamagishi Y; Togawa M; Shiomi M. Septic arthritis and acute hematogenous osteomyelitis in childhood at a tertiary hospital in Japan. Pediatr Int. 2009;51:371-6.[CrossRef][PubMed]

Sattler CA; Mason EO Jr; Kaplan SL. Prospective comparison of risk factors and demographic and clinical characteristics of community-acquired, methicillin-resistant versus methicillin-susceptible Staphylococcus aureus infection in children. Pediatr Infect Dis J. 2002;21:910-7.[CrossRef][PubMed]

Kaplan SL; Hulten KG; Gonzalez BE; Hammerman WA; Lamberth L; Versalovic J; Mason EO Jr. Three-year surveillance of community-acquired Staphylococcus aureus infections in children. Clin Infect Dis. 2005;40:1785-91.[CrossRef][PubMed]

Miller LG; Perdreau-Remington F; Bayer AS; Diep B; Tan N; Bharadwa K; Tsui J; Perlroth J; Shay A; Tagudar G; Ibebuogu U; Spellberg B. Clinical and epidemiologic characteristics cannot distinguish community-associated methicillin-resistant Staphylococcus aureus infection from methicillin-susceptible S. aureus infection: a prospective investigation. Clin Infect Dis. 2007;44:471-82.[CrossRef][PubMed]

Skiest DJ; Brown K; Cooper TW; Hoffman-Roberts H; Mussa HR; Elliott AC. Prospective comparison of methicillin-susceptible and methicillin-resistant community-associated Staphylococcus aureus infections in hospitalized patients. J Infect. 2007;54:427-34.[CrossRef][PubMed]

Crawford SE; David MZ; Glikman D; King KJ; Boyle-Vavra S; Daum RS. Clinical importance of purulence in methicillin-resistant Staphylococcus aureus skin and soft tissue infections. J Am Board Fam Med. 2009;22:647-54.[CrossRef][PubMed]

Young DM; Harris HW; Charlebois ED; Chambers H; Campbell A; Perdreau-Remington F; Lee C; Mankani M; Mackersie R; Schecter WP. An epidemic of methicillin-resistant Staphylococcus aureus soft tissue infections among medically underserved patients. Arch Surg. 2004;139:947-53.[CrossRef][PubMed]

Saavedra-Lozano J; Mejías A; Ahmad N; Peromingo E; Ardura MI; Guillen S; Syed A; Cavuoti D; Ramilo O. Changing trends in acute osteomyelitis in children: impact of methicillin-resistant Staphylococcus aureus infections. J Pediatr Orthop. 2008;28:569-75.[CrossRef][PubMed]

Kocher MS; Zurakowski D; Kasser JR. Differentiating between septic arthritis and transient synovitis of the hip in children: an evidence-based clinical prediction algorithm. J Bone Joint Surg Am. 1999;81:1662-70.[PubMed]

Kocher MS; Mandiga R; Zurakowski D; Barnewolt C; Kasser JR. Validation of a clinical prediction rule for the differentiation between septic arthritis and transient synovitis of the hip in children. J Bone Joint Surg Am. 2004;86:1629-35.[PubMed]

Kocher MS; Mandiga R; Murphy JM; Goldmann D; Harper M; Sundel R; Ecklund K; Kasser JR. A clinical practice guideline for treatment of septic arthritis in children: efficacy in improving process of care and effect on outcome of septic arthritis of the hip. J Bone Joint Surg Am. 2003;85:994-9.[CrossRef][PubMed]

Topics

methicillin ; osteomyelitis ; methicillin-resistant staphylococcus aureus ; methicillin susceptible staphylococcus aureus ; methicillin-resistant staphylococcus aureus infection

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

Print
PDF
Email

Recipient(s) will receive an email with a link (good for 5 days) to 'Differentiating Between Methicillin-Resistant and Methicillin-Sensitive Staphylococcus aureus Osteomyelitis in Children: An Evidence-Based Clinical Prediction Algorithm'.
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: 'Differentiating Between Methicillin-Resistant and Methicillin-Sensitive Staphylococcus aureus Osteomyelitis in Children: An Evidence-Based Clinical Prediction Algorithm'
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

Kevin L. Ju, David Zurakowski, Mininder S. Kocher; Differentiating Between Methicillin-Resistant and Methicillin-Sensitive Staphylococcus aureus Osteomyelitis in ChildrenAn Evidence-Based Clinical Prediction Algorithm. The Journal of Bone & Joint Surgery. 2011 Sep;93(18):1693-1701.

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