Skip to main content
NCBI home page
Clinical Orthopaedics and Related Research logoLink to Clinical Orthopaedics and Related Research
. 2017 Dec 21;476(1):153–159. doi: 10.1007/s11999.0000000000000019

Validating an Algorithm to Predict Adjacent Musculoskeletal Infections in Pediatric Patients With Septic Arthritis

Benjamin D Welling 1,2, Lee S Haruno 1,2, Scott B Rosenfeld 1,2,✉
PMCID: PMC5919251  PMID: 29389760

Abstract

Background

Septic arthritis is frequently associated with adjacent infections including osteomyelitis and subperiosteal and intramuscular abscesses. While often clinically indiscernible from isolated septic arthritis, the diagnosis of adjacent infections is important in determining the need for additional surgical intervention. MRI has been used as the diagnostic gold standard for assessing adjacent infection. Routine MRI, however, can be resource-intensive and delay surgical treatment. In this context, there is need for additional diagnostic tools to assist clinicians in determining when to obtain preoperative MRI in children with septic arthritis. In a previous investigation by Rosenfeld et al., an algorithm, based on presenting laboratory values and symptoms, was derived to predict adjacent infections in septic arthritis. The clinical applicability of the algorithm was limited, however, in that it was built from and applied on the same population. The current study was done to address this criticism by evaluating the predictive power of the algorithm on a new patient population.

Questions/purposes

(1) Can a previously created algorithm used for predicting adjacent infection in septic arthritis among pediatric patients be validated in a separate population?

Methods

Records for all pediatric patients (1-18 years old) surgically treated for suspected septic arthritis during a 3-year period were retrospectively reviewed (109 patients). Of these patients, only those with a diagnosis of septic arthritis confirmed by synovial fluid analysis were included in the study population. Patients without confirmation of septic arthritis via synovial fluid analysis, Gram stain, or culture were excluded (34 patients). Patients with absence of MRI, younger than 1 year, insufficient laboratory tests, or confounding concurrent illnesses also were excluded (18 patients), resulting in a total of 57 patients in the study population. Five variables which previously were shown to be associated with risk of adjacent infection were collected: patient age (older than 4 years), duration of symptoms (> 3 days), C-reactive protein (> 8.9 mg/L), platelet count (< 310 x 103 cells/µL), and absolute neutrophil count (> 7.2 x 103 cells/µL). Adjacent infections were determined exclusively by preoperative MRI, with all patients in this study undergoing preoperative MRI. MR images were read by pediatric musculoskeletal radiologists and reviewed by the senior author. According to the algorithm we considered the presence of three or more threshold-level variables as a “positive” result, meaning the patient was predicted to have an adjacent infection. Comparing against the gold standard of MRI, the algorithm’s accuracy was evaluated in terms of sensitivity, specificity, positive predictive value, and negative predictive value.

Results

In the new population, the sensitivity and specificity of the algorithm were 86% (95% CI, 0.70-0.95) and 85% (95% CI, 0.64-0.97), respectively. The positive predictive value was determined to be 91% (95% CI, 0.78-0.97), with a negative predictive value of 77% (95% CI, 0.61-0.89). All patients meeting four or more algorithm criteria were found to have septic arthritis with adjacent infection on MRI.

Conclusions

Critical to the clinical applicability of the above-mentioned algorithm was its validation on a separate population different from the one from which it was built. In this study, the algorithm showed reproducible predictive power when tested on a new population. This model potentially can serve as a useful tool to guide patient risk stratification when determining the likelihood of adjacent infection and need of MRI. This better-informed clinical judgement regarding the need for MRI may yield improvements in patient outcomes, resource allocation, and cost.

Level of Evidence

Level II, diagnostic study.

Introduction

Septic arthritis is a common form of pediatric musculoskeletal infection (affecting between four and 13 per 100,000 children). Up to 60% of septic arthritis cases are complicated by adjacent infections such as intramuscular and subperiosteal abscesses, pyomyositis, and osteomyelitis [5, 7, 10, 12, 23, 24, 27, 30]. Isolated and concomitant septic arthritis often are clinically indistinguishable, with both manifesting as a painful, swollen, and erythematous joint with impaired weightbearing and motion [15, 17, 23]. Treatment of septic arthritis involves urgent surgical irrigation and débridement of the joint and proper intravenous antibiotic therapy [2, 9, 15, 23, 26]. However, if an adjacent infection is not recognized, the surgical treatment of the joint may be inadequate, with an unaddressed intramuscular or intraosseous abscess. This may result in the need for repetitive operations, longer hospital stay, and additional complications [12, 13, 16, 23]. Accurate diagnosis and treatment of adjacent infection in pediatric septic arthritis poses a substantial clinical challenge for treating physicians.

MRI is highly sensitive and specific for identifying adjacent infections in septic arthritis [3, 4, 13, 16, 18, 19, 21, 22]. Currently, MRI is not considered standard in the workup of suspected septic arthritis [23, 24]. However, the noted prevalence of adjacent infections in the setting of septic arthritis has led some to recommend routine MRI during workup [23]. Such has been the trend at our institution. However, obtaining preoperative MRI for all patients with suspected septic arthritis may not be appropriate. Such a practice places a resource burden on healthcare institutions and patients [8, 25, 29, 32]. Additionally, forestalling surgical intervention to obtain MRI for all patients with septic arthritis has the potential to delay treatment in those with isolated septic arthritis. Finally, depending on the age and comfort level of the patient, the use of sedation or general anesthesia during MRI is not without risk, particularly in children younger than 2 to 3 years old [1, 6, 11, 20, 31, 32]. It is in this context that the need for additional diagnostic tools with which to guide the decision for advanced imaging workup becomes evident.

In a previous study by Rosenfeld et al. [30], a predictive algorithm was derived to identify patients with a high likelihood of adjacent infection in septic arthritis and who thereby would benefit from preoperative MRI. Five factors were found to be predictive of adjacent infection: older than 4 years, C-reactive protein (CRP) level greater than 8.9 mg/L, duration of symptoms (including fever, effusion, or impaired weightbearing) more than 3 days, platelet count less than 310 x 103 cells/µL, and absolute neutrophil count greater than 7.2 x 103 cells/µL. Patients with three or more of these factors were found to be at greater risk for having septic arthritis with adjacent infection (Fig. 1). The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and receiver operating characteristic (ROC) c-statistic of this algorithm when three or more factors were met in predicting the presence of adjacent infection were 90%, 67%, 80%, 83%, and 0.891, respectively [30].

Fig. 1.

Fig. 1

Open in a new tab

The proposed algorithm for predicting the presence of septic arthritis with adjacent musculoskeletal infection in pediatric patients is shown. The diagnosis is based on the presence of joint fluid aspiration with one or more of the following: (1) greater than 50,000 nucleated cells per milliliter; (2) bacterial growth in cultures; or (3) positive Gram stain.

A major criticism of the original study, however, was that it used the same population to create and test the algorithm, potentially resulting in an exaggerated predictive ability [30]. Without validation, the integrity of the algorithm was in question. As such, the purpose of the current study is to validate this original algorithm in a separate pediatric patient population, one from which the original algorithm was not derived. Successful validation of the algorithm on a new patient population would strengthen its clinical utility. A validated algorithm could aid the treating physician when determining the need for preoperative MRI with subsequent improvements in patient care, resource allocation, and cost.

Patients and Methods

In this institutional review board-approved, retrospective case series, we examined all pediatric patients (range, 1-18 years old) surgically treated for suspected septic arthritis at a single tertiary care pediatric hospital in the southwestern United States during a period of 3 years. One hundred nine patients were identified through the institution’s surgery database as having undergone operative irrigation and débridement. Of these, only patients with septic arthritis confirmed by synovial fluid analysis were included in the study population. Three variables were used as criteria for positive synovial fluid aspiration: greater than 50,000 nucleated cells (WBC) per milliliter, positive bacterial growth in cultures, or a positive Gram stain. Of the 109 original patients, 35 who underwent irrigation and débridement but did not have septic arthritis as evidenced by negative synovial fluid were excluded. Owing to age-related differences in normal laboratory values, 10 were excluded secondary to being younger than 1 year [14, 24]. Two patients were excluded for not having preoperative MRI. Three were excluded secondary to confounding cancer (Hodgkin’s lymphoma) or infection (sepsis and pneumonia). There was no exclusion based on location of the septic arthritis. Ultimately, a total of 57 patients who underwent surgical open irrigation and débridement for synovial fluid-confirmed septic arthritis were included in this study.

All patients included in the study population were risk-stratified using the five-variable algorithm proposed by Rosenfeld et al. [30] (Fig. 1). Patients with three or more threshold level variables were considered to have an adjacent infection. The algorithm then was tested for sensitivity, specificity, PPV, and NPV using that cutoff. As a secondary aim, we evaluated the rates of adjacent infection corresponding to each of the number of variables satisfied.

The five variables (Fig. 1) previously identified as predictive factors for adjacent infection (age, CRP, duration of symptoms, platelet count, and absolute neutrophil count) were obtained for all patients from the time of presentation. All patients in this study underwent MRI. All MRI reports (including pre- and postcontrast studies) initially were read by pediatric musculoskeletal radiologists and subsequently reviewed by the senior author (SR). MRI interpretations for each patient were reviewed to determine the presence of adjacent infection. Adjacent infection was defined as any of the following findings on MRI: adjacent intramuscular abscess, subperiosteal abscesses, or osteomyelitis.

Intravenous gadolinium was used per institutional imaging protocol whenever an abnormality was noted on precontrast imaging. Osteomyelitis was diagnosed with fluid intensity signals consistent with marrow and/or adjacent cortical soft tissue infiltration or evidence of sinus tracts, cloaca, or sequestrum. Abscesses were indicated by focal hyperintense loculated fluid collections in the intramedullary, subperiosteal, or periarticular soft tissue spaces on fluid-sensitive precontrast sequences. Similarly, postcontrast imaging with gadolinium showing focal hyperintense fluid collections with rim enhancement but absent central enhancement also was consistent with the diagnosis of an abscess. Myositis was evidenced by an infiltrative fluid intensity signal in subcutaneous fat or muscles [17].

We attempted to reduce selection bias resulting from missing information by ensuring all patients included in the analysis had the relevant clinical and imaging data. To ensure a true representation of septic arthritis, it was important that the algorithm be applied to patients who had confirmed septic arthritis via synovial fluid analysis so as not to be confounded by noninfectious entities. For example, it is entirely possible a patient may undergo joint aspiration, MRI, and even subsequent joint irrigation for a transient synovitis. As these patients are likely to satisfy fewer of the aforementioned threshold-level variables, and will not show adjacent infections, to include these patients in the analysis may have incorrectly increased the predictive power of the algorithm. As a result, there were 35 patients who clinically appeared to have septic arthritis and were treated as such, but could not be included in the study because they lacked confirmation on joint fluid aspirate. Therefore, we cannot rule out all spectrum bias because the results of our analysis are applicable only to patients with confirmed septic arthritis. Patients with equivocal joint aspirates may represent a large minority of patients treated for septic arthritis in the hospital setting.

Statistical Analysis

Twenty of 57 (35%) patients had isolated septic arthritis while 37 of the 57 patients (65%) had septic arthritis with adjacent infection. The average age of the patients was 6 ± 5 years Thirty-four of the 57 patients (60%) were male.

Patients were stratified by the number of algorithm factors present. Patients with three or more threshold level variables were considered to have an adjacent infection and need for advanced imaging (MRI) [31]. Sensitivity, specificity, NPV, PPV, and area under the curve (AUC) of the algorithm for predicting septic arthritis with adjacent infection were computed with corresponding 95% CIs. The data analysis was conducted using SAS, Version 9.3 software (SAS Institute, Cary, NC, USA).

Results

The sensitivity and specificity of the algorithm were 86% (95% CI, 0.70-0.95) and 85% (95% CI, 0.64-0.97), respectively. The PPV was determined to be 91% (95% CI, 0.78-0.97) with a NPV of 77% (95% CI, 0.61-0.89) (Table 1). The ROC curve was plotted and the AUC was calculated as a summary measure of diagnostic accuracy yielding a value of 0.93 (Fig. 2).

Table 1.

Algorithm performance

graphic file with name abjs-476-153-g002.jpg

Open in a new tab

Fig. 2.

Fig. 2

Open in a new tab

The ROC curve shows the true positive rate against the https://en.wikipedia.org/wiki/False_positive_ratefalse positive rate at different threshold settings.

As a secondary aim, we evaluated the rates of adjacent infection corresponding to each number of variables satisfied. Of the three patients (5% of the total study population) who met zero algorithm criteria, none had adjacent infection observed on MRI. For those who met one criterion (eight patients [14%]), seven (88%) had isolated septic arthritis and one (13%) had septic arthritis with an adjacent infection. In patients who met two criteria (11 patients [19%]), seven (64%) had an isolated infection and four (36%) had an adjacent infection. For patients with three positive criteria (13 patients [23%]), three (21%) had isolated septic arthritis and 10 (77%) had an adjacent infection. Finally, patients who met four criteria (16 patients [28%]) or five criteria (six patients [11%]) as described by the algorithm had a 100% rate of adjacent infection (Fig. 3).

Fig. 3.

Fig. 3

Open in a new tab

The graph shows the percentages of patients treated for septic arthritis with and without adjacent infection based on the total number of algorithm factors met.

Discussion

Pediatric septic arthritis may be either isolated or associated with adjacent sites of infection. Although the clinical presentations are nearly identical, septic arthritis with adjacent infection may require more-extensive treatment approaches [22]. Given the difficulty in clinical diagnosis, many physicians are turning to routine MRI to assess for adjacent infection. This trend toward routine MRI, while diagnostically sensitive and specific, may delay surgical care, place patients at sedation risks, unduly consume hospital resources, and increase healthcare costs. An algorithm, created by Rosenfeld et al. [30], is a method for risk stratifying patients based on presenting laboratory values and symptoms. Patients who met a threshold of three or more criteria showed higher likelihood of having an adjacent infection with their septic arthritis. A major criticism of this algorithm’s predictive ability was that it was tested on the same population from which it was built, potentially exaggerating its predictive power and calling into question its clinical applicability. The purpose of the current study is to test the predictive ability of the algorithm on a distinct patient population. Our results indicate that the presence of three or more algorithm risk factors predicts an increased likelihood of having adjacent infection. We believe now that the algorithm has been tested on a unique population and the results confirmed that this clinical decision tool could guide physicians regarding whether to obtain preoperative MRI in patients with septic arthritis to better characterize adjacent foci before surgery.

We acknowledge that this algorithm is not perfect and that our study has some limitations. First, although certainly predictive in most cases, there are instances in which adherence to the algorithm may result in some patients who do not undergo MRI having missed adjacent infections. In our study five of 22 patients who had fewer than three criteria did have adjacent infections and use of this algorithm would have resulted in these patients not getting MRI. Although not perfect, we feel that the algorithm remains useful as it can provide objective guidance in what previously has been a completely subjective decision (to obtain MRI or not) with less ambiguity. These patients who fall in the low-risk category (fewer than three risk factors) and do not undergo MRI before surgical débridement should be closely monitored during the postoperative period. If symptoms fail to improve, advanced imaging is warranted to examine for adjacent infection.

Second, in an effort to ensure the presented cohort of patients is representative of having true septic arthritis, only patients with septic arthritis confirmed by synovial fluid analysis, Gram stain, or culture were included in the study. To do otherwise would risk inclusion of patients who did not have septic arthritis. This resulted in the exclusion of 34 patients who underwent irrigation and débridement for suspected septic arthritis but who did not have confirmatory synovial fluid studies. While necessary, we acknowledge the possibility of spectrum bias as patients with negative synovial fluid findings may represent a minority of patients surgically treated for septic arthritis.

Additionally, we acknowledge the possibility of transfer bias. The patients in this study were only followed for the course of their hospital stay. Subsequent treatments for possible missed adjacent infections after discharge were not evaluated. Thus, there is an implicit assumption in this study that a patient with synovial fluid-confirmed septic arthritis with MR images showing no adjacent infection who undergoes surgical irrigation and débridement and has a postoperative course of improvement culminating in discharge is unlikely to later have an adjacent infection. While it is our opinion that such a scenario is unlikely, it cannot be ruled out.

We also acknowledge likely regional bias. As Refakis et al. [28] showed in a validation study of the proposed algorithm conducted in the northeastern United States, this algorithm may perform differently when subjected to regional variations in bacterial epidemiology. Although we have now validated the algorithm in a novel population, regional influence should be further explored before this clinical tool can be widely applied. The findings described from a different region support the need for a prospective multicenter study to increase regional diversity and obtain further data regarding the widespread adoptability of this algorithm [28].

Finally we recognize the inherent limitations of a retrospective study. We believe however, that although retrospective, this study substantially adds to the medical literature in that it addresses that principal criticism of the original study [30], that the algorithm was derived from and then statistically tested on the same population. By assessing it in a new population, naive to the factors that produced the algorithm, and confirming its predictability, we increase the strength of its clinical applicability.

The algorithm performed well when we set a threshold of three positive criteria to indicate high risk of adjacent infection (PPV = 91%). Patients meeting fewer than three criteria therefore are considered at reduced risk for adjacent infection (NPV = 77%). It is notable that in this group, those meeting fewer than three variables, the majority of false-negatives were identified in patients meeting two criteria. We propose that this group be considered “intermediate risk” with the need for MRI determined on a case-by-case basis. Patients meeting fewer than two criteria can be classified as “low risk” and proceed directly to the operating room for joint débridement, with careful observation during the postoperative period to ensure resolution of symptoms.

While formal statistical analysis between the current study and that performed previously [30] are not valid, general comparisons can be made. The current and previous studies showed similar sensitivities at 86% and 90% respectively and improved specificities at 85% and 67% respectively. The current and previous PPVs and NPVs are 91% and 80% respectively and 77% and 83% respectively. The AUC was comparable, with the current study yielding a value of 0.93 and the previous study showing a value of 0.89. When compared with the previous study, the current study showed similar predictive ability of the algorithm when tested on a novel population.

The ability of the algorithm to predict the risk of adjacent infection in patients with synovial fluid-confirmed septic arthritis was confirmed when tested on a new population, thus validating the results of the study by Rosenfeld et al. [30]. Pediatric patients with synovial fluid-confirmed septic arthritis who meet three of five predictive factors (based on patient age, duration of symptoms, CRP, platelet count, and absolute neutrophil count) are at greater risk of having adjacent infections and should undergo preoperative MRI before surgical débridement. Patients with zero or one predictive factor may benefit from immediate surgical treatment without preoperative MRI. Patients with two variables are of intermediate risk and should be approached case by case. While additional studies will be of benefit and improve wide applicability, this algorithm provides the physician with an additional tool in the clinical challenge of treating of pediatric musculoskeletal infections.

Acknowledgments

We thank Ifeoma Inneh MPH, Sha'Tia Brownell MPH, and Theodora Browne BA (Division of Orthopedic Surgery, Texas Children’s Hospital) for assistance in preparation and critical review of the manuscript. We also thank Wei Zhang PhD (Outcomes and Services Department, Texas Children’s Hospital) for assistance with the statistical analysis.

Footnotes

Each author certifies that neither he, nor any member of his immediate family, have funding or commercial associations (consultancies, stock ownership, equity interest, patent/licensing arrangements, etc) that might pose a conflict of interest in connection with the submitted article.

All ICMJE Conflict of Interest Forms for authors and Clinical Orthopaedics and Related Research® editors and board members are on file with the publication and can be viewed on request.

Each author certifies that his institution approved the human protocol for this investigation and that all investigations were conducted in conformity with ethical principles of research.

This work was performed at Texas Children's Hospital, Houston, TX, USA.

References

  • 1.Andropoulos DB, Greene MF. Anesthesia and developing brains: implications of the FDA warning. N Engl J Med. 2017;376:905–907. [DOI] [PubMed] [Google Scholar]
  • 2.Bremell T, Abdelnour A, Tarkowsky A. Histopathological and serological progression of experimental Staphylococcus aureus arthritis. Infect Immun. 1992;60:2976–2985. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Browne LP, Guillerman RP, Orth RC, Patel J, Mason EO, Kaplan SL. Community-acquired staphylococcal musculoskeletal infection in infants and young children: necessity of contrast-enhanced MRI for the diagnosis of growth cartilage involvement. AJR Am J Roentgenol. 2012;198:194–199. [DOI] [PubMed] [Google Scholar]
  • 4.Browne LP, Mason EO, Kaplan SL, Cassady CI, Krishnamurthy R, Guillerman RP. Optimal imaging strategy for community-acquired Staphylococcus aureus musculoskeletal infections in children. Pediatr Radiol. 2008;38:841–847. [DOI] [PubMed] [Google Scholar]
  • 5.Carrillo-Marquez MA, Hulten KG, Hammerman W, Mason EO, Kaplan SL. USA300 is the predominant genotype causing Staphylococcus aureus septic arthritis in children. Pediatr Infect Dis J. 2009;28:1076–1080. [DOI] [PubMed] [Google Scholar]
  • 6.Cauldwell C. Anesthesia risks associated with pediatric imaging. Pediatr Radiol. 2011;41:949–950. [DOI] [PubMed] [Google Scholar]
  • 7.Copley LA. Recent advances in the evaluation and treatment of pediatric musculoskeletal infection. Curr Orthop Pract. 2013;24:586–589. [Google Scholar]
  • 8.Copley LA, Kinsler MA, Gheen T, Shar A, Sun D, Browne R. The impact of evidence-based clinical practice guidelines applied by a multidisciplinary team for the care of children with osteomyelitis. J Bone Joint Surg Am. 2013;95:686–693. [DOI] [PubMed] [Google Scholar]
  • 9.Dodwell ER. Osteomyelitis and septic arthritis in children: current concepts. Curr Opin Pediatr. 2013;25:58–63. [DOI] [PubMed] [Google Scholar]
  • 10.Dohin B, Gillet Y, Kohler R, Gerard L, Vandenesch F, Vanhems P, Floret D, Etienne J. Pediatric bone and joint infections caused by Panton-Valentine leukocidin-positive Staphylococcus aureus. Pediatr Infect Dis J. 2007;26:1042–1048. [DOI] [PubMed] [Google Scholar]
  • 11.Edwards AD, Arthurs OJ. Paediatric MRI under sedation: is it necessary? What is the evidence for the alternatives? Pediatr Radiol. 2011;41:1353–1364. [DOI] [PubMed] [Google Scholar]
  • 12.Gafur OA, Copley LA, Hollmig ST, Browne RH, Thornton LA, Crawford SE. The impact of the current epidemiology of pediatric musculoskeletal infection on evaluation and treatment guidelines. J Pediatr Orthop. 2008;28:777–785. [DOI] [PubMed] [Google Scholar]
  • 13.Guillerman RP. Osteomyelitis and beyond. Pediatr Radiol. 2013;43(suppl 1):S193–S203. [DOI] [PubMed] [Google Scholar]
  • 14.Jackson M, Burry VF, Olson L. Pyogenic arthritis associated with adjacent osteomyelitis. Pediatr Infect Dis J. 1992;11:9–13. [DOI] [PubMed] [Google Scholar]
  • 15.Johnson AH, Campbell WG, Callahan BC. Infection of rabbit knee joints after intra-articular injection of Staphylococcus aureus. Am J Pathol. 1970;60:165–203. [PMC free article] [PubMed] [Google Scholar]
  • 16.Kan JH, Hilmes MA, Martus JE, Yu C, Hernanz-Schulman M. Value of MRI after recent diagnostic or surgical intervention in children with suspected osteomyelitis. AJR Am J Roentgenol. 2008;191:1595–1600. [DOI] [PubMed] [Google Scholar]
  • 17.Kan JH, Young RS, Yu C, Hernanz-Schulman M. Clinical impact of gadolinium in the MRI diagnosis of musculoskeletal infection in children. Pediatr Radiol. 2010;40:1197–1205. [DOI] [PubMed] [Google Scholar]
  • 18.Kim EY, Kwack KS, Cho JH, Lee DH, Yoon SH. Usefulness of dynamic contrast-enhanced MRI in differentiating between septic arthritis and transient synovitis in the hip joint. AJR Am J Roentgenol. 2012;198:428–433. [DOI] [PubMed] [Google Scholar]
  • 19.Kirkhus E, Flato B, Oystein R, Reiseter T, Smith HJ. Differences in MRI findings between subgroups of recent-onset childhood arthritis. Pediatr Radiol. 2011;41:432–440. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Malviya S, Voepel-Lewis T, Eldevik OP, Rockwell DT, Wong JH, Tait AR. Sedation and general anaesthesia in children undergoing MRI and CT: adverse events and outcomes. Br J Anaesth. 2000;84:743–748. [DOI] [PubMed] [Google Scholar]
  • 21.Marin C, Sanchez-Alegre ML, Gallego C, Ruiz Y, Collado E, Garcia JA, Mardones GG. Magnetic resonance imaging of osteoarticular infections in children. Curr Probl Diagn Radiol. 2004;33:42–59. [DOI] [PubMed] [Google Scholar]
  • 22.Mazur JM, Ross G, Cummings RJ, Hahn GA, McCluskey WP. Usefulness of magnetic resonance imaging for the diagnosis of acute musculoskeletal infections in children. J Pediatr Orthop. 1995;15:144–147. [PubMed] [Google Scholar]
  • 23.Monsalve J, Kan JH, Schallert EK, Bisset GS, Zhang W, Rosenfeld SB. Septic arthritis in children: frequency of coexisting unsuspected osteomyelitis and implications on imaging work-up and management. AJR Am J Roentgenol. 2015;204:1289–1295. [DOI] [PubMed] [Google Scholar]
  • 24.Montgomery CO, Siegel E, Blasier RD, Suva LJ. Concurrent septic arthritis and osteomyelitis in children. J Pediatr Orthop. 2013;33:464–467. [DOI] [PubMed] [Google Scholar]
  • 25.Muller AJ, Kwon JK, Steiner JW, Mittal VS, Fernandes NJ, Jo C, Lindsay EA, Copley LA. Improved magnetic resonance imaging utilization for children with musculoskeletal infection. J Bone Joint Surg Am. 2015;97:1869–1876. [DOI] [PubMed] [Google Scholar]
  • 26.Pääkkönen M, Peltola H. Treatment of acute septic arthritis. Pediatr Infect Dis J. 2013;32:684–685. [DOI] [PubMed] [Google Scholar]
  • 27.Perlman MH, Patzakis MJ, Kumar JP, Holtom P. The incidence of joint involvement with adjacent osteomyelitis in pediatric patients. J Pediatr Orthop. 2000;20:40–43. [PubMed] [Google Scholar]
  • 28.Refakis CA, Arkader A, Baldwin KD, Spiegel DA, Sankar WN. Predicting periarticular infection in children with septic arthritis of the hip: regionally derived criteria may not apply to all populations. J Pediatr Orthop. 2017 Jan 5. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]
  • 29.Robinson JC, Whaley C, Brown TT. Reference pricing, consumer cost-sharing, and insurer spending for advanced imaging tests. Med Care. 2016;54:1050–1055. [DOI] [PubMed] [Google Scholar]
  • 30.Rosenfeld S, Bernstein D, Daram S, Dawson J, Zhang W. Predicting the presence of adjacent infections in septic arthritis in children. J Pediatr Orthop. 2016;36:70–74. [DOI] [PubMed] [Google Scholar]
  • 31.Schulte-Uentrop L, Goepfer MS. Anaesthesia or sedation for MRI in children. Curr Opin Anaesthesiol. 2010;23:513–517. [DOI] [PubMed] [Google Scholar]
  • 32.Vanderby SA, Babyn PS, Carter MW, Jewell SM, McKeever PD. Effect of anesthesia and sedation on pediatric MR imaging patient flow. Radiology. 2010;256:229–237. [DOI] [PubMed] [Google Scholar]

Articles from Clinical Orthopaedics and Related Research are provided here courtesy of The Association of Bone and Joint Surgeons

RESOURCES