Secondary Logo

Journal Logo

Original Studies

The Contemporary Epidemiology, Microbiology and Management of Chronic Osteomyelitis in US Children

McNeil, J. Chase MD; Joseph, Marritta RN; Sommer, Lauren M. MS; Vallejo, Jesus G. MD

Author Information
The Pediatric Infectious Disease Journal: June 2021 - Volume 40 - Issue 6 - p 518-524
doi: 10.1097/INF.0000000000003067


Osteomyelitis is among the most common serious bacterial infections encountered by the pediatric infectious disease specialist. The majority of pediatric osteomyelitis present acutely and are of presumed hematogenous origin. A number of recent studies have examined the diagnosis, management and outcomes of acute hematogenous osteomyelitis (AHO) in children.1–5 While a number of pathogens may contribute to AHO, approximately 60% of cases are caused by Staphylococcus aureus.3,6 Data indicate that early transition to oral antimicrobial therapy is associated with good outcomes in children with AHO.1,2 Moreover, the European Society for Pediatric Infectious Diseases has recently published evidence-based guidelines for the management of AHO which support early oral transition in the majority of patients.7

While tremendous strides have been made in the understanding and management of pediatric acute osteomyelitis, a small subset present with a prolonged duration of symptoms often associated with substantial morbidity. Chronic osteomyelitis (CO) in children may result from extension from a contiguous focus of infection (eg, infected decubiti), remote penetrating trauma or surgery or as a late complication of AHO.8 Given the variable pathogenesis, there is the potential for more diverse microbiology than seen with AHO. There are unfortunately very limited data to guide clinicians in the care of these patients, particularly in the era of community-acquired methicillin-resistant S. aureus (MRSA). We examined the clinical features, management and outcomes of pediatric CO at a tertiary children’s hospital.


Patient Identification

We reviewed hospital admissions associated with a diagnosis of CO from January 2011 to December 2018 at Texas Children’s Hospital, a 973-bed tertiary care center. Cases were initially identified by searching for records of hospital admissions with the ICD10 codes M86.4X-6X. Cases were included if: (1) symptoms suggestive of osteomyelitis (eg, pain, swelling, warmth, erythema, drainage, loss of function, etc.) lasted ≥28 days on presentation or (2) there was a clearly documented history of acute osteomyelitis in a patient who received at least 4 weeks of effective antimicrobial therapy along with (a) new or worsening drainage, swelling, erythema, pain or loss of function; (b) radiographic evidence of sequestrum or permeative lucencies; or (c) readmission for the management of osteomyelitis.4 Following the review of medical records, those that did not meet the above criteria and/or were given an alternative diagnosis by the infectious diseases consultant and/or orthopedic surgeon caring for the patient were excluded. Cases with radiologic abnormalities or the presence of exposed bone (either as a result of decubiti or trauma) alone without meeting other clinical criteria were deemed insufficient for diagnosis/inclusion.9 Cases of chronic recurrent multifocal osteomyelitis were excluded. Patients who received their initial diagnosis before the start of the study period or who received initial care at an outside institution were also excluded due to lack of availability of comprehensive medical records. All medical and surgical interventions were at the discretion of the providers of record; while protocols exist at our institution for the initial management of AHO, practice is individualized for CO. When available, bone biopsy/histology results were reviewed but were not utilized as inclusion/exclusion criteria. A subset of patients had tissue specimens sent to the University of Washington Molecular Microbiology Laboratory (Seattle, WA) for 16S ribosomal RNA (rRNA) polymerase chain reaction (PCR); the utilization of this study was at the discretion of the treating clinician. This study was approved by the IRB of Baylor College of Medicine.


Cases were classified into mutually exclusive categories as follows: those associated with: (1) a contiguous focus (CoF, including adjacent decubitus ulcers or chronic wounds); (2) penetrating or open trauma; (3) orthopedic hardware; (4) postacute CO (PACO, those occurring after ≥28 days of therapy for documented acute osteomyelitis); and (5) primary hematogenous CO (PHCO, those with ≥28 days of symptoms at the time of presentation without clear risk factors or documented preceding AHO). Cases of Brodie abscess were included in the category of PHCO assuming they otherwise meet inclusion criteria. A subset of the patients who developed PACO were presented and compared with patients with uncomplicated acute osteomyelitis in previous publications.4,10 Early transition to oral (PO) antibiotics was considered in patients who received ≤2 weeks of parenteral therapy prior to initiating oral antibiotics.11 The presence and type of health insurance was used as a surrogate for socioeconomic status. Treatment failure was regarded as the persistence of signs/symptoms of CO at the time of last follow-up with either infectious diseases or orthopedics. Medical records were reviewed for all patients from the date of admission through September 30, 2020. For patients with PACO, clinical, laboratory and treatment data were extracted from the time of diagnosis with CO. For cases with PACO, the causative organism was regarded as that identified during the index admission provided a different etiology was not identified during subsequent admissions. Coagulase-negative staphylococci were only considered pathogens in those patients with orthopedic hardware in situ or if isolated in ≥2 culture specimens.

Statistical Analyses

Categorical variables were examined with Fisher exact test and continuous variables with Wilcoxon Rank Sum and Kruskal–Wallis tests. The category of CO, early transition to oral antibiotics and total treatment duration ≤25th%-tile for the study group were included a priori in a stepwise logistic regression model for treatment failure; other variables associated with treatment failure at P <0.2 were also included in the model. Statistical analyses were conducted using STATA ver. 13.


During the study period, 279 hospitalizations were associated with ICD10 codes for CO. Following review of medical records, 114 nonduplicate cases were ultimately identified meeting all inclusion criteria. The median patient age was 11.8 years and 41 (35.9%) patients had underlying comorbidities (Table 1). The median duration of symptoms at the time of presentation was 122 days (interquartile range [IQR]: 60–244 days). The most common sites of disease were the femur (n = 28, 24.6%) followed by the tibia (n = 23, 20.2%), feet/toes (n = 21, 18.4%) and pelvis (n = 14, 12.3%).

TABLE 1. - Overall Characteristics of Study Population
N (%)
Median age, y (IQR)* 11.8 (6.2–14.4)
Medical comorbidities 41 (35.9)
 Neurologic comorbidities 21 (18.4)
 Malignancy 4 (3.5)
 Slipped capital femoral epiphysis 3 (2.6)
 Diabetes 2 (1.8)
 End-stage renal disease 2 (1.8)
 Rheumatologic diagnosis 2 (1.8)
 Eczema 2 (1.8)
 Other comorbidities 20 (17.5)
Duration of symptoms on presentation, d 122 (60–244)
Health insurance status
 Private 53 (46.5)
 Public 59 (51.7)
 Uninsured 2 (1.8)
Most common sites of disease
 Femur 28 (24.6)
 Tibia 23 (20.2)
 Feet/toes 21 (18.4)
 Pelvis 14 (12.3)
Physical examination findings
 Fever on admission 33 (28.9)
 Drainage 36 (31.6)
 Local inflammation 83 (72.8)
WBC at admission, ×103 cells/mm3 8.8 (6.9–12.2)
CRP at admission, mg/dL 1.3 (0.5–3.3)
ESR at admission, mm/h 28 (10–60)
MRI performed 98 (85.9)
CT performed 2 (1.8)
Radiograph performed without additional imaging 14 (12.3)
Surgery performed 80 (70.2)
≥2 surgical procedures 51 (44.7)
No. surgical procedures 2 (1–3)
Blood culture performed 67 (58.7)
Blood culture positive 5/67 (7.4)
Aerobic culture from bone tissue performed 103 (90.4)
Aerobic culture positive 73/103 (70.9)
Anaerobic culture performed 78 (68.4)
Anaerobic cultures positive 9/78 (11.5)
Fungal cultures performed 72 (63.2)
Fungal culture positive 2/72 (2.8)
Mycobacteria culture performed 71 (62.3)
Mycobacteria culture positive 3/71 (4.2)
Synovial fluid culture performed 3 (2.6)
Synovial fluid culture positive 0
16s PCR performed 22 (19.3)
16S PCR identified a pathogen 5/22 (22.7)
*Continuous variables presented as medians with IQRs. Categorical variables presented as n (%).
Neurologic comorbidities included cerebral palsy (n = 8), traumatic brain injury (n = 2), stroke (n = 1), epilepsy (n = 1), spina bifida (n = 12) and congenital pain insensitivity syndrome (n = 1); diagnoses are not mutually exclusive.
CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; WBC, white blood cell.

Types of CO

Cases were diverse in terms of pathogenesis with the most common being PHCO (n = 40, 35.1%), followed in descending order by those associated with CoF (n = 22, 19.3%), PACO (n = 21 18.4%), trauma-associated (n = 18, 15.8%) and hardware-associated (n=13, 11.4%). The minority of cases of PACO occurred in individuals with comorbidities (1/21, 4.8%), whereas medical comorbidities were common in those with CoF (19/22, 86.4%) and hardware-associated infection (8/13, 61.5%, P < 0.001, Table 2). Wound drainage was more common among those with CoF (11/22, 50%) and hardware-associated infection (9/13, 69.2%, P = 0.001) than other disease forms. CoF infection was more often associated with disease involving the pelvis/sacrum (P = 0.001) and feet/toes (P = 0.01) than other disease categories. For patients with PACO, treatment regimens utilized at the time of their acute infection are summarized in Table, Supplemental Digital Content 1,

TABLE 2. - Select Clinical Features by CO Category
Postacute, n = 21 Contiguous Focus, n = 22 Hardware Associated, n = 13 Trauma Associated, n = 18 Presumed Hematogenous Chronic, n = 40 P
Age, y 9.9 (6.8–12.2) 13.4 (12.3–16.8) 14.4 (12.5–14.6) 6.3 (3.2–13.3) 8.9 (3.8–14.2) <0.001
Underlying conditions 1 (4.8) 19 (86.4) 8 (61.5) 4 (22.2) 9 (22.5) <0.001
Health insurance status 0.046
 Private 10 (47.6) 7 (31.8) 7 (53.8) 4 (22.2) 25 (62.5)
 Public 11 (52.4) 15 (68.1) 6 (46.2) 13 (72.2) 14 (35)
 Uninsured 0 0 0 1 (5.6) 1 (2.5)
Most common sites of disease
 Femur 5 (23.8) 4 (18.1) 6 (46.1) 1 (5.6) 12 (30) 0.09
 Tibia/fibula 5 (23.8) 0 1 (7.6) 6 (33.3) 11 (27.5) 0.01
 Pelvis/sacrum 3 (14.2) 8 (36.4) 1 (7.6) 0 1 (2.5) 0.001
 Foot/toes 1 (4.7) 9 (40.9) 0 4 (27.2) 7 (17.5) 0.01
 Monomicrobial 20 (95.2) 8 (36.4) 7 (33.8) 6 (33.3) 17 (42.5) <0.001
 Polymicrobial 0 10 (45.4) 4 (30.7) 6 (33.3) 4 (10) <0.001
 No organism identified 1 (4.8) 4 (18.2) 2 (15.4) 6 (33.3) 19 (47.5) 0.21
S. aureus 20 (95.6) 6 (27.2) 1 (7.7) 7 (38.8) 11 (27.5) <0.001
 MRSA 6 (28.6) 1 (4.5) 0 3 (16.7) 7 (17.5) 0.12
 Anaerobes 0 4 (18.2) 3 (23.1) 3 (16.7) 2 (5) 0.048
Enterobacter 0 0 2 (15.1) 3 (10.7) 0 0.002
Pseudomonas 0 7 (31.8) 0 1 (5.6) 0 <0.001
Fever on presentation 8 (38.1) 5 (22.7) 0 6 (33.3) 14 (35) 0.08
Wound drainage 2 (9.5) 11 (50) 9 (69.2) 6 (33.3) 8 (20) 0.001
Local inflammation on examination 15 (71.4) 17 (77.7) 10 (76.9) 16 (85.9) 25 (62.5) 0.33
Blood culture positive/blood culture performed* 0/9 0/11 0/6 1/11 (9.1) 4/30 (13.3) 0.78
16S PCR Positive/16S PCR performed 0/1 0/4 1/4 (25) 2/5 (40) 2/8 (25) 0.86
WBC at admission, ×103 cells/mm3 7.2 (6.4–7.9) 9.4 (7.5–15) 9.6 (6.2–12.6) 9.8 (8.1–14.6) 9.4 (7.2–12.2) 0.03
CRP at admission, mg/dL 0.6 (0.3–2) 1.8 (0.7–6.6) 1.1 (0.5–2.6) 1.3 (0.6–2) 1.6 (0.5–5.2) 0.1
ESR at admission, mm/h 11.5 (6–37) 30 (11–79) 34 (15–36) 16 (8.5–59.5) 32 (10–72) 0.22
Duration of total treatment, d 312.5 (220–375 180 (150–365) 230 (113–335) 105 (60–330) 210 (151–365) 0.056
Route of therapy 0.59
 IV therapy only 2 (9.5) 1 (4.5) 1 (7.7) 0 4 (10)
 ≥2 wks of IV therapy then transition to PO therapy 11 (52.4) 7 (31.8) 5 (38.4) 5 (27.8) 13 (32.5)
 Early transition to PO therapy 8 (38.1) 14 (63.6) 7 (53.8) 13 (72.2) 23 (57.5)
Surgical drainage/debridement 9 (42.8) 16 (72.7) 13 (100) 15 (83.3) 27 (67.5) 0.004
≥2 Surgeries 6 (28.6) 8 (36.3) 9 (69.2) 9 (50) 19 (47.5) 0.18
Percutaneous drainage/biopsy 0 1 (4.5) 0 0 2 (5) 0.9
Number of surgeries 2 (1–3) 1 (1–3) 2 (1–4) 2 (1–3) 2 (1–2) 0.45
Treatment failure 6 (28.6) 9 (40.9) 3 (23.1) 6 (33.3) 6 (15) 0.21
*Organisms isolated in blood culture included S. aureus (n=2), Klebsiella pneumoniae (n = 1), Escherichia coli (n = 1) and Propionibacterium acnes (n = 1, this patient also had P. acnes isolated from bone).


A microbial etiology was identified in 82 cases (71.9%) with 24 cases (21%) being polymicrobial (Fig. 1A). Blood cultures were positive in 7.4% of patients, primarily among PHCO patients (Table 2). Aerobic cultures from bone identified at least one pathogen in 70.9% of cases (73/103). 16S rRNA PCR was performed from bone specimens in 22 cases and identified pathogens in 22.7% in which the test was performed; in 2 cases (9.1%), the pathogen was only identified by 16S rRNA PCR. Eight of the patients with negative cultures also had negative 16S rRNA PCR.

Microbiology of chronic osteomyelitis in children. A: Relative frequency of isolated organisms. Named organisms included if ≥2 isolates identified. B: Relative Frequency of isolation of organisms by category of chronic osteomyelitis. “Other” includes one isolate of each of the following: Nocardia spp., Actinomyces spp., Group G Streptococcus, Comamonas testosteroni, Serratia marcescens, Kingella kingae, Klebsiella pneumoniae, Proteus mirabilis, Citrobacter spp., Dermabacter hominis, Propionibacterium acnes, Pasteurella canis and Haemophilus parainfluenzae.

S. aureus was the single most commonly isolated organism (n = 45, 39.4%) of which 37.8% were MRSA. The next most commonly isolated organisms included anaerobic bacteria as a group (n = 14, 12.3%), Pseudomonas spp. (n = 9, 7.8%) and Enterobacter spp. (n = 5, 4.4%). The microbial etiology differed significantly by disease presentation (Fig. 1B and Table 2). The majority of PACO cases were caused by S. aureus (20/21, 95.2%), while the most commonly isolated organism in CoF infection cases was Pseudomonas spp. (7/22, 31.8%). Among those with PHCO, 47.5% had no organism identified and S. aureus was isolated in 27.5%. Polymicrobial infections were more commonly noted in CoF (10/22, 45.4%), hardware-associated (4/13, 30.7%) and trauma-associated infections (6/18, 33.3%) than other disease forms (0/21, PACO and 4/40, 10%, PHCO, P < 0.001).


Seventy-two subjects (63.2%) had samples of bone taken for histologic examination (Table, Supplemental Digital Content 2, The most common findings were bone necrosis (22/72, 30.6%), chronic inflammatory changes (17/72, 23.6%) and normal/viable bone (15/22, 20.8%). Seven patients had only acute inflammatory changes noted on biopsy (9.8%); the median duration of symptoms on presentation for these patients was 180 days (IQR: 90–367 days). Five patients had evidence of microorganisms on Gram-stain or fungal stain of bone tissue and 4 exhibited granulomas on histopathology. Among the 15 patients with normal histology, in 7 cases, a microorganism was identified by culture and/or molecular methods (46.7%) from the bone tissue.


Eighty patients (70.2%) underwent at least 1 surgical procedure and 51 underwent ≥2 surgeries (44.7%). A greater proportion of patients with hardware-associated infection (100%) underwent surgical debridement than other forms of disease (P = 0.004). In the subset of patients with hardware-associated infection, 8 underwent complete hardware removal (61.5%), 2 underwent partial removal (15.4%), and in 3 cases, debridement and irrigation were performed with hardware retention (23.1%). Among those patients with decubitus ulcers, 10/17 (58.8%) underwent wound coverage with either a vacuum-assisted closure or muscle flap.

All patients received a period of intravenous (IV) antibiotic therapy and the median duration of IV therapy was 12 days (IQR: 4–42 days); the median duration of total antibiotic therapy was 210 days (IQR: 150–367 days). Sixty-eight (59.6%) ultimately underwent early transition to oral antibiotics; these patients received a median of 5 days of IV antibiotics before oral step-down (IQR: 4–12 days). Among the 3 patients with hardware-associated CO with implant retention, the total durations of therapy ranged from 91 to 1421 days. Overall, the most commonly used single agents for definitive therapy were clindamycin (34, 29.8%), trimethoprim-sulfamethoxazole (23, 20.2%) and cephalexin (21, 18.4%). Combination antibiotics were used as definitive therapy in 26 cases (22.8%), of which 3 were hardware-associated and 9 were polymicrobial infections.

Among 32 patients in whom no pathogen was identified, 24 (75%) underwent early transition to oral therapy. The most commonly used agent for definitive therapy in these patients was clindamycin (11/32, 34.3%) followed by trimethoprim-sulfamethoxazole (10/32, 31.2%).


Twelve patients (10.5%) required readmission for additional surgical management after the index hospitalization; one diabetic patient with recurrent CO of the foot due to MRSA ultimately required amputation. Five patients with CoF disease (5/22, 22.7%) developed reinfection with a new organism during the study period; all of these patients had disease associated with decubitus ulcers. No infections with new organisms occurred in any of the other forms of CO (P < 0.001).

Thirty patients (26.3%) experienced treatment failure as defined for this study. A relationship was observed between the presence of neurologic comorbidities as well as the presence of decubiti and treatment failure (Table 3). There was no correlation between use of oral antibiotics, timing of transition to oral therapy or duration of IV therapy and likelihood of treatment success/failure. Patients who experienced treatment failure received a longer duration of total antibiotic therapy than those who experienced treatment success (295 days [IQR: 180–394] vs. 180 days [IQR: 97–356], P = 0.03). Among 3 patients with hardware-associated infection with implant retention, one experienced treatment failure. Relationships between treatment modalities and treatment success/failure by individual category of CO are presented in Table, Supplemental Digital Content 3,

TABLE 3. - Characteristics Associated With Treatment Failure Versus Success
Treatment Failure, n = 30 Treatment Success, n = 84 P
Age, y 12.7 (6–14.9) 11.6 (6.2–14.1) 0.36
Disease category 0.3
 Contiguous focus 9 (30) 13 (15.4)
 Primary hematogenous chronic osteomyelitis 7 (23.3) 33 (39.2)
 Postacute chronic osteomyelitis 5 (16.7) 16 (19.1)
 Trauma associated 6 (20) 12 (14.2)
 Hardware associated 3 (10) 10 (11.9)
Public insurance or uninsured 20 (66.7) 41 (48.8) 0.37
Neurologic comorbidities 12 (40) 9 (10.7) 0.001
Decubiti present 9 (30) 8 (9.5) 0.01
 Coverage of decubiti 6/9 (66.7) 4/8 (50) 0.64
Disease of pelvis/sacrum 7 (23.3) 8 (9.5) 0.06
Surgery performed 25 (83.3) 55 (65.4) 0.1
 Partial/total hardware retention* 1/3 (33.3) 4/12 (33.3) 1
Early transition to PO antibiotics 15 (50) 50 (59.5) 1
Duration of IV antibiotics 21 (5–42) 10 (4–37) 0.37
Duration of total antibiotics, d 295 (180–394) 180 (97–356) 0.03
Treatment >120 d 25/30 (83.3) 54/73 (73.9) 0.4
S. aureus 10 (33.3) 44 (52.4) 0.09
MRSA 4 (13.3) 13 (15.5) 1
Anaerobes 5 (16.7) 7 (8.30 0.29
Pseudomonas 4 (13.3) 4 (4.7) 0.2
Enterobacter spp. 1 (3.3) 4 (4.7) 1
Polymicrobial infection 7 (23.3) 17 (20.2) 0.79
No. organisms if polymicrobial 4 (3-5) 3 (2-3) 0.08
*Among those with hardware-associated CO.
Represented the bottom quartile for duration of therapy for the study group.

The category of CO, early transition to PO antibiotics, total treatment duration <120 days (the bottom quartile), S. aureus infection, Pseudomonas spp. infection, pelvic/sacral disease, decubitus ulcers and neurologic comorbidities were all included in a stepwise logistic regression model for treatment failure. Among these many variables, only neurologic comorbidities remained significantly associated with treatment failure (P = 0.001).


While the majority of cases of osteomyelitis in children are acute in nature and resolve with treatment, a subset of patients experience chronic symptoms and significant morbidity. We described a relatively large group of US children with CO in the MRSA era; these infections were heterogeneous in terms of pathogenesis and microbiology.

Similar to previous work,12–15 we noted S. aureus to be the most commonly recovered organism occurring in 39.4% of cases. Importantly, the microbiology varied significantly based on the presumed pathogenesis of disease. For patients with confirmed PACO, S. aureus was responsible for 95.6% of cases with MRSA causing disease in >25% of these patients. These findings are likely influenced by the predominance of S. aureus as a cause of acute osteomyelitis3,6 as well as its virulence potential. Previous studies at our own institution found that CO complicates ~5% of S. aureus AHO.4 By contrast, among our subset of patients with PHCO, only 27% were caused by S. aureus and in nearly half of cases an organism was not identified. CO in children is regarded by some authors as a disease associated with poverty, stemming from limited access to medical care and lack of or delayed treatment of AHO. One of the largest studies on this subject was conducted in Malawi of 167 children with hematogenous CO in which 61% of cases were found to be caused by S. aureus.16 While it is believed that PACO and PHCO (as defined for this study) have identical pathogenesis, we delineated these categories to evaluate differences between those patients who have a delay in diagnosis compared with those diagnosed acutely who then experienced a poor outcome. The lower rate of pathogen recovery in PHCO cases may suggest a lower inoculum infection or infection with fastidious organisms that were not cultivatable in this subset of patients. Interestingly, we found that use of private health insurance, a surrogate marker for socioeconomic status, was higher among children with PHCO compared with other forms of CO; these findings suggest subtle difference in disease epidemiology between developed and developing nations.

Studies among adults with pelvic osteomyelitis associated with decubiti have revealed a high rate of polymicrobial infection. In one series from the United Kingdom, while S. aureus was the predominant pathogen, polymicrobial infection occurred in 44%; Gram-negative bacilli were co-isolated in 56.5% and anaerobes in 34.7%.17 Most studies of children with osteomyelitis extending from a CoF are small series or represent a small subset of patients within studies of all osteomyelitis. In a series from the 1980s by Dubey of CoF and posttraumatic osteomyelitis in children, 46.7% of cases were polymicrobial containing a combination of skin flora and Gram-negative bacilli.18 In our study population, while S. aureus was commonly identified, Pseudomonas spp. were the most frequently isolated organisms in CoF cases, occurring in 31.8%. These findings highlight the need to provide broad empiric antimicrobial coverage in such patients. With the exception of perhaps PACO disease, antimicrobial coverage for both Gram-positive cocci as well as Gram-negative bacilli and anaerobes should be provided while awaiting culture results in children with CO.

Numerous authors place a high value on bone histology in the diagnosis and management of CO, particularly in the setting of osteomyelitis underlying decubiti.8,9,15,19 In principle, bone biopsy can confirm the diagnosis or guide therapy based on the presence of acute versus chronic inflammatory cells or microorganisms on tissue stain or through providing specimens for culture. In our study, however, over 20% of bone biopsy samples exhibited normal histology and in nearly half of these cases a pathogen was identified. Furthermore, patients with only evidence of acute inflammation on histopathology had a median duration of symptoms of 180 days. These seemingly counterintuitive findings likely are reflective of sampling error and highlight some of the potential limitations of bone biopsy. It is possible that in at least a subset of samples the core bone biopsy was obtained from an area adjacent to the focus of infection and thus provided incomplete data to the pathologist. These limitations underscore some of the challenges in diagnosing CO, particularly in light of the often very modest elevation of inflammatory markers in this disease.

Very little high-quality data exist to guide clinical management of children with CO, particularly with regards to the appropriate duration of therapy. Many experts recommend surgical debridement combined with prolonged antibiotic courses ranging from 6 weeks to 6–12 months.8,13 Moreover, in surveys of adult providers, practices are quite variable with respect to the diagnosis and treatment of osteomyelitis associated with decubiti.19 In a small clinical trial from the 1980s of nafcillin versus nafcillin plus rifampin in CO, 66% of patients had a favorable response after 6 weeks of therapy with no statistically significant difference between those receiving monotherapy versus combination therapy.20 Nevertheless, there was a high rate of treatment failure and relapse in this study. In a case series of 4 children from Israel with CO, all were successfully treated with a surgical debridement, instillation of antibiotic impregnated cement and systemic antibiotics for a median of 16 weeks (with a maximum of 37 weeks).12 Notably, in the previously mentioned series from Malawi, patients were treated with 6 weeks of oral antibiotics following sequestrectomy with only 16% requiring readmission for additional intervention at 1 year of follow-up.16 Based on the limited available literature, the optimal duration of therapy in children with CO is unclear. In our series, we found that the rate of treatment failure was 26%; the proportion of cases with treatment failure ranged from 17.5% in PHCO to 40.9% in CoF disease. Notably, the proportion of patients who received the bottom quartile of antibiotic duration was similar between those with treatment success versus failure. Additionally, early transition to oral antibiotics did not appear to increase the likelihood of treatment failure. This is consistent with a randomized trial of adults with either acute or CO, in which early transition to oral antibiotics was found to be noninferior to 6 weeks of IV therapy.21 Overall, our data suggest that prolonged therapy (>120 days), as well as >14 days of IV therapy, may not necessarily reduce rates of treatment failure in children with CO. These findings are consistent with studies in adults which failed to illustrate a relationship between prolonged antibiotic therapy and likelihood of treatment success.17 Spellberg and Lipsky22 performed an extensive review of the published literature regarding management of CO in adults and reported durations of therapy ranging from 2 weeks to >6 months with no clear relationship between treatment duration and cure. Unfortunately, because of our sample size, the retrospective design and the local management practices, we are unable to define the lower boundary for treatment duration to achieve success. Additionally, given the relatively high rate of hardware removal in our study, caution should be exercised in attempting to extrapolate these findings to all patients with orthopedic hardware-associated CO in which prolonged antibiotic suppression may be appropriate.23–25 Furthermore, some providers may have elected to extend therapy duration based on poor response potentially biasing results regarding the impact of duration of therapy on outcomes. Finally, the heterogenous nature of our study population makes generation of firm treatment recommendations challenging. Interestingly, some authors have suggested that antibiotics may not be completely necessary in children with CO, particularly in resource limited settings, if good surgical source control can be achieved.26

It is important to note that we did find an association with treatment failure and the presence of neurologic comorbidities as well as decubitus ulcers in univariable analyses. The findings imply that the premorbid state of the child may have a larger impact on the likelihood of treatment success or failure than duration of antibiotics. In an observational study of adults with sacral decubitus ulcers with and without osteomyelitis (based on histology), subjects with CO received only a 5- to 7-days course of antibiotics; all patients were treated with pressure off-loading, debridement and muscle flap coverage.15 These authors noted a similar rate of ulcer recurrence/graft failure in patients with and without CO suggesting that a short course of antibiotics may be sufficient if adequate source control is obtained in this specific population.

There are additional limitations to our study which should be acknowledged. The reliance on ICD10 coding may have potentially underestimated the number of cases of CO. Additionally, the inclusion of patients with normal bone histology and/or negative cultures may have led to misclassification of patients who in fact did not have CO. This does, however, provide an element of real world applicability as it is common for a pathogen to not be identified in musculoskeletal infections.3,16 Finally, our results are from a single center and may not be generalizable to the overall pediatric population.

In conclusion, CO is a relatively rare but important infection in children and contributes to substantial morbidity. While S. aureus is the most common etiology, the microbiology is diverse as are the potential modes of acquisition. Rates of treatment failure are high and while the optimal duration of antimicrobials has not been defined, prolonged therapy may not necessarily improve outcomes. Further study is needed to better understand the optimal management of this disease.


1. Keren R, Shah SS, Srivastava R, et al.; Pediatric Research in Inpatient Settings Network. Comparative effectiveness of intravenous vs oral antibiotics for postdischarge treatment of acute osteomyelitis in children. JAMA Pediatr. 2015;169:120–128.
2. Zaoutis T, Localio AR, Leckerman K, et al. Prolonged intravenous therapy versus early transition to oral antimicrobial therapy for acute osteomyelitis in children. Pediatrics. 2009;123:636–642.
3. McNeil JC, Forbes AR, Vallejo JG, et al. Role of operative or interventional radiology-guided cultures for osteomyelitis. Pediatrics. 2016;137:e20154616.
4. McNeil JC, Vallejo JG, Kok EY, et al. Clinical and microbiologic variables predictive of orthopedic complications following Staphylococcus aureus acute hematogenous osteoarticular infections in children. Clin Infect Dis. 2019;69:1955–1961.
5. Peltola H, Pääkkönen M, Kallio P, et al.; Osteomyelitis-Septic Arthritis Study Group. Short- versus long-term antimicrobial treatment for acute hematogenous osteomyelitis of childhood: prospective, randomized trial on 131 culture-positive cases. Pediatr Infect Dis J. 2010;29:1123–1128.
6. Arnold SR, Elias D, Buckingham SC, et al. Changing patterns of acute hematogenous osteomyelitis and septic arthritis: emergence of community-associated methicillin-resistant Staphylococcus aureus. J Pediatr Orthop. 2006;26:703–708.
7. Saavedra-Lozano J, Falup-Pecurariu O, Faust SN, et al. Bone and joint infections. Pediatr Infect Dis J. 2017;36:788–799.
8. Krogstad P. Cherry JD, Harrison GJ, Kaplan SL, et al, eds. Osteomyelitis. In: Feigin and Cherry’s Textbook of Pediatric Infectious Diseases. 2019:Vol 1. 8th ed. Elsevier; 516–529.
9. Wong D, Holtom P, Spellberg B. Osteomyelitis complicating sacral pressure ulcers: whether or not to treat with antibiotic therapy. Clin Infect Dis. 2019;68:338–342.
10. McNeil JC, Sommer LM, Boyle M, et al. Cefazolin inoculum effect and methicillin-susceptible Staphylococcus aureus osteoarticular infections in children. Antimicrob Agents Chemother. 2020;64:e00703–e00720.
11. Kaplan J, Ikeda S, McNeil JC, et al. Microbiology of osteoarticular infections in patients with sickle hemoglobinopathies at Texas Children’s Hospital, 2000-2018. Pediatr Infect Dis J. 2019;38:1251–1253.
12. Bar-On E, Weigl DM, Bor N, et al. Chronic osteomyelitis in children: treatment by intramedullary reaming and antibiotic-impregnated cement rods. J Pediatr Orthop. 2010;30:508–513.
13. Lauschke FH, Frey CT. Hematogenous osteomyelitis in infants and children in the northwestern region of Namibia. Management and two-year results. J Bone Joint Surg Am. 1994;76:502–510.
14. Foster CE, Taylor M, Schallert EK, et al. Brodie abscess in children: a 10-year single institution retrospective review. Pediatr Infect Dis J. 2019;38:e32–e34.
15. Marriott R, Rubayi S. Successful truncated osteomyelitis treatment for chronic osteomyelitis secondary to pressure ulcers in spinal cord injury patients. Ann Plast Surg. 2008;61:425–429.
16. Beckles VL, Jones HW, Harrison WJ. Chronic haematogenous osteomyelitis in children: a retrospective review of 167 patients in Malawi. J Bone Joint Surg Br. 2010;92:1138–1143.
17. Russell CD, Tsang SJ, Simpson AHRW, et al. Outcomes, microbiology and antimicrobial usage in pressure ulcer-related pelvic osteomyelitis: messages for clinical practice. J Bone Jt Infect. 2020;5:67–75.
18. Dubey L, Krasinski K, Hernanz-Schulman M. Osteomyelitis secondary to trauma or infected contiguous soft tissue. Pediatr Infect Dis J. 1988;7:26–34.
19. Kaka AS, Beekmann SE, Gravely A, et al. Diagnosis and management of osteomyelitis associated with stage 4 pressure ulcers: report of a query to the Emerging Infections Network of the Infectious Diseases Society of America. Open Forum Infect Dis. 2019;6:ofz406.
20. Norden CW, Bryant R, Palmer D, et al. Chronic osteomyelitis caused by Staphylococcus aureus: controlled clinical trial of nafcillin therapy and nafcillin-rifampin therapy. South Med J. 1986;79:947–951.
21. Li HK, Rombach I, Zambellas R, et al.; OVIVA Trial Collaborators. Oral versus intravenous antibiotics for bone and joint infection. N Engl J Med. 2019;380:425–436.
22. Spellberg B, Lipsky BA. Systemic antibiotic therapy for chronic osteomyelitis in adults. Clin Infect Dis. 2012;54:393–407.
23. Kowalski TJ, Berbari EF, Huddleston PM, et al. The management and outcome of spinal implant infections: contemporary retrospective cohort study. Clin Infect Dis. 2007;44:913–920.
24. Keller SC, Cosgrove SE, Higgins Y, et al. Role of suppressive oral antibiotics in orthopedic hardware infections for those not undergoing two-stage replacement surgery. Open Forum Infect Dis. 2016;3:ofw176.
25. Osmon DR, Berbari EF, Berendt AR, et al.; Infectious Diseases Society of America. Diagnosis and management of prosthetic joint infection: clinical practice guidelines by the Infectious Diseases Society of America. Clin Infect Dis. 2013;56:e1–e25.
26. Jones HW, Beckles VL, Akinola B, et al. Chronic haematogenous osteomyelitis in children: an unsolved problem. J Bone Joint Surg Br. 2011;93:1005–1010.

chronic osteomyelitis; contiguous focus; decubitus ulcer; pediatrics; Staphylococcus aureus

Supplemental Digital Content

Copyright © 2021 Wolters Kluwer Health, Inc. All rights reserved.