Diagnosing Diabetic Osteomyelitis in the Emergency Department

November 2010

Diagnosing Diabetic Osteomyelitis in the Emergency Department

Search Strategy: You conduct a PUBMED Clinical Query using a broad search strategy for diagnosis and the search term “diabetic osteomyelitis” to obtain 32 systematic review citations. To reproduce this search strategy see http://tinyurl.com/2e4rfya. From these systematic reviews, you obtain all of the studies reviewed below.

After finishing another helping of the hospital chicken fingers with the spicy mustard, you open the electronic medical record to assess the new patient awaiting an empty bed in the hallway. You note that the 55 year old homeless man has a long-standing history of poorly controlled diabetes with one prior myocardial infarction and chronic renal insufficiency. Today, he is complaining about his sore feet, not surprising looking at his rain-soaked shoes after 2-weeks of a rainy fall season. The nurses have just found an open room for him in the ED when you walk in to introduce yourself and evaluate his feet. You immediately note the repugnant odor of rotting flesh permeating the room before you’ve even removed his shoes.

When he pulls off his shoes, the smell grows stronger and you immediately note an open wound on the sole of his feet that may extend to the bone. It is draining the yellowish-brown fluid that you smell and seems to be surrounded by erythema and slight swelling of his foot. His pulses are 1+ symmetrically and you note no popliteal cord or adenopathy. A quick exam suggests that he has diabetic neuropathy with decreased sensation in the distal lower extremities bilaterally. He is uncertain of his last tetanus booster and does not recall any specific injury to the foot including no penetrating nail or other foreign body injury. He has no prior history of diabetic foot infections, but has not been taking his Insulin or monitoring his glucose levels for several months. He has not sought medical care for this foot wound before today. He notes no prior history of bone infections or surgery on his foot. On review of systems he has had no fevers, vomiting, diarrhea, or other constitutional symptoms. As you sit at the computer devising your diagnostic and therapeutic plan for him, you wonder whether this is a simple diabetic foot infection or a more complicated osteomyelitis. You contemplate the diagnostic accuracy of your bedside exam, screening labs, x-rays, and more time-intensive and expensive tests like MRI or bone scans to distinguish soft-tissue diabetic foot infections from osteomyelitis

PICO Question

Population: ED diabetic patients with foot infections and suspected osteomyelitis

Intervention: N/A

Comparison: N/A

Outcome: Diagnostic accuracy (sensitivity, specificity, likelihood ratio) for bedside physical exam, lab tests (WBC, ESR, CRP), plain film imaging, and other imaging tests


First years: Probing to bone in infected pedal ulcers: A clinical sign of underlying osteomyelitis in diabetic patients, JAMA 1995; 273: 721-723. (http://pmid.us/7853630)

Second years: Correlation of imaging techniques to histopathology in patients with diabetic foot syndrome and clinical suspicion of chronic osteomyelitis: The role of high resolution ultrasound, Diabetes Care 1999; 22: 294-299. (http://pmid.us/10333948)

Third years: Magnetic resonance imaging for diagnosing foot osteomyelitis: A meta-analysis, Arch Intern Med 2007; 167: 125-132. (http://pmid.us/17242312)

Fourth years: Does this patient with diabetes have osteomyelitis of the lower extremity? JAMA 2008; 299: 806-813. (http://pmid.us/18285592)


Article 1: Probing to Bone in Infected Pedal Ulcers: A Clinical Sign of Underlying Osteomyelitis in Diabetic Patients, JAMA 1995; 273:721-723

Article 2: Correlation of imaging techniques to histopathology in patients with diabetic foot syndrome and clinical suspicion of chronic osteomyelitis, Diabetes Care 1999; 22: 294-299

Article 3: Magnetic Resonance Imaging for Diagnosing Foot Osteomyelitis: A Metaanalysis, Arch Intern Med 2007; 167:125-132

Article 4: Does This Patient With Diabetes Have Osteomyelitis of the Lower Extremity? JAMA 2008 299: 806 – 813

Bottom Line

Perhaps the most pertinent question is whether emergency physicians need to diagnose osteomyelitis at all. There are no ED-based diagnostic randomized controlled trials assessing patient-centric outcomes of “keep in ED until diagnosis established” versus “defer definitive diagnostic testing to inpatient/outpatient health care teams”. In fact, there are no ED-based therapeutic RCT’s evaluating different management approaches for these patients, either. To make an informed decision about whether or not to pursue this diagnosis in the ED (or to modify your disposition/follow-up decisions based upon the likelihood of underlying osteomyelitis, one should understand the epidemiology and prognosis of diabetic patients with foot ulcers. The discussion and diagnostic algorithm below are an attempt to simplify the data we have available based upon input from our local Radiology, Podiatry, ID, Orthopedic Surgery, and Wound Care experts. In addition, recent IDSA guidelines emphasize that most diabetic foot ulcer patients, including most of those with osteomyelitis, do NOT require admission. Instead, they emphasize that clinicians should try to make the distinction between limb-threatening and non limb-threatening infections using parameters that we have incorporated into the algorithm below. Another important distinction is acute versus chronic osteomyelitis. Chronic osteo involves formation of isolated, dead bone that antibiotics will not penetrate and with advanced erosive changes on x-rays with reduction of most of the bone mass. These usually require excision surgically.

Osteomyelitis of the foot or ankle is the primary or secondary reason for 75000 hospitalizations in the United States each year and foot-related complications account for 20% of all diabetes-related admissions in North America. One large retrospective cohort demonstrated that 15% of diabetic foot ulcer patients ultimately require an amputation and the 3-year survival of diabetic foot ulcer patients is 72% (compared with 87% for age & gender matched diabetic controls). Even when the primary ulcer has healed, the 3-year foot ulcer recurrence rate is 61% with a 10% amputation rate during the same interval. Only 59% of patients who heal after amputation are alive at 3-years, whereas 73% of those who heal with amputation are alive at that time. Furthermore, the attributable cost for a 40-year-old male with a new foot ulcer is $27,987 (US$ 1999) for the 2 years after diagnosis. Medicare data from 1995 revealed that $1.45 billion were spent that year on diabetic foot ulcer care. The presence of pedal osteomyelitis in diabetics increases the risk of amputation and the 30-day perioperative mortality following amputation ranges from 7.4% to 15.5%. Unfortunately, clinicians often underestimate the likelihood of osteomyelitis in this population.

One reason that clinicians fail to recognize osteomyelitis early in the disease course is that the diagnosis is extremely difficult confounded by an incomplete understanding of the differential diagnosis and the test characteristics for elements of the history, physical exam, lab tests, and imaging studies. Ulcers can be venous, arterial, or diabetic and the distinction can be challenging. In general, venous ulcers are proximal to the malleoli with irregular borders whereas arterial ulcers involve the toes or shins with pale & punched-out appearance. On the other hand, diabetic ulcers occur in areas of increased pressure (usually the soles). Charcot neuropathy can mimic diabetic osteomyelitis but has a completely different therapy. Instead of antibiotics, Charcot neuropathy is treated with a cast and non-weight bearing status. Recognizing that Charcot is a possibility is important for emergency physicians since these can collapse “almost overnight”, especially in the ankle joint, with catastrophic outcomes.

Previous diagnostic cost-effectiveness models for diabetic foot osteomyelitis assessments have yielded diverging recommendations. Eckman et al suggested that “noninvasive testing adds significant expense to the treatment of patients with NIDDM in whom pedal osteomyelitis is suspected, and such testing may result in little improvement in health outcomes. In patients without systemic toxicity, a 10-week course of culture-guided oral antibiotic therapy following surgical debridement may be as effective as and less costly than other approaches.” On the other hand, Mushlin et al. suggested that “when the likelihood of osteomyelitis is higher (10-20%), scanning results in outcomes and cost-effectiveness ratios comparable to those of immediate biopsy and is less invasive. When the probability of osteomyelitis is 50%, biopsy is quite cost-effective compared with all the other strategies…and is preferred to the scan strategy.” Our Washington University EM-ID expert recommends withholding antibiotics in the ED pending culture results. The manuscripts we reviewed will not guide cost-effectiveness decision-making, but should enlighten clinicians regarding pre-test probability estimates and the diagnostic accuracy for common findings in the history, physical exam, lab tests, and imaging studies ordered when contemplating the diagnosis of osteomyelitis.

Ulcer size > 2 cm, a positive probe-to-bone test, or ESR > 70 mm/h each significantly increases the likelihood of diabetic osteomyelitis. A positive probe test is recognized when “on gentle probing, the evaluator detected a rock-hard, often gritty structure at the ulcer base without the apparent presence of any intervening soft tissue.” Although abnormal x-rays can increase the probability of osteomyelitis, only MRI substantially reduces the likelihood of lower extremity osteomyelitis in diabetic patients. Except for fever, no studies have evaluated the diagnostic accuracy or precision for symptoms or signs in the diagnosis of lower extremity osteomyelitis. In the case of fever, one low quality study suggested that an elevated temperature has a sensitivity of 19% for diabetic osteomyelitis, but the same study did not define the specificity and likelihood ratios cannot be computed.

The best-estimate pre-test probability for osteomyelitis in diabetics with suspicious foot ulcers is 15%. The following diagnostic test characteristics have been delineated from moderate quality research and were used to guide development of the diagnostic algorithm presented below.

Diagnostic Test Likelihood RatioPositive Likelihood RatioNegative
Probe to bone6.40.39
Bone exposure9.20.70
Ulcer area > 2cm27.20.48
Ulcer inflammation1.50.84
Clinical judgment (Wagner >2)5.50.54
ESR > 70 mm/h110.34
Swab culture11

Future research is needed to understand the usefulness and reliability of combinations of history/physical exam/lab/imaging tests for the risk-stratification of osteomyelitis in ED patients with diabetic foot lesions. Ultimately, a clinical decision rule may help to efficiently and reliably risk stratify diabetic foot lesions for osteomyelitis in the ED.