Commentary by Judith Brenner MD, Associate Program Director, NYU Internal Medicine Residency Program
The most recent installment in JAMA’s Rational Clinical Exam Series seeks to determine the accuracy of the history, physical exam, radiology and laboratory in making the diagnosis of osteomyelitis in diabetics. This is relevant given its frequency of occurrence and its cost and since the gold standard for diagnosis, namely a bone biopsy and culture, is less than optimal for a variety of reasons.
Less than 10% of the nearly 300 articles reviewed were included for review by the authors and of the 10%, most were graded in the mid categories for quality. Nonetheless, some important findings were reported (highlighted in green).
|Finding||Likelihood Ratio||Negative Likelihood Ratio|
|Ulcers >2 cm2||7.2 (CI 1.1-49)||0.48 (CI 0.31-.076)|
|Positive “probe to bone” test*||6.4 (CI 3.6-11)||0.39 (CI 0.20-0.76)|
|ESR >70||11 (CI 1.6-79)||0.34 (CI 0.06-1.9)|
|Abnormal plain X-ray**||2.3 (CI 1.6-3.3)||0.63 (CI 0.51-0.78)|
|Abnormal MRI||3.8 (CI 2.5-5.8)||0.14 (CI 0.08-0.26)|
*Probe to bone test: the examiner gently and in a sterile fashion, probes the ulcer with a steel probe to determine if the probe can advance to bone
**abnormal X-Ray findings include: focal loss of trabecular pattern, periosteal reaction, and frank bone destruction. 2 or 3 views can be selected
For example, the interpretation is as follows:
The LR (sensitivity/1-specificity) tells you how much the odds of a disease will increase based on a positive result.
So, if you determine the pre-test probability to be 15% and the patient has an ulcer >2.0cm2 then the post-test probability would rise to 56%.
Pre-test odds = pre-test probability /(1-pre-test probability)= 0.15/0.85=0.18 Post-test odds=pre-test odds * LR = 0.18 * 7.2 = 1.3
Post-test probability = post-test odds / (post test odds+1) = 1.3/2.3= 0.56 = 56%
or, more simply, use a Likelihood ratio nomogram
Similarly, a negative LR (1-sensitivity/specificity) tells you how much the odds of a disease decrease based on a negative test rest. So, for a patient whose pre-test probability is 15%, a normal MRI (ie a negative result), would change the post-test probability to 2.4%.
The article specifically noted that the studies that exist have not studied the significance of multiple findings in a patient. (As a personal commentary, I believe that it stands to reason that multiple positive findings would make a diagnosis of osteomyelitis more likely.)
Of note, there are no features of the medical history or other elements of the physical which were found to be helpful in the diagnosis, including presence of pain, fever or other systemic signs. Because it is established that diabetic ulcers tend to occur in patients with more advanced and poorly controlled diabetics, a history focusing on the diabetes history, including the presence of retinopathy, neuropathy, and nephropathy would be appropriate.
Reference: The Rational Clinical Examination: Does This Patient with Diabetes Have Osteomyelitis of the Lower Extremity?JAMA. 2008;299(7):806-13.