OBSERVATIONS Probing the Validity of the Probe-to-Bone Test in the Diagnosis of Osteomyelitis of the Foot in Diabetes

T he ability to probe the base of a wound to periosteum or bone (the “probe-to-bone” test) is increasingly used to indicate the likelihood of underlying osteomyelitis. The original study (1) reported sensitivity, specificity, and positive (PPV) and negative (NPV) predictive values of 66, 85, 89, and 56%, respectively. However, this work has been criticized on the grounds of the high pretest probability of the disease (2), since the prevalence of osteomyelitis in the chosen sample (in-patients with clinically overt infection) was 66%. It follows that the usefulness of the test may be very different in less-selected populations. We have therefore determined the validity of the probe-to-bone test in a consecutive series of outpatients attending our own multidisciplinary service. A total of 81 patients (with a total 104 foot ulcers) attended the clinic over a 5-week period in May–June 2005. Ulcers were probed by one of two specialist podiatrists following debridement. The diagnosis of osteomyelitis was determined by one of two expert diabetologists, who were blind to the results of the probe to bone test. The diagnosis of osteomyelitis was based in each case on the presence of clinical signs of infection (inflammation with or without serous or purulent discharge) in association with radiologic evidence of bone destruct ion with interruption of the cortex (either at presentation or at any stage over the ensuing 8 weeks), supported when necessary by magnetic resonance imaging and microbiologic analysis of deep tissue samples. Those who were diagnosed with osteomyelitis included those in whom the diagnosis had already been made at the time of probing and those in whom the diagnosis was made later. Nineteen (23.5%) patients were diagnosed with osteomyelitis complicating foot ulcers, in two of whom bone infection complicated two separate nonadjacent ulcers. Three patients had two or more nonadjacent ulcers, of which only one was associated with osteomyelitis. A total of 14 patients had osteomyelitis complicating a single ulcer. A total of 21 ulcers (20.2% of 104) were associated with osteomyelitis. The probe-to-bone test was positive in 8 of these 21 ulcers and in 7 of 83 without associated bone infection (sensitivity 38%, specificity 91%). While the NPV was 85%, the PPV (the probability that a patient with a positive test would have osteomyelitis) was only 53%. It is possible that the calculation of both sensitivity and NPV might be in part explained by the fact that some cases of osteomyelitis may already have been responding to treatment at the time of probing, but this would not have affected the calculation of either the specificity or the PPV. These data emphasize that the predictive value of a positive probe-to bone test in the original report was influenced by the high prevalence of osteomyelitis in the population studied. The prevalence of osteomyelitis in the present population was still high at 23.5% patients (20.2% ulcers) but was only approximately one-third of that in the earlier study, and the PPV was correspondingly lower. It is likely that the PPV would be lower still in patients managed in a less-specialized service.