DIAGNOSING PEDAL OSTEOMYELITIS - TESTING CHOICES AND THEIR CONSEQUENCES

Objective: To compare the efficacies and cost-effectivenesses of four strategies for the management of suspected pedal osteomyelitis in the setting of vascular impairment: 1) therapeutic trial of short-term ant ibiotics for presumed cellulitis without osteomyelitis (short); 2) tec hnetium bone scanning followed by either short-term therapy if negativ e or either a biopsy or aggressive long-term intravenous therapy if po sitive (scan); 3) bone biopsy followed by long-term intravenous therap y if positive or short-term therapy if negative (biopsy); and 4) immed iate long-term intravenous antibiotics for presumed osteomyelitis (lon g). Design: Decision analysis and cost-effectiveness analysis with sen sitivity analyses. The main outcomes states are amputation and the res ource expenditures associated with bone scans, biopsies, and therapies . Data sources: The authors obtained estimates of test accuracy from l iterature review and summarized them using newly developed metaanalyti c techniques. Main results: The optimal decision depends heavily on th e estimated probability of osteomyelitis at presentation. At very low probabilities, the short-term strategy is preferred. When the probabil ity of osteomyelitis is from 2% to 8%, the lowest amputation rate occu rs when one does a diagnostic scan. From 8% to 50%, the best outcomes follow biopsy. At probabilities higher than 50%, the preferred strateg y is long term antibiotics. However, the differences in outcomes are q uite small even when osteomyelitis is a virtual certainty. Conclusions : Over the whole range of prior probabilities, the short-term strategy is the least expensive. At very low probabilities, it dominates the o ther strategies. When the likelihood of osteomyelitis is higher( 10-20 %), scanning results in outcomes and cost-effectiveness ratios compara ble to those of immediate biopsy and is less invasive. When the probab ility of osteomyelitis is 50%, biopsy is quite cost-effective compared with all the other strategies (cost-effectiveness ratio = $l5,502 per amputation averted) and is preferred to the scan strategy. When the c onfidence that a patient has osteomyelitis is very high (> 90% probabi lity), the improved outcomes associated with long-term antibiotics are achieved with little additional expense and with favorable cost-effec tiveness ratios compared with those of the other strategies.