Review: Technetium bone scanning is poor for detecting osteomyelitis
ACP J Club. 1992 Sept-Oct;117:53. doi:10.7326/ACPJC-1992-117-2-053
Littenberg B, Mushlin AI, and the Diagnostic Technology Assessment Consortium. Technetium bone scanning in the diagnosis of osteomyelitis: a meta-analysis of test performance. J Gen Intern Med. 1992 Mar-Apr;7:158-63.
To determine the performance, using meta-analysis, of technetium bone scanning for the diagnosis of osteomyelitis of the lower extremity in patients with underlying vascular insufficiency or peripheral neuropathy.
4 independent literature searches were done using the MEDLARS databases. The 4 searches were merged into a single list of references, and bibliographies were scanned to obtain additional references.
Potential articles for inclusion must have reported on the use of intravenous technetium-99m methylene diphosphonate (99mTc MDP) or a similar agent in at least 10 patients > 16 years old in diagnosing osteomyelitis of the distal lower extremity with ulcer or soft-tissue inflammation with coexisting diabetes, neuropathy, or vasculopathy. The results must have allowed for the generation of a 2 × 2 table.
Data were extracted on the types of patients studied, the number of bone scan phases, the dose of 99mTc, the diagnostic standard, the study design, and the results of the investigation.
10 of 296 studies met the selection criteria. Sample size ranged from 10 to 94 patients. 6 studies used retrospective designs. 9 studies used 99mTc MDP or hydroxy-MDP in doses ranging from 18 to 22 mCi. 7 studies used a 3-phase method, 1 used a 4-phase method, and 1 used a single-phase method. Most studies used a combination of pathologic findings after surgery or biopsy and clinical follow-up as a diagnostic standard. Pooled data were used to generate a summary receiver operating characteristic (SROC) curve. The false-positive rate (FPR) of the bone scan was at best 20%, occurring at sensitivities of 70% to 80%. The studies with increased sensitivities reported sizable increases in the FPRs ranging from 20% to over 90%, with small increases in sensitivity leading to large sacrifices in specificity. 7 of the 10 studies reported FPRs over 30%. To achieve a sensitivity of 90%, a FPR of 54% would be necessary, and to hold the FPR at 25% would result in a sensitivity of 77%. Variations in study designs did not have a statistically significant effect on the SROC curve.
The performance of technetium bone scanning for the diagnosis of osteomyelitis in the impaired foot is poor.
Source of funding: John A. Hartford Foundation.
Address for article reprint: Not available.
Differentiating soft-tissue infection from osteomyelitis is a common diagnostic dilemma. Although textbooks may recommend technetium bone scanning to resolve the issue, they seldom provide test operating characteristics such as sensitivity or specificity. This information has been given by the authors of this well-conducted meta-analysis.
It is a revealing comment on the state of the diagnostic technology assessment in this field that only 10 of 296 (3%) published articles met the criteria for review, and even these studies had major methodologic flaws.
To summarize these varied studies, the investigators combined estimates of sensitivity and specificity into a single ROC curve, which they call the summary ROC. This curve shows that technetium bone scanning is less accurate than previously thought: It has a low specificity (high FPR) that ranges from 46% to 75% at sensitivities varying from 90% to 70%, respectively.
How can we best use this information in a clinical setting? The test's high FPR (low specificity) suggests that, if we suspect osteomyelitis, a positive bone scan would be unhelpful. Clearly, a better way of identifying osteomyelitis is required. Conversely, some value may be found in a negative scan if we use liberal criteria to label a scan as positive (i.e., to increase the sensitivity); a negative test would indicate that osteomyelitis is unlikely, justifying a trial of conservative treatment.
A recent study reported that a 24-hour indium-111 leukocyte scan (with a sensitivity of 89% and specificity of 69%) outperformed a bone scan for the diagnosis of osteomyelitis associated with diabetic foot ulcers (1). The bone scan appears to be superfluous, at least for this predication.
Ramon Velez, MD
Wake Forest University School of Medicine Bowman Gray CampusWinston-Salem, North Carolina, USA
Ramon Velez, MD
Wake Forest University School of Medicine Bowman Gray Campus
Winston-Salem, North Carolina, USA