Author reply

We thank Sadun et al for their letter. We are in agreement that inappropriate testing leads not only to delays in final diagnosis and definitive treatment, but can result in further unnecessary testing and intervention. We would also amplify the point they raise regarding pretest probabilities. Pretest probability refers to the probability of the target disorder before a diagnostic test result is known. This can then be used to calculate likelihood ratios, which have advantages over sensitivity and specificity in that they are less likely to change with the prevalence of the disorder, and can be used to combine the results of multiple diagnostic tests to create a posttest probability for the disease in question. Many tests are most useful in intermediate ranges of pretest probability, but are of little help at the extremes. For example, an 85-year-old patient presents with jaw claudication, headache, and sudden loss of vision to the range of light perception and a swollen optic nerve has such a high pretest probability for giant cell arteritis (GCA) that a laboratory test such as C-reactive protein—both highly sensitive and specific for GCA—will not influence the decision to begin treatment or obtain a temporal artery biopsy. Conversely, a 49-year-old patient presenting with ischemic optic neuropathy and no systemic symptoms has such a low pretest probability of GCA that an elevated C-reactive protein would not significantly alter the likelihood ratio or posterior probability of GCA.