TEST THRESHOLDS FOR OBTAINING PEDIATRIC NEUROIMAGING IN CLINICAL PRACTICE GUIDELINES

Purpose: We sought to determine how often clinical practice guidelines related to pediatric neuroimaging included information about test thresholds, and to determine the level of those thresholds.   

Methods: Current guidelines related to pediatric neuroimaging identified through the National Guideline Clearinghouse, Medline, and websites of relevant professional societies were reviewed to identify those with explicit test thresholds, and those from which we were able to infer implicit test thresholds. Implicit thresholds were determined when a guideline contained the pre-test probability, or risk of a condition, in a given patient population as well as a recommendation about whether neuroimaging was appropriate in that population. If the recommendation was to perform neuroimaging, the implicit test threshold was considered to be at or below the pre-test probability in the population. If the guideline recommended against performing neuroimaging the implicit test threshold was considered to be somewhere above the pre-test probability. For the purposes of this analysis, weak or optional recommendations were considered recommendations.    Test thresholds were grouped into the following four outcome categories for comparison: any abnormality; findings that provide information regarding diagnosis or prognosis; findings that change management; and findings that require surgery. If a guideline contained recommendations pertaining to more than one outcome, an implicit threshold was determined for each outcome.   

Results: None of the 32 relevant guidelines contained an explicit test threshold, and 19 (59%) contained some quantitative pre-test probability information. There were 15 (47%) guidelines with enough information to determine one or more implicit thresholds. The implicit thresholds were: any abnormality (>0.9%-1.2%, ≤30%, ≤43-80% ≤50%, ≤65.5%, ≤89%); diagnosis or prognosis (≤3.9%, ≤12%, ≤15%-20%); management (>0%, >0.18%, >1%-≤25%; ≤2%-4%; ≤3%-8%; ≤3.6%); and surgery (≤2.3%, ≤5%, ≤5%).   

Conclusions: Although test thresholds are an essential aspect of evidence-based decision-making regarding diagnosis, they are not explicitly discussed in guidelines that include recommendations for pediatric neuroimaging, a class of diagnostic testing which often involves small but real risks to patients and significant costs. There is a lack of high quality evidence regarding pre-test probability of disease in populations of interest to guideline developers.  Publication of explicit thresholds in guidelines might facilitate planning of future research that could determine with confidence whether the pre-test probability of conditions is above or below the threshold for performing neuroimaging.