Accuracy of Physician Estimates of Pulmonary Nodule Pretest Probability of Malignancy
A recent study sought to compare physician estimates of nodule pretest probability of malignancy with previously validated malignancy prediction calculators.
Widespread use of computed tomography (CT) scans is leading to a sharp rise in the identification of incidentally detected pulmonary nodules. A recent prospective multicenter observational study published in Chest sought to compare physician estimates of nodule pretest probability of malignancy (pCA) with previously validated malignancy prediction calculators. The researchers also sought to determine how often appropriate guideline-recommended diagnostic testing was used based on clinician assessment of malignancy.
A total of 337 physician assessments of pCA and 2 validated pulmonary nodule malignancy prediction models (Veterans Administration [VA] model and Mayo model) were evaluated and compared in 337 patients with pulmonary nodules. Clinician-assessed pCA was categorized as low-, intermediate-, or high-risk, and the next diagnostic test ordered was evaluated.
At 1 year, the prevalence of malignancy was 47% (158 of 337 patients). Physicians were significantly better at predicting malignancy than did either the VA model (area under the curve [AUC], 0.85 vs 0.75, respectively; 95% CI, 0.70-0.79; P <.001) or the Mayo model (AUC, 0.85 vs 0.78, respectively; 95% CI, 0.73-0.82;
Although clinicians' prediction of malignancy was more accurate than either prediction model, in 61% of the cases physicians did not follow guideline recommendations when selecting the next test in nodule management. In patients with a low pCA, management strategies were more aggressive (positron emission tomography [PET] scanning; biopsy) than the recommended CT imaging in 52% of patients, whereas in patients with a high pCA, 75% of patients were managed more conservatively than recommended.
The investigators concluded that although physicians are adept at differentiating between benign and malignant pulmonary nodules, they do not consistently follow guideline-based recommendations when selecting the next diagnostic test. Enhanced focus on guideline refinement, implementation, and dissemination is warranted in order to provide optimal patient care.
Tanner NT, Porter A, Gould MK, Li X-J, Vachani A, Silvestri GA. Physician assessment of pretest probability of malignancy and adherence with guidelines for pulmonary nodule evaluation. Chest. 2017;152(2):263-270.