Does this patient have Pheochromocytoma? a systematic review of clinical signs and symptoms

Context Pheochromocytoma is a rare disease but with high mortality if it is not being diagnosed early. Several biochemical tests with high accuracy have been obtained, but the clinical threshold for request of these tests is not determined clearly. Objectives To determine the Likelihood Ratios of clinical symptoms and signs in diagnosing pheochromocytoma. And also meta-analysis of their sensitivity in this disease. Data sources MEDLINE was searched for relevant English-language articles dated 1960 to February 2014. Bibliographies were searched to find additional articles. Study selection We included original studies describing the sensitivity and/or likelihood ratios of signs and symptoms in clinical suspicion of pheochromocytoma. Their method of diagnosis should have been based on pathology. We excluded specific subtypes or syndromes related to pheochromocytoma, or specific ages or gender. Also we excluded studies before 1993 (JNC5) which no definition of hypertension was presented. 37 articles were chosen finally. Data extraction Two authors reviewed data from articles independently and gave discrepancies to third author for decision. The aim was extraction of raw numbers of patients having defined signs or symptoms, and draw 2 × 2 tables if data available. We meta-analyzed sensitivities by Statsdirect and Likelihood Ratios by Meta-disc soft wares. Because our data was heterogeneous based on I 2 > 50 % (except negative Likelihood ratio of hypertension), we used random effect model for doing meta-analysis. We checked publication bias by drawing Funnel plot for each sign/symptom, and also Egger test. Data synthesis The most prevalent signs and symptoms reported were hypertension (pooled sensitivity of 80.7 %), headache (pooled sensitivity of 60.4 %), palpitation (pooled sensitivity of 59.3 %) and diaphoresis (pooled sensitivity of 52.4 %). The definition of orthostatic hypotension was different among studies. The sensitivity was 23–50 %. Paroxysmal hypertension, chest pain, flushing, and weakness were the signs/symptoms which had publication bias based on Funnel plot and Egger test (P value