Case Presentation: A 75-year-old man with diabetes mellitus and recently detected pancreatic “cancer” presented with abdominal pain, vomiting and significant weight loss. Two weeks prior he presented to another hospital with laboratory values indicating obstructive jaundice and initial concern for pancreatic cancer. No mass was seen by Computed Tomography (CT) of the abdomen and pelvis. Patient underwent Endoscopic Retrograde Cholangiopancreatography (ERCP), Endoscopic Ultrasound with fine needle aspiration, and placement of a common bile duct stent to relieve obstruction. Pathology showed atypical cells favored to be reactive and deemed nondiagnostic. He was transferred to definitively obtain a tissue diagnosis for presumed malignancy. Gastroenterology consultants recommended repeat CT scans and ERCP, surgical consult, and oncological consult for further management of patient’s “pancreatic cancer”. Patient underwent repeat CT abdomen/pelvis and in addition a CT chest for staging. CT again showed no mass but “prominence of the pancreatic head in a patient with known pancreatic malignancy”. Pathology from repeat ERCP showed tissue consistent with nonmalignant cells once again. At this time a differential was considered. IgG4 levels were sent and returned elevated to >1000 supporting the diagnosis of autoimmune pancreatitis. He was started on steroids, had significant improvement of his symptoms during his hospitalization, and continued to improvement upon follow-up.
Discussion: A common principle in medical education is that the single, most common diagnosis is usually the correct one. Occam’s razor, while helpful, often creates the perfect setup for anchoring bias. Hospitalists are often guided by emergency room physicians, radiographic imaging, and laboratory values that point them to the most obvious diagnosis. In this case, multiple physicians continued to ascertain that the patient had malignancy and implored further workup without exploring other differentials even when repeat imaging showed no mass. Between 40,000 and 80,000 U.S. hospital deaths occur annually from missed, delayed, or wrong diagnoses. Anchoring causes a delay in diagnosis, lengthy and costly hospital admissions, abundant use of resources including imaging modalities, and multiple repetitive procedures. This causes emotional uncertainty and anxiety for the patient and family members. Multiple methods have been suggested to reduce anchoring bias including understanding the limitations of imaging, improved communication among physicians, continuing education, and systems based interventions.
Conclusions: While it is imperative to exclude the most life-threatening and most prevalent diagnoses, it is prudent to continue to think broadly and reexamine data frequently.