A 65-year-old female, Spanish speaking, Jehovah’s Witness, with a past medical history of insulin-dependent diabetes, PUD 2/2 H pylori treated but not tested for cure, and osteoarthritis who presents with hematemesis of one-day duration and abdominal pain of 5 days duration.The patient woke up vomiting blood on the morning of admission. While the patient was being admitted, the patient developed frank epistaxis with roughly 500 cc of output. GI and ENT where consulted.
While in the ED, the patient had a rigid nasopharyngeal endoscopy which showed some clot in the posterior nasopharynx but no clear source of bleeding.On hospital day 2 the patient received an EGD which demonstrated clot in the oropharynx, upper esophagus, and stomach with no clear source of bleeding. After speaking with the GI fellow it was felt that the bleeding was coming from the oropharynx. ENT was re-consulted and performed a flexible fiber optic nasolaryngoscopy, which demonstrated pink tinged sputum in the left nares, but again no clear source of bleeding. At this point the patient’s hemoglobin had dropped from 12.7 to 10.2. Of note she was never tachycardic and did not have an oxygen requirement.
After further questioning, it was discovered that the patient grew up in Mexico and had known past TB exposure. This case was discussed in a multidisciplinary fashion, and after speaking with radiology about the possibility of doing a tagged RBC scan; however, a CTA was suggested first, and if it was negative a tagged RBC scan would be done.
On hospital day 3 the patient underwent a CTA which revealed 3 lobar pulmonary embolisms. As the patient was still bleeding, she was transferred to the MICU to start a heparin drip.
On hospital day 4 the patient had increased bleeding, thought to be hemoptysis, and heparin was held. The patient had a Doppler of her lower extremities which showed a LLE DVT. An IVC filter was placed later that afternoon.
On hospital day 5 the patient underwent a bronchoscopy which did not show any sources of bleeding and no areas amendable to repair. The patient was restarted on heparin and was successfully bridged to Coumadin without any further instances of bleeding. The patient was discharged home on Coumadin therapy.
Discussion:
This case is a perfect example of anchoring bias with an unusual presentation that led the primary team to the wrong initial diagnosis. The patient presented with abdominal pain and risk factors for an upper GI bleed, but only after a prolonged investigation was the proper diagnosis of PE reached. This process was aided by a multidisciplinary approach and a reassessment of the patient’s presentation.
Conclusions:
Early recognition of physician biases, in this case anchoring bias, is critical in preventing the medical errors that come from making the wrong diagnosis. Delayed diagnosis can injure the patient physically but these errors also negatively impact both patient care and the health care system by increasing costs.
The issues regarding bias can be divided into two groups: cognitive bias and information bias. Cognitive bias includes anchoring and framing effects, while information bias includes personality traits such as tolerance to uncertainty and aversion to ambiguity.
Early recognition of bias is crucial for the practicing physician to optimize their medical decision-making, provide more realistic patient expectations, prevent medical errors, and reduce the cost of hospitalizations.