Background: The observation unit is frequently a destination for patients with yet undifferentiated conditions. The goal of the observation stay is to complete a crucial diagnostic test or treatment trial that should help specify the diagnosis, but that is not always the case. Understanding the frequency of diagnoses that remain unspecified after an observation visit can help identify areas for improvement in diagnostic protocols. We sought to identify the highest frequency unspecified presenting diagnoses to the observation unit and how often those diagnoses remained unspecified at discharge.
Methods: All patient encounters to the hospitalist-run observation unit of a large tertiary urban referral center connected to multiple community emergency departments from September 2014 until August 2015 were reviewed. Surgical and post-procedural patients managed by their respective operators were excluded from analysis. Diagnoses listed on presentation and on discharge were collected. Additional demographic information, such as age, gender, length of stay (LOS), number of consults ordered, and need for inpatient admission were also collected.
Results: A total of 4456 observation counters were reviewed during the study period. The most common overall presenting diagnosis was “Chest pain, unspecified” with 1058 (24%) encounters. At discharge, 63% of these encounters maintained the same unspecified chest pain diagnosis and were associated with a length of stay (LOS) of 26 hours, a 1% rate of admission, and an average of 0.16 consults/patient. In contrast, the specific diagnosis of coronary atherosclerosis, which only accounted for 2% of chest pain diagnoses, had an average LOS of 28.7 hours, admission rate of 78%, and 0.83 consults/patient. Overall there were 106 different discharge diagnoses for unspecified chest pain patients. “Abdominal pain, unspecified” was the second most common unspecified diagnosis and the 6th most common overall in the unit with 219 (5%) encounters, and whose most common discharge diagnosis was still unspecified for 18% of encounters, which had an average LOS of 29.6 hours, a 0% admission rate, and an average of 0.40 consults per encounter. Of abdominal pain patients, persistent vomiting, was the most common specific discharge diagnosis at 6% of encounters, and overall there were 101 different discharge diagnoses for unspecified abdominal pain patients.
Conclusions: The observation unit is often a site to determine the etiology of undifferentiated presenting complaints, with over 30% of all encounters carrying an “unspecified” initial diagnosis. However, these unspecified diagnoses lack further differentiation even at the time of discharge from the observation unit for many patients. This identifies an area for improvement in documentation within the unit, which will potentially be improved with a higher degree of diagnosis specificity capable with ICD10 codes. There is also the opportunity to identify more comprehensive diagnosis pathways leading to more accurate differntiation, to reduce the number of patients leaving the unit still with an unspecified diagnosis.