A paradox is a statement or proposition that, despite sound (or apparently sound) reasoning from acceptable premises, leads to a conclusion or action that is senseless. In the ammonia paradox, measuring ammonia levels in confused cirrhotic patients with hepatic encephalopathy (HE) is, to paraphrase Adrian Reuben, MD, an action that creates greater confusion in those ordering
the test than the confusion present in the patient (1).

Uncertainty creates, in the provider, a desire to resolve it. Tests are ordered to improve certainty and reasons to test include establishing a diagnosis or prognosis, to guide management based upon the result, or to exclude a diagnosis. With ammonia levels, the clinician is hoping the level will provide any or all of these.

In noncirrhotic patients with fulminant hepatic failure, when the ammonia level is greater than 150 µg/dL it does predict prognosis (cerebral edema or death) and a need for a higher level of care like a transplant program. However, in those with cirrhosis, the ammonia level has little, if any, predictive or clinical value in HE for multiple reasons (2) (Table 1).

The Ammonia Level Project

Until recently, providers ordered approximately 14,000 ammonia level tests annually in the Intermountain Healthcare system. This is despite several publications and numerous local educational endeavors about the absence of the test’s value in patients with HE. Further analysis confirmed that approximately 90% of the 14,000 were ordered inappropriately.

With the ammonia level project, the idea for change started with the observation that providers frequently referred patients for ammonia level assessment. Preliminary analysis identified that the emergency department (ED) appeared to order the test most frequently. Further analysis showed ammonia level results were often a prerequisite before admission and low levels appeared to generate additional tests, like head CT scans.

Some patients admitted with HE underwent serial ammonia levels during their hospitalization. In certain instances, despite clinical resolution of HE, patient discharges were delayed while awaiting normalization of the level. When asked, all parties involved stated that someone associated with the patient’s evaluation or management needed the result—although hepatology, the consulting service in several cases, was not interested in the result.

Designing and Testing an Intervention

We approached the process in three phases (Table 2). Initially we deployed educational programs for the ED and hospitalist services, shared reading material on HE and ammonia levels, and conducted interviews with involved parties seeking to understand the reason for ordering the test. The impact of this first phase on tests ordered was limited and reduced the number of tests minimally.

For the second phase, we identified the electronic medical record (EMR) as a common step in placing ammonia level orders. Given the objective to reduce inappropriate ordering, we selected an order time decision support solution.

The intervention consists of two parts. First, upon ordering an ammonia level test, an informational alert suggests the clinician consider the appropriateness of the test. Second, if the clinician chooses to proceed with testing, a prompt appears to enter the indication for testing and is required to proceed (Figure 1).

In reviewing provider ordering patterns before and after implementation of the EMR based intervention, it was noted that some providers were influenced by the solution more than others (Figure 2, online) and some groups more than others (ER physicians ~45% reduction versus outpatient physicians 3.5%). This suggested an opportunity to further reduce inappropriate orders with additional interventions, such as report cards to providers showing their ordering frequency as compared to their peers.

The average number of tests dropped from 1150 to 650 per month after implementation of the EMR based intervention. At a cost of $42 per test, savings approach $250,000 a year. Additional benefits yet to be confirmed include reduced ED length of stay (LOS)—increased ED bed availability—as well as shorter time to treatment initiation, reduced hospital LOS, and reduced subsequent testing, such as head CT.

In an era that requires prioritization of value-based care principals, initiatives to reduce inappropriate testing are increasingly important. At Intermountain, reduced inappropriate ammonia level testing has yielded cost savings without diminishing. The initiative has enhanced value. There are many opportunities, across multiple clinical settings, to which these principals can be applied to generate substantial savings while maintaining the highest quality of care. We only ask that you look for these and then tell us your story of success.

James Hellewell, MD, MS, is a clinical informatics medical director at Intermountain Healthcare in Salt Lake City. +Email: [email protected]

Richard K. Gilroy, MD, is the medical director of liver transplant and hepatology for the Intermountain Health System. +Email: [email protected]


  1. Reuben A. There is nothin' like a Dame. Hepatology 2002;35:983−5.
  2. Ge PS, Runyon BA. Serum ammonia level for the evaluation of hepatic encephalopathy. JAMA 2014;312:643−4.