Impact of Hub & Spoke organization on the measurement of plasma ammonia

To the Editor,

We read with great interest the UK National audit of the measurement of ammonia by Aitkenhead. ¹ As indicated by the author, laboratories should improve all processes to avoid delaying the diagnosis and management of a medical emergency due to hyperammonaemia.

Increased blood ammonia can be caused by inherited defects of the urea cycle, hepatic or kidney disease and valproate administration. However, pre-analytical errors may also spuriously increase ammonia levels.^2,3 Among these, storage conditions and the time of analysis are crucial to prevent an ammonia increase due to in vitro production, spontaneous release from or breakdown of erythrocytes. ³ To reduce the artefactual increase of ammonia, samples must be transported in ice water and analysed as soon as possible.

We conducted a real-life retrospective study to evaluate the difference between the results of ammonia assayed in a Hub laboratory versus a Spoke laboratory. Indeed, the organization of Clinical Pathology laboratories in our hospital network dictates that outpatient samples taken in the secondary hospitals are sent for testing to the Hub laboratory, while inpatient samples are analysed in the internal (Spoke) laboratory.

We evaluated all EDTA samples collected from November 2022 to November 2023 and ammonia measurements were performed on the Alinity platform (Abbott). Ammonia was requested for 84 outpatients (110 samples, age range 0–79, median 5 years) and 278 inpatients (328 samples, age range 0–16, median 1 year).

The haemolysis index (HI) was automatically measured on all samples, and haemolysed test results were not reported for outpatients. Otherwise, for inpatient samples, results were reported whilst also reporting the degree of haemolysis as a free haemoglobin concentration, and a comment recommending a repeat sample. ⁴ The HI threshold was set at 50 (free haemoglobin concentration ≥0.50 g/L). The Mann–Whitney test (independent samples) or Wilcoxon test (paired samples) were performed using MedCalc software. A P value <.05 denoted statistical significance.

The haemolysis index was higher in outpatients compared to inpatient samples, with a median of 28 (interquartile range, IQR: 19–42) versus 16 (IQR: 11–37), respectively (P < .0001).

Comparison between HI determined on plasma collected in EDTA tubes for ammonia determination and on serum from the same blood collection (n = 99) showed a median of 27 (IQR = 19-42) versus 5 (IQR = 3-11) in outpatients (P < .0001). HI on plasma in EDTA tubes from inpatients (N = 276) showed a median of 16 (IQR = 11-36) compared to a median of 8 (IQR = 4-21) on the same blood collected in lithium heparin tubes for a different test (P < .0001).

When we consider samples with HI ≥50, results were not reported for 18 (16.4%) outpatient samples. In contrast, for inpatients’ samples (n = 54, 16.5%) the results were always released with an appropriate comment. Although the percentage of haemolyzed samples was equal, the two populations differed in the proportion of patients less than 1 year of age: 7% in outpatients compared to 64% in inpatients.

Based on our analysis and the suggestion reported by Aitkenhead, ¹ we revised the management of ammonia testing within our ‘Hub-and-Spoke’ organization. To ensure quick delivery of samples, rapid analysis of all ammonia requests and the release of results with an appropriate comment, the measurement of plasma ammonia is now also performed at the spoke laboratory for outpatients. Finally, we will also be implementing age-related reference intervals in the near future.