Background: The serum anion gap (AG) is commonly used as a screening tool for acid-base disorders. With modern laboratory techniques using ion-selective electrodes to measure the main electrolyte components of the AG, our definition high AG (HAGMA) should be reviewed.

Aim: This study aims to assess the diagnostic value of AG and to determine a diagnostic threshold for HAGMA in a high-prevalence clinical setting.

Method: Computerized extraction of anonymised data from electronic medical records was performed. A pre-defined criteria included all inpatients of an acute-care hospital who had measurements for organic acids (lactate, ketone or salicylate) paired with a serum urea, electrolyte and creatinine panel.

BACKGROUND: The urea & electrolyte (U&E) panel is a common blood test that is often performed for any ill patients in a hospital. Evaluation of this panel can reveal serious underlying acid-base disturbances, often a harbinger of further haemodynamic compromise. The detection of a gap acidosis often is the first indicator of a serious underlying condition.

The AG is a reflection of the unmeasured anions. The anion gap (AG) is derived from the U&E panel by subtracting measured anions from measured cations:

AG = [Sodium] + [Potassium] - [Chloride] - [Bicarbonate] The serum potassium can be included in the calculation, but due to low variability of serum potassium, most clinicians omit serum potassium in the calculation of AG.

Accumulation of organic acids (e.g. lactic acid, ketones, uremic toxins) results in a high AG metabolic acidosis (HAGMA) and such patients have a significant risk of further clinical deterioration. It has however been challenging to assess for gap acidosis as the range for normal AG can vary greatly due to historical differences in the laboratory techniques used to measure the constituents of AG as well as a component of interindividual variability. In the past, AG were reported (in milliequivalents per liter, mEq/L) in the range of 9-17 milliequivalents/L or 8-16 milliequivalents/L with older laboratory techniques such as flame photometry, colorimetric method. In more recent years with the advent of use of ion-selective electrodes (ISE), published literature suggests that the AG range may be lower as the ISE may give higher readings of serum chloride. Some clinical texts and laboratories have listed the reference range of AG as 3-11 milliequivalents/L or 5-12 milliequivalents/L.

Despite so, the reference range for serum sodium, potassium and chloride themselves have remained the same over the years despite changes in laboratory techniques. With such conflicting information, it is important to re-evaluate the screening or diagnostic threshold of the AG based on modern laboratory techniques.

OBJECTIVES: To assess the diagnostic threshold of AG to detect gap acidosis due to organic acidosis.

METHODS: Study protocol was reviewed and approved by the institutional ethics committee (SingHealth centralised institutional review board: 2019/2463). A waiver of consent was obtained as data was anonymised.

This is a retrospective study. Data will be extracted from electronic medical and laboratory records from all inpatient admissions to an acute hospital. Data extracted will be anonymized.

The search criteria will include the following parameters:

• all inpatient admissions 2017 - 2020 and
• patients with serum lactate, ketones and salicylates performed, and
• patients with urea, electrolytes and creatinine at the same time.

Data to be extracted include baseline demographic data, admission date, and laboratory results including urea / electrolytes / creatinine, albumin, glucose, calcium, phosphate, magnesium, albumin, liver function test, arterial blood test and hemoglobin.

If patients had multiple blood samples taken, only laboratory results from a single sample will be included. This will be selected by an automated algorithm which selects the sample with the most complete information.

Statistical Analysis Plan: The patients will be divided into 2 groups - with / without organic acidosis (elevated serum lactate, ketones and/or salicylates). An area under the curve (AUC) of receiver operating characteristics (ROC) analysis of AG for the detection of organic acidosis will be performed. The optimal cut-off threshold will be explored using the Youden index. Analysis of the diagnostic threshold of AG at different cut-off levels will be performed and the performance of the AG as a screening test at different cut-off thresholds will be assessed by reviewing the sensitivity, specificity, positive/negative predictive values and related indicators. A recommended cut-off value will be determined based on a test sensitivity adjudicated to be acceptable.

Inclusion Criteria:

• Admitted from 01 July 2017 to 31 December 2019
• Patients who have had blood lactate, blood ketones and serum salicylates performed during this time period
• Patient who also had urea / creatinine and a comprehensive panel of electrolytes performed up to +/-12h of blood lactate, blood ketones and serum salicylates being performed

Exclusion Criteria:

• Patient who did not have blood lactate, blood ketones and serum salicylates performed during this time period
• Patient who did not have urea / creatinine and a comprehensive panel of electrolytes performed up to +/-12h of blood lactate, blood ketones and serum salicylates being performed

Inpatients in acute care hospital admitted from 01 July 2017 to 31 December 2019 who were being assessed for organic acidosis