The significance of ketosis in perioperative care

Management of hyperglycaemia in the perioperative setting

Perioperative hyperglycaemia is largely driven by the physiological stress response to surgery. Activation of the neuroendocrine axis increases circulating concentrations of counter-regulatory hormones, including cortisol, catecholamines, glucagon and growth hormone, promoting hepatic gluconeogenesis and glycogenolysis [1]. In individuals with diabetes, pre-existing insulin resistance and impaired glucose regulation amplify this response [2].

Preoperatively, glycaemic control is achieved through the use of both insulin and non-insulin glucose-lowering agents, with adjustments made during fasting. Glycaemic targets vary across professional bodies. However, the American Diabetes Association (ADA) recommends maintaining a target between 5.6 and 10 mmol/L. A 2023 systematic review demonstrated that intensive glycaemic control offered little or no reduction in all-cause mortality, infection rates or length of hospital or intensive care unit (ICU) stay compared with conventional strategies, while significantly increasing the incidence of hypoglycaemia [3]. These findings support a more balanced approach rather than aggressive normalisation of capillary blood glucose levels.

Non-insulin agents are typically withheld on the day of surgery. Sodium-glucose cotransporter 2 (SGLT-2) inhibitors should be discontinued 3–4 days preoperatively due to the risk of hypovolaemia from osmotic diuresis and the development of euglycaemic DKA. As such, several authors have advocated for perioperative ketone monitoring in patients actively on SGLT-2 inhibitors [4].

On the day of surgery, prandial insulin is omitted whilst basal insulin doses are reduced. Perioperative management involves regular capillary blood glucose monitoring with supplemental short-acting insulin administration where required.

Perioperative BOHB monitoring

The increasing use of ketone monitoring perioperatively reflects the importance of early recognition of diabetic emergencies. BOHB forms a central component in diagnosing DKA. According to the National Institute for Health and Care Excellence (NICE) guidelines, DKA is defined by blood ketones ≥3 mmol/L or urine ketones ≥2+, alongside hyperglycaemia and acidosis [5]. Elevated ketones reflect relative or absolute insulin deficiency leading to increased lipolysis and hepatic ketogenesis.

Although ketones provide an alternative oxidative substrate during carbohydrate deprivation – such as in perioperative fasting – excessive accumulation may result in metabolic acidosis and haemodynamic instability [6]. In diabetic patients undergoing surgery, the combined effects of operative stress and metabolic vulnerability further increase the risk of decompensation [7]. Consequently, elective procedures are generally deferred in patients with overt metabolic instability, particularly in established DKA [7].

The perioperative period confers an increased risk of diabetic crises. Surgical stress and general anaesthesia may precipitate metabolic decompensation in susceptible individuals, justifying a low diagnostic threshold [8]. This concern is especially relevant in patients pretreated with SGLT-2 inhibitors, in whom euglycaemic ketoacidosis has emerged as an important complication. A 2023 retrospective analysis reported that patients on SGLT-2 inhibitors developed varying degrees of ketoacidosis despite a preoperative cessation period of 1.5 days [9]. These findings reinforce the role of ketone surveillance in these cohorts.

However, reliance on BOHB measurements in isolation may result in excessive caution in otherwise clinically stable patients. Kilpatrick et al. [10] demonstrated that a bicarbonate level of 15 mmol/L – consistent with the NICE diagnostic criteria – corresponded to a BOHB level of 4.7 mmol/L, well above the diagnostic threshold of 3 mmol/L [10]. This discordance illustrates that ketone concentrations and acid–base parameters do not always align precisely and, as such, BOHB levels should be interpreted alongside venous pH and bicarbonate to provide a more comprehensive assessment of metabolic status.

Inconsistencies also exist in defining resolution of DKA. The Joint British Diabetes Societies for Inpatient Care (2022) define resolution as plasma ketones <0.6 mmol/L, and venous pH ≥7.3 or bicarbonate ≥18 mmol/L [11]. In contrast, Tremblay et al. [12] reported that a plasma ketone threshold ≤1.5 mmol/L demonstrated adequate sensitivity and specificity for resolution [12]. Although ketone monitoring remains integral to both diagnosis and resolution, more evidence is required to harmonise current practice and inconsistencies underscore the importance of clinical context.

Starvation ketosis in the perioperative context

When ketone levels do not meet diagnostic criteria for DKA, alternative causes must be considered. These include starvation ketosis, chronic alcohol use and persistent vomiting, particularly in pregnancy [13]. Starvation ketosis may develop during prolonged preoperative fasting and is more frequently observed in critically ill patients [14].

In otherwise healthy individuals, mild ketosis typically develops after 12–14 hours of fasting [15]. However, a 2022 retrospective cohort study of 1,831 perioperative patients demonstrated that urinary ketosis and even metabolic acidosis may occur following relatively short fasting intervals [16]. Younger patients and those with lower body mass index and lower American Society of Anesthesiologists physical status classification appeared more susceptible, possibly due to higher basal metabolic rates and rapid glycogen depletion resulting in ketogenesis acceleration [16].

In critically ill patients, untreated starvation ketoacidosis may contribute to haemodynamic instability. A retrospective observational ICU study reported that perioperative surgical patients with ketonuria had a higher incidence of shock compared with medical patients [14]. Progressive acidaemia may impair myocardial contractility, promote vasodilation and reduce responsiveness to catecholamines, thereby contributing to haemodynamic compromise [17].

It is, however, important to distinguish biochemical ketosis from clinically significant acidosis. Reports of prolonged ICU admission attributable solely to perioperative starvation ketosis are largely confined to case reports, most of which documented decompensated metabolic acidosis on arterial blood gas analyses [18, 19]. In the absence of acidaemia, mild-to-modest ketonaemia is unlikely to carry equivalent physiological consequences. From an anaesthetic perspective therefore, it is the presence of metabolic acidosis rather than isolated ketonaemia that is the principal determinant of increased cardiovascular and ventilatory risk.

Balancing metabolic optimisation and surgical urgency

The central clinical question is not whether metabolic abnormalities should be recognised, but whether their presence necessitates deferral of surgery. Whilst careful glycaemic optimisation is desirable, postponement solely on the basis of isolated biochemical parameters may not always be justified.

The ADA advises that surgery should not be deferred exclusively due to HbA1c or glucose targets in the absence of compelling interventional evidence, as such delays may deny patients timely or life-saving treatment [20]. Decision-making should therefore integrate the overall clinical picture, including haemodynamic stability, acid–base status and the urgency of surgical source control.

Poor perioperative management is associated with adverse surgical outcomes and may precipitate diabetic emergencies. Hyperglycaemia and significant hyperketonaemia may alter physiological responses to anaesthesia and increase operative risk. However, it is equally important not to conflate starvation ketosis with DKA, sepsis-associated acidosis or lactic acidosis. While prolonged fasting commonly results in elevated ketones, clinically significant metabolic acidosis is uncommon in otherwise stable individuals.

Perioperative management must also be contextualised within institutional constraints, including laboratory access, staffing and theatre availability. Optimisation strategies should therefore be pragmatic, balancing metabolic stability against the risks of delaying definitive treatment.

In the case vignette described, the absence of overt metabolic decompensation and haemodynamic compromise suggests that the isolated elevation of BOHB should not have delayed timely surgical source control of the carbuncle in the context of staphylococcal bacteraemia. Careful interpretation of BOHB alongside acid–base parameters would have supported proceeding with surgical intervention without unnecessary deferment.

Research ethics statement

Not applicable to this article.

Consent for publication statement

This narrative article is based on a fictional clinical vignette and does not describe a real patient or clinical encounter so consent for publication is not necessary. Any images used are solely for illustrative purposes and are not related to the vignette. Where real patient images are included, the authors confirm that written informed consent for publication has been obtained for this purpose in accordance with the journal’s patient consent requirements.