A series of patients with hospital-acquired diabetic ketoacidosis (HADKA): a descriptive analysis

Discussion

In this retrospective analysis, the characteristics of 25 patients who developed HADKA over a 6-year period were described. Interestingly, almost half of the cases were in patients with T2D. This is an important observation as there is often a misconception around the need for insulin in people with T2D and even the risk of DKA in people with T2D. There has been a recent focus on DKA risk in people with T2D, especially in those prescribed sodium-glucose co-transporter-2 (SGLT-2) inhibitors.⁵ Indeed, in this study four (16%) patients who developed HADKA were concurrently prescribed a SGLT-2 inhibitor, which may have contributed to these patients developing HADKA, though only one (4%) case was attributed to this in the case notes (see Table 2).⁵ Moreover, with increasing T2D duration, pancreatic beta cell function declines and so patients have a greater risk of hyperglycaemic emergencies, such as DKA and HHS with their relative insulinopenia.⁶ The risk of DKA is greater still in those with higher HbA_1c, suggestive of poor glycaemic control and insulin deficiency.⁷ Indeed, the patients presented in this study had a high mean pre-admission HbA_1c of 84.7 mmol/mol suggestive of an at-risk population.

In our cohort, the most common reason for developing HADKA were errors of insulin administration, including omission (six (24.0%)), inappropriate dose reduction (one (4.0%)), and interrupted insulin infusions (two (8.0%)). These are events that should never happen in hospital, and there are an increasing number of UK guidelines and resources available to support inpatient clinical teams in the management of patients with diabetes in a range of settings. During periods of sickness, there is a hormone response (including secretion of endogenous glucocorticoids, adrenaline and glucagon) that drives gluconeogenesis, often resulting in ketone formation. In people with T1D or advanced stages of T2D, the insulin response is frequently insufficient to regulate ketone formation and DKA may occur.⁸ Such patients often require initiation or dose escalation of insulin, though sometimes insulin is inappropriately held or the dose is reduced if, for example, the patient is not eating or drinking.⁸ Improving the awareness of doctors and nurses who are involved with the inpatient care of people with diabetes of the guidelines produced by the JBDS for the management of inpatients with diabetes would likely reduce the incidence of HADKA and other serious or life-threatening diabetes-related adverse clinical events, which are available on the JBDS website.⁹

To date, we are aware of just one previous study exploring the incidence and causes of HADKA, undertaken at an Australian centre.¹⁰ In that study, 12 patients with HADKA were identified across two centres over a 4-year period, with DKA defined as a pH of ≤7.3, bicarbonate of ≤18 mmol/L and positive ketones (urine or plasma). In this group, most patients (75%) had T2D with a mean HbA_1c of 68 mmol/mol. Most patients (83.3%) required intensive care admission, and 16.6% of patients in this cohort died. The results from our cohort are similar, except fewer patients required intensive care admission compared with this Australian study (20.0% vs 83.3%). Our findings add significantly to these previous observations as a more stringent definition of DKA was applied in our cohort, a greater number of patients were included and this is the first description of a UK-based group of patients with HADKA that we are aware of. In the UK, the NaDIA has previously observed a consistent prevalence of HADKA at around 4%.² However, the causes of HADKA were not described by the NaDIA at a national level and the group suggest that, for each event, a serious incident is reported and a root-cause analysis should be undertaken.

A significant proportion of this group (20.0%) required intensive care admission, and the length of hospital stay in the cohort was also significant. This highlights the associated economic burden of HADKA on healthcare resources at a time when such resources are increasingly precious. Of course, the major mortality associated with HADKA in our cohort (and in previous cohorts) may reflect these patients’ general condition and the concurrent illness that precipitated the development of HADKA.¹⁰ Indeed, while DKA did not appear to directly cause death in these patients, its occurrence naturally reflects the severity of their underlying sepsis and comorbidity.

Moving forward in our practice, improving the awareness of clinicians involved with the treatment of people with diabetes in hospital is important; for example, highlighting recent JBDS guidelines to medical and nursing staff for patients undergoing surgery, enteral feeding, long-term steroid treatment and general sickness is important. Additionally, given the recent focus on SGLT-2 inhibitor-associated DKA, it is essential to highlight the need to review these medications and likely hold during periods of illness.^5,11 Of course, clinician education by diabetes specialist teams to reiterate the absolute need to give insulin to people with T1D is needed, and to be cautious with withdrawal or omission of insulin in people with T2D who are admitted with significant concurrent illness, or the risk factors described earlier, because insulin requirements generally increase in times of significant illness to counteract the catabolic state associated with the stress hormone response. Locally this has been achieved through updating and adding to local guidelines in keeping with the JBDS updates, formal and informal teaching to medical and nursing teams in various inpatient specialties, and presenting these data at local meetings.

Given the high prevalence of diabetes in hospitalised patients, inpatient review of all patients with diabetes is not usually possible on a routine basis. However, clinicians should always consider referring patients with complex medical issues or severe non-diabetes-related illness to the diabetes team to support diabetes management, if concerned, and prevent life-threatening inpatient complications, such as HADKA. Further guidance to support inpatient teams in a number of settings is available from the JBDS.⁹