Out of target glycemic control and severe hypoglycemia in older adults with diabetes presenting to the emergency department

Abstract

Background

Severe hypoglycemia is a frequent and high-risk complication to treatment of diabetes in older adults. It is often related to strict glycemic targets, treatment intensification, and age-related physiological vulnerability. Long-term patterns of HbA1c concentrations before hypoglycemia-related hospital admissions remain poorly described. This study aimed to characterize pre-admission HbA1c concentration value trajectories in older individuals with diabetes and to explore if these patterns indicate overtreatment.

Methods

We performed a retrospective cohort study at the Emergency Department of North Zealand Hospital, Denmark. Patients aged ≥ 65 years admitted with severe hypoglycemia between 2021 and 2024 were included. Data were obtained from electronic systems. HbA1c concentration values from 0 to 5 years before admission were analyzed. Linear regression was used to estimate HbA1c concentration changes over time, stratified by diabetes type, HbA1c category ranges, and age.

Results

A total of 212 patients were included (68 with type 1 diabetes and 144 with type 2 diabetes), accounting for 301 hypoglycemia-related admissions and 2,574 HbA1c measurements. HbA1c distributions suggested overtreatment in both groups, particularly in type 2 diabetes, where 49% of admissions occurred with HbA1c < 53 mmol/mol and 26% <43 mmol/mol. Conversely, 27% of type 1 diabetes and 23% of type 2 diabetes were admitted with HbA1c > 70 mmol/mol. Longitudinal analyses demonstrated significant declines in HbA1c prior to admission among patients with low HbA1c at admittance. In type 1 diabetes, HbA1c declined markedly in the lowest two HbA1c groups (− 2.85 and − 2.75 mmol/mol per 100 days), while higher baseline groups showed stable-modestly increasing trajectories. In type 2 diabetes, HbA1c declined in admissions with baseline values ≤ 53 mmol/mol (− 0.46 to − 1.86 mmol/mol per 100 days), whereas admissions with higher baseline HbA1c showed increasing trajectories, most pronounced in those with HbA1c > 53 mmol/mol (+ 0.85 mmol/mol per 100 days).

Conclusions

Our data suggest that a large fraction of admissions of older adults with severe hypoglycemia are associated with overtreatment, and could potentially have been prevented by due time treatment de-intensification.

Clinical trial number

Not applicable.

Background

Hypoglycemia is the most serious side effect to glucose-lowering treatment of diabetes mellitus (DM) with insulin or sulfonylureas [1, 2]. Due to progressive beta-cell failure with increasing diabetes duration, older adults with type 2 diabetes (T2DM) are frequently treated with insulin or sulfonylureas (SU), therapies associated with an increased risk of severe hypoglycemia [3]. In type 1 diabetes (T1DM), lifelong dependence on exogenous insulin and impaired counter-regulatory responses similarly confer a high risk of severe hypoglycemia, as defined by the American Diabetes Association [4, 5]. These events are associated with serious clinical consequences such as falls, fractures and risk of cardiovascular events in older individuals [6–8] and represent a substantial global burden on health and health expenditures [9]. As the global populations age, and treatment options have become more effective and improve survival, the prevalence of diabetes in older age groups continues to expand.

Current international guidelines recommend that glycemic targets in older adults should be individualized and generally relaxed as comorbidity and frailty increase. Rather than pursuing tight glycemic control, targets are often set in higher hemoglobin A1c (HbA1c) concentration ranges, for example around 53–58 mmol/mol in otherwise healthy older adults and up to approximately 64 mmol/mol or higher in those with complex comorbidity, limited life expectancy, or high treatment burden [1, 10]. However, there is considerable clinical experience suggesting that glucose-lowering treatment in older adults is not always revised in a timely manner, and that many hospitalizations for severe hypoglycemia are preceded by a decline in HbA1c concentration values to well within the appropriate range [1]. However, there remains a lack of longitudinal data describing HbA1c concentration value patterns before emergency department (ED) admissions for hypoglycemia in older adults.

We hypothesized that overtreatment is a common cause of severe hypoglycemia leading to ED admission in elderly adults with diabetes. Therefore, we determined admission HbA1c concentrations and characterized pre-admission HbA1c concentration value trajectories among older adults with diabetes, hospitalized with severe hypoglycemia.

Methods

This is a retrospective single-center study conducted at the ED of North Zealand Hospital, Copenhagen University Hospital, Denmark. The hospital is a large tertiary care center, provides acute care across all medical specialties, except gynecology and obstetrics, and the ED treats approximately 150 patients per day, with roughly two third of patients aged 65 years or older. We included patients aged ≥ 65 years who were admitted with severe hypoglycemia between 2021 and 2024. Data were extracted from the electronic health record system Epic, and the laboratory information system Labka.

The inclusion criteria were as follows; Age ≥ 65 years, diagnosed with DM, with documented hypoglycemia at, or shortly after arrival. Exclusion criteria were absence of a Danish Civil Registration Number (CPR), or residence abroad. Data collected included demographics, prior hypoglycemia, antidiabetic medications, laboratory results, i.e., plasma HbA1c and glucose concentrations.

We collected data for a total period of 4 years from January 1, 2021, to December 31, 2024 to ensure that current treatment guidelines were relevant for the period. The electronic journal system and Labka reported 231 unique patients admitted with hypoglycemia, as defined by a blood glucose of 4 or below, measured in mmol/L or a registered hypoglycemia diagnosis. Of the 231 patients, we were able to find matching HbA1c concentration values for 212 patients who thus, were included in the study.

We stratified patients by their last HbA1c concentration values prior to admission, age, and diabetes type to allow clinically meaningful comparisons across levels of glycemic control, age-related physiology, and diabetes pathophysiology.

HbA1c concentration range groups (mmol/mol): Group 1 ≤ 43, Group 2 (43–48), Group 3 (48–53), Group 4 (> 53) [4]. These groups were defined according to internationally recognized glycemic targets, representing significantly below target (≤ 43), below target (43–48), most probably below target (48–53), possibly below target (53–58), probably in target range (58–70), and possibly above target range (> 70) and used to assess the prevalence of suspected overtreatment. Note that the same patient can be counted multiple times, when admitted more than once.

Age groups (65–75, 75–85, ≥ 85 years) were chosen to reflect distinct geriatric risk profiles. Patients aged 65–75 years are typically younger seniors with a somewhat preserved homeostatic reserve, while those aged 75–85 and ≥ 85 often have increased comorbidity, polypharmacy, and physiological vulnerability to hypoglycemia.

Diabetes type was divided into T1DM and T2DM, as the disease mechanisms, insulin dependency, and treatment regimens differ markedly, influencing both HbA1c concentration dynamics and hypoglycemia risk. Furthermore, to discriminate between patients treated in primary care (mainly T2DM) and at specialized diabetes outpatient clinics (T1DM).

Statistical analyses

All statistical analyses were performed in Python (version 3.11), using the library Statsmodels for regression modeling, and interaction analyses. For plotting regression trajectories, confidence intervals, and bar charts, we used the Matplotlib and Seaborn libraries.

In our descriptive data we summarized categorical variables with counts, percentages, means ± standard deviation (SD). We plotted regression trajectories for both T1DM and T2DM and stratified them into the HbA1c concentration subgroups. The regression lines of HbA1c concentrations over time were plotted with shaded 95% confidence bands.

Medication use was reported as counts and percentages within each diabetes type. Differences in the distribution of pharmacological treatments between T1DM and T2DM were assessed using the χ² test for categorical variables, and age differences were analyzed with the Student’s t-test. Statistical significance was assessed using confidence intervals and p values < 0.05.

The main analysis was linear regression of HbA1c concentrations against time to admission. Slopes were reported as HbA1c concentration change per 100 days. Because different patients had different numbers of HbA1c measurements prior to admission and because error variance might not be constant across patients, heteroskedasticity-consistent HC3 standard errors were computed.

We approached the examination of changes in HbA1c concentration values by performing stratified regressions: Separate models fitted within each subgroup (DM type, HbA1c concentration range group, age range group).

Associations between medication use and HbA1c concentration values at admission were examined with linear regression, adjusting for age and sex. Models were fitted separately for T1DM and T2DM.

Ethics

Data access was granted by the Institutional Review Bord (IRB) of North Zealand Hospital to assess departmental health care quality. Head of quality assurance and patient safety at North Zealand Hospital, Karen Gliese Nielsen approved the project on behalf of the IRB, her signature is attached in supplementary files. The reference number for this project is: 25,000,839. The data for this article was processed in an anonymized fashion.

Results

A total of 212 patients (Table 1) were admitted during the 4-year period that matched our inclusion criteria. Within this group, 68 patients (32%) had T1DM, and 144 (68%) had T2DM. The mean age was 77 years, and 42% were female.

Table 1.

Patient demographics. Data are presented as n (%) or mean ± SD

Characteristic	T1DM (N = 68)	T2DM (N = 144)	p-value
Gender
Male	31 (45.6%)	93 (64.6%)	0.013*
Female	37 (54.4%)	51 (35.4%)
Age, mean ± SD (years)	75.6 ± 6.2	78.0 ± 6.4	0.010*
Medication use
Short-acting insulin	63 (93%)	47 (33%)	< 0.001*
Long-acting insulin	61 (90%)	93 (65%)	< 0.001*
Combination insulin	4 (6%)	16 (11%)	0.335
Sulfonylurea	1 (1%)	11 (8%)	0.135
SGLT2 inhibitor	5 (7%)	44 (31%)	< 0.001*
GLP-1 analogue	2 (3%)	29 (20%)	0.002*
DPP-4 inhibitor	0 (0%)	23 (16%)	0.001*
Metformin	5 (7%)	72 (50%)	< 0.001*

P-values refer to comparisons between T1DM and T2DM groups

In T1DM, 90% were on a basal bolus regimen, 6% were on a premix regimen and 4% were on basal insulin only. Use of non-insulin glucose-lowering agents was rare among patients with T1DM, with only a few receiving metformin (n = 5), sodium-glucose cotransporter-2 (SGLT2) inhibitors (n = 5), or glucagon-like peptide-1 receptor (GLP-1) analogues (n = 2). Among the 68 patients with T1DM, 60 (88.2%) were treated with a basal–bolus regimen including both short- and long-acting insulin, whereas only 3 patients (4.4%) were prescribed short-acting insulin without concomitant long-acting insulin. As information regarding insulin delivery modality was unavailable, the use of hybrid closed-loop (HCL) insulin pumps could not be determined directly.

In contrast, patients with T2DM demonstrated a broader treatment profile. A total of 65% were treated with long-acting insulin while 33% received short-acting insulin. Of the patients receiving short-acting insulin, the vast majority would also receive long-acting insulin. 11.1% were treated with combination insulin formulations. Sulfonylureas were used by 8% of patients, and no overlap between sulfonylurea use and insulin therapy was observed. Besides insulin and SU, a substantial proportion were treated with oral agents, including metformin (n = 72), SGLT2 inhibitors (n = 44), dipeptidyl peptidase-4 (DPP-4) inhibitors (n = 23), and SU (n = 11). Injectable GLP-1 analogues were used by 29 patients, and a smaller subset received combination therapies (metformin/DPP-4 inhibitor, n = 5; combination insulin, n = 18).

Further characterization of glucose lowering therapy among patients with T2DM demonstrated that 124 of 144 patients (86.1%) admitted with severe hypoglycemia were treated with either insulin or sulfonylurea therapy. Sulfonylurea use was uncommon (11/144, 7.6%), and only one patient received sulfonylurea monotherapy, whereas the remaining sulfonylurea treated patients received combination therapy with metformin, SGLT2 inhibitors, and/or DPP-4 inhibitors. No patients were treated concomitantly with both insulin and sulfonylureas.

GLP-1 receptor agonists were not used as monotherapy and were not prescribed in combination with sulfonylureas. Of the 29 patients treated with GLP-1 receptor agonists, 25 (86.2%) received at least one additional oral glucose-lowering medication, and nearly all also received concomitant insulin therapy. Also among the T2DM patients, 20 (13,9%) were admitted with severe hypoglycemia despite receiving neither insulin nor sulfonylurea. In this subgroup, the most commonly prescribed medications were metformin (16/20), DPP-4 inhibitors (8/20), and SGLT2 inhibitors (6/20).

A total of 301 admissions were recorded related to the 212 patients included in the study during the 4-year period of data collection. HbA1c concentration values up to five years prior to admission were collected which amounted to 2,574 HbA1c concentration values that were available for statistical analyses.

Prevalence of HbA1c concentrations below target

Distribution of admissions into HbA1c concentration range groups, as defined in the method section, is shown in Fig. 1. In T1DM, the Hba1c concentration value closest to admission was below 53 mmol/mol in 27% of the patients, which could be consistent with potential overtreatment. In T2DM this figure was much higher, 49% and with a large fraction of 26% with HbA1c concentration values below 43 mmol/mol, indicating severe overtreatment. The fraction of patients closer to a more appropriate target (53–70 mmol/mol) was 46% and 26% in T1DM and T2DM, respectively. Finally, a significant proportion of 27% and 23% in T1DM and T2DM, respectively, had HbA1c concentration values above 70mmol/mol, suggestive of treatment possibly above target range. The T2DM group had 48 admissions with HbA1c concentration values > 70 mmol/mol. These admissions came from 30 unique patients, of which 18 (60%) were on fast acting insulin.

Fig. 1.

Distribution of patients into HbA1c groups, divided by type of DM. Above each column is the percentage that this group constitutes, out of all admittances from their respective DM type

HbA1c concentration value trajectories pre admission

To determine if there was a significant decline in HbA1c concentration values prior to admission, trajectories were analyzed within each HbA1c group, by linear regression (Fig. 2). The T1DM group 1 showed a significant decrease in HbA1c concentration value, (-2.85 mmol/mol per 100 days (95% CI -3.56 to -2.13, p < 0.001)). For T1DM group 2, HbA1c concentration values also declined significantly (-2.75 mmol/mol per 100 days (95% CI -3.62 to -1.89, p < 0.001)). T1DM group 3 showed a non-significant, but downward trend (-0.25 mmol/mol per 100 days (95% CI -0.61 to + 0.12, p = 0.179)). T1DM group 4 showed a non-significant, but rising trend in HbA1c concentration values (+ 0.36 mmol/mol per 100 days (95% CI -0.02 to + 0.75, p = 0.065)).

Fig. 2.

The correlations between HbA1c concentrations and time to emergency department admittance. The column on the left shows correlations seen in T1DM for HbA1c range group 1–4. The column on the right shows correlations seen in T2DM for HbA1c range group 1–4

The T2DM group 1 had a significant decrease in HbA1c concentration values (-0.46 mmol/mol per 100 days (95% CI -0.84 to -0.09, p = 0.016)). For T2DM group 2, HbA1c concentration values also significantly declined (-1.86 mmol/mol per 100 days (95% CI -2.37 to -1.35, p < 0.001). T2DM group 3 showed a significant decline as well (-0.90 mmol/mol per 100 days (95% CI -1.37 to -0.43, p < 0.001)). And lastly for T2DM group 4, HbA1c concentration values increased significantly prior to admission (+ 0.85 mmol/mol per 100 days (95% CI + 0.57 to + 1.14, p < 0.001)).

Analyses according to age groups

In T1DM, the proportion of cases with HbA1c concentration values below 53 mmol/mol increased to 50% in the group aged 85 years or older. For ages ranging from 65 to 74, 30% had a HbA1c concentration below 53 mmol/mol. Ages ranging from 75 to 84 years amounted to 19% with a HbA1c concentration below 53 mmol/mol (Fig. 3). In T2DM the age group ranging from 65 to 74 years of age had 48% with an HbA1c concentration below 53 mmol/mol. For ages ranging from 75 to 84 years this proportion increased to 52%. In the age group ranging from 85 + years of age 48% had a HbA1c concentration below 53 mmol/mol. Because a large proportion of these patients were distributed into group 4 (HbA1c > 53 mmol/mol), and because the number of patients with T2DM admitted to the ED was much larger than the patients with T1DM, we further expanded this group. To better characterize this group group 4 was further divided into the following groups: HbA1c concentrations: 53–58 mmol/mol, 58–70 mmol/mol, and 70 + mmol/mol (Fig. 4).

Fig. 3.

Distribution of patients by HbA1c group, in patients with T1DM. Age group is displayed on the X-axis. Number of patients is displayed on the Y-axis. The percentage of patients distributed in each group is displayed above each column

Fig. 4.

Distribution of patients by HbA1c group, in patients with T2DM. Age group is displayed on the X-axis. Number of patients is displayed on the Y-axis. The percentage of patients distributed in each group is displayed above each column

Discussion

This study shows that most elderly people with diabetes admitted to hospital for severe hypoglycemia were out of HbA1c target range at admission. This was particularly true for T2DM, where only 26% were in a reasonable target range, whereas 50% had HbA1c values lower than recommended by guidelines [4], suggesting overtreatment. Furthermore, analysis of HbA1c trajectories prior to admission showed a clinically significant decline during the years up to admission in those with low HbA1c values. Together these findings suggest there are many challenges in balancing glycemic control and hypoglycemia risk in older adults, and a need for more attention on timely reduction of glucose-lowering therapy in the elderly with diabetes. Many admissions could potentially have been prevented with earlier changes in HbA1c target values and treatment plans.

HbA1c concentration trajectories

Our analyses of HbA1c concentration trajectories revealed differences by diabetes type and baseline glycemic control. Among patients with T1DM, the declining HbA1c concentrations observed in those with initially tight control (≤ 48 mmol/mol) likely reflect overly intensive management resulting from age-related declining insulin needs, not accommodated by reduction of insulin dosage. In contrast, T1DM patients with higher HbA1c values demonstrated stable or slightly rising trends, implying that factors other than treatment intensity, such as erroneous or inappropriate insulin use, comorbidity, or cognitive impairment, could be the underlying reason for their hypoglycemia. Although information regarding insulin delivery modality was not available, only 3 of 68 patients with T1DM were prescribed short-acting insulin without concomitant long-acting insulin. In modern clinical practice, this pattern would most commonly reflect insulin pump therapy, as patients receiving multiple daily injections generally require both short- and long-acting insulin. Consequently, patients potentially treated with HCL insulin pumps appear to constitute only a small proportion of T1DM admissions in our cohort. While this observation should be interpreted with caution, it raises the possibility that HCL insulin pump therapy may be underrepresented among older adults admitted with severe hypoglycemia. Future studies including direct information on insulin delivery modality are needed to evaluate whether HCL systems reduce the risk of severe hypoglycemia in cognitively intact older adults with T1DM.

In T2DM, different HbA1c concentration trajectories emerged. Patients with strict or moderate baseline control (≤ 53 mmol/mol) exhibited consistent HbA1c concentration reductions prior to admission, whereas those with a higher baseline value (> 53 mmol/mol) showed rising HbA1c concentration levels before hospitalization. These findings are in accordance with previous evidence [11, 12] suggesting a U-shaped association between HbA1c concentration and hypoglycemia risk, in which both excessively tight control and unstable relaxed control increase vulnerability to severe events of hypoglycemia. Both near-normal HbA1c concentration values, often reflecting recent treatment intensification, and very high values, indicating unstable or advanced disease, are associated with increased hospitalizations for hypoglycemia [11, 13]. These earlier findings support the current results, suggesting that severe hypoglycemia, for both T1DM and T2DM, may be preceded by either a rapid decline or an increase in HbA1c due to erratic insulin use.

There are significant differences between DM types, which could be due to how different their supporting setting are. T1DM patients are more often followed by a diabetes specialist, and more rarely receive direct changes of their medication. In contrast, T2DM patients are more commonly managed by general practitioners, and are more prone to fluctuations in concentrations of plasma glucose and HbA1c, as well as alterations in medications.

Age, frailty, and individualized targets

DM is no longer predominantly a middle-aged disease, as nearly one in four people aged ≥ 70 years are living with the disease globally, reflecting both increased longevity and improved survival with the disease [14].As a result, an increasing number of older people with diabetes will need insulin. The pathophysiology of hypoglycemia in older adults is multifactorial. Ageing impairs normal counter-regulatory hormonal responses against falling glucose including reduced glucagon and adrenergic responses, reduces renal clearance of insulin and certain oral agents, and increases the likelihood of reduced nutritional intake, acute illness, and polypharmacy [15, 16]. Cognitive impairment and functional limitations further impair the ability to self-adjust insulin or carbohydrate intake [1, 17]. These mechanisms collectively amplify both the frequency and the severity of hypoglycemia in older adults receiving glucose-lowering treatment. Hypoglycemia is associated with serious clinical consequences in older individuals [6–8], including falls, traumatic injury, cardiovascular events, cognitive decline, dementia, and excess mortality [7, 8, 6]. In addition to individual harm and morbidity, hypoglycemia in older people represents a substantial global burden as diabetes already accounts for an estimated 3.4 million deaths annually and approximately 1 trillion USD in global health expenditures [9], and hospital presentations for hypoglycemia remain a persistent contributor to emergency department admittance [2, 18].

Medication patterns reflected current clinical practice with SU and combination insulin therapies being associated with lower HbA1c levels, supporting their role in treatment intensification [19], and their known propensity to cause hypoglycemia [4, 20].

Multiple studies have found that a substantial proportion of older adults with DM would be classified as overtreated when individualized targets are applied, with many continuing intensive regimens despite HbA1c concentration levels near or below 48 mmol/mol [21–23]. This is supported by our data showing that at least half of those with type 2 diabetes, and one fourth of those with type 1 diabetes, admitted with severe hypoglycemia had HbA1c values suggestive of overtreatment. Interestingly, 60% of T2DM patients with very high HbA1c concentration levels (> 70 mmol/mol) were still receiving fast-acting insulin. This pattern may reflect underlying issues such as poor adherence, cognitive impairment, or erratic intake as reasons for sudden hypoglycemia, highlighting the importance of assessing functional status and treatment appropriateness in an aging population.

This study strongly supports the principle that glycemic targets should be regularly revisited as patients grow older, develop comorbidities, and frailty increases. In frail older adults, the potential harms of tight control, such as falls, cardiovascular events, and cognitive decline, often outweigh its benefits [24, 25]. Our findings therefore reinforce international recommendations advocating individualized and less stringent glycemic goals and de-intensification of therapy in high-risk populations as their physiological needs change [12, 15, 23, 24].

Limitations

This study has several limitations. Prehospital data were incomplete, likely excluding milder episodes managed outside the hospital, or severe episodes given proper prehospital treatment, with insufficient documentation. We were unable to account for confounders such as diet, adherence, or self-monitoring behavior. Furthermore, the absence of renal function data limits our ability to assess potential confounding factors. Impaired renal function is a well-established contributor to hypoglycemia risk through reduced clearance of insulin and certain glucose-lowering medications. Inclusion of eGFR measurements could therefore have provided additional insight into factors contributing to severe hypoglycemia in this cohort. Also, medication exposure was derived from active prescriptions and in-hospital documentation, which may not fully reflect real adherence or changes made by primary care physicians shortly before admission. In addition, even though we stratified by age group, diabetes type, and HbA1c concentration category, our sample size of 212 patient was modest, particularly within the oldest age group (≥ 85 years) of T1DM (n = 8). Furthermore, the absence of formal frailty assessment meant we were restricted to age-based assessment. Because frailty was not systematically captured in the electronic records, we were unable to account for a multitude of factors, including functional status, cognitive impairment, gait instability, polypharmacy-related vulnerability, or overall physiological reserve.

Clinical implications and future perspectives

Our findings underscore the need for regular medication review, timely treatment de-intensification, and individualized treatment targets in older adults with diabetes to prevent severe hypoglycemia. Greater emphasis on patient education, assessment of appropriateness of self-administration of (particularly rapid-acting) insulin, even in people running at very high HbA1c levels, and multidisciplinary medication review could mitigate hypoglycemia risk and reduce preventable hospitalizations.

Future research should employ prospective, multicenter designs to validate these observations and explore predictive markers of overtreatment. Integration of frailty indices, such as Clinical frailty score, cognitive assessments, and continuous glucose monitoring data may further enhance the precision of individualized diabetes care in the elderly.

Although the majority of T2DM patients admitted with severe hypoglycemia were treated with insulin or sulfonylurea therapy, a smaller subgroup was receiving neither of these medications. As metformin, DPP-4 inhibitors, SGLT2 inhibitors, and GLP-1 receptor agonists have a low intrinsic risk of severe hypoglycemia when used without concomitant insulin or sulfonylureas, these findings suggest that factors beyond glucose-lowering therapy alone may contribute to hypoglycemic events in older adults. Potential explanations include reduced oral intake, malnutrition, acute intercurrent illness, impaired renal function, frailty, or medication administration errors. The multifactorial nature of severe hypoglycemia in this population highlights the importance of comprehensive clinical assessment beyond medication review alone.

Conclusion

The majority of older adults with diabetes, admitted with severe hypoglycemia, had HbA1c values out of target range. Hypoglycemic events occurred both in individuals with declining HbA1c values suggestive of overtreatment, and in those with persistently high HbA1c levels, indicating inappropriate or erroneous therapy. Such events are potentially preventable by timely reduction of glucose-lowering therapy, or by revision of therapy avoiding rapid-acting insulin if possible, or provision of assistance for daily medication.

Abbreviations

DM

Diabetes mellitus

T1DM

Type 1 diabetes mellitus

T2DM

Type 2 diabetes mellitus

IDF

International Diabetes Federation

ADA

American Diabetes Association

HbA1c

Hemoglobin A1c

SU

Sulfonylureas

ED

Emergency Department

IRB

Institutional Review Bord

SGLT2

Sodium-glucose cotransporter-2

GLP-1

Glucagon-like peptide-1 receptor

Author contributions

All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by RP, TAS, and UPB. The first draft of the manuscript was written by RP and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding

Open access funding provided by Copenhagen University. The authors declare that no funds, grants, or other support were received during preparation of this manuscript.

Data availability

The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request. Note that all datasets provided will be anonymized.

Declarations

Ethics approval and consent to participate

Collection of individual consent was deemed unnecessary by an impartial third party. The data was processed in an anonymized fashion, and we received only retrospective relevant data from patient files, thus no files were directly accessed. Data access was granted by the Institutional Review Bord (IRB) of North Zealand Hospital to assess departmental health care quality. Head of quality assurance and patient safety at North Zealand Hospital approved the project on behalf of the IRB. The reference number for this project is: 25000839.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.