Hospitalization for hypoglycaemia in people with diabetes in Denmark, 1997–2017: Time trends in incidence and HbA1c and glucose‐lowering drug use before and after hypoglycaemia

Abstract Objective To assess incidence trends of first hospitalization for hypoglycaemia in Denmark and to examine HbA1c levels and glucose‐lowering drug use before and after hospitalization among individuals with type 1 or type 2 diabetes. Research Design and Methods We performed a population‐based study linking diagnosis, prescription and laboratory data. Standardized incidence of first hospitalization for hypoglycaemia in Denmark was assessed for each calendar year 1997–2017. HbA1c and glucose‐lowering drug use was compared with age‐ and sex‐matched diabetes comparisons without hospitalization for hypoglycaemia. Results The annual age‐ and sex‐standardized incidence rate of first hospitalization for hypoglycaemia per 100,000 person‐years increased during 1997–2003 (from 17.7 to 30.3 per 100,000 person‐years), remained stable until 2010 (30.4) and gradually declined until 2017 (22.0). During this period, we identified 3,479 people with type 1 diabetes and 15,329 people with type 2 diabetes experiencing first hospitalization for hypoglycaemia. Both diabetes groups experienced a mean HbA1c decrease of ~12%–15% in the months preceding first hospitalization, followed by a gradually increasing HbA1c afterwards. People with type 1 diabetes and hospitalization used similar insulin therapies as those without hospitalization. People with type 2 diabetes and hospitalization more often received insulin (55%) than comparisons (45%), and 45% discontinued insulin or stopped all glucose‐lowering therapy after first hospitalization. Conclusions Incidence of hospitalizations for hypoglycaemia has declined by one fourth the last decade in the Danish population. A HbA1c decrease precedes first hospitalization for hypoglycaemia in individuals with diabetes, and profound changes in glucose‐lowering drug therapy for type 2 diabetes occur after hospitalization.


| INTRODUC TI ON
Hypoglycaemia is a common and feared condition in people with type 1 and type 2 diabetes, remaining a hindrance for optimal glycemic management. Consequently, some people intentionally reduce doses or discontinue glucose-lowering drugs trying to prevent episodes of hypoglycaemia, 1 possibly leading to HbA 1c dysregulation. 1 While non-severe episodes of hypoglycaemia are self-manageable, severe hypoglycaemia requires external assistance and possibly leads to hospitalization. Hospitalization for hypoglycaemia represents a tremendous economic burden, and treatment in hospital setting is several fold higher than in community setting. 1 Hospitalization for hypoglycaemia is associated with increased morbidity and mortality, 1 and the first event is the strongest predictor of future events of severe hypoglycaemia. 2 Recent years have witnessed newer glucose-lowering drugs being introduced to the market, and increased availability of technological devices designed to improve diabetes treatment (eg insulin pump, continuous glucose monitoring). 3 However, the extent to which these changes in diabetes management correlate with reduced incidence of hospitalization for hypoglycaemia in adults with type 1 and type 2 diabetes is unclear. Previous studies examining incidence rates of hospitalization for hypoglycaemia have reported opposing results, [4][5][6][7][8] and few studies have focused on the first event.
Understanding the current incidence trends of first hospitalization for hypoglycaemia in real-world diabetes populations and understanding which individuals are at increased risk is essential.
We aimed to assess: 1) incidence trends in first hospitalization for hypoglycaemia 1997-2017 in the country of Denmark and 2) trends in HbA 1c and glucose-lowering drug use before and after first hospitalization for hypoglycaemia in people with type 1 diabetes and type 2 diabetes.

| Setting
We conducted a population-based study based using prospectively collected health care data. The Danish National Health Service provides universal, tax-funded healthcare, guaranteeing access to primary and secondary sectors and partial reimbursement for prescribed drugs. The unique personal registry number assigned to all Danish residents at birth or immigration allows for unambiguous data linkage. 9 We linked several existing population-based medical databases, as described below. The Danish National Prescription Registry was established 1994 covering all redeemed prescriptions at pharmacies in Denmark. 10 The Danish National Patient Registry (DNPR) contains information on admissions and discharges from all Danish non-psychiatric hospitals since 1977. Since 1995, records of emergency and outpatient specialist clinic visits are included. 11 The coding used is the International Classification of Diseases, Eighth

| Study population: Incidence
From the DNPR, we identified all people with a first hospitalization for hypoglycaemia (using ICD-10 codes: E100, E110, E120, E130, E140, E159, E160, E161 and E162) between 1JAN1977 and 31DEC2018. To restrict our population to people with incident hypoglycaemia hospitalization, we excluded people diagnosed with hypoglycaemia before 1997.

| Study population: Individuals with diabetes
Similar to other studies, 13 we identified individuals with incident diabetes using either the date of their first ever redemption of a glucose-lowering drug prescription (Anatomical Therapeutic Chemical classification system [ATC] code starting with A10) or their first ever DNPR hospital-coded diabetes (ICD-8 or ICD-10   code starting with 249-250, 2515, E10-E15, O24, T383A, M142,   G590, G632, H280, H334, H450, H360, N083), whichever came first. This algorithm for identifying diabetes has been found to have a positive predictive value (PPV) of 97% for hospital-based diagnoses and 95% for prescription-based diagnoses, whereas estimated sensitivity of this combined approach for detecting known

Bulleted novelty statement What is already known?
Hospitalization for hypoglycaemia is associated with increased morbidity and mortality in people with diabetes, and severe hypoglycaemia is the strongest predictor of future events of severe hypoglycaemia.

What this study has found?
The incidence of hospitalization for hypoglycaemia has declined during the last decade in the Danish population. An observable HbA 1c decrease precedes hospitalization, and treatment discontinuation is frequent among people with type 2 diabetes after first hospitalization.

What are the clinical implications of the study?
The declining incidence of hospitalization for hypoglycaemia is likely caused by a number of improvements in diabetes management. The observed decline in HbA1c preceding hospitalization for hypoglycaemia may reflect intensification of treatment and implies that good monitoring is important when intensifying pharmaceutical therapy.
diabetes is higher than 80%. 14 We excluded individuals who had not resided in Denmark for at least one year prior to this date.
To ensure a high specificity for type 1 diabetes, people who redeemed insulin before age 30 years (ATC code starting with A10A) or any glucose-lowering drug before age 15 years were considered as likely having type 1 diabetes. Women giving birth within nine months after diabetes diagnosis were excluded as likely having gestational diabetes mellitus. Women with pre-existing hospital diagnosed polycystic ovarian syndrome or who redeemed any metformin prescription (ATC code A10BA02) in combination with clomifene (ATC code G03 GB02) within 12 months following diagnosis were excluded as likely having polycystic ovarian syndrome.
We defined the remaining people with diabetes as having type 2 diabetes.

| Patient characteristics
Data on sex and age at date of first hospitalization for hypoglycaemia/matched index date were obtained from the Danish Civil Registration System (CRS). 16   or uncertain (none of the above). Treatment categories for people with type 2 diabetes were defined as: metformin monotherapy, metformin +other non-insulin glucose-lowering drug (NI-GLD), sodium-glucose transport protein 2 inhibitors (SGLT2) monotherapy, no treatment, glucagon-like peptide-1 receptor analogues (GLP-1RA) monotherapy, dipeptidyl peptidase-4 inhibitors (DPP4i) monotherapy, sulfonylurea (SU) monotherapy, other NI-GLD monotherapy, other NI-GLD combinations, insulin monotherapy, metformin +insulin, insulin +other NI-GLD and metformin +insulin + other NI-GLD (as described in more detail elsewhere 18 ).

| Statistical Analysis
First, we computed and plotted the annual age-and sex-standard-  (Table 1).
Third, we calculated the mean HbA 1c with three-month intervals five years before and five years after first hospitalization for hypo-

| SIRs of hospitalization for hypoglycaemia
The overall SIR of hospitalization for hypoglycaemia per 100,000 person-years (  Figure 1: B and C.

| HbA 1c before and after hospitalization for hypoglycaemia
People with type 1 diabetes and hospitalization for hypoglycaemia had a stable HbA 1c from five years to one month prior to the hypoglycaemic event (Figure 2: A). HbA 1c was at 73 mmol/mol  I G U R E 1 Incidence trends. A: Overall age-and sex-standardized incidence rates (SIR) from 1997 to 2017 of first hospitalization for hypoglycaemia in Denmark, and B and C: SIR by age groups. Line with each SIR indicates 95% CI.  Note: Categories of comorbidity were based on Charlson comorbidity index scores of 0 (no comorbidity), 1 (moderate), 2 (severe), ≥3 (very severe). a Results are limited to those that residing in Northern Denmark where laboratory data were available: Regions Midt +Nord upon diabetes diagnosis.
The population of Northern Denmark is considered representative of the nationwide population.
( Figure 3: B). After hospitalization, a clear trend in drug discontinuation was present with 45% of people with type 2 diabetes receiving no treatment one year after the event. Overall, people with type 2 diabetes and hospitalization more often received insulin-based regimens (either monotherapy or insulin combined with one glucoselowering drug) (55%) than did comparisons (45%). Furthermore, people with type 2 diabetes and hospitalization for hypoglycaemia appeared to receive less SU monotherapy (10%) compared with comparisons (20%).

| DISCUSS ION
In our study, we recorded a gradual increase in population-based SIRs of first hospitalization for hypoglycaemia in Denmark from 1997 to 2003 followed by a stable period till 2010 and succeeded by a decline to 2017. We observed a similar HbA 1c reduction in the months preceding hospitalization for hypoglycaemia in people with type 1 and type 2 diabetes. To our knowledge, this is the first population-based study investigating HbA 1c trends before and after hospitalization for hypoglycaemia in a real-world setting.
The overall inclining incidence rates in our study observed from 1997 to 2003 resemble the findings from the Retrospective Cohort Study from UK examining type 1 diabetes and type 2 diabetes from 1998 to 2013. 4 The authors reported a subtle incline in incidence rates of hospitalization for hypoglycaemia in adults.
A subgroup analysis revealed declining incidence rates in people with type 2 diabetes older than 64 years during 2009-2013. In contrast, our incline appears to be driven mainly by the age group We showed that people with type 1 diabetes and hospitalization received similar insulin regimens as type 1 diabetes comparisons and found no clear treatment differences before and after the event.
This important finding reduces the chances of uncovering a specific insulin treatment regimen as an important contributor to hospitalization for hypoglycaemia. It is beyond the scope of this study to investigate changes within each treatment regimen, though alterations in insulin dosage could be an important factor to take into account.
Other studies have found irregular adherence to insulin therapy, irregular eating and physical activity alterations as possible contributors to non-severe hypoglycaemic events, 29  Of all severe hypoglycaemic events, few are hospitalized.
Accordingly, most severe events are treated at home and prehospital 30 and thus not included in our study, although they can be as serious as the ones resulting in hospitalization.

| CON CLUS ION
The population-based incidence rate of first hospitalization for hypoglycaemia in Denmark has declined by one fourth the last decade. In both diabetes groups, a clearly observable HbA 1c decrease forecasts first hospitalization, and profound glucose-lowering drug discontinuation occurred among people with type 2 diabetes after hospitalization.

E TH I C S A PPROVA L
Not needed for purely registry-based studies in Denmark.

PATI ENT I N VO LV EM ENT
Patients were not involved in setting the research question, the outcome measures or the design or implementation of the study.
There are no plans to involve patients in dissemination of the results.

TR A N S PA R EN C Y
The senior author, RWT, affirms that the manuscript is an honest, accurate and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

CO PY R I G HT
The corresponding author has the right to grant on behalf of all au- include electronic links from the Contribution to third party material wherever it may be located and vi) licence any third party to do any or all of the above.

DATA AVA I L A B I L I T Y S TAT E M E N T
Danish law does not allow researchers to share raw data from the registries with third parties. Data can be accessed by researchers