Medication-Related Hospital Admissions and Emergency Department Visits in Older People with Diabetes: A Systematic Review

Limited data are available regarding adverse drug reactions (ADRs) and medication-related hospitalisations or emergency department (ED) visits in older adults with diabetes, especially since the emergence of newer antidiabetic agents. This systematic review aimed to explore the nature of hospital admissions and ED visits that are medication-related in older adults with diabetes. The review was conducted according to the PRISMA guidelines. Studies in English that reported on older adults (mean age ≥ 60 years) with diabetes admitted to the hospital or presenting to ED due to medication-related problems and published between January 2000 and October 2023 were identified using Medline, Embase, and International Pharmaceutical Abstracts databases. Thirty-five studies were included. Medication-related hospital admissions and ED visits were all reported as episodes of hypoglycaemia and were most frequently associated with insulins and sulfonylureas. The studies indicated a decline in hypoglycaemia-related hospitalisations or ED presentations in older adults with diabetes since 2015. However, the associated medications remain the same. This finding suggests that older patients on insulin or secretagogue agents should be closely monitored to prevent potential adverse events, and newer agents should be used whenever clinically appropriate.


Introduction
As the global population is now living longer, the world is experiencing an increase in both the number and proportion of older people [1,2].The World Health Organization reported that there were 1 billion people aged 60 years and older in 2019 [3].This number is predicted to grow considerably to 1.4 billion by 2030, accounting for around 15% of the world's population [3].Notably, older adults are more likely to have physiological deterioration and chronic medical conditions, including diabetes mellitus [4][5][6].
The prevalence of diabetes among older adults has increased significantly in recent decades, affecting approximately 33% of this demographic group worldwide [3].Older individuals with diabetes, whether newly diagnosed or long-standing, encounter an elevated risk of complications [4,[6][7][8].As a result, many of them are prescribed complex medication regimens, which may include potentially inappropriate medications (PIMs) and polypharmacy, typically defined as the use of five or more medicines [9].In addition to this, age-related physiological changes further elevate the risk of adverse medication outcomes in this group.Impaired renal function, reduced cognitive function, and increased frailty are among the factors contributing to this heightened risk [6,7].
Treating older adults with diabetes presents numerous challenges linked to medicationrelated problems (MRPs), or "events or circumstances involving a patient's drug treatment that actually, or potentially, interfere with the achievement of an optimal outcome" [10].Such challenges encompass the risks of hypoglycaemia, neurocognitive decline, falls, and even mortality [7].Currently, limited data are available regarding medication-related hospitalisations or ED visits in older individuals with diabetes.The two antidiabetic drug groups that have been commonly linked with hypoglycaemia in older adults are sulfonylureas and insulin [11,12].However, with the emergence of newer antidiabetic agents, such as dipeptidyl peptidase-4 (DPP4) inhibitors and sodium-glucose cotransporter-2 (SGLT2) inhibitors, reported to be safer and more effective in older adults [13][14][15], the use of these older agents (e.g., sulfonylureas and insulin/analogues) has decreased [16][17][18][19].As a result, the trends in MRPs that result in hospital admissions or ED visits might also have changed.This study, therefore, aimed to elucidate the trends and incidence of hospital admissions and ED visits in older patients with diabetes that are medication-related.Additionally, the research aimed to analyse the medications and other risk factors commonly associated with these hospital admissions or ED visits.

Materials and Methods
This systematic review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guideline [20].The review protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO: CRD42022353864).

Eligibility Criteria
In this study, we reviewed original research that reported hospitalisations or ED visits associated with MRPs in older people with diabetes.The medication involved did not have to be explicitly used for diabetes.Original research studies were eligible for inclusion if they fulfilled the following criteria: studies of hospital admissions or ED presentations, with data including older patients (mean age of 60 years and above) with diabetes mellitus whose admission or presentation was due to a MRP.The MRP reported could relate to (but was not limited to) overtreatment, undertreatment, drug interactions, poor adherence, medication error or adverse drug reaction (ADR).Included were observational studies designed to collect data on real-world patients and studies drawing on existing patient registries, insurance databases, and electronic medical records.
Studies were excluded if they were not in English, were written reporting the outcome as MRPs that occurred subsequent to hospitalisation or ED presentation, were published as a thesis, conference abstract, protocol, review, commentary, case report or case series, were designed in a trial or intervention setting, had less than 100 participants, or had inclusion criteria that targeted patients with specific medication (e.g., only patients treated with certain insulins or sulfonylureas in combination with antibiotics).

Search Strategy
Comprehensive, systematic searches of Ovid Medline, Ovid Embase, and Ovid International Pharmaceutical Abstracts (IPA) were performed to locate relevant studies published between 1 January 2000 and 20 October 2023 (Tables S1-S3).Subject headings and truncated search terms related to MRPs (e.g., side effect, ADR, medication error, polypharmacy), antidiabetic agents (e.g., sulfonylureas, insulin, DPP4 inhibitor), hospitalisation (e.g., hospital admission, ED presentation), and diabetes (e.g., diabetes mellitus, T1DM, T2DM) were combined.Type 1 diabetes mellitus was included, considering that an increased proportion of individuals with type 1 diabetes are living into the later decades of life [21].From the search results, a citation analysis and hand search were performed using Google Scholar and Web of Science.A flowchart of the search strategy is depicted in Figure 1.

Data Extraction
The initial title screening was performed by AV, followed by abstract and full-text review by two reviewers independently (AV, GP/MS), using Covidence (www.covidence.org)(accessed from 27 March to 7 November 2023).Using the pre-specified inclusion/exclusion criteria, two independent reviewers had to agree on all inclusions and exclusions at the abstract and full-text stages.A third reviewer made the final decision when discrepancies could not be resolved between the first two reviewers.Data from published studies were extracted into a standardised format that described: study characteristics (e.g., study design, setting, country, number of patients, year), patients' characteristics, types of MRPs identified, and risk factors associated with medication-related hospital admissions or ED presentations.

Quality Assessment
Two reviewers (AV, GP/MS) independently evaluated each study's quality using the Joanna Briggs Institute (JBI) list, which has tools for different types of studies (prevalence, analytical cross-sectional, case-control, cohort studies).Discrepancies in judgment were solved by discussion between reviewers to reach a consensus.The items covered by each JBI tool are shown in Tables S4-S7.No studies were excluded based on the quality assessment outcome.

Discussion
The objective of this review was to expand understanding of the overall trend and incidence of medication-related hospitalisation or ED presentation in older adults with diabetes and to identify medications or other factors associated with these events.Our review detected studies performed in multiple countries, except for countries in Africa, South America, and Oceania.One study from Australia underwent full-text screening but was eventually removed from the review due to the outcome of the study (ambulance calls due to hypoglycaemia) [57].Therefore, more studies are still required to provide robust data across different geographic and cultural contexts.

Interpretation of Findings in Relation to Other Studies
Other systematic review studies with different populations and conditions [58][59][60] listed various types of MRPs (e.g., ADRs, PIMs, drug-drug interaction) associated with hospital admission.However, our review in diabetes only discovered admissions or ED presentations due to hypoglycaemia episodes.The limited number of studies that explicitly mention medication or drug-related hospitalisation in older adults with diabetes raises concerns that perhaps the events are being under-reported or unnoticed by healthcare professionals.
Most hypoglycaemia events identified in this review were based on diagnosis codes documented in administrative or electronic health record data.Compared to self-report, this source of data could minimise the bias of detecting hypoglycaemia that sometimes can be misinterpreted in the older adult due to their overlapping features with geriatric syndromes [2].However, in many of the studies [24,27,30,33,36,38,42,43,48,49,51], there was no differentiation made between the outcomes of ED visits and hospitalisations.Consequently, for these studies, it remains unclear whether the ED visits ultimately led to hospitalisation or not.
Overall, this review shows that hypoglycaemia-related hospitalisations or ED visits have declined since 2015.Although the trends in hospitalisations or ED visits seem to portray a similar pattern, there was considerable diversity in incidence rates between countries, with the lowest incidence rate (22 events per 100,000 people in 2017) being reported in Denmark [45].Various studies that also observed the trend in antidiabetic drug use during the time found a decrease in the use of medications that have a high contribution to hypoglycaemia incidents [34,46,48].This might explain the drop in hypoglycaemia events during the observation period.One study that assessed the trend of hospitalisation and the use of sulfonylureas indicated that the decreased use of the drugs might have contributed to the lower incidence of hospitalisation in recent years [34].Two studies conducted in the US observed an increasing trend when they utilised a combination of ICD-9 and ICD-10 diagnosis codes to identify hypoglycaemia-related ED visits or hospitalisations [36,49].However, it is important to note that these two studies employed different diagnostic codes (ICD-9 and ICD-10) to determine the outcomes, making it challenging to describe the observed trend over the study period conclusively.Additionally, some studies have suggested that using ICD-10 codes can increase the specificity in describing medical conditions [61,62], thus identifying and recording more hypoglycaemia cases.
Despite the seemingly decreasing rate of hypoglycaemia-related hospitalisation, the older antidiabetic agents (insulin and sulfonylureas) were still reported to be associated with hypoglycaemia-related hospital admissions or ED visits in older adults.Insulin, whether in the form of analogues or human, basal or mixed, was commonly reported (14 studies) to be associated with admission or ED presentation in older adults with diabetes [26,[28][29][30]36,38,[41][42][43][47][48][49]53,55].The updated 2019 Beers Criteria only listed insulin given in a sliding-scale dosage as a PIM for older adults.However, our review could not confirm whether the patients in the studies above were receiving sliding-scale doses.Another antidiabetic class linked to hypoglycaemia-related admissions are sulfonylureas, with the risk reported to be up to 13 times higher in one study [30].The 2019 Beers Criteria updated its recommendation in regard to the use of sulfonylureas from avoiding only long-acting sulfonylureas to avoiding all sulfonylureas in older adults.If considering a sulfonylurea for older adult patients, several guidelines suggest opting for a short-acting one, such as glipizide [63,64].A multitude of factors complicate the management of diabetes in older adults [6].The patient's overall health status, coexisting illnesses, and degree of cognitive function must be thoroughly considered prior to starting any form of glucose-lowering therapy for the older adult [7].These findings further emphasise the need for extreme caution and close monitoring when prescribing insulin and secretagogue agents to older adults.
In this review, we found that metformin and SGLT2 inhibitors were consistently linked to decreased odds of hospitalisation or ED visits due to hypoglycaemia [26,36,46], presumably because of their different mechanisms of action.However, more observational real-world data are required to comprehensively assess the risk of hypoglycaemia from the newer drugs.Clinical trials have reported severe hypoglycaemia episodes occurring in 0.7% to 2.4% of patients treated with newer antidiabetic medications (SGLT2 inhibitors, GLP-1 analogues, gliptins) [2].Considering the nature of clinical trials, it is suspected that the rate of events in the real world could be higher.Among non-antidiabetic medications associated with an increased risk of hospitalisation or ED visits due to hypoglycaemia, diazoxide exhibited the highest risk [46].Since diazoxide is indicated to counter hypoglycaemia in conditions like insulinoma and congenital hyperinsulinism [65], it is possible that patients with these conditions had pre-existing hypoglycaemia and were treated with diazoxide before hospital admission.
In terms of gender, females were found to be at higher risk of experiencing hypoglycemia-related hospitalisation or ED visits [30,40,43,49,55].Several findings from different studies may shed light on this association.For instance, women tend to have reduced intrinsic counter-regulatory responses to hypoglycaemia [66].Several research studies have consistently pointed out an increased likelihood of subsequent hypoglycaemia-related hospitalisations in individuals who have previously experienced such events [26,36,38,41,42,55].This highlights the urgent importance of conducting thorough medication reviews and providing comprehensive patient education for those who have experienced hypoglycaemiarelated hospitalisation or ED visits.
Among significant comorbidities reported to elevate the odds of hypoglycaemiarelated hospitalisations and ED visits were cardiovascular disease, hypertension, dementia or cognitive impairment, and chronic kidney disease [28,38,40].Diabetes, hypertension and cardiovascular disease are closely intertwined due to the sharing of risk factors, such as endothelial dysfunction, arterial remodelling, vascular fibrosis, atherosclerosis, dyslipidemia, and obesity [67].Individuals with diabetes who have normal blood pressure tend to exhibit reduced insulin secretion, resulting in a lower risk of hypoglycemia [68].
A previous study has reported an elevated risk of hypoglycaemia in older adult patients with diabetes who also have dementia [24].Older adults with dementia often have a reduced dietary intake, a decline in cognitive and functional capacity, and a heightened likelihood of hypoglycaemic unawareness [69].These characteristics can mask hypoglycemia symptoms, potentially leading to the development of more severe hypoglycemia episodes requiring hospital admission or an ED visit.Kidney disease, whether in the form of a reduced estimated glomerular filtration rate (eGFR) or chronic kidney disease, has consistently been reported to elevate the odds of hypoglycaemia-related hospitalisations or ED visits [26,28,30,31,33,38,40,43,47,49,50,53].Impaired kidney function affects the clearance of many antidiabetic medications, which in turn prolongs the drugs' presence in the body, subsequently increasing the risk of hypoglycaemia [6].

Figure 1 .
Figure 1.PRISMA flow diagram of the study selection process and citation analysis.

Table 1 .
Study and participant characteristics, trends or events of medication-related hospital admissions and emergency department (ED) visits in older adult patients with diabetes.
ICD, International Classification of Diseases, T1DM, Type 1 Diabetes Mellitus, T2DM, Type 2 Diabetes Mellitus, NA: not available.* Included are patients with diabetes and end-stage kidney disease.** Included are patients with Alzheimer's dementia (AD) and with concomitant T2DM.

Table 2 .
Medications associated with hospital admission or ED presentation due to hypoglycaemia in older adult patients with diabetes.

Table 3 .
Other risk factors associated with hospital admission or ED presentation due to hypoglycaemia in older adult patients with diabetes.