Type 2 diabetes mellitus treatment patterns in US nursing home residents

Abstract

Background. The prevalence of type 2 diabetes mellitus (diabetes) in nursing home residents (NHRs) is increasing, concurrently with obesity and other comorbid conditions. NHR would benefit greatly from antidiabetic medications that would improve glycemic control and give a lower risk of hypoglycemia but that do not contribute to weight gain in obese individuals. Objective. To examine the prescription patterns to NHRs with diabetes, including the use of newer injectable therapies such as glucagon-like peptide-1 (GLP-1) receptor agonists. Methods. Treatment patterns of diabetes in NHR were analyzed using Minimum Data Set records and prescription claims from the Omnicare Senior Health Outcomes data repository (May 2011–September 2012). Results. The prevalence of diabetes in this population of 229,283 NHRs was 35.4%. Among the 44,665 NHRs with diabetes and prescription claims data, the prevalence of obesity (40.3%) and multiple comorbidities (100%) was high. Approximately 20% of the NHRs with diabetes were aged <65 years. Overall, 20% of NHRs had diabetes that was untreated with medications during the study period. Insulin was the mainstay of treatment (>80%), followed by oral agents (54%). GLP-1 receptor agonist use was low (0.5%) and associated with poor treatment persistence. Conclusion. Considerations other than glycemic control may drive prescribing decisions, contrary to recommendations from the American Diabetes Association, American Medical Directors Association, and European Association for the Study of Diabetes.

Keywords :

• Type 2 diabetes
• insulin treatment
• elderly nursing home residents
• obesity
• prescription patterns
• minimum data set

Introduction

Initiation of basal insulin therapy is often recommended for patients with type 2 diabetes mellitus (diabetes) who are unable to achieve or maintain glycemic control on oral antidiabetic drugs (OADs) [1]. However, ∼50% of all patients are unable to achieve glycemic control on basal insulin alone [2]. The current treatment algorithm for the management of diabetes, issued as a joint position statement by the American Diabetes Association (ADA) and the European Association for the Study of Diabetes (EASD) in 2012, emphasizes an individualized approach [1]. For example, recently diagnosed patients with few or no (diabetes-related) comorbidities would benefit most from hyperglycemia management approaches that aim at preventing or delaying vascular complications over time and modifying the disease process. Therefore, more stringent efforts to lower hemoglobin A1C values are justified in these generally younger and healthier patients with long life expectancies [1]. Older adults, especially those living long term in a nursing home, have a shorter life expectancy and are more vulnerable to hypoglycemia, which may lead to falls and fractures [3]. As a consequence, the American Geriatrics Society (AGS) recommends that glycemic targets for elderly patients with diabetes for a longer duration, multiple comorbidities, and an overall poorer health status are less stringent than for younger patients [4]. The ADA/EASD as well as the American Medical Directors Association (AMDA) clinical practice guidelines recommend that choosing treatment strategies for older individuals with diabetes should focus on personalized goals, drug safety, and minimizing the risk of hypoglycemia [1,5,6].

The ADA/EASD position statement provides several options for escalating glucose-lowering therapy in patients not well controlled on a basal insulin [1,5]. Three classes of oral agent (thiazolidinediones, dipeptidyl peptidase-4 inhibitors, and sodium-glucose co-transporter 2 [SGLT2] inhibitors, all three of which are associated with low risk of hypoglycemia) and two injectables (glucagon-like peptide-1 [GLP-1] receptor agonists and rapid-acting insulins [RAIs]) have been proposed [5]. GLP-1 receptor agonists (or incretin mimetics) may be used to intensify basal insulin therapy as an alternative to prandial RAIs – the use of these treatments in combination results in improved glycemic control, a decreased risk of hypoglycemia, and the potential added benefit of weight loss [1,5,7,8].

Diabetes is disproportionately prevalent in the elderly population. The International Diabetes Federation and the US Centers for Disease Control have estimated that 25.9% of individuals aged ≥65 years have diabetes, compared with 9.3% of the total US population [9,10]. Diabetes is prevalent in elderly nursing home residents (NHRs), but much of the current data are outdated [11-13], and data on the demographical and clinical characteristics of these patients are sparse. Use of sliding-scale insulin in NHRs is still common [14], despite evidence that these regimens are associated with poorer glycemic control and with hypoglycemia rates that are similar to non-sliding-scale insulin regimens; they are discouraged by the ADA and the EASD [15], as well as the AMDA [6]. Therefore, an assessment of current treatment patterns in the elderly is particularly useful.

The objective of this study was to investigate real-world diabetes treatment patterns among US NHRs and to assess demographical and clinical differences and functional differences, if any, among patients treated with traditional therapies versus newer therapies such as GLP-1 receptor agonists. Information on real-world prescribing patterns is useful for understanding the utilization of newly emerging therapeutic options such as GLP-1 receptor agonists in the nursing home setting.

Research design and methods

Study design and patients

This was a retrospective study of data from NHRs with diabetes in the United States. The data included linked and de-identified Minimum Data Set (MDS) records (version 3.0) and prescription claims data from the Omnicare Senior Health Outcomes data repository. MDS is a resident assessment instrument that includes a core set of screening, clinical, and functional status elements that form the basis for comprehensive assessment of all residents in long-term care facilities certified to participate in Medicare and/or Medicaid plans, irrespective of the resident’s individual payment source [16]. This assessment is administered and completed by a registered nurse. The MDS is a standardized tool that has demonstrated usability as a source of resident status outcomes in the nursing home setting [17-19].

Inclusion criteria were defined as being an NHR with a diagnosis of diabetes in the MDS records (item I2900, or manually entered International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] diagnosis codes 250.xx and subtypes) [20]. NHRs were excluded from the study if their MDS elements included ICD-9-CM diagnosis codes 250.x1 or 250.x3, indicating that they had type 1 diabetes mellitus. The study interval was a 17-month period from 1 May 2011 to 30 September 2012. The demographical, clinical, and functional characteristics were assessed for all eligible NHRs with diabetes, and there was no age restriction.

Among the subpopulation of diabetic patients with prescription claims, the characteristics of NHRs who received diabetes treatment including GLP-1 receptor agonists were compared with NHRs who received diabetes treatment that did not include GLP-1 receptor agonists. The prescription patterns, payer source, and treatment persistence were determined for these NHRs. Treatment persistence was defined as the duration of time from initiation to discontinuation of therapy [21]. Antidiabetic therapy treatment persistence was calculated by counting the number of sequential days of therapy dispensed to each unique resident prior to discontinuation, if use of the medication was discontinued, or through the entire 360-day study period, if continued throughout. Residents were required to receive refills of their medication at least every 42 days to be considered persistent. Individual NHRs were de-identified, in accordance with the Health Insurance Portability and Accountability Act of 1996. Sterling Institutional Review Board (Atlanta, GA, USA) reviewed the study, and issued a waiver of authorization as well as an exemption for full review.

Measures

In this study, we included assessment of the degrees of both cognitive and functional impairments of the NHRs. Cognitive function was assessed via one of two instruments. The first of these was the Brief Interview for Mental Status (BIMS) [22], a seven-item interview administered to NHRs who are able to participate in the interview process. The BIMS is scored on a scale of 0–15, with lower scores indicating greater cognitive impairment. A score of 13–15 indicates intact cognition, 8–12 represents moderate cognitive impairment, and a score of 0–7 indicates severe cognitive impairment. The second assessment used the Cognitive Performance Scale (CPS) [23]. The CPS was assessed when the BIMS score was unavailable due to the NHR’s inability to participate in the BIMS interview. The CPS is used to characterize cognitive impairment on a scale of 0 (no impairment/intact) to 6 (very severe impairment). The functional status of the NHR was assessed using a composite of activities of daily living (ADL) score, based on four MDS categories: bed mobility; transferring; toilet use; and eating. ADL scores range from 4 to 18, with a score of 4 indicating independence and a score of 18 indicating total dependence [16,24].

Statistical analyses

Standard summary measures were used to describe the data, including mean values, standard deviation (SD), ranges, and percentages. Significance of the differences was determined using the Student’s t-test and χ² test. Analyses were conducted using SAS, version 9.3 (SAS Institute, Inc., Cary, NC, USA).

Results

Demographical and clinical characteristics

A total of 229,283 NHRs in the United States were identified from the MDS database. Of these, 81,087 (35.4%) had a diabetes diagnosis (Table 1). The majority of NHRs with diabetes were elderly (mean [SD] age: 75.7 [12.3] years), white (68.6%), and women (57.6%). Almost half (43.7%) were aged ≥80 years, and almost 20% were aged <65 years. The mean (SD) body mass index (BMI) was 30.0 (8.5) kg/m² and 41.0% of NHRs with diabetes were obese (i.e. had a BMI of ≥30 kg/m²). NHRs included in the analysis had moderate cognitive impairment (mean [SD] BIMS score: 10.8 [4.4]; and mean [SD] CPS score: 4.1 [1.9]) and were highly dependent on others for daily functioning (mean [SD] ADL score: 13.7 [3.9]). Comorbidities were common; many comorbid conditions had a prevalence of >20%. Falls were common (28.0%), and approximately half of the falls were injurious (43.9%) (Table 1). Other comorbid conditions occurring at a prevalence of <20% included renal insufficiency (16.9%), dysrhythmia (16.8%), pneumonia (12.5%), Alzheimer’s disease (9.5%), osteoporosis (8.8%), non-hip fracture (7.6%), and hip fracture (5.0%).

Table 1. Baseline demographics and clinical characteristics of all NHRs with a diagnosis of type 2 diabetes (diabetes) in the MDS and the subset with prescription claims data.

Characteristic	All NHRs with diabetes (n = 81,087)	NHRs with diabetes and prescription claims data (n = 44,665)
Age, mean (SD), years	75.71 (12.3)	75.38 (12.3)
Gender, n (%)
Female	46,735 (57.6)	25,924 (58.0)
Male	34,344 (42.4)^a	18,741 (42.0)
Age category, n (%)
<65 years	14,968 (18.5)	8649 (19.4)
65–69 years	8552 (10.5)	4820 (10.8)
70–74 years	10,143 (12.5)	5686 (12.7)
75–79 years	12,009 (14.8)	6592 (14.8)
80–84 years	13,764 (17.0)	7422 (16.6)
85–89 years	12,496 (15.4)	6635 (14.9)
≥90 years	9155 (11.3)	4861 (10.9)
Race, n (%)
White	55,662 (68.6)	28,366 (63.5)
Black/African American	11,082 (13.7)	6870 (15.4)
Hispanic/Latino	8746 (10.8)	6041 (13.5)
Asian	2911 (3.6)	1633 (3.7)
Other	2686 (3.3)	1755 (3.9)
Cognitive pattern, mean (SD)
BIMS summary score	10.78 (4.4)	10.88 (4.3)
CPS score^b	4.06 (1.9)	4.12 (1.8)
Functional status, mean (SD)
Composite ADL score	13.67 (3.9)	13.96 (3.8)
Active comorbid disease diagnosis, n (%)
Hypertension	69,581 (85.8)	38,700 (86.6)
Hyperlipidemia	43,313 (53.4)	24,230 (54.2)
Depression	36,444 (44.9)	20,987 (47.0)
Dementia (non-Alzheimer’s disease)	26,412 (32.6)	14,920 (33.4)
Heart failure	23,967 (29.6)	13,170 (29.5)
GERD or ulcer	23,809 (29.4)	13,655 (30.6)
Asthma, COPD, or chronic lung disease	22,378 (27.6)	11,927 (26.7)
Urinary tract infection	20,281 (25.0)	11,423 (25.6)
Anxiety disorder	20,245 (25.0)	11,433 (25.6)
Coronary artery disease	19,184 (23.7)	10,099 (22.6)
Cerebrovascular accident, TIA, or stroke	16,897 (20.8)	9761 (21.9)
Renal insufficiency, renal failure	13,725 (16.9)	7413 (16.6)
Falls, n (%)	22,716 (28.0)	13,215 (29.6)
Falls leading to non-major injury	8480 (37.3)	4820 (36.5)
Falls leading to major injury	1491 (6.6)	850 (6.4)
BMI category, n (%)
≥30 kg/m^2	33,221 (41.0)	18,022 (40.3)
<30 kg/m^2	47,866 (59.0)	26,643 (59.7)
BMI, mean (SD), kg/m^2	29.97 (8.5)	29.89 (8.7)
Payer source, n (%) ^c
Medicare Part A (facility paid)		35,369 (72.6)
Medicare Part D		18,926 (38.8)
Medicaid		2847 (5.8)
Commercial insurance		1952 (4.0)
Other (e.g. private pay, VA)		973 (2.0)

^aEight patients had no data available.

^bFor NHRs with no BIMS summary scores available.

^cBased on claims data from 48,747 claims for OAD GLP-1 RA, or insulin therapy. Percentages add up to >100% because NHRs may be beneficiaries of multiple payers during their admission.

Abbreviations: ADL = Activities of daily living; BIMS = Brief Interview for Mental Status; BMI = Body mass index; COPD = Chronic obstructive pulmonary disease; CPS = Cognitive Performance Scale; GERD = Gastroesophageal reflux disease; GLP-1 RA = Glucagon-like peptide-1 receptor agonist; MDS = Minimum Data Set; NHR = Nursing home residents; OAD = Oral antidiabetic drug; SD = Standard deviation; TIA = Transient ischemic attack; VA = Veteran Affairs.

Of the 81,087 NHR with diabetes in the MDS database, 44,665 (55%) had prescription claims data available for analyses. This subset of NHRs was similar to the entire diabetes cohort in terms of age, cognitive function, functional status, and comorbid conditions (Table 1).

Treatment patterns, persistence, and concomitant medications

Of the NHRs with diabetes who had prescription claims data available (n = 44,665), 79.9% had claims data related to antidiabetic therapy. The remaining 20.1% of NHRs with a diagnosis of diabetes did not receive antidiabetic medication during the study period. Treatment with OADs and injectable therapy was common: 81.8% of patients received injectable therapy, 18.2% received OAD therapy only, and 54.4% received OADs with or without concomitant injectable therapy. Metformin and sulfonylureas were the most commonly prescribed OAD treatments (53.6% and 51.0%, respectively); α-glucosidase inhibitors (0.7%) were the least commonly prescribed (Figure 1a ). RAIs and basal insulin were the most commonly prescribed injectable therapies (59.5% and 53.2%, respectively), whereas GLP-1 receptor agonists (0.5%) were the least commonly used (Figure 1b ). Only 169 patients received GLP-1 receptor agonists, either alone, or in combination with OADs or basal insulin.

Figure 1. Treatment received by NHRs with a diagnosis of type 2 diabetes mellitus: (a) oral antidiabetic therapy; and (b) injectable therapies are shown.

Treatment persistence was significantly lower among NHRs receiving antidiabetic therapy including GLP-1 receptor agonists than among NHRs receiving other treatment regimens (Figure 2). Concomitant prescription medications among patients with prescription claims data available included statins, antidepressants, and angiotensin-converting enzyme (ACE) inhibitors/angiotensin-receptor blockers (ARB) (Table 2).

Figure 2. Treatment persistence among nursing home residents with a diagnosis of type 2 diabetes mellitus is shown.

Table 2. Concomitant prescription medication utilization in NHRs with a diagnosis of type 2 diabetes.

Concomitant prescription medication, %	NHR with prescription claims (n = 44,665)
Statins	53.2
Antidepressants	52.4
ACE inhibitors/ARB^a	51.3
Antiemetics	25.5
Antipsychotics	23.4
Antiplatelet therapy^b	17.8
Acetylcholinesterase inhibitors	16.1
Glucagon	11.5
Memantine	9.3
Phosphate binders^c	5.5
Vitamin-D analogs^d	2.1

^aIncludes fixed combinations.

^bIncludes clopidogrel, prasugrel, ticagrelor, and aspirin/extended-release dipyridamole.

^cIncludes calcium acetate, sevelamer hydrochloride, sevelamer carbonate, and lanthanum carbonate.

^dIncludes calcitriol, doxercalciferol, and paricalcitol.

Abbreviations: ACE = Angiotensin-converting enzyme; ARB = Angiotensin-receptor blockers; NHR = Nursing home residents.

The payer sources for OADs, insulin, and GLP-1 receptor agonists were most commonly Medicare Part A (facility paid), followed by Medicare Part D, and Medicaid (Table 1). No striking differences in prescription patterns were observed between payers.

Analysis of baseline demographics and clinical characteristics by treatment group

Analysis of the demographics and clinical features on treatment selection revealed that, compared with NHRs who did not receive GLP-1 receptor agonist treatment, NHRs treated with GLP-1 receptor agonists were younger (aged <80 years); more often female; less often black or African American; and more cognitively intact (higher mean BIMS score and lower mean CPS score). They also tended to have less functional impairment (lower mean ADL score). However, these residents had a higher BMI (indicative of more severe obesity) and were prone to depression, although they were less likely to be diagnosed with dementia. NHRs who received GLP-1 receptor agonist therapy in combination with basal insulin were more likely to have comorbid hypertension, dyslipidemia, or heart failure (Table 3, Figure 3).

Table 3. Baseline demographics and clinical characteristics of NHRs with a diagnosis of type 2 diabetes who received antidiabetic medications (20% of NHRs had no claims for antidiabetic medications during the study period), by treatment group.

Characteristic	No GLP-1 RA therapy			GLP-1 RA therapy
	OAD only (n = 6467)	Basal insulin (n = 5084)	Basal + prandial insulin (n = 6340)	GLP-1 RA (n = 169)	GLP-1 RA + OAD (n = 40)	GLP-1 RA + basal insulin (n = 52)
Age, mean (SD), years	77.43 (11.6)	74.54 (12.3)	73.07 (12.3)	67.30 (10.5)	65.85 (10.4)	67.77 (11.5)
Gender, n (%)
Female	3928 (60.7)	2954 (58.1)	3654 (57.6)	109 (64.5)	30 (75.0)	32 (61.5)
Male	2539 (39.3)	2130 (41.9)	2686 (42.4)	60 (35.5)	10 (25.0)	20 (38.5)
Race, n (%)
White	4627 (71.5)	2837 (55.8)	3994 (63.0)	121 (71.6)	30 (75.0)	37 (71.2)
Black/African American	711 (11.0)	827 (16.3)	1127 (17.8)	9 (5.3)	2 (5.0)	2 (3.8)
Hispanic/Latino	673 (10.4)	991 (19.5)	763 (12.0)	31 (18.3)	6 (15.0)	11 (21.2)
Asian	233 (3.6)	230 (4.5)	206 (3.2)	1 (0.6)	0	1 (1.9)
Other	223 (3.5)	199 (4.0)	250 (4.0)	7 (4.1)	2 (5.0)	1 (1.9)
Cognitive pattern, mean (SD)
BIMS summary score	10.93 (4.3)	10.54 (4.4)	10.52 (4.2)	12.74 (3.0)	13.29 (2.7)	11.86 (3.6)
CPS score^a	4.01 (1.7)	4.37 (1.7)	4.44 (1.6)	3.71 (1.4)	3.60 (1.7)	4.00 (0.0)
Functional status, mean (SD)
Composite ADL score	13.11 (4.1)	14.11 (4.0)	14.38 (3.6)	12.83 (4.0)	12.80 (3.9)	13.12 (4.2)
Active comorbid disease diagnosis^b, n (%)
Hypertension	5559 (86.0)	4457 (87.7)	5725 (90.3)	151 (89.3)	34 (85.0)	51 (98.1)
Hyperlipidemia	3551 (54.9)	2762 (54.3)	4001 (63.1)	102 (60.4)	22 (55.0)	36 (69.2)
Depression	2917 (45.1)	2490 (49.0)	3624 (57.2)	106 (62.7)	24 (60.0)	38 (73.1)
Dementia (non-Alzheimer’s disease)	2187 (33.8)	1837 (36.1)	2417 (38.1)	29 (17.2)	9 (22.5)	9 (17.3)
Heart failure	1417 (21.9)	1568 (30.8)	2325 (36.7)	54 (32.0)	10 (25.0)	22 (42.3)
GERD or ulcer	1763 (27.3)	1528 (30.1)	2333 (36.8)	60 (35.5)	12 (30.0)	22 (42.3)
Asthma, COPD, or chronic lung disease	1414 (21.9)	1372 (27.0)	1884 (29.7)	48 (28.4)	7 (17.5)	16 (30.8)
Urinary tract infection	1395 (21.6)	1272 (25.0)	2077 (32.8)	38 (22.5)	3 (7.5)	16 (30.8)
Anxiety disorder	1538 (23.8)	1305 (25.7)	1814 (28.6)	49 (29.0)	9 (22.5)	17 (32.7)
Coronary artery disease	1279 (19.8)	1127 (22.2)	1805 (28.5)	39 (23.1)	6 (15.0)	14 (26.9)
BMI, mean (SD), kg/m^2	29.49 (7.8)	30.47 (8.4)	32.08 (9.6)	39.69 (11.0)	39.96 (10.8)	41.37 (11.4)

^aFor NHR with no BIMS summary scores available.

^bCerebrovascular accident, TIA, and stroke not included here as numbers/percentages are much lower than for other comorbidities.

Abbreviations: ADL = Activities of daily living; BIMS = Brief interview for mental status; BMI = Body mass index; COPD = Chronic obstructive pulmonary disease; CPS = Cognitive Performance Scale; GERD = Gastroesophageal reflux disease; GLP-1 RA = Glucagon-like peptide-1 receptor agonist; NHR = Nursing home residents; OAD = Oral antidiabetic drug; SD = Standard deviation; TIA = Transient ischemic attack.

Figure 3. Analysis of baseline demographics and clinical characteristics by treatment group for NHRs with a diagnosis of type 2 diabetes mellitus is shown.

Discussion

Our findings provide insight into the treatment patterns for diabetes in NHR. There is a high prevalence of insulin use, which could arise from the need to achieve better glycemic control or could reflect the advanced stage of diabetes in NHRs. The high utilization of short-acting insulin may also be an indicator of the implementation of sliding-scale treatment, despite contrary recommendations from the ADA [25] and AMDA [6]. The finding that 20% of NHRs with diabetes did not receive any antidiabetic medications during the study period suggests that some clinicians may be relying on non-pharmacologic therapies, such as dietary measures, to maintain blood glucose levels in this population. Despite the potential advantages of weight loss and reduced risk of hypoglycemia [1,5,7,8,25,26], our findings indicate a lower usage of GLP-1 receptor agonists in elderly NHRs with diabetes. Low GLP-1 receptor agonist use may be attributed to gastrointestinal side effects [2,27] or to the necessity for injections and the relatively high costs compared to insulin and non-branded OADs.

This retrospective analysis of the MDS and prescription claims data in the US NHRs revealed the prevalence of diabetes to be 35% in this population. This finding is consistent with a 24%–33% prevalence among NHRs aged ≥65 years, as reported by others over the past decade [11-13,28,29]. Increases in diabetes prevalence in NHRs have been accompanied by an increased prevalence of obesity. In turn, the increasing prevalence of obesity in NHRs in the United States has been associated with several comorbidities, including diabetes [30,31], pressure ulcers [32], and recurrent venous thromboembolism [33], and obese NHRs or hospitalized patients experience poorer outcomes of infections compared with normal-weight individuals [34]. In this study, >40% of NHRs with diabetes were obese. It is possible that the increasing prevalence of obesity in NHRs may be contributing to increases in diabetes in this care environment. Nearly one-fifth of NHRs with a diagnosis of diabetes were aged <65 years, which is younger than the average age of the general nursing home population [24]. Despite being younger than average, the NHRs in this study had a high prevalence of comorbidities as well as significant cognitive and functional impairments.

Given the increasing prevalence of diabetes and obesity, it is possible that the patient profile observed in this study (i.e. obese, younger than average, with many comorbidities) may become representative of the diabetes population in nursing homes in the future, with significant chronic treatment implications. For example, this population had a high level of cardiovascular comorbidity; this resulted in many concomitant medications being prescribed as part of a risk-reduction approach, in line with current quality metrics (i.e. adherence to statin and ACE inhibitor/ARB therapy) [35]. However, many diabetic nursing home residents may be ineligible for treatment with ACE inhibitor/ARB therapy due to absolute or relative contraindications. Only 51.3% of NHRs with diabetes received ACE inhibitor/ARB therapy, even though a hypertension diagnosis was present in 86.6% of these individuals; this may indicate a gap in care that can negatively affect Centers for Medicare and Medicaid Services quality metrics (i.e. the proportion of patients with diabetes and concomitant hypertension who are receiving ACE inhibitor/ARB therapy is low) or contraindications to ACE inhibitor/ARB therapy in the untreated population [35].

Additionally, there were noticeable differences in the characteristics of NHRs who received different antidiabetic treatments, thus suggesting that considerations other than glycemic control may have played a role in therapy selection. Examinations of the demographical, cognitive, functional, and clinical characteristics revealed that, compared with NHR who were not treated with GLP-1 receptor agonists, those who did receive GLP-1 receptor agonist treatment were less cognitively and functionally impaired, were younger, and were more obese.

Furthermore, NHRs in this group demonstrated lower treatment persistence than those receiving other treatment regimens. It may be possible that younger, more obese, and less cognitively and functionally challenged NHRs were more likely to be considered by clinicians for GLP-1 receptor agonist treatment due to the association of this treatment option with weight loss [26]. Issues relating to poor treatment persistence with GLP-1 receptor agonists may have also played a part in the observed reluctance of clinicians to prescribe these medications to NHRs, along with a lack of experience in the use of these newer agents in this patient population. The favorable safety profile of GLP-1 receptor agonists, the factors driving treatment choice, and issues relating to persistence, all deserve further exploration in obese, older, and more cognitively impaired NHRs. Metformin use rates were 53.6%, despite the fact that metformin is the medication of choice in several guidelines. Low metformin use rates may reflect the suggested advanced disease state of the studied population of NHRs, as well as continued reluctance to prescribe this medication in an elderly population because of underlying contraindications such as age >80 years, severe chronic kidney disease (i.e. creatinine clearance <30 mL/min), symptomatic New York Heart Association class III or class IV heart failure, or the inability to ingest oral medications due to swallowing difficulties or G-tube placement.

Treatment persistence was found to be low among NHRs who received treatment that included a GLP-1 receptor agonist. In NHRs on basal insulin therapy to which a GLP-1 receptor agonist was added, the number of persistent days was 1.5 months shorter in comparison with NHRs in whom an RAI was added to basal insulin. Another recent, real-world study demonstrated lower treatment persistence in patients with diabetes when a GLP-1 receptor agonists or insulin was added, compared with adding another OAD to combination OAD therapy. Almost 40% of patients in whom a GLP-1 receptor agonist was added switched therapy during the first year, compared with 25% of patients in whom insulin was added and 9% of those in whom a third OAD was added. However, the greater treatment persistence among patients with triple-OAD therapy did not translate into better glycemic control [36].

From a health policy perspective, this study assists in understanding the current state of affairs with respect to hyperglycemia management in a long-term care and elderly population. Additionally, this study provides information on persistence of therapies in an elderly institutionalized population seldom seen before. With the revelation of high obesity rates in the long-term care setting and advent of new treatment option, the results assist decision makers in choosing the most appropriate option for their patients in changing population of long-term care.

The primary limitations of this study arise from the fact that it was a retrospective database analysis; the data may have been subject to selection bias, confounding, and coding errors. Because of the observational nature of this study, data cannot be used to establish causality of the observed outcomes. As the MDS version 3.0 does not distinguish between type 1 and type 2 diabetes, it is possible, although unlikely, that some of the included NHRs had type 1 diabetes. However, the prevalence of type 2 diabetes appears to be correct, as 90%–95% of all diabetes cases in adults are type 2 diabetes, and the prevalence of type 2 diabetes is approximately seven times higher in those aged ≥65 years than in younger patients [37]. The fact that a more comprehensive definition of diabetes (using a combination of MDS and ICD-9-CM diagnostic codes [20]) was used in this report could be considered a particular strength of this study. However, it is possible that data discrepancies might have occurred due to miscoding. Agreement rates for several data elements have been reportedly poor; however, they were reliably adequate when calculated using a weighted kappa statistic for any unequal categories [18].

Other limitations include the fact that the presence of a claim for a filled-prescription can only be used as a proxy for estimating treatment persistence. Additionally, results obtained using the Omnicare Senior Health Outcomes data repository may not be representative of other populations with diabetes and prescription claims, and prescription data for NHRs who used pharmacies other than Omnicare to fill prescriptions were not captured in this study. It is possible that very few of those NHRs who had no prescription claims for antidiabetic therapy had prescriptions filled elsewhere. The number of NHRs treated with GLP-1 receptor agonists in this analysis was small, which may have also impacted the reliability of the data. Newer diabetes drugs that have recently become available (including the SGLT2 inhibitors) were not included in the analysis nor were drugs that are less frequently used for the treatment of diabetes (e.g. colesevelam or pramlintide). Finally, the study was unable to obtain data on clinical and medical outcomes. Future studies are needed to gain a more comprehensive understanding of the impact of current treatment patterns.

Conclusions

The prevalence of diabetes in NHRs in the United States was high – >1 in 3 NHRs – as was the prevalence of comorbidities and obesity, with >4 of 10 NHRs with diabetes having a BMI ≥30 kg/m². Given the increasing prevalence of diabetes and obesity, it is possible that the patient profile observed in the current study – which includes patients who are obese, younger than average, and have a high level of comorbidities – may be representative of the nursing home diabetes population of the future, with significant long-term treatment implications. Variances seen in the characteristics of NHRs who received different treatments suggest that considerations other than glycemic control may drive prescribing decisions. Additionally, despite guidelines and recommendations to the contrary, some use of sliding-scale treatment regimens appears to persist. Guidelines from the ADA/EASD [1] and the AGS [4] focus on personalization of diabetes treatment and relaxing glycemic goal attainment. Certain treatment options should be used with caution; the 2015 update to the position statement of the ADA/EASD guidelines specifically warns against the use of SGLT2 inhibitors in elderly patients [5]. The AMDA guidelines also encourage personal treatment goals and aim for a decline in hypoglycemia, hyperglycemia, and progression of diabetic complications [6].

Regardless of the high level of cardiovascular comorbidity, prescriptions of medications for risk reduction were lower than expected. GLP-1 receptor agonists were used infrequently in the studied NHR population, even though they offer potential advantages such as lower risk of hypoglycemia and weight loss. The utilization of newly emerging antihyperglycemic drugs, such as GLP-1 receptor agonists, as well as patient outcomes should be further investigated in longitudinal studies to guide healthcare providers’, payers’, and other stakeholders’ selection of the most appropriate and effective treatment strategies for older adults.

Although 20% of NHRs in this study did not receive any type of antidiabetic treatment, it is possible that they may rely on lifestyle and dietary glycemic regulation. Education on the approach to diabetes care in this population is warranted to assist prescribers in the development of appropriately age- and condition-individualized treatment plans for their patients.

Declaration of interest

Funding for this study was provided by Sanofi US, Inc. The authors received editorial and writing support in the preparation of this manuscript from Tessa Hartog, PhD, of Excerpta Medica, funded by Sanofi US, Inc. B Zarowitz, C Allen, and T O’Shea are employees of Omnicare Senior Health Outcomes, Inc., contracted on behalf of Sanofi. M Dalal and M Haumschild are employees of Sanofi. M Dalal is a stockholder of Sanofi and T O’Shea and B Zarowitz are stockholders of Omnicare. T O’Shea and B Zarowitz have received grant funding from Sanofi. A DiGenio was an employee of Sanofi at the time this study was performed and is currently an employee of Isis Pharmaceuticals. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Notes