National Trends in Treatment Initiation for Nursing Home Residents With Diabetes Mellitus, 2008 to 2010

Prescribing Trends of Glucose-Lowering Medications Near End of Life Among Adults With Type 2 Diabetes: A Cohort Study

OBJECTIVE

To assess prescribing trends of glucose-lowering medications in the last year of life among older adults with type 2 diabetes (T2D) and explore whether frailty is associated with differential prescribing.

RESEARCH DESIGN AND METHODS

In this observational cohort study of Medicare beneficiaries aged ≥67 years (2015-2019) with T2D, we assessed temporal trends in prescribing a glucose-lowering medication, stratified by frailty. The main outcome included glucose-lowering medication fills within 1 year of death.

RESULTS

Among 975,407 community-dwelling Medicare beneficiaries with T2D, the use of glucose-lowering medications within 1 year of death slightly increased from 71.4% during the first 6-month period in 2015 to 72.9% (standardized mean difference [SMD] -0.03) during the second 6-month period in 2019. The most pronounced increase in use was observed for metformin (40.7% to 46.5%, SMD -0.12), whereas the largest decrease was observed for sulfonylureas (37.0% to 31.8%, SMD 0.11). Overall glucose-lowering medication use decreased from 66.1% in the 9 to 12 months before death to 60.8% in the last 4 months of life (SMD 0.11; P < 0.01), driven by reduced noninsulin medication use. The use of short-acting and long-acting insulin both increased near death, with frailer individuals more likely to receive insulin. Sodium-glucose cotransporter-2 inhibitors and glucagon-like peptide 1 receptor agonists, although less common, became more frequent in more recent years.

CONCLUSIONS

The use of glucose-lowering medications declined in the last year of life, mainly due to reduced noninsulin use. Insulin use increased near death, particularly among frailer individuals, highlighting the need for careful end-of-life management.

A Bayesian method for adverse effects estimation in observational studies with truncation by death

Death among subjects is common in observational studies evaluating the causal effects of interventions among geriatric or severely ill patients. High mortality rates complicate the comparison of the prevalence of adverse events between interventions. This problem is often referred to as outcome "truncation" by death. A possible solution is to estimate the survivor average causal effect, an estimand that evaluates the effects of interventions among those who would have survived under both treatment assignments. However, because the survivor average causal effect does not include subjects who would have died under one or both arms, it does not consider the relationship between adverse events and death. We propose a Bayesian method which imputes the unobserved mortality and adverse event outcomes for each participant under the intervention they did not receive. Using the imputed outcomes we define a composite ordinal outcome for each patient, combining the occurrence of death and the adverse event in an increasing scale of severity. This allows for the comparison of the effects of the interventions on death and the adverse event simultaneously among the entire sample. We implement the procedure to analyze the incidence of heart failure among geriatric patients being treated for Type II diabetes with sulfonylureas or dipeptidyl peptidase-4 inhibitors.

Comparative safety and cardiovascular effectiveness of sodium-glucose cotransporter-2 inhibitors and glucagon-like peptide-1 receptor agonists in nursing homes

AIM

Studies examining the safety and effectiveness of sodium-glucose cotransporter-2 inhibitors (SGLT2is) versus glucagon-like peptide-1 receptor agonists (GLP-1RAs) among community-dwelling adults may not generalize to nursing home (NH) residents, who are typically older and more multimorbid. We compared the safety and cardiovascular effectiveness of SGLT2is and GLP-1RAs among US NH residents.

MATERIALS AND METHODS

Eligible individuals were aged ≥66 years with type 2 diabetes mellitus and initiated an SGLT2i or GLP-1RA in an NH between 2013 and 2018. Safety outcomes included fall-related injuries, hypoglycaemia, diabetic ketoacidosis (DKA), urinary tract infection or genital infection, and acute kidney injury in the year following treatment initiation. Cardiovascular effectiveness outcomes included death, major adverse cardiovascular events and hospitalization for heart failure. Per-protocol adjusted hazard ratios (HR) were calculated using stabilized inverse probability of treatment and censoring weighted cause-specific hazard regression models accounting for 127 covariates.

RESULTS

The study population included 7710 residents (31.08% SGLT2i, 68.92% GLP-1RA). Compared with GLP-1RA initiators, SGLT2i initiators had higher rates of DKA (HR 1.95, 95% confidence limits 1.27, 2.99) and death (HR 1.18, 95% confidence limits 1.02, 1.36). Rates of urinary tract infection or genital infection, acute kidney injury, major adverse cardiovascular events, and heart failure were also elevated, while rates of fall-related injuries and hypoglycaemia were reduced, but all estimates were imprecise and highly compatible with no difference.

CONCLUSIONS

SGLT2is do not have superior, and may have inferior, effectiveness compared with GLP-1RAs for cardiovascular and mortality outcomes in NH residents. Residents initiating SGLT2is should be monitored closely for DKA.

National trends in utilisation of glucose lowering medicines by older people with diabetes in long-term care facilities

AIMS

To examine national trends in glucose lowering medicine (GLM) use among older people with diabetes in long-term care facilities (LTCFs) during 2009-2019.

METHODS

A repeated cross-sectional study of individuals ≥65 years with diabetes in Australian LTCFs (n = 140,322) was conducted. Annual age-sex standardised prevalence of GLM use and number of defined daily doses (DDDs)/1000 resident-days were estimated. Multivariable Poisson or Negative binomial regression models were used to estimate adjusted rate ratios (aRRs) and 95 % confidence intervals (CIs).

RESULTS

Prevalence of GLM use remained steady between 2009 (63.9%, 95 %CI 63.3-64.4) and 2019 (64.3%, 95 %CI 63.9-64.8) (aRR 1.00, 95 %CI 1.00-1.00). The percentage of residents receiving metformin increased from 36.0% (95 %CI 35.3-36.7) to 43.5% (95 %CI 42.9-44.1) (aRR 1.01, 95 %CI 1.01-1.01). Insulin use also increased from 21.5% (95 %CI 21.0-22.0) to 27.0% (95 %CI 26.5-27.5) (aRR 1.02, 95 %CI 1.02-1.02). Dipeptidyl peptidase-4 inhibitor use increased from 1.0% (95 %CI 0.9-1.1) to 21.1% (95 %CI 20.7-21.5) (aRR 1.24, 95 %CI 1.24-1.25), while sulfonylurea use decreased from 34.4% (95 %CI 33.8-35.1) to 19.3% (95 %CI 18.9-19.7) (aRR 0.93, 95 %CI 0.93-0.94). Similar trends were observed in DDDs/1000 resident days.

CONCLUSIONS

The increasing use of insulin and ongoing use of sulfonylureas suggests a need to implement evidence-based strategies to optimise diabetes care in LTCFs.

Continuous Glucose Monitoring-Guided Insulin Administration in Long-Term Care Facilities: A Randomized Clinical Trial

OBJECTIVES

To evaluate the efficacy of real-time continuous glucose monitoring (rt-CGM) in adjusting insulin therapy in long-term care facilities (LTCF).

DESIGN

Prospective randomized clinical trial.

SETTINGS AND PARTICIPANTS

Insulin-treated patients with type 2 diabetes (T2D) admitted to LTCF.

METHODS

Participants in the standard of care wore a blinded CGM with treatment adjusted based on point-of-care capillary glucose results before meals and bedtime (POC group). Participants in the intervention (CGM group) wore a Dexcom G6 CGM with treatment adjusted based on daily CGM profile. Treatment adjustment was performed by the LTCF medical team, with a duration of intervention up to 60 days. The primary endpoint was difference in time in range (TIR 70-180 mg/dL) between treatment groups.

RESULTS

Among 100 participants (age 74.73 ± 11 years, 80% admitted for subacute rehabilitation and 20% for nursing home care), there were no significant differences in baseline clinical characteristics between groups, and CGM data were compared for a median of 17 days. There were no differences in TIR (53.38% ± 30.16% vs 48.81% ± 28.03%, P = .40), mean daily mean CGM glucose (184.10 ± 43.4 mg/dL vs 190.0 ± 45.82 mg/dL, P = .71), or the percentage of time below range (TBR) <70 mg/dL (0.83% ± 2.59% vs 1.18% ± 3.54%, P = .51), or TBR <54 mg/dL (0.23% ± 0.85% vs 0.56% ± 2.24%, P = .88) between rt-CGM and POC groups.

CONCLUSIONS AND IMPLICATIONS

The use of rtCGM is safe and effective in guiding insulin therapy in patients with T2D in LTCF resulting in a similar improvement in glycemic control compared to POC-guided insulin adjustment.

Adoption of sodium-glucose cotransporter-2 inhibitors among prescribers caring for nursing home residents

BACKGROUND

Sodium-glucose cotransporter-2 inhibitor (SGLT2I) use has increased among community-dwelling populations, but little is known about how clinicians have prescribed them for US nursing home (NH) residents. We described the adoption of SGLT2Is by prescribers caring for long-stay NH residents by clinician specialty and over time, compared with sulfonylureas, an older diabetes medication class.

METHODS

We conducted a retrospective cohort study of prescribers of SGLT2Is and sulfonylureas for all long-stay US NH residents aged 65 years or older (2017-2019). Using 100% of Medicare Part D claims linked to prescriber characteristics data, we identified all dispensings of SGLT2Is and sulfonylureas for long-stay NH residents and their associated prescribers. We described the distribution of prescriber specialties for each drug class over time as well as the number of NH residents prescribed SGLT2s versus sulfonylureas. We estimated the proportions of prescribers who prescribed both drug classes versus only sulfonylureas or only SGLT2Is.

RESULTS

We identified 36,427 unique prescribers (SGLT2I: N = 5811; sulfonylureas: N = 35,443) for 117,667 NH residents between 2017 and 2019. For both classes, family medicine and internal medicine physicians accounted for most prescriptions (75%-81%). Most clinicians (87%) prescribed only sulfonylureas, 2% prescribed SGLT2Is only, and 11% prescribed both. Geriatricians were least likely to prescribe only SGLT2Is. We observed an increase in the number of residents with SGLT2I use from n = 2344 in 2017 to n = 5748 in 2019.

CONCLUSIONS

Among NH residents, most clinicians have not incorporated SGLT2Is into their prescribing for diabetes, but the extent of use is increasing. Family medicine and internal medicine physicians prescribed the majority of diabetes medications for NH residents, and geriatricians were the least likely to prescribe only SGLT2Is. Future research should explore provider concerns regarding SGLT2I prescribing, particularly adverse events.

Comparative safety of sulfonylureas among U.S. nursing home residents

BACKGROUND

The comparative safety of sulfonylureas (SUs) in nursing home (NH) residents remains understudied despite widespread use. We compared the effects of three SU medications and initial SU doses on adverse glycemic and cardiovascular events among NH residents.

METHODS

This national retrospective cohort study linked Medicare claims with Minimum Data Set 2.0 assessments for long-stay NH residents aged ≥65 years between January 2008 and September 2010. Exposures were the SU medication initiated (glimepiride, glipizide, or glyburide) and doses (standard or reduced). One-year outcomes were hospitalizations or emergency department visits for severe hypoglycemia, heart failure (HF), stroke, and acute myocardial infarction (AMI). After the inverse probability of treatment and inverse probability of censoring by death weighting, we estimated hazard ratios (HR) using Cox regression models with robust 95% confidence intervals (CI).

RESULTS

The cohort (N = 6821) included 3698 new glipizide, 1754 glimepiride, and 1369 glyburide users. Overall, the mean (standard deviation) age was 81.4 (8.2) years, 4816 (70.6%) were female, and 5164 (75.7%) were White non-Hispanic residents. The rates of severe hypoglycemia were 30.3 (95% CI 22.3-40.1), 49.0 (95% CI 34.5-67.5), and 35.9 (95% CI 22.2-54.9) events per 1000 person-years among new glipizide, glimepiride, and glyburide users, respectively (glimepiride versus glipizide HR 1.6, 95% CI 1.0-2.4, p = 0.04; glyburide versus glipizide HR 1.2, 95% CI 0.7-1.9, p = 0.59). The rates of severe hypoglycemia were 27.1 (95% CI 18.6-38.0) and 42.8 (95% CI 33.6-53.8) events per 1000 person-years among new users of reduced and standard SU doses, respectively (HR 2.2, 95% CI 1.4-3.5, p < 0.01). Rates of HF, stroke, and AMI were similar between medications and doses.

CONCLUSIONS

Among long-stay NH residents, new use of glimepiride and standard SU doses resulted in higher rates of severe hypoglycemic events. Cardiovascular outcomes may not be affected by the choice of SU medication or dose.

Electronic Health Records to Rapidly Assess Biosimilar Uptake: An Example Using Insulin Glargine in a Large U.S. Nursing Home Cohort

Large healthcare administrative databases, like Medicare claims, are a common means to evaluate drug policies. However, administrative data often have a lag time of months to years before they are available to researchers and decision-makers. Therefore, administrative data are not always ideal for timely policy evaluations. Other sources of data are needed to rapidly evaluate policy changes and inform subsequent studies that utilize large administrative data once available. An emerging area of interest in both pharmacoepidemiology and drug policy research that can benefit from rapid data availability is biosimilar uptake, due to the potential for substantial cost savings. To respond to the need for such a data source, we established a public-private partnership to create a near-real-time database of over 1,000 nursing homes’ electronic health records to describe and quantify the effects of recent policies related to COVID-19 and medications. In this article, we first describe the components and infrastructure used to create our EHR database. Then, we provide an example that illustrates the use of this database by describing the uptake of insulin glargine-yfgn, a new exchangeable biosimilar for insulin glargine, in US nursing homes. We also examine the uptake of all biosimilars in nursing homes before and after the onset of the COVID-19 pandemic. We conclude with potential directions for future research and database infrastructure.

Effects of dipeptidyl peptidase-4 inhibitors and sulphonylureas on cognitive and physical function in nursing home residents

AIMS

Dipeptidyl peptidase-4 inhibitors (DPP4Is) may mitigate hypoglycaemia-mediated declines in cognitive and physical functioning compared with sulphonylureas (SUs), yet comparative studies are unavailable among older adults, particularly nursing home (NH) residents. We evaluated the effects of DPP4Is versus SUs on cognitive and physical functioning among NH residents.

MATERIALS AND METHODS

This new-user cohort study included long-stay NH residents aged ≥65 years from the 2007-2010 national US Minimum Data Set (MDS) clinical assessments and linked Medicare claims. We measured cognitive decline from the validated 6-point MDS Cognitive Performance Scale, functional decline from the validated 28-point MDS Activities of Daily Living scale, and hospitalizations or emergency department visits for altered mental status from Medicare claims. We compared 180-day outcomes in residents who initiated a DPP4I versus SU after 1:1 propensity score matching using Cox regression models.

RESULTS

The matched cohort (N = 1784) had a mean ± SD age of 80 ± 8 years and 73% were women. Approximately 46% had no or mild cognitive impairment and 35% had no or mild functional impairment before treatment initiation. Compared with SU users, DPP4I users had lower 180-day rates of cognitive decline [hazard ratio (HR) = 0.61, 95% confidence interval (CI) 0.31-1.19], altered mental status events (HR = 0.71, 95% CI 0.39-1.27), and functional decline (HR = 0.89, 95% CI 0.51-1.56), but estimates were imprecise.

CONCLUSIONS

Rates of cognitive and functional decline may be reduced among older NH residents using DPP4Is compared with SUs, but larger studies with greater statistical power should resolve the remaining uncertainty by providing more precise effect estimates.

Comparative safety of dipeptidyl peptidase-4 inhibitors and sulfonylureas among frail older adults

BACKGROUND

Studies comparing dipeptidyl peptidase-4 inhibitors (DPP4Is) to sulfonylureas (SUs) are unavailable for frail older adults, especially nursing home (NH) residents. We examined the effects of DPP4Is versus SUs on severe adverse glycemic events, cardiovascular events, and death among NH residents.

METHODS

We conducted a national retrospective cohort study of long-stay NH residents aged ≥65 years using 2008-2010 national US Minimum Data Set clinical assessment data and linked Medicare claims. Exposure was new DPP4I versus new SU use assessed via Medicare Part D drug claims. One-year outcomes were severe hypoglycemia, severe hyperglycemia, acute myocardial infarction (AMI), heart failure (HF), major adverse cardiovascular events plus HF (MACE+HF), and death. We compared outcomes after propensity score matching using Cox proportional hazards regression models.

RESULTS

The cohort (N = 2016) had a mean (SD) age of 81 (8.1) years and was 72% female. Compared with SU users, DPP4I users had a lower 1-year rate of severe hypoglycemic events (HR = 0.57, 95% CI 0.34-0.94), but statistically similar rates of severe hyperglycemic events (HR = 0.94, 95% CI 0.52-1.72), AMI (HR = 0.76, 95% CI 0.44-1.30), HF (HR = 1.01, 95% CI 0.79-1.30), MACE+HF (HR = 0.90, 95% CI 0.72-1.12), and death (HR = 0.97, 95% CI 0.86-1.10).

CONCLUSIONS

DPP4Is should be a preferred treatment option over SUs for NH residents and other frail older adults given the importance of avoiding hypoglycemia.

Outcomes of "diabetes-friendly" vs "diabetes-unfriendly" β-blockers in older nursing home residents with diabetes after acute myocardial infarction

AIMS

To assess whether nursing home (NH) residents with type 2 diabetes mellitus (T2D) preferentially received "T2D-friendly" (vs "T2D-unfriendly") β-blockers after acute myocardial infarction (AMI), and to evaluate the comparative effects of the two groups of β-blockers.

MATERIALS AND METHODS

This new-user retrospective cohort study of NH residents with AMI from May 2007 to March 2010 used national data from the Minimum Data Set and Medicare system. T2D-friendly β-blockers were those hypothesized to increase peripheral glucose uptake through vasodilation: carvedilol, nebivolol and labetalol. Primary outcomes were hospitalizations for hypoglycaemia and hyperglycaemia in the 90 days after AMI. Secondary outcomes were functional decline, death, all-cause re-hospitalization and fracture hospitalization. We compared outcomes using binomial and multinomial logistic regression models after propensity score matching.

RESULTS

Of 2855 NH residents with T2D, 29% initiated a T2D-friendly β-blocker vs 24% of 6098 without T2D (P < 0.001). For primary outcomes among residents with T2D, T2D-friendly vs T2D-unfriendly β-blockers were associated with a reduction in hospitalized hyperglycaemia (odds ratio [OR] 0.45, 95% confidence interval [CI] 0.21-0.97), but unassociated with hypoglycaemia (OR 2.05, 95% CI 0.82-5.10). For secondary outcomes, T2D-friendly β-blockers were associated with a greater rate of re-hospitalization (OR 1.26, 95% CI 1.01-1.57), but not death (OR 1.06, 95% CI 0.85-1.32), functional decline (OR 0.91, 95% CI 0.70-1.19), or fracture (OR 1.69, 95% CI 0.40-7.08).

CONCLUSIONS

In older NH residents with T2D, T2D-friendly β-blocker use was associated with a lower rate of hospitalization for hyperglycaemia, but a higher rate of all-cause re-hospitalization.

Metformin Safety Warnings and Diabetes Drug Prescribing Patterns for Older Nursing Home Residents

OBJECTIVE

Diabetes mellitus is common in US nursing homes (NHs), and the mainstay treatment, metformin, has US Food and Drug Administration (FDA) boxed warnings indicating safety concerns in those with advanced age, heart failure, or renal disease. Little is known about treatment selection in this setting, especially for metformin. We quantified the determinants of initiating sulfonylureas over metformin with the aim of understanding the impact of FDA-labeled boxed warnings in older NH residents.

DESIGN AND SETTING

National retrospective cohort in US NHs.

PARTICIPANTS

Long-stay NH residents age ≥65 years who initiated metformin or sulfonylurea monotherapy following a period of ≥6 months with no glucose-lowering treatment use between 2008 and 2010 (n = 7295).

MEASUREMENTS

Measures of patient characteristics were obtained from linked national Minimum Data Set assessments; Online Survey, Certification and Reporting (OSCAR) records; and Medicare claims. Odds ratios (ORs) comparing patient characteristics and treatment initiation were estimated using univariable and multivariable multilevel logistic regression models with NH random intercepts.

RESULTS

Of the 7295 residents in the study population, 3066 (42%) initiated metformin and 4229 (58%) initiated a sulfonylurea. In multivariable analysis, several factors were associated with sulfonylurea initiation over metformin initiation, including heart failure (odds ratio [OR] 1.2, 95% confidence interval [CI] 1.1-1.4) and renal disease (OR 2.1, 95% CI 1.7-2.5). Compared with those aged 65 to <75 years, residents 75 to <85 (OR 1.3, 95% CI 1.2-1.5), 85 to <95 (OR 2.0, 95% CI 1.7-2.3), and ≥95 (OR 4.3, 95% CI 3.2-5.8) years were more likely to initiate sulfonylureas over metformin.

CONCLUSIONS

In response to FDA warnings, providers initiated NH residents on a drug class with a known, common adverse event (hypoglycemia with sulfonylureas) over one with tenuous evidence of a rare adverse event (lactic acidosis with metformin).