Approximately one-quarter of the US population die in nursing homes,1 where end-of-life care is of variable quality.2 In particular, it is unknown whether patients with chronic illness, such as diabetes, continue to receive burdensome testing and treatment after transitioning to hospice care in nursing homes. Experts and the American Diabetes Association recommend relaxing glycemic control target levels for patients with diabetes and advanced disease and eventual discontinuation of medications as patients near death to avoid hypoglycemia.3,4 Hypoglycemia causes symptoms of weakness, diaphoresis, confusion, shakiness, and dizziness,5 and is a potentially preventable cause of suffering among hospice patients. Whether nursing home patients with type 2 diabetes on hospice are assessed for dysglycemia, receive insulin or oral hypoglycemic medications, or experience hypoglycemia and hyperglycemia has not previously been described.
We conducted a retrospective cohort study of patients older than 65 years with type 2 diabetes admitted to Veterans Affairs (VA) nursing homes between January 1, 2006, and June 30, 2015, using linked laboratory, pharmacy, and administrative data. We identified patients with type 2 diabetes by International Classification of Diseases, Ninth Revision (ICD-9) code or glycated hemoglobin values greater than 6.5%. We excluded patients with type 1 diabetes by ICD-9 code. For patients with multiple admissions during the study period, we chose the last admission. We used descriptive statistics to analyze demographic variables, comorbidities, and diabetes management (laboratory testing and drug administration), and stratified the cohort by whether patients received insulin while on hospice. We analyzed the cumulative incidence of hypoglycemia (glucose <70 mg/dL [to convert to mmol/L, multiply by 0.0555]), severe hypoglycemia (glucose <50 mg/dL), hyperglycemia (glucose ≥400 mg/dL), and the competing risk of death among all hospice patients and among patients treated with insulin vs patients not treated with insulin. This study was reviewed and approved by the University of California, San Francisco Committee on Human Research.
The study cohort included 20 329 hospice patients (Table), 98% of whom were men (n = 19 991). Hospice patients had an 83% 100-day mortality rate (n = 16 791 deaths), and a median length of stay of 10 days. Eight percent of patients in the cohort received insulin (n = 1687). Among patients treated with insulin, mean baseline glycated hemoglobin levels were higher than patients not treated with insulin (7.4% vs 6.8%; P < .001), and the mortality rate at 100 days was lower (61% vs 85%; P < .001). Patients treated with insulin had more frequent glucose tests (mean 1.7 glucose tests/d, vs 0.6 glucose tests/d among patients not treated with insulin; P < .001). The cumulative incidence of hypoglycemia (glucose <70 mg/dL) among all patients, accounting for the competing risk of death, was 12% at 180 days, and that of severe hypoglycemia (glucose <50 mg/dL) was 5% (Figure). Among patients treated with insulin, 38% experienced hypoglycemia and 18% experienced severe hypoglycemia at 180 days. The highest risk of hypoglycemia occurred in the first 20 days of admission. The cumulative incidence of hyperglycemia (glucose >400 mg/dL) at 180 days was 9% in all patients, higher in the group treated with insulin (35%).
Despite guidelines that stress avoiding hypoglycemia in hospice patients with diabetes,4 we found that 1 in 9 nursing home patients with type 2 diabetes experienced hypoglycemia (glucose <70 mg/dL) while 1 in 20 experienced severe hypoglycemia (glucose <50 mg/dL) while on hospice. The risk of hypoglycemia was highest among patients treated with insulin, one-third of whom experienced hypoglycemia. Patients treated with insulin lived longer and experienced more hyperglycemia than patients not treated with insulin, which suggests that clinicians may be choosing to continue insulin for those hospice patients with a longer life expectancy and more severe diabetes at hospice admission. Nevertheless, hypoglycemia is not consistent with a goal of comfort, and these data demonstrate suboptimal avoidance of dysglycemia among patients with type 2 diabetes on hospice in nursing homes. Further research is needed to establish optimal timing of diabetes medication titration and cessation and characterize the effect of hypoglycemia and hyperglycemia on the symptom burden of patients with diabetes on hospice.
Corresponding Author: Laura A. Petrillo, MD, Department of Medicine, Massachusetts General Hospital, 55 Fruit St, Founders 600, Boston, MA 02114 (firstname.lastname@example.org).
Accepted for Publication: November 6, 2017.
Published Online: December 26, 2017. doi:10.1001/jamainternmed.2017.7744
Author Contributions: Dr Petrillo had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Petrillo, Lee.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Petrillo.
Critical revision of the manuscript for important intellectual content: Gan, Jing, Lang-Brown, Boscardin, Lee.
Statistical analysis: Petrillo, Gan, Jing, Boscardin, Lee.
Obtained funding: Lee.
Administrative, technical, or material support: Lang-Brown, Lee.
Study supervision: Lee.
Conflict of Interest Disclosures: None reported.
et al. Severe hypoglycemia symptoms, antecedent behaviors, immediate consequences and association with glycemia medication usage: Secondary analysis of the ACCORD clinical trial data. BMC Endocr Disord
. 2012;12(1):5. doi:10.1186/1472-6823-12-5PubMedGoogle ScholarCrossref