Diabetes Care e1

Insulin Detemir Use Is Associated With Irit Hochberg Higher Occurrence of Hypoglycemia in Hospitalized Patients With Hypoalbuminemia https://doi.org/10.2337/dc17-1957

The insulin analogs detemir and glargine Data were collected for all patients hospitalization, minimal albumin, maximal are the two basal insulins commonly used over 18 years old hospitalized between creatinine, admission length, and ward. in hospitalized patients (1). Their slow- 1 February 2012 and 31 December 2016, Next, a forward and backward stepwise release reservoir is differentdserum treated with detemir or glargine, and with Akaike information criterion regression albumin for detemir and subcutaneous at least one hospitalization albumin blood method was performed; the depen- microcrystals for glargine (2,3). test. The date of lowest albumin level was dent variable was glucose #70 mg/dL. There is a common belief that detemir set as the index event. Exclusion criteria An identical logistic analysis was done should be avoided in hypoalbuminemic were “obstetric ward” (where glargine for glucose #54 mg/dL. Last, a stepwise patients because they have less predict- is not used) and glucose ,15 mg/dL or forward and backward regression method able levels of free insulin, raising a con- albumin ,1 g/dL (owing to the scarcity of with second order interactions was cern for higher risk of hypoglycemia (4,5). such events). The time frame for hypogly- performed. Nevertheless, the guidelines recommend cemia was defined as 5 days around the To further test the “common belief” using either insulin with impartiality index event (2 days before and 2 days after). regarding the effect of detemir on hypo- to hypoalbuminemia (1). There are no Patient variables extracted were de- glycemia risk in hypoalbuminemic pa- human studies on pharmacokinetics or mographics; hospitalization length; ward; tients, we used the stepwise regression outcomes of using detemir in hypoalbu- relevant medications; and minimal, maxi- model results. Independent variables minemic patients. Randomized controlled mal, and standard deviation of albumin, were fixed using the mean (continuous studies of detemir versus glargine have creatinine (spectrophotometric autoana- variables) or highest frequency (categori- not found a difference in hypoglycemia lyzer [Dimension; Siemens, IL]), and glu- cal variables) in the cohort. Albumin levels (6–9), but these trials included only a few cose (acquired by StatStrip central were set between 1 and 4 g/dL in 0.1 hundred patients who probably were not glucometer [Nova Biomedical] or the intervals, and hypoglycemia risk was pre- hypoalbuminemic. A recent retrospective same spectrophotometric analyzer). dicted using the above models.

study found a higher rate of hypoglycemia Glargine- or detemir-treated patients Overall, 4,677 inpatients met inclusion e-LETTERS (,70 mg/dL) in determir-treated com- were first compared using simple bivari- criteria, 82.2% treated with glargine and pared to glargine-treated inpatients (10). ate analysis. Logarithmic transformation 17.8% with detemir. Glargine-treated pa- We took advantage of the large Rambam was applied for variables with a skewed tients differed from detemir-treated pa- –

Health Care Campus inpatient database to distribution. tients in some baseline characteristics: OBSERVATIONS retrospectively compare hypoglycemia in To assess the relationship between they were older (67.2 6 13.9 vs. 65.3 6 inpatients with low albumin treated with albumin, insulin glargine/detemir, and 13.6 years), had a higher male percentage detemir or glargine. Data were retrieved hypoglycemia, the following regression (56.9% vs. 51.2%), had lower weight (84.3 6 using MDClone (https://www.mdclone models were used. First, univariate regres- 20.4 vs. 86.4 6 20.5 kg), had a longer .com), a query tool that provides compre- sion coefficients measured associations hospital stay (12.9 6 17.2 vs. 11.4 6 hensive patient-level data of wide-ranging between hypoglycemia and variables 14.9 days) (P , 0.05), and had a lower variables in a defined time frame around an that impact hypoglycemia: age, sex, minimal albumin measurement (2.6 6 index event. The study was approved by weight, insulin dose, home insulin before 0.74 vs. 2.8 6 0.7 g/dL) (P , 0.0001) the Rambam Institutional Review Board. hospitalization, short-acting insulin during but no difference in albumin variability.

Endocrinology, Diabetes, and Metabolism Institute, Rambam Health Care Campus, Haifa, Israel Corresponding author: Irit Hochberg, [email protected]. Received 19 September 2017 and accepted 2 January 2018. © 2018 by the American Diabetes Association. Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license. Diabetes Care Publish Ahead of Print, published online February 1, 2018 e2 Hypoglycemia in Inpatients With Hypoalbuminemia Diabetes Care

Figure 1—Illustration of logistic model for prediction of the risk for hypoglycemia according to albumin measurements. Model is based on fixed-profile patients with the following characteristics: age = 67 years, weight = 84 kg, creatinine = 1.6 mg/dL, basal insulin dose = 21 units, treated with insulin at home, hospitalized in an internal medical ward, length of stay = 12 days.

The rate of glucose ,70 mg/dL was 16.8% internal medicine increased hypoglyce- patients found that there is an excess risk under glargine and 20.9% under detemir mia by 35% compared with all other for hypoglycemia in inpatients with hypo- (P , 0.05), and the rate of glucose wards, and each additional hospitaliza- albuminemia treated with detemir. This ,54 mg/dL was 6.3% under glargine tion day increased hypoglycemia by supports the long-held belief that insulin and 8.4% under detemir (P , 0.05). Me- 15%. Minimal albumin level was nega- detemir levels are less predictable in hypo- dian and maximal glucose were slightly tively associated with hypoglycemia albuminemia. On the other hand, there higher under glargine (191.3 6 63.8 vs. (odds ratio = 0.68, 95% CI 0.58–0.80) was no difference in occurrence of hypo- 186.1 6 62.2 mg/dL and 306.8 6 114.1 such that for 1 g/dL decrease in minimal glycemia between the two insulins in vs. 296.4 6 107.7 mg/dL, respectively; albumin, hypoglycemia increased by 32%. individuals with albumin over 3 g/dL, con- P , 0.05), and mean glucose was similar Results for hypoglycemia at glucose firming the equivalent safety of both in- (198.7 6 61.4 vs. 194.4 6 60.2 mg/dL). ,54 mg/dL were very similar. sulins in the majority of people. Fewer glargine-treated patients were The difference in albumin levels of The main limitations of this study are treated with insulin at home (51.3% vs. glargine- versus detemir-treated inpa- its retrospective design and the unavoid- 66.5%), and more were treated in paral- tients demonstrates that physicians tend able differences between glargine and lel with rapid-acting insulin (84.4% vs. to avoid detemir in hypoalbuminemia. To detemir groups. We made an effort to 80.3%) (P , 0.0001). Glargine-treated evaluate the reliability of the “common overcome these differences by using a patients had a lower basal insulin dose belief” that detemir should be avoided multivariate model that corrects for all (20.8 6 11.7 vs. 23.1 6 13.3 units) (P , in hypoalbuminemia, we evaluated the these differences. 0.0001), but there was no difference in effect of albumin level and insulin type The results suggest that it may be units/kg (P =0.7). on hypoglycemia, while overcoming advisable to consider avoiding the use Univariate logistic regression analysis physicians’ bias, by generating a multivar- of detemir (and perhaps other albumin- found that risk of hypoglycemia was iate model taking into account only albu- bound insulins) in individuals with hypo- associated with age, weight, length of min while fixing all other variables. This albuminemia. stay, maximal creatinine, minimal albu- model is presented in Fig. 1, which illus- min, home insulin treatment, basal insulin trates that for individuals with low albumin, , dose, and use of rapid insulin (P 0.005). the risk for hypoglycemia under detemir Acknowledgments. The author wishes to thank These variables were entered as possible treatment is significantly higher compared Sara Tzafrir, Idan Sipori, Ram Ben-Mayor, and predictors into the Akaike informat- with glargine treatment, but for normal- Marianna Sherman from the Rambam Health ion criterion analysis. Treatment with range albumin, the same risk for hypoglyce- CareCampusInformation, Computerization, and Communications Department for assistance in detemir increased the odds for hypogly- mia is present for the two insulins. retrieving the data. cemia by 35%. Home insulin increased In conclusion, this retrospective analy- Duality of Interest. No conflicts of interest rele- hypoglycemia by 62%. Hospitalization in sis of a large database of hospitalized vant to this article were reported. care.diabetesjournals.org Hochberg e3

Author Contributions. I.H. researched data pharmacology of insulin detemir. Int J Obes Relat 7. Raskin P, Gylvin T, Weng W, Chaykin L. Com- and wrote and edited the manuscript. I.H. is the Metab Disord 2004;28(Suppl. 2):S23–S28 parison of insulin detemir and insulin glargine guarantor of this work and, as such, had full 3. Coppolino R, Coppolino S, Villari V. Study of the using a basal-bolus regimen in a randomized, con- access to all the data in the study and takes re- aggregation of insulin glargine by light scatter- trolled clinical study in patients with type 2 diabe- sponsibility for the integrity of the data and the ing. J Pharm Sci 2006;95:1029–1034 tes. Diabetes Metab Res Rev 2009;25:542–548 accuracy of the data analysis. 4. Goldman-Levine JD, Lee KW. Insulin detemird 8. Davis S, Friece C, Roderman N, Newcomer D, Prior Presentation. This study was presented a new basal insulin analog. Ann Pharmacother Castaneda E. Comparison of insulin detemir and as a lecture and abstract the annual meeting of 2005;39:502–507 insulin glargine for hospitalized patients on a the Israeli Association for Medical Informatics, 5. Reilly JB, Berns JS. Selection and dosing of basal-bolus protocol. Pharmacy (Basel) 2017;5: Tel Aviv, 22–23 October 2017. medications for management of diabetes in pa- 22 tients with advanced kidney disease. Semin Dial 9. Zhang T, Lin M, Li W, et al. Comparison of the References 2010;23:163–168 efficacy and safety of insulin detemir and insulin 1. Umpierrez GE, Hellman R, Korytkowski MT, 6. Swinnen SG, Dain MP, Aronson R, et al. A glargine in hospitalized patients with type 2 dia- et al.; Endocrine Society. Management of hyper- 24-week, randomized, treat-to-target trial com- betes: a randomized crossover trial. Adv Ther glycemia in hospitalized patients in non-critical paring initiation of insulin glargine once-daily 2016;33:178–185 care setting: an endocrine society clinical practice with insulin detemir twice-daily in patients with 10. Galindo RJ, et al. Comparison of efficacy and guideline. J Clin Endocrinol Metab 2012;97:16–38 type 2 diabetes inadequately controlled on oral safety of glargine and determir insulin in the man- 2. Kurtzhals P. Engineering predictability and glucose-lowering drugs. Diabetes Care 2010;33: agement of inpatient hyperglycemia and diabe- protraction in a basal insulin analogue: the 1176–1178 tes. Endocr Pract 2017;23:1059–1066