Insulin Degludec for the Management of Patient With Recurrent Diabetic Ketoacidosis

Hypothesis: Use of a longer-lasting basal insulin (degludec) in patients with recurrent Diabetic Ketoacidosis (DKA) will reduce the rate of recurrent DKA and improve adherence to daily administration in patients who have had 2 or more episode of DKA in the past 2 years. STUDY DESIGN AND METHODS: Patients admitted with Diabetic Ketoacidosis (DKA) to Hennepin County Medical Center (HCMC) who have a prior episode of DKA within 2 years will be offered enrollment in a prospective, randomized unblinded trial to determine whether long acting insulin degludec will reduce the rate of recurrent DKA and improve compliance over the period of 1 year when compared with the control group who will stay on their pre-study long acting insulin (glargine or detemir). Primary Outcomes: To determine the difference in rate of recurrent Diabetic Ketoacidosis (DKA) over 6 months and 12 months in participants who have a history of previous DKA and medication adherence between groups treated with degludec vs other long-acting insulins (glargine or detemir). Secondary outcomes: BG control (Hba1c during study period) tested q 3 months Discharge long-acting insulin dose compared to admission dose LOS for DKA admission Number of readmissions in one year Number of ER visits Hypoglycemia -self reported, documented by glucometer or ER visit for severe hypoglycemia Adherence to study regimen from phone call information and clinic visits Inclusion criteria: Patients age \>18 year with previous diagnosis of Diabetes Mellitus and previous admission for DKA in any facility within the prior 2 years who meet the criteria for DKA at the time of admission and are willing to participate in the study protocol will be enrolled in the study. DKA is defined as: 1. Serum bicarbonate below 18 mg/dl 2. Anion Gap over greater than or equal to 16 3. Serum ketones or beta hydroxybutyrate elevation \> 3 mmol/L 4. Serum glucose greater than or equal to 250 mg/dl Exclusion Criteria: Patients with no previous DKA, patients who are not willing to consent to participate in the study protocol for 1 year, pregnant women, dementia, and end stage renal disease on dialysis. Study Protocol: Investigators will obtain informed consent and enroll participants admitted to the Hennepin County Medical Center (HCMC) in this study from Jan 2016 until investigators have adequate enrollment. Investigators will collect the initial data as following: 1. Age, sex, height, weight, BMI 2. Age at diagnosis of diabetes 3. Number previous DKA admissions in past 2 years, lifetime (even if the admission was to other hospitals) 4. Preadmission insulin dosing - long acting 5. Reason for DKA 6. Alcohol, drug use, homelessness, mental illness 7. Admission labs - HCO3, AG, BOHB, glucose, creatinine 8. Length of stay with each DKA episode( LOS) The participants will be randomly assign to the intervention or control groups. Intervention Group: Discharge on once a day degludec insulin by pen (supplied by Novo) at dose 0.3 U/kg with reduction to 0.2 U/kg for GFR \<30 ml/hr. Degludec insulin will be started at study initiation during the hospitalization if the participant is randomized to this arm. Fast acting insulin that participant is using will be adjusted so that total daily dose given to cover 2 or 3 meals is equal to the long-acting insulin dose. Participants will have basal bolus regimen either using carbohydrate counting or with set doses for meals, depending on their ability. Participants will use the fast acting insulin they were on at the time of admission. Insulin adjustment if hypoglycemia (less than 60 mg/dl) or if fasting BG over 140 despite insulin compliance. For hypoglycemia, dose will be lowered 0.05 U/kg and for elevated fasting glucose it will be raised 0.05 U/kg. Phone calls every month to check on dose, compliance, hospital admissions, and hypoglycemia. Compliance with medication will be verified verbally. Clinic visits every 3 months for 1 year with dose adjustment as above. Supply adequate insulin degludec until next visit. Control Group: Discharge on 0.3 U/kg of admission long acting insulin glargine or detemir (provided by diabetes and endocrine clinic) regardless of participant's home insulin regimen at the time of admission with reduction to 0.2 U/kg in CKD with GFR \<30 ml/hr. This will be done at the time of study initiation during the hospitalization if the participant is randomized to this arm. Fast acting insulin adjustment with meals and insulin adjustment for hypoglycemia will be done similarly as in the intervention group. Phone calls every month to check on dose, compliance, hospital admissions, and hypoglycemia. Compliance will be assessed verbally. Clinic visits every 3 months for 1 year with insulin dose adjustment as above. Supply adequate insulin glargine or detemir until next visit. Study site: Hennepin County Medical Center (HCMC), Minneapolis, MN. Data Analysis: Sample size, power calculation and statistical plan Investigators determined that considering incidence of DKA 100 % (since all patients admitted with DKA would be eligible for screening), moreover, in order to observe 30 % reduction in DKA related hospitalization over 2 years period in intervention group and by keeping power at 80 % with confidence interval (CI) of 0.05, study will need to enroll 21 subjects in each group or 42 total sample size. In order to mitigate the effects of loss to follow up, noncompliance and protocol deviations, intention to treat (ITT) analysis will be performed on primary and secondary outcome measure for both groups. A separate analysis will be done including those patient who will complete the trial in both groups. Baseline data of intervention and control group will be compared. In case of significant differences, further analysis will be done to adjust for those differences prior to final analysis. ITT analysis will be based on in ITT comparison. Descriptive and inferential statistical methods will be used for analyzing the data using Statistical Analysis software (SAS) 9.4. For continuous variables, mean, median and variance will be reported, while for discrete variables rate, risk ratios, relative risk, rate of occurrence and percentage differences will be reported. For primary outcome measure, investigators will use χ2 test to assess relationship between the primary outcome and some of the binary nature secondary measures. Stratification will be used for analysis as needed. In that case, chi-square tests for overall and within relevant stratifications will be conducted. To assess the difference between continuous variables (such as hemoglobin A1C, bicarbonate, anion gap, Creatinine etc.), t-test will be used for normally distributed data. Otherwise Mann-Whitney U test will be used. P-values at 95 % CI will be reported. Since it is a superiority trial, we will use one-tail test for significance. .