Probability of Achieving Glycemic Control with Basal Insulin in Patients with Type 2 Diabetes in Real-World Practice in the USA

Probability of Achieving Glycemic Control with Basal Insulin in Patients with Type 2 Diabetes in... Diabetes Ther (2018) 9:1347–1358 https://doi.org/10.1007/s13300-018-0413-5 BRIEF REPORT Probability of Achieving Glycemic Control with Basal Insulin in Patients with Type 2 Diabetes in Real-World Practice in the USA . . . Lawrence Blonde Luigi Meneghini Xuejun Victor Peng . . . . Anders Boss Kyu Rhee Alka Shaunik Supriya Kumar . . Sidhartha Balodi Claire Brulle-Wohlhueter Rory J. McCrimmon Received: March 2, 2018 / Published online: March 29, 2018 The Author(s) 2018 least one valid glycated hemoglobin (HbA1c) ABSTRACT result recorded both within 90 days before and 720 days after BI initiation were selected. We Introduction: Basal insulin (BI) plays an estimated the changes from baseline in HbA1c important role in treating type 2 diabetes (T2D), every 6 months, the quarterly conditional especially when oral antidiabetic (OAD) medi- probabilities of reaching HbA1c \ 7% if a cations are insufficient for glycemic control. We patient had not achieved glycemic control prior conducted a retrospective, observational study to each quarter (Q), and the cumulative proba- using electronic medical records (EMR) data bility of reaching glycemic control over from the IBM Explorys database to evaluate 24 months. the probability of achieving glycemic control Results: Our cohort was representative of over 24 months after BI initiation in patients patients with T2D who initiated BI from OADs with T2D in the USA. in the USA. The average HbA1c was 9.1% at BI Methods: A cohort of 6597 patients with T2D initiation, and decreased robustly (1.5%) in the who started BI following OAD(s) and had at first 6 months after initiation with no further reductions thereafter. The conditional proba- Enhanced content To view enhanced content for this bility of reaching glycemic control decreased article go to https://doi.org/10.6084/m9.figshare. rapidly in the first year (26.6% in Q2; 17.6% in Q3; 8.6% in Q4), and then remained low (B 6.1%) for each quarter in the second year. L. Blonde (&) Ochsner Medical Center, New Orleans, LA, USA Cumulatively, about 38% of patients reached e-mail: lawrence.blonde@gmail.com HbA1c \ 7% in the first year; only approxi- mately 8% more did so in the second year. L. Meneghini University of Texas, Southwestern Medical Center Conclusion: Our study of real-world data from and Parkland Health & Hospital System, Dallas, TX, a large US EMR database suggested that among USA patients with T2D who initiated BI after OADs, the likelihood of reaching glycemic control X. V. Peng  A. Boss  A. Shaunik C. Brulle-Wohlhueter diminished over time, and remained low from Sanofi, Bridgewater, NJ, USA 12 months onwards. Additional treatment options should be considered if patients do not K. Rhee  S. Kumar  S. Balodi Watson Health, IBM, Cambridge, MA, USA reach glycemic control within 12 months of BI initiation. R. J. McCrimmon Funding: Sanofi Corporation. University of Dundee, Dundee, UK 1348 Diabetes Ther (2018) 9:1347–1358 Keywords: Basal insulin; IBM Explorys database; proportions of patients achieving target HbA1c Real-world evidence; Type 2 diabetes levels (HbA1c \ 7%) in these studies have ran- ged from as low as 11% after 1 year to as high as 58% over 2.5 years [11–19]. The likelihood of INTRODUCTION reaching glycemic control with BI treatment in the real world remains a question of interest to Type 2 diabetes (T2D) affects 30.3 million peo- patients, clinicians, healthcare systems, and ple (* 9.4% of the population) in the USA as of payers. TM 2017 [1]. The economic and social costs of T2D The IBM Watson Health Explorys data- are considerable because of its associated serious base (later referred to as Explorys) is a large, US short- and long-term complications, particu- population-based, commercial database that larly in patients who do not achieve and/or contains an aggregate of electronic medical maintain glycemic control [2, 3], as well as its record (EMR) data from over 54 million unique significant contribution to overall mortality patients from 39 major integrated healthcare (approximately 12% of all deaths) [4]. Hyper- systems covering over 344,000 clinicians across glycemia, which may result from insufficient all 50 US states. It contains de-identified, treatment intensification, significantly increa- structured, longitudinal patient data (including ses the risks of microvascular complications, diagnoses, laboratory results, biometric mea- such as retinopathy and nephropathy, as well as sures, and procedures) from clinical encounters macrovascular complications including at participating institutions. Collected data are myocardial infarction, heart failure, stroke, and standardized and normalized by IBM Watson their composite [3, 5]. The American Diabetes Health. The data are automatically updated at Association (ADA) Standards of Medical Care in least once every 24 h. This database offers a Diabetes recommend that most adult patients good means of studying the real-world evidence with T2D maintain a glycated hemoglobin for the effectiveness of different treatment (HbA1c) level at \ 7% [6]. modalities (including BI) in the USA. Over the past three decades, there has been We conducted a retrospective, observational considerable progress in the development of study using the Explorys database to evaluate new antihyperglycemic medications for the probability of achieving glycemic control patients with T2D. Therapeutic options now over 24 months after BI initiation in patients include novel agents such as dipeptidyl pepti- with T2D. dase 4 (DPP4) inhibitors, sodium–glucose cotransporter 2 (SGLT-2) inhibitors, as well as METHODS injectable glucagon-like peptide 1 receptor agonists (GLP-1 RAs). While oral antidiabetics Study Cohort Construction (OADs) are usually the first- or second-line options, many patients with T2D will eventu- ally require therapy with injectables including Within the Explorys dataset (from 2000 to insulin therapy. Basal insulin (BI) has been November 3, 2017), we created a diabetes shown to be effective in reducing HbA1c levels cohort by selecting all patients with at least one and attaining glycemic control, with between encounter with a diagnostic code (primary or 40% and 70% of patients reaching a target of secondary) for T2D (ICD9 codes 250.x0, 250.x2, \ 7% in randomized controlled trials (RCTs) and ICD10 code E11.*), an HbA1c mea- [7, 8]. Initiating insulin replacement with BI has sure C 6.5%, or a prescription for an antidia- been endorsed in professional guidelines for betic medication. We excluded those who had a patients who have not attained target HbA1c diagnosis of type 1 diabetes, gestational dia- levels with the use of non-insulin therapies betes, or polycystic ovarian syndrome. This [9, 10]. resulted in a set of * 4.27 million patients, Studies of the effectiveness of BI in real- among whom * 2.3 million had at least one world practice have shown mixed results. The valid prescription for antidiabetic medication. Diabetes Ther (2018) 9:1347–1358 1349 Among the * 2.3 million patients, 69% had at same regimen. Only outpatient BI prescriptions least one OAD prescription and 24% had at least were included in the analysis. Data were strati- one prescription of BI [neutral protamine fied by the number of concomitant OADs at the Hagedorn (NPH), glargine, detemir, or deglu- time of BI initiation (i.e., BI only, BI ? 1 OAD, dec]. Approximately 85% of BI prescriptions BI ? 2 OADs, and BI ?C 3 OADs), creating four identified for this study were filled after 2011. subcohorts. A flow diagram showing inclusion For this analysis, we were interested in and exclusion of the patients in the study patients who progressed from any OAD regimen cohort is shown in Fig. 1. to BI. To assess treatment regimens we calcu- This article is based on the existing EMR lated prescription length as the time between database and does not contain any studies with the prescription start date and prescription end human participants or animals performed by date. Where a valid prescription end date was any of the authors. given, that date was used as provided. An end date was considered valid if it occurred: Statistical Approach • Between 2000 and the analysis date and • After the prescription start date and Descriptive statistics were calculated to describe • A maximum of 12 months after the prescrip- the baseline demographics and clinical charac- tion date. teristics of the study cohort and subcohorts. For oral prescriptions without end dates, we Means and standard deviations (SD) are repor- used a proxy of twice as many days as the ted for continuous variables and percentages are median population valid prescription duration provided for categorical variables. past the start date. A valid prescription is a Each quarter post-index date was defined to prescription with a start date which is not the be 90 days, i.e., 0–90 days as the first quarter same as the end date (if it exists), and where the (0–3 months); 91–180 days as the second quar- prescription is not flagged as erroneous or can- ter (3–6 months), etc. The descriptive statistics celled. For insulin prescriptions without valid of HbA1c change from baseline were calculated end dates, we inferred the end date to be 1 year semi-annually (i.e., every 180 days) following BI after the prescription start date. initiation. Our analysis cohort included patients who To assess the patients’ response to BI treat- had progressed to BI from one or more OAD(s). ment within 24 months after the index date, we Patients who did not have any encounters defined the target goal of glycemic control as recorded within Explorys for the year prior to achieving an HbA1c \ 7%. We calculated the the first diabetes prescription were excluded. percentage of patients who reached glycemic The date of first prescription of BI was defined as control for the first time among those who had the index date. Patients were included if they not reached glycemic control and were still on had at least one valid HbA1c test recorded (i.e., BI and had any EMR record in the correspond- positive HbA1c value) within 90 days prior to ing periods. and including the index date AND at least one We estimated the probability of achieving valid HbA1c test recorded within 24 months glycemic control after BI initiation in two ways: (720 days) after the index date. If a patient had 1. The conditional probability was estimated multiple valid HbA1c records prior to (and as the proportion of patients who reached including) the index date, the last value was their first glycemic control within a specific used as the baseline. Patient follow-up contin- quarter among those patients who had not ued until the first BI regimen ended or was previously achieved glycemic control, who changed, until they reached 2 years after BI were still taking their BI regimen, and who initiation, or until the database cutoff date for had a valid HbA1c test recorded in that this analysis was reached. Overlapping BI pre- quarter. As this denominator changed at scriptions or BI prescriptions with \ 90 days of each quarter, conditional probability was gap between the end of one prescription and the start of another were collapsed into the 1350 Diabetes Ther (2018) 9:1347–1358 Fig. 1 Patient selection from the database. BI basal insulin, HbA1c glycated hemoglobin, OAD oral antidiabetic, T2D type 2 diabetes calculated quarterly, not cumulatively over 13% African-American), insurance coverage time. (47% private, 36% Medicare, 7% Medicaid) and 2. The cumulative probability of patients most common comorbidities (81% hyperten- reaching first glycemic control over time sion, 70% obesity, 50% dyslipidemia, 24% heart was estimated via Kaplan–Meier curves for disease, and 20% anemia, which may be asso- the whole study cohort as well as the four ciated with diabetic chronic kidney disease). subcohorts. Log-rank tests were done to The most common OADs taken were metformin compare the subcohorts. Censoring (79%), sulfonylureas (63%), DPP4 inhibitors occurred at the end of the BI regimen (30%), and thiazolidinediones (24%). At BI ini- (including switching to a new non-BI regi- tiation, the mean (SD) of HbA1c was 9.1% men), loss of record in the database, or the (2.1%), with 3219 (48.8%) of the 6597 patients cutoff date for the analysis. having an HbA1c [ 9%. Before BI initiation, 3856 (58.5%) of the 6597 patients were on one OAD only, 2032 (30.8%) were on two OADs, RESULTS and 709 (10.7%) were on at least three OADs. At BI initiation, about 20% of patients were pre- Baseline Demographics and Clinical scribed BI alone, while 40%, 31%, and 9% were Characteristics prescribed BI together with one OAD, two OADs, and at least three OADs, respectively Our study cohort included a total of 6597 (Table 2). The four subcohorts appeared to be patients selected from the clinical EMR database similar in terms of age, sex, HbA1c, and body (Fig. 1). Patient demographic and clinical char- mass index at baseline. Diabetes duration could acteristics data at baseline (i.e., last value within not be provided, as the diagnosis of T2D for 90 days prior to and including the index date) some patients predated their entry into the are shown in Table 1. The study cohort was Explorys database. representative of the US population with T2D in terms of age (62 ± 12.7 years), race (75% white, Diabetes Ther (2018) 9:1347–1358 1351 Table 1 Baseline demographics and characteristics of the Table 1 continued study cohort and overall T2D cohorts in the Explorys T2D patients in Study cohort at database the US IBM time of BI T2D patients in Study cohort at Explorys database initiation the US IBM time of BI as of November 3, (n = 6597) Explorys database initiation as of November 3, (n = 6597) (N = 4.27 M) Prescription medication, n (%) (N = 4.27 M) OADs N/A 6597 (100) Female, n (%) 2.27 M (53) 3042 (46) Metformin N/A 5206 (79) Mean 58 ± 15.7 62 ± 12.7 age ± SD, Sulfonylureas N/A 4135 (63) years DPP4 N/A 2006 (30) Median age, 60 62 inhibitors years TZDs N/A 1561 (24) C65 years, 1.65 M (38.7) 3042 (43) SGLT-2 N/A 206 (3) n (%) inhibitors Race, n (%) BI N/A 6597 (100) White 2.6 M (61) 4979 (75) BI basal insulin, DPP4 dipeptidyl peptidase 4, K thousand, African- 574 K (14) 882 (13) M million, OAD oral antidiabetic, SD standard deviation, American SGLT-2 sodium–glucose cotransporter 2, T2D type 2 diabetes, TZD thiazolidinedione Asian 85 K (2) 67 (1) Ethnicity, n (%) HbA1c Change Over Time Hispanic/ 243 K (6) 466 (7) Latino Based on the available HbA1c data in the cohort in each 6-month interval, the change from Unknown 1.1 M (27) 318 (5) baseline in HbA1c was estimated. Mean (SD) Insurance, n (%) HbA1c decreased 1.49 (2.63) percentage points Private 1.6 M (38) 3099 (47) from baseline to 6 months, with no further reductions thereafter (Table 3). As not all Medicare 1.1 M (27) 2377 (36) patients had an HbA1c value in each period, the Self-pay 185 K (5) 192 (3) members of the cohort differed in each period. Medicaid 241 K (6) 425 (7) Patients Achieving Glycemic Control Unknown 1 M (23) 205 (3) (HbA1c < 7%) Comorbidity, n (%) The percentage of patients who registered an Hypertension 2.6 M (62) 5336 (81) HbA1c \ 7% for the first time, out of the num- Dyslipidemia 1.3 M (30) 3308 (50) ber of those patients who were still in the cohort, was calculated for each quarter after the Obesity 2.1 M (50) 4671 (70) index date (Fig. 2, Table 4). During the second Anemia 1 M (24) 1352 (20) quarter after the index date, there were 6086 Heart disease 1 M (23) 1584 (24) patients who had HbA1c C 7% prior to the beginning of the period and were continuing on 1352 Diabetes Ther (2018) 9:1347–1358 Table 2 Baseline characteristics of subcohorts at time of BI initiation BI only BI 1 1 BI 1 2 BI 1 ‡3 Total OAD OADs OADs Patients,n (%) 1319 (20) 2658 (40) 2024 (31) 596 (9) 6597 (100) Female, % 50 48 44 38 46 Mean ± SD age, years 64.5 ± 12.5 61.6 ± 13.1 60.8 ± 12.4 61.3 ± 11.5 61.9 ± 12.7 Mean ± SD HbA1c at BI initiation, % 8.7 ± 2.0 9.3 ± 2.2 9.3 ± 2.0 9.1 ± 2.0 9.1 ± 2.1 Mean ± SD BMI at BI initiation, kg/m 33.2 ± 7.8 33.5 ± 7.6 33.8 ± 7.6 33.8 ± 7.3 33.6 ± 7.6 BI basal insulin, BMI body mass index, HbA1c glycated hemoglobin, OAD oral antidiabetic, SD standard deviation At baseline, 3219 (48.8%) of the 6597 patients had an HbA1c [ 9.0% Table 3 HbA1c change over time in the first 2 years post- BI initiation Duration Number of patients HbA1c change post-BI with HbA1c records from baseline, initiation in this period mean (SD) 0–6 months 5679 - 1.49 (2.63) 6–12 months 3600 - 1.43 (2.69) 12–18 months 861 - 1.44 (2.69) 18–24 months 331 - 1.49 (2.88) Fig. 2 Percentage of patients reaching their first Only those with a valid HbA1c record at baseline and the HbA1c \ 7% post-BI initiation. The numerator is defined corresponding time period were included for the calcula- as the number of patients who reached glycemic control tions. Few patients had multiple HbA1c records across two (HbA1c \ 7%) for the first time during each correspond- or more periods post-BI initiation ing quarter; the denominator is defined as the number of BI basal insulin, HbA1c glycated hemoglobin, SD standard patients who did not reach glycemic control prior to that deviation quarter AND were still on BI treatment AND had at least one valid electronic medical records entry in that quarter. BI treatment, among whom a total of 1311 BI basal insulin, HbA1c glycated hemoglobin (21.5%) patients had reached HbA1c \ 7% for we estimated the conditional probabilities of the first time in that quarter. Similar calcula- reaching the glycemic target of HbA1c \ 7% for tions were done for every subsequent quarter the first time beginning at the second quarter until 24 months. The percentages of patients after BI initiation (Table 4, column F). For those reaching HbA1c \ 7% for the first time in that patients who had not reached HbA1c \ 7% in quarter are shown in Fig. 2 and Table 4, column the first quarter (90 days) following BI initia- E. tion, the probability of reaching first glycemic control in the second quarter was 26.6%. This Estimation of Conditional Probabilities conditional probability decreased to 17.6% during the third quarter and to 8.6% in the In each quarter post-index date, approximately fourth quarter after BI initiation. After 2/3 of the patients had a valid HbA1c mea- 12 months post-BI initiation, it diminished surement (Table 4, columns B and C). From further to B 6.1% during any quarter in the patients who had a valid HbA1c measurement, second year. Diabetes Ther (2018) 9:1347–1358 1353 Table 4 Conditional probability of reaching first glycemic control (HbA1c \ 7%) A: time after B: number of C: number of D: number of E: percentage F: estimated BI initiation patients who had not patients who had not patients who (%) of conditional reached glycemic reached glycemic reached their patients in probability (% of control previously control previously first glycemic column D patients in column AND were still on AND were still on control within among the D among patients BI treatment within BI treatment AND this quarter patients in in column C), % this quarter had at least one valid column B (95% CI) HbA1c record within this quarter 3–6 months 6086 4933 1311 21.5 26.6 (25.4, 27.8) 6–9 months 4102 2767 487 11.9 17.6 (16.2, 19.0) 9–12 months 2423 1668 143 5.9 8.6 (7.3, 9.9) 12–15 months 1679 684 37 2.2 5.4 (3.7, 7.1) 15–18 months 597 361 22 3.7 6.1 (3.6, 8.6) 18–21 months 365 260 14 3.8 5.4 (2.7, 8.1) 21–24 months 216 147 5 2.3 3.4 (0.5, 6.3) BI basal insulin, CI confidence interval, HbA1c glycated hemoglobin Time to Reach First Glycemic Control Analyses The Kaplan–Meier curves of the time to reach HbA1c\ 7% showed similar trends for the overall cohort (Fig. 3, dotted line) as well as across all four subcohorts of concomitant OAD usage (Fig. 3, solid lines). All curves rose gradually over the first 3 months, with a sharper increase over the remainder of the first year, followed by a much more gradual increase over the second year. Fig. 3 Kaplan–Meier curves for time to reach glycemic Overall, approximately 25%, 38%, 42%, and 46% control (HbA1c \ 7%) for the overall study cohort and of eligible patients achieved their first HbA1c\7% the four subcohorts. BI basal insulin, HbA1c glycated by 6, 12, 18, and 24 months after BI initiation, hemoglobin, OAD oral antidiabetic respectively (estimated from the overall cohort Kaplan–Meier curve). Patients in the two sub- BI from OAD(s) had a mean HbA1c of 9.1%, and cohorts treated with BI only or BI ? 1 OAD appeared to have a slightly higher rate of approximately 49% had an HbA1c [ 9.0% at baseline. This underscores that patients may achieving HbA1c \ 7% than those treated with BI ? 2 or at least 3 OADs, but the differences remain on OAD(s) for extended periods, despite being significantly above recommended gly- did not reach statistical significance (p = 0.27). cemic control targets. Further, it indicates that the patients selected for RCTs of BI are generally DISCUSSION in better glycemic control than patients who initiate BI in real life; in fact, the average base- In this real-world study based on a large US EMR line HbA1c was higher in our study than in 36 database, our study cohort of patients initiating of 44 RCTs of BI included in a systematic review 1354 Diabetes Ther (2018) 9:1347–1358 [7]. However, the results presented here are in regimen for 6 or 12 months and has not yet line with previous real-world studies in which reached their glycemic target, the likelihood of HbA1c at initiation of BI was well above the achieving success on the same regimen is low, recommended target, showing that early which should prompt consideration of treat- intensification to the recommended BI treat- ment modification or intensification. While the ment is uncommon in routine clinical care estimated probabilities can be affected by the [11, 17, 18]. Factors underlying such delays are relative robustness of the EMR database and complex [20–22] and may reflect physicians’ need further confirmation, the overall decreas- perceptions of patients’ lifestyles and non-ad- ing trend and the very low probabilities of herence to existing medications; and patients’ reaching HbA1c targets after the first year concerns about out-of-pocket expenses, and highlight the need for paying attention to reluctance to consider BI, as well as fear of patients who have failed to achieve glycemic hypoglycemia and concern about weight gain targets in the first year following BI initiation. [11, 23]. There is often a delay in treatment intensifica- After initiation of BI (with or without con- tion despite persistently elevated glucose levels comitant OADs), there was a robust drop in [13, 14, 24]. In a real-world observational study mean HbA1c of about 1.5% over the first published in 2016, the median time to treat- 6 months, but with no additional change from ment intensification in patients with elevated baseline over subsequent quarters. Few real- HbA1c following BI initiation was 4.3 years [24]. world studies have examined both HbA1c Multiple factors may contribute to such delays change after 3–6 months of BI treatment and [25] including concerns related to some treat- after long-term follow-up [11, 12, 17]. A US ment options beyond BI. Basal-bolus and pre- retrospective analysis of patients with T2D mix insulin regimens are potential options, but found that in 4387 new initiators of BI with an hypoglycemia, weight gain, and the need to average HbA1c of 9.5% at baseline, the HbA1c take multiple daily injections can be major decrease of 1.3% at 3 months was the same as concerns [26–28]. With the recent development the decrease from baseline at 12 months [11]. of medications such as GLP-1 RAs, fixed-ratio Similarly, a retrospective longitudinal analysis combinations of BIs and GLP-1 RAs, and SGLT2 of EMR from five European countries and the inhibitors, which do not increase the risk of USA showed a similar trend, with the majority hypoglycemia or induce weight gain [29–31], of the mean HbA1c decline within the initial prescribers now have more options to consider. 6 months, and no further improvement after The cumulative probability of reaching gly- 12 months [17]. As these data represent a cross- cemic control over time (Fig. 3) revealed that sectional analysis of a cohort, no consideration about 38% of patients reached glycemic control is given to the impact of patients who may in the first 12 months but only about 8% more achieve control initially, and subsequently did so in the second year. These results are also return to an HbA1c C 7% again. in line with other real-world research To better mimic the situation of the health- [11, 17, 18] which has generally found little care practitioner (HCP), we estimated the con- further increase in rates of patients with HbA1c ditional probability of reaching glycemic below the glycemic target with extended treat- control if continuing BI treatment, beginning in ment. The previously mentioned US-based ret- the second quarter post-BI initiation until rospective analysis [11] found an increase in 2 years, given the condition that glycemic con- patients below target for new initiators of BI, trol has not been achieved up to the beginning from 11% at baseline to 27% at 3 months after of the specific quarter. This provides a poten- BI initiation; however, this fell to 25% after tially useful perspective for prescribers who 12 months. The aforementioned retrospective initiate their patients on BI and need to esti- longitudinal EMR analysis from five European mate the likelihood over time that a patient will countries and the USA [17] found that 20.9% achieve glycemic goals. Our study results sug- had an HbA1c B 7% at 3 months after BI initi- gest that if a patient has been using the same BI ation; by 24 months post-initiation this had Diabetes Ther (2018) 9:1347–1358 1355 only increased to 27.8%. An additional US EMR measurements were not consistently performed analysis found that 44% of its cohort achieved every 3 months in many patients, reducing the HbA1c B 7% within 1 year after BI initiation, assessable number of patients included in the with 58% reaching this goal over the entire 2.5- analysis. In addition, it is possible that HbA1c year follow-up [18]. Differences in the patient measurements may have occurred outside of group (39.2% of this cohort had HbA1c [ 9% at the data capture infrastructure, and it cannot be baseline, compared with 48.8% in our study) guaranteed that those patients with missing and the specific glycemic goal may play a role in HbA1c records behave the same as those with the results. available HbA1c results. The paucity of fasting In the current study, the subcohorts of plasma glucose data and of dosing information patients taking no or one concomitant OAD at in the database are other limitations which BI initiation had a slightly higher rate of make it difficult to evaluate titration practice achieving HbA1c \ 7% than those taking two after BI initiation, or any possible relationship or more OADs, which might reflect differences between insulin dose and glycemic response. in stage of disease progression. While the Given the difficulty of achieving glycemic goals Explorys database cannot definitively provide reported in real-world literature, clinicians may disease duration for the majority of patients, it find value in understanding which factors pre- is possible that those patients who continued dict achievement of an HbA1c target. For treatment with two or more OADs while initi- example, assessing durability of glycemic con- ating BI were considered more progressed and trol in those treated with BI, along with treat- therefore had lower residual beta cell function, ment patterns after BI initiation, can add which could impact glycemic response to BI important further information on factors asso- treatment. Corresponding broadly to results ciated with significantly higher rates of reach- seen here, a retrospective database analysis of ing the goal of HbA1c B 7% [18]. We did not do 1830 patients found greater achievement of this analysis in our current study because of glycemic targets among those patients taking limitations of the data, but it is our plan to do so fewer OADs at baseline (38.2%, 26.7%, and in ongoing studies. 19.6% for patients taking one, two, and at least three OADs, respectively; p \ 0.0001) [15]. CONCLUSIONS There could have been other contributing fac- tors; for example, patients on more medications Our study of real-world data from a large US may have more challenges to be compliant with database suggests that, among patients with the prescriptions. T2D who initiated BI after the use of OADs, the Our study has some important limitations. likelihood of newly reaching glycemic control In the USA, patients change insurance coverage diminished over time, regardless of whether BI and/or HCPs from time to time because of was taken alone or together with OADs. The employer decisions, job changes, relocations, ADA or American Association of Clinical personal choices, etc.; the Health Insurance Endocrinologists guidelines recommend some Portability and Accountability Act rules typi- change to treatment if patients do not achieve cally preclude the database from linking multi- control within 3 months. Our data suggest that ple records from different sources using maintaining a BI regimen over 12 or 24 months personal information. The EMR data collected in a patient who has not yet reached the in the Explorys database provides only a snap- HbA1c \ 7% target provides little additional shot of patients’ medical histories, and cannot benefit towards reaching this goal, and that provide the rigor and completeness of data that prescribers should consider additional medica- is typically expected from a prospective longi- tion options if a patient does not reach glycemic tudinal clinical trial. Laboratory variables are control within 12 months of BI initiation. assessed for clinical practice instead of research Information generated from a large real-world purposes, and are from multiple sources, EMR database offers complementary obtained as per local clinical practices. HbA1c 1356 Diabetes Ther (2018) 9:1347–1358 information to that provided by RCTs, with the received consultancy fees from Lilly, Novo potential for additional insights of use for HCPs, Nordisk, and Sanofi; and speaker bureau fees healthcare systems, payers, and industry. from Novo Nordisk and Sanofi. Compliance with Ethics Guidelines. This article is based on an existing EMR database and ACKNOWLEDGEMENTS does not contain any studies with human par- ticipants or animals performed by any of the The authors would like to acknowledge Sharon authors. Hensley Alford for her contributions to the manuscript. Data Availability. These analyses were conducted on medical records data provided Funding. This study and article processing under a commercial licence, which the authors charges for this journal article were sponsored are unable to share. by Sanofi. Open Access. This article is distributed Medical Writing, Editorial, and Other under the terms of the Creative Commons Assistance. Editorial assistance in the prepara- Attribution-NonCommercial 4.0 International tion of this article was provided by Rob Coover License (http://creativecommons.org/licenses/ of Caudex (New York, NY, USA). Support for by-nc/4.0/), which permits any non- this assistance was funded by Sanofi. commercial use, distribution, and reproduction in any medium, provided you give appropriate Authorship. All named authors meet the credit to the original author(s) and the source, International Committee of Medical Journal provide a link to the Creative Commons license, Editors (ICMJE) criteria for authorship for this and indicate if changes were made. manuscript, take responsibility for the integrity of the work as a whole, and have given final approval for the version to be published. All authors had full access to all of the data in this REFERENCES study and take complete responsibility for the integrity of the data and accuracy of the data 1. Centers for Disease Control. National diabetes analysis. statistics report, 2017. [article online], 2018. https:// www.cdc.gov/diabetes/data/statistics-report/index. html. Accessed 22 Jan 2018. Disclosures. Lawrence Blonde has received grant/research support from AstraZeneca, Jans- 2. Dandona P. Minimizing glycemic fluctuations in sen, Lexicon, Merck, Novo Nordisk, and Sanofi; patients with type 2 diabetes: approaches and speaker honoraria from AstraZeneca, Janssen, importance. Diabetes Technol Ther. 2017;19(9):498–506. and Merck; and consultancy fees from AstraZe- neca, GlaxoSmithKline, Intarcia Therapeutics, 3. Paul SK, Klein K, Thorsted BL, Wolden ML, Khunti Janssen, Merck, and Novo Nordisk. Luigi K. Delay in treatment intensification increases the Meneghini has received advisory board/consul- risks of cardiovascular events in patients with type 2 diabetes. Cardiovasc Diabetol. 2015;14:100. tancy fees from Novo Nordisk, Sanofi, and Intarcia. Anders Boss is an employee of Sanofi. 4. Stokes A, Preston SH. Deaths attributable to dia- Alka Shaunik is an employee of Sanofi. Claire betes in the United States: comparison of data Brulle-Wohlhueter is an employee of Sanofi. sources and estimation approaches. PLoS One. 2017;12(1):e0170219. Xuejun Victor Peng is an employee of Sanofi. Kyu Rhee is an employee of Watson Health, 5. Stratton IM, Adler AI, Neil HA, et al. Association of IBM. Supriya Kumar is an employee of Watson glycaemia with macrovascular and microvascular Health, IBM. Sidhartha Balodi is an employee of complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. Watson Health, IBM. Rory J McCrimmon has 2000;321(7258):405–12. Diabetes Ther (2018) 9:1347–1358 1357 6. American Diabetes Association. 6. Glycemic targets: 16. Lin SD, Tsai ST, Tu ST, et al. Glycosylated hemo- standards of medical care in diabetes—2018. Dia- globin level and number of oral antidiabetic drugs betes Care. 2018;41(Suppl 1):S55–64. predict whether or not glycemic target is achieved in insulin-requiring type 2 diabetes. Prim Care 7. Esposito K, Chiodini P, Bellastella G, Maiorino MI, Diabetes. 2015;9(2):135–41. Giugliano D. Proportion of patients at HbA1c tar- get \ 7% with eight classes of antidiabetic drugs in 17. Mauricio D, Meneghini L, Seufert J, et al. Glycaemic type 2 diabetes: systematic review of 218 random- control and hypoglycaemia burden in patients with ized controlled trials with 78 945 patients. Diabetes type 2 diabetes initiating basal insulin in Europe Obes Metab. 2012;14(3):228–33. and the USA. Diabetes Obes Metab. 2017;19(19):1155–64. 8. Hermansen K, Davies M, Derezinski T, Martinez RG, Clauson P, Home P. A 26-week, randomized, par- 18. Wu N, Aagren M, Boulanger L, Friedman M, Wilkey allel, treat-to-target trial comparing insulin detemir K. Assessing achievement and maintenance of gly- with NPH insulin as add-on therapy to oral glucose- cemic control by patients initiating basal insulin. lowering drugs in insulin-naive people with type 2 Curr Med Res Opin. 2012;28(10):1647–56. diabetes. Diabetes Care. 2006;29(6):1269–74. 19. Kostev K, Dippel FW, Rathmann W. Glycemic 9. Garber AJ, Abrahamson MJ, Barzilay JI, et al. Con- control after initiating basal insulin therapy in sensus statement by the American Association of patients with type 2 diabetes: a primary care data- Clinical Endocrinologists and American College of base analysis. Diabetes Metab Syndr Obes. Endocrinology on the comprehensive type 2 dia- 2015;8:45–8. betes management algorithm—2016 executive summary. Endocr Pract. 2016;22(1):84–113. 20. Kim SG, Kim NH, Ku BJ, et al. Delay of insulin ini- tiation in patients with type 2 diabetes mellitus 10. Inzucchi SE, Bergenstal RM, Buse JB, et al. Man- inadequately controlled with oral hypoglycemic agement of hyperglycemia in type 2 diabetes: a agents (analysis of patient- and physician-related patient-centered approach: position statement of factors): a prospective observational DIPP-FACTOR the American Diabetes Association (ADA) and the study in Korea. J Diabetes Investig. European Association for the Study of Diabetes 2017;8(3):346–53. (EASD). Diabetes Care. 2012;35(6):1364–79. 21. Peyrot M, Rubin RR, Lauritzen T, et al. Resistance to 11. Dalal MR, Grabner M, Bonine N, Stephenson JJ, insulin therapy among patients and providers: DiGenio A, Bieszk N. Are patients on basal insulin results of the cross-national Diabetes Attitudes, attaining glycemic targets? Characteristics and goal Wishes, and Needs (DAWN) study. Diabetes Care. achievement of patients with type 2 diabetes mel- 2005;28(11):2673–9. litus treated with basal insulin and physician-per- ceived barriers to achieving glycemic targets. 22. Strain WD, Cos X, Hirst M, et al. Time to do more: Diabetes Res Clin Pract. 2016;121:17–26. addressing clinical inertia in the management of type 2 diabetes mellitus. Diabetes Res Clin Pract. 12. Gordon J, Pockett RD, Tetlow AP, McEwan P, Home 2014;105(3):302–12. PD. A comparison of intermediate and long-acting insulins in people with type 2 diabetes starting 23. Khunti K, Millar-Jones D. Clinical inertia to insulin insulin: an observational database study. Int J Clin initiation and intensification in the UK: a focused Pract. 2010;64(12):1609–18. literature review. Prim Care Diabetes. 2017;11(1):3–12. 13. Harris SB, Kapor J, Lank CN, Willan AR, Houston T. Clinical inertia in patients with T2DM requiring 24. Khunti K, Nikolajsen A, Thorsted BL, Andersen M, insulin in family practice. Can Fam Physician. Davies MJ, Paul SK. Clinical inertia with regard to 2010;56(12):e418–24. intensifying therapy in people with type 2 diabetes treated with basal insulin. Diabetes Obes Metab. 14. Heintjes EM, Thomsen TL, Penning-van Beest FJ, 2016;18(4):401–9. Christensen TE, Herings RM. Glycemic control and long-acting insulin analog utilization in patients 25. Berard L, Bonnemaire M, Mical M, Edelman S. with type 2 diabetes. Adv Ther. 2010;27(4):211–22. Insights into optimal basal insulin titration in type 2 diabetes: results of a quantitative survey. Diabetes 15. Levin PA, Zhou S, Durden E, Farr AM, Gill J, Wei W. Obes Metab. 2018;20(2):301–8. Clinical and economic outcomes associated with the timing of initiation of basal insulin in patients 26. McFarlane SI. Insulin therapy and type 2 diabetes: with type 2 diabetes mellitus previously treated management of weight gain. J Clin Hypertens with oral antidiabetes drugs. Clin Ther. (Greenwich). 2009;11(10):601–7. 2016;38(1):110–21. 1358 Diabetes Ther (2018) 9:1347–1358 27. Polonsky WH, Thompson S, Wei W, et al. Greater glargine/lixisenatide treatment in type 2 diabetes: a fear of hypoglycaemia with premixed insulin than systematic review and meta-analysis. Expert Opin with basal-bolus insulin glargine and glulisine: Pharmacother. 2017;18(17):1789–98. patient-reported outcomes from a 60-week ran- domised study. Diabetes Obes Metab. 30. Steen O, Goldenberg RM. The role of sodium-glu- 2014;16(11):1121–7. cose cotransporter 2 inhibitors in the management of type 2 diabetes. Can J Diabetes. 28. Vijan S, Hayward RA, Ronis DL, Hofer TP. Brief 2017;41(5):517–23. report: the burden of diabetes therapy: implications for the design of effective patient-centered treat- 31. Valentine V, Goldman J, Shubrook JH. Rationale ment regimens. J Gen Intern Med. for, initiation and titration of the basal insulin/ 2005;20(5):479–82. GLP-1RA fixed-ratio combination products, IDe- gLira and iGlarLixi, for the management of type 2 29. Cai X, Gao X, Yang W, Ji L. Comparison between diabetes. Diabetes Ther. 2017;8(4):739–52. insulin degludec/liraglutide treatment and insulin http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Diabetes Therapy Springer Journals

Probability of Achieving Glycemic Control with Basal Insulin in Patients with Type 2 Diabetes in Real-World Practice in the USA

Free
12 pages
Loading next page...
 
/lp/springer_journal/probability-of-achieving-glycemic-control-with-basal-insulin-in-sS0BKiEa3j
Publisher
Springer Healthcare
Copyright
Copyright © 2018 by The Author(s)
Subject
Medicine & Public Health; Internal Medicine; Diabetes; Cardiology; Endocrinology
ISSN
1869-6953
eISSN
1869-6961
D.O.I.
10.1007/s13300-018-0413-5
Publisher site
See Article on Publisher Site

Abstract

Diabetes Ther (2018) 9:1347–1358 https://doi.org/10.1007/s13300-018-0413-5 BRIEF REPORT Probability of Achieving Glycemic Control with Basal Insulin in Patients with Type 2 Diabetes in Real-World Practice in the USA . . . Lawrence Blonde Luigi Meneghini Xuejun Victor Peng . . . . Anders Boss Kyu Rhee Alka Shaunik Supriya Kumar . . Sidhartha Balodi Claire Brulle-Wohlhueter Rory J. McCrimmon Received: March 2, 2018 / Published online: March 29, 2018 The Author(s) 2018 least one valid glycated hemoglobin (HbA1c) ABSTRACT result recorded both within 90 days before and 720 days after BI initiation were selected. We Introduction: Basal insulin (BI) plays an estimated the changes from baseline in HbA1c important role in treating type 2 diabetes (T2D), every 6 months, the quarterly conditional especially when oral antidiabetic (OAD) medi- probabilities of reaching HbA1c \ 7% if a cations are insufficient for glycemic control. We patient had not achieved glycemic control prior conducted a retrospective, observational study to each quarter (Q), and the cumulative proba- using electronic medical records (EMR) data bility of reaching glycemic control over from the IBM Explorys database to evaluate 24 months. the probability of achieving glycemic control Results: Our cohort was representative of over 24 months after BI initiation in patients patients with T2D who initiated BI from OADs with T2D in the USA. in the USA. The average HbA1c was 9.1% at BI Methods: A cohort of 6597 patients with T2D initiation, and decreased robustly (1.5%) in the who started BI following OAD(s) and had at first 6 months after initiation with no further reductions thereafter. The conditional proba- Enhanced content To view enhanced content for this bility of reaching glycemic control decreased article go to https://doi.org/10.6084/m9.figshare. rapidly in the first year (26.6% in Q2; 17.6% in Q3; 8.6% in Q4), and then remained low (B 6.1%) for each quarter in the second year. L. Blonde (&) Ochsner Medical Center, New Orleans, LA, USA Cumulatively, about 38% of patients reached e-mail: lawrence.blonde@gmail.com HbA1c \ 7% in the first year; only approxi- mately 8% more did so in the second year. L. Meneghini University of Texas, Southwestern Medical Center Conclusion: Our study of real-world data from and Parkland Health & Hospital System, Dallas, TX, a large US EMR database suggested that among USA patients with T2D who initiated BI after OADs, the likelihood of reaching glycemic control X. V. Peng  A. Boss  A. Shaunik C. Brulle-Wohlhueter diminished over time, and remained low from Sanofi, Bridgewater, NJ, USA 12 months onwards. Additional treatment options should be considered if patients do not K. Rhee  S. Kumar  S. Balodi Watson Health, IBM, Cambridge, MA, USA reach glycemic control within 12 months of BI initiation. R. J. McCrimmon Funding: Sanofi Corporation. University of Dundee, Dundee, UK 1348 Diabetes Ther (2018) 9:1347–1358 Keywords: Basal insulin; IBM Explorys database; proportions of patients achieving target HbA1c Real-world evidence; Type 2 diabetes levels (HbA1c \ 7%) in these studies have ran- ged from as low as 11% after 1 year to as high as 58% over 2.5 years [11–19]. The likelihood of INTRODUCTION reaching glycemic control with BI treatment in the real world remains a question of interest to Type 2 diabetes (T2D) affects 30.3 million peo- patients, clinicians, healthcare systems, and ple (* 9.4% of the population) in the USA as of payers. TM 2017 [1]. The economic and social costs of T2D The IBM Watson Health Explorys data- are considerable because of its associated serious base (later referred to as Explorys) is a large, US short- and long-term complications, particu- population-based, commercial database that larly in patients who do not achieve and/or contains an aggregate of electronic medical maintain glycemic control [2, 3], as well as its record (EMR) data from over 54 million unique significant contribution to overall mortality patients from 39 major integrated healthcare (approximately 12% of all deaths) [4]. Hyper- systems covering over 344,000 clinicians across glycemia, which may result from insufficient all 50 US states. It contains de-identified, treatment intensification, significantly increa- structured, longitudinal patient data (including ses the risks of microvascular complications, diagnoses, laboratory results, biometric mea- such as retinopathy and nephropathy, as well as sures, and procedures) from clinical encounters macrovascular complications including at participating institutions. Collected data are myocardial infarction, heart failure, stroke, and standardized and normalized by IBM Watson their composite [3, 5]. The American Diabetes Health. The data are automatically updated at Association (ADA) Standards of Medical Care in least once every 24 h. This database offers a Diabetes recommend that most adult patients good means of studying the real-world evidence with T2D maintain a glycated hemoglobin for the effectiveness of different treatment (HbA1c) level at \ 7% [6]. modalities (including BI) in the USA. Over the past three decades, there has been We conducted a retrospective, observational considerable progress in the development of study using the Explorys database to evaluate new antihyperglycemic medications for the probability of achieving glycemic control patients with T2D. Therapeutic options now over 24 months after BI initiation in patients include novel agents such as dipeptidyl pepti- with T2D. dase 4 (DPP4) inhibitors, sodium–glucose cotransporter 2 (SGLT-2) inhibitors, as well as METHODS injectable glucagon-like peptide 1 receptor agonists (GLP-1 RAs). While oral antidiabetics Study Cohort Construction (OADs) are usually the first- or second-line options, many patients with T2D will eventu- ally require therapy with injectables including Within the Explorys dataset (from 2000 to insulin therapy. Basal insulin (BI) has been November 3, 2017), we created a diabetes shown to be effective in reducing HbA1c levels cohort by selecting all patients with at least one and attaining glycemic control, with between encounter with a diagnostic code (primary or 40% and 70% of patients reaching a target of secondary) for T2D (ICD9 codes 250.x0, 250.x2, \ 7% in randomized controlled trials (RCTs) and ICD10 code E11.*), an HbA1c mea- [7, 8]. Initiating insulin replacement with BI has sure C 6.5%, or a prescription for an antidia- been endorsed in professional guidelines for betic medication. We excluded those who had a patients who have not attained target HbA1c diagnosis of type 1 diabetes, gestational dia- levels with the use of non-insulin therapies betes, or polycystic ovarian syndrome. This [9, 10]. resulted in a set of * 4.27 million patients, Studies of the effectiveness of BI in real- among whom * 2.3 million had at least one world practice have shown mixed results. The valid prescription for antidiabetic medication. Diabetes Ther (2018) 9:1347–1358 1349 Among the * 2.3 million patients, 69% had at same regimen. Only outpatient BI prescriptions least one OAD prescription and 24% had at least were included in the analysis. Data were strati- one prescription of BI [neutral protamine fied by the number of concomitant OADs at the Hagedorn (NPH), glargine, detemir, or deglu- time of BI initiation (i.e., BI only, BI ? 1 OAD, dec]. Approximately 85% of BI prescriptions BI ? 2 OADs, and BI ?C 3 OADs), creating four identified for this study were filled after 2011. subcohorts. A flow diagram showing inclusion For this analysis, we were interested in and exclusion of the patients in the study patients who progressed from any OAD regimen cohort is shown in Fig. 1. to BI. To assess treatment regimens we calcu- This article is based on the existing EMR lated prescription length as the time between database and does not contain any studies with the prescription start date and prescription end human participants or animals performed by date. Where a valid prescription end date was any of the authors. given, that date was used as provided. An end date was considered valid if it occurred: Statistical Approach • Between 2000 and the analysis date and • After the prescription start date and Descriptive statistics were calculated to describe • A maximum of 12 months after the prescrip- the baseline demographics and clinical charac- tion date. teristics of the study cohort and subcohorts. For oral prescriptions without end dates, we Means and standard deviations (SD) are repor- used a proxy of twice as many days as the ted for continuous variables and percentages are median population valid prescription duration provided for categorical variables. past the start date. A valid prescription is a Each quarter post-index date was defined to prescription with a start date which is not the be 90 days, i.e., 0–90 days as the first quarter same as the end date (if it exists), and where the (0–3 months); 91–180 days as the second quar- prescription is not flagged as erroneous or can- ter (3–6 months), etc. The descriptive statistics celled. For insulin prescriptions without valid of HbA1c change from baseline were calculated end dates, we inferred the end date to be 1 year semi-annually (i.e., every 180 days) following BI after the prescription start date. initiation. Our analysis cohort included patients who To assess the patients’ response to BI treat- had progressed to BI from one or more OAD(s). ment within 24 months after the index date, we Patients who did not have any encounters defined the target goal of glycemic control as recorded within Explorys for the year prior to achieving an HbA1c \ 7%. We calculated the the first diabetes prescription were excluded. percentage of patients who reached glycemic The date of first prescription of BI was defined as control for the first time among those who had the index date. Patients were included if they not reached glycemic control and were still on had at least one valid HbA1c test recorded (i.e., BI and had any EMR record in the correspond- positive HbA1c value) within 90 days prior to ing periods. and including the index date AND at least one We estimated the probability of achieving valid HbA1c test recorded within 24 months glycemic control after BI initiation in two ways: (720 days) after the index date. If a patient had 1. The conditional probability was estimated multiple valid HbA1c records prior to (and as the proportion of patients who reached including) the index date, the last value was their first glycemic control within a specific used as the baseline. Patient follow-up contin- quarter among those patients who had not ued until the first BI regimen ended or was previously achieved glycemic control, who changed, until they reached 2 years after BI were still taking their BI regimen, and who initiation, or until the database cutoff date for had a valid HbA1c test recorded in that this analysis was reached. Overlapping BI pre- quarter. As this denominator changed at scriptions or BI prescriptions with \ 90 days of each quarter, conditional probability was gap between the end of one prescription and the start of another were collapsed into the 1350 Diabetes Ther (2018) 9:1347–1358 Fig. 1 Patient selection from the database. BI basal insulin, HbA1c glycated hemoglobin, OAD oral antidiabetic, T2D type 2 diabetes calculated quarterly, not cumulatively over 13% African-American), insurance coverage time. (47% private, 36% Medicare, 7% Medicaid) and 2. The cumulative probability of patients most common comorbidities (81% hyperten- reaching first glycemic control over time sion, 70% obesity, 50% dyslipidemia, 24% heart was estimated via Kaplan–Meier curves for disease, and 20% anemia, which may be asso- the whole study cohort as well as the four ciated with diabetic chronic kidney disease). subcohorts. Log-rank tests were done to The most common OADs taken were metformin compare the subcohorts. Censoring (79%), sulfonylureas (63%), DPP4 inhibitors occurred at the end of the BI regimen (30%), and thiazolidinediones (24%). At BI ini- (including switching to a new non-BI regi- tiation, the mean (SD) of HbA1c was 9.1% men), loss of record in the database, or the (2.1%), with 3219 (48.8%) of the 6597 patients cutoff date for the analysis. having an HbA1c [ 9%. Before BI initiation, 3856 (58.5%) of the 6597 patients were on one OAD only, 2032 (30.8%) were on two OADs, RESULTS and 709 (10.7%) were on at least three OADs. At BI initiation, about 20% of patients were pre- Baseline Demographics and Clinical scribed BI alone, while 40%, 31%, and 9% were Characteristics prescribed BI together with one OAD, two OADs, and at least three OADs, respectively Our study cohort included a total of 6597 (Table 2). The four subcohorts appeared to be patients selected from the clinical EMR database similar in terms of age, sex, HbA1c, and body (Fig. 1). Patient demographic and clinical char- mass index at baseline. Diabetes duration could acteristics data at baseline (i.e., last value within not be provided, as the diagnosis of T2D for 90 days prior to and including the index date) some patients predated their entry into the are shown in Table 1. The study cohort was Explorys database. representative of the US population with T2D in terms of age (62 ± 12.7 years), race (75% white, Diabetes Ther (2018) 9:1347–1358 1351 Table 1 Baseline demographics and characteristics of the Table 1 continued study cohort and overall T2D cohorts in the Explorys T2D patients in Study cohort at database the US IBM time of BI T2D patients in Study cohort at Explorys database initiation the US IBM time of BI as of November 3, (n = 6597) Explorys database initiation as of November 3, (n = 6597) (N = 4.27 M) Prescription medication, n (%) (N = 4.27 M) OADs N/A 6597 (100) Female, n (%) 2.27 M (53) 3042 (46) Metformin N/A 5206 (79) Mean 58 ± 15.7 62 ± 12.7 age ± SD, Sulfonylureas N/A 4135 (63) years DPP4 N/A 2006 (30) Median age, 60 62 inhibitors years TZDs N/A 1561 (24) C65 years, 1.65 M (38.7) 3042 (43) SGLT-2 N/A 206 (3) n (%) inhibitors Race, n (%) BI N/A 6597 (100) White 2.6 M (61) 4979 (75) BI basal insulin, DPP4 dipeptidyl peptidase 4, K thousand, African- 574 K (14) 882 (13) M million, OAD oral antidiabetic, SD standard deviation, American SGLT-2 sodium–glucose cotransporter 2, T2D type 2 diabetes, TZD thiazolidinedione Asian 85 K (2) 67 (1) Ethnicity, n (%) HbA1c Change Over Time Hispanic/ 243 K (6) 466 (7) Latino Based on the available HbA1c data in the cohort in each 6-month interval, the change from Unknown 1.1 M (27) 318 (5) baseline in HbA1c was estimated. Mean (SD) Insurance, n (%) HbA1c decreased 1.49 (2.63) percentage points Private 1.6 M (38) 3099 (47) from baseline to 6 months, with no further reductions thereafter (Table 3). As not all Medicare 1.1 M (27) 2377 (36) patients had an HbA1c value in each period, the Self-pay 185 K (5) 192 (3) members of the cohort differed in each period. Medicaid 241 K (6) 425 (7) Patients Achieving Glycemic Control Unknown 1 M (23) 205 (3) (HbA1c < 7%) Comorbidity, n (%) The percentage of patients who registered an Hypertension 2.6 M (62) 5336 (81) HbA1c \ 7% for the first time, out of the num- Dyslipidemia 1.3 M (30) 3308 (50) ber of those patients who were still in the cohort, was calculated for each quarter after the Obesity 2.1 M (50) 4671 (70) index date (Fig. 2, Table 4). During the second Anemia 1 M (24) 1352 (20) quarter after the index date, there were 6086 Heart disease 1 M (23) 1584 (24) patients who had HbA1c C 7% prior to the beginning of the period and were continuing on 1352 Diabetes Ther (2018) 9:1347–1358 Table 2 Baseline characteristics of subcohorts at time of BI initiation BI only BI 1 1 BI 1 2 BI 1 ‡3 Total OAD OADs OADs Patients,n (%) 1319 (20) 2658 (40) 2024 (31) 596 (9) 6597 (100) Female, % 50 48 44 38 46 Mean ± SD age, years 64.5 ± 12.5 61.6 ± 13.1 60.8 ± 12.4 61.3 ± 11.5 61.9 ± 12.7 Mean ± SD HbA1c at BI initiation, % 8.7 ± 2.0 9.3 ± 2.2 9.3 ± 2.0 9.1 ± 2.0 9.1 ± 2.1 Mean ± SD BMI at BI initiation, kg/m 33.2 ± 7.8 33.5 ± 7.6 33.8 ± 7.6 33.8 ± 7.3 33.6 ± 7.6 BI basal insulin, BMI body mass index, HbA1c glycated hemoglobin, OAD oral antidiabetic, SD standard deviation At baseline, 3219 (48.8%) of the 6597 patients had an HbA1c [ 9.0% Table 3 HbA1c change over time in the first 2 years post- BI initiation Duration Number of patients HbA1c change post-BI with HbA1c records from baseline, initiation in this period mean (SD) 0–6 months 5679 - 1.49 (2.63) 6–12 months 3600 - 1.43 (2.69) 12–18 months 861 - 1.44 (2.69) 18–24 months 331 - 1.49 (2.88) Fig. 2 Percentage of patients reaching their first Only those with a valid HbA1c record at baseline and the HbA1c \ 7% post-BI initiation. The numerator is defined corresponding time period were included for the calcula- as the number of patients who reached glycemic control tions. Few patients had multiple HbA1c records across two (HbA1c \ 7%) for the first time during each correspond- or more periods post-BI initiation ing quarter; the denominator is defined as the number of BI basal insulin, HbA1c glycated hemoglobin, SD standard patients who did not reach glycemic control prior to that deviation quarter AND were still on BI treatment AND had at least one valid electronic medical records entry in that quarter. BI treatment, among whom a total of 1311 BI basal insulin, HbA1c glycated hemoglobin (21.5%) patients had reached HbA1c \ 7% for we estimated the conditional probabilities of the first time in that quarter. Similar calcula- reaching the glycemic target of HbA1c \ 7% for tions were done for every subsequent quarter the first time beginning at the second quarter until 24 months. The percentages of patients after BI initiation (Table 4, column F). For those reaching HbA1c \ 7% for the first time in that patients who had not reached HbA1c \ 7% in quarter are shown in Fig. 2 and Table 4, column the first quarter (90 days) following BI initia- E. tion, the probability of reaching first glycemic control in the second quarter was 26.6%. This Estimation of Conditional Probabilities conditional probability decreased to 17.6% during the third quarter and to 8.6% in the In each quarter post-index date, approximately fourth quarter after BI initiation. After 2/3 of the patients had a valid HbA1c mea- 12 months post-BI initiation, it diminished surement (Table 4, columns B and C). From further to B 6.1% during any quarter in the patients who had a valid HbA1c measurement, second year. Diabetes Ther (2018) 9:1347–1358 1353 Table 4 Conditional probability of reaching first glycemic control (HbA1c \ 7%) A: time after B: number of C: number of D: number of E: percentage F: estimated BI initiation patients who had not patients who had not patients who (%) of conditional reached glycemic reached glycemic reached their patients in probability (% of control previously control previously first glycemic column D patients in column AND were still on AND were still on control within among the D among patients BI treatment within BI treatment AND this quarter patients in in column C), % this quarter had at least one valid column B (95% CI) HbA1c record within this quarter 3–6 months 6086 4933 1311 21.5 26.6 (25.4, 27.8) 6–9 months 4102 2767 487 11.9 17.6 (16.2, 19.0) 9–12 months 2423 1668 143 5.9 8.6 (7.3, 9.9) 12–15 months 1679 684 37 2.2 5.4 (3.7, 7.1) 15–18 months 597 361 22 3.7 6.1 (3.6, 8.6) 18–21 months 365 260 14 3.8 5.4 (2.7, 8.1) 21–24 months 216 147 5 2.3 3.4 (0.5, 6.3) BI basal insulin, CI confidence interval, HbA1c glycated hemoglobin Time to Reach First Glycemic Control Analyses The Kaplan–Meier curves of the time to reach HbA1c\ 7% showed similar trends for the overall cohort (Fig. 3, dotted line) as well as across all four subcohorts of concomitant OAD usage (Fig. 3, solid lines). All curves rose gradually over the first 3 months, with a sharper increase over the remainder of the first year, followed by a much more gradual increase over the second year. Fig. 3 Kaplan–Meier curves for time to reach glycemic Overall, approximately 25%, 38%, 42%, and 46% control (HbA1c \ 7%) for the overall study cohort and of eligible patients achieved their first HbA1c\7% the four subcohorts. BI basal insulin, HbA1c glycated by 6, 12, 18, and 24 months after BI initiation, hemoglobin, OAD oral antidiabetic respectively (estimated from the overall cohort Kaplan–Meier curve). Patients in the two sub- BI from OAD(s) had a mean HbA1c of 9.1%, and cohorts treated with BI only or BI ? 1 OAD appeared to have a slightly higher rate of approximately 49% had an HbA1c [ 9.0% at baseline. This underscores that patients may achieving HbA1c \ 7% than those treated with BI ? 2 or at least 3 OADs, but the differences remain on OAD(s) for extended periods, despite being significantly above recommended gly- did not reach statistical significance (p = 0.27). cemic control targets. Further, it indicates that the patients selected for RCTs of BI are generally DISCUSSION in better glycemic control than patients who initiate BI in real life; in fact, the average base- In this real-world study based on a large US EMR line HbA1c was higher in our study than in 36 database, our study cohort of patients initiating of 44 RCTs of BI included in a systematic review 1354 Diabetes Ther (2018) 9:1347–1358 [7]. However, the results presented here are in regimen for 6 or 12 months and has not yet line with previous real-world studies in which reached their glycemic target, the likelihood of HbA1c at initiation of BI was well above the achieving success on the same regimen is low, recommended target, showing that early which should prompt consideration of treat- intensification to the recommended BI treat- ment modification or intensification. While the ment is uncommon in routine clinical care estimated probabilities can be affected by the [11, 17, 18]. Factors underlying such delays are relative robustness of the EMR database and complex [20–22] and may reflect physicians’ need further confirmation, the overall decreas- perceptions of patients’ lifestyles and non-ad- ing trend and the very low probabilities of herence to existing medications; and patients’ reaching HbA1c targets after the first year concerns about out-of-pocket expenses, and highlight the need for paying attention to reluctance to consider BI, as well as fear of patients who have failed to achieve glycemic hypoglycemia and concern about weight gain targets in the first year following BI initiation. [11, 23]. There is often a delay in treatment intensifica- After initiation of BI (with or without con- tion despite persistently elevated glucose levels comitant OADs), there was a robust drop in [13, 14, 24]. In a real-world observational study mean HbA1c of about 1.5% over the first published in 2016, the median time to treat- 6 months, but with no additional change from ment intensification in patients with elevated baseline over subsequent quarters. Few real- HbA1c following BI initiation was 4.3 years [24]. world studies have examined both HbA1c Multiple factors may contribute to such delays change after 3–6 months of BI treatment and [25] including concerns related to some treat- after long-term follow-up [11, 12, 17]. A US ment options beyond BI. Basal-bolus and pre- retrospective analysis of patients with T2D mix insulin regimens are potential options, but found that in 4387 new initiators of BI with an hypoglycemia, weight gain, and the need to average HbA1c of 9.5% at baseline, the HbA1c take multiple daily injections can be major decrease of 1.3% at 3 months was the same as concerns [26–28]. With the recent development the decrease from baseline at 12 months [11]. of medications such as GLP-1 RAs, fixed-ratio Similarly, a retrospective longitudinal analysis combinations of BIs and GLP-1 RAs, and SGLT2 of EMR from five European countries and the inhibitors, which do not increase the risk of USA showed a similar trend, with the majority hypoglycemia or induce weight gain [29–31], of the mean HbA1c decline within the initial prescribers now have more options to consider. 6 months, and no further improvement after The cumulative probability of reaching gly- 12 months [17]. As these data represent a cross- cemic control over time (Fig. 3) revealed that sectional analysis of a cohort, no consideration about 38% of patients reached glycemic control is given to the impact of patients who may in the first 12 months but only about 8% more achieve control initially, and subsequently did so in the second year. These results are also return to an HbA1c C 7% again. in line with other real-world research To better mimic the situation of the health- [11, 17, 18] which has generally found little care practitioner (HCP), we estimated the con- further increase in rates of patients with HbA1c ditional probability of reaching glycemic below the glycemic target with extended treat- control if continuing BI treatment, beginning in ment. The previously mentioned US-based ret- the second quarter post-BI initiation until rospective analysis [11] found an increase in 2 years, given the condition that glycemic con- patients below target for new initiators of BI, trol has not been achieved up to the beginning from 11% at baseline to 27% at 3 months after of the specific quarter. This provides a poten- BI initiation; however, this fell to 25% after tially useful perspective for prescribers who 12 months. The aforementioned retrospective initiate their patients on BI and need to esti- longitudinal EMR analysis from five European mate the likelihood over time that a patient will countries and the USA [17] found that 20.9% achieve glycemic goals. Our study results sug- had an HbA1c B 7% at 3 months after BI initi- gest that if a patient has been using the same BI ation; by 24 months post-initiation this had Diabetes Ther (2018) 9:1347–1358 1355 only increased to 27.8%. An additional US EMR measurements were not consistently performed analysis found that 44% of its cohort achieved every 3 months in many patients, reducing the HbA1c B 7% within 1 year after BI initiation, assessable number of patients included in the with 58% reaching this goal over the entire 2.5- analysis. In addition, it is possible that HbA1c year follow-up [18]. Differences in the patient measurements may have occurred outside of group (39.2% of this cohort had HbA1c [ 9% at the data capture infrastructure, and it cannot be baseline, compared with 48.8% in our study) guaranteed that those patients with missing and the specific glycemic goal may play a role in HbA1c records behave the same as those with the results. available HbA1c results. The paucity of fasting In the current study, the subcohorts of plasma glucose data and of dosing information patients taking no or one concomitant OAD at in the database are other limitations which BI initiation had a slightly higher rate of make it difficult to evaluate titration practice achieving HbA1c \ 7% than those taking two after BI initiation, or any possible relationship or more OADs, which might reflect differences between insulin dose and glycemic response. in stage of disease progression. While the Given the difficulty of achieving glycemic goals Explorys database cannot definitively provide reported in real-world literature, clinicians may disease duration for the majority of patients, it find value in understanding which factors pre- is possible that those patients who continued dict achievement of an HbA1c target. For treatment with two or more OADs while initi- example, assessing durability of glycemic con- ating BI were considered more progressed and trol in those treated with BI, along with treat- therefore had lower residual beta cell function, ment patterns after BI initiation, can add which could impact glycemic response to BI important further information on factors asso- treatment. Corresponding broadly to results ciated with significantly higher rates of reach- seen here, a retrospective database analysis of ing the goal of HbA1c B 7% [18]. We did not do 1830 patients found greater achievement of this analysis in our current study because of glycemic targets among those patients taking limitations of the data, but it is our plan to do so fewer OADs at baseline (38.2%, 26.7%, and in ongoing studies. 19.6% for patients taking one, two, and at least three OADs, respectively; p \ 0.0001) [15]. CONCLUSIONS There could have been other contributing fac- tors; for example, patients on more medications Our study of real-world data from a large US may have more challenges to be compliant with database suggests that, among patients with the prescriptions. T2D who initiated BI after the use of OADs, the Our study has some important limitations. likelihood of newly reaching glycemic control In the USA, patients change insurance coverage diminished over time, regardless of whether BI and/or HCPs from time to time because of was taken alone or together with OADs. The employer decisions, job changes, relocations, ADA or American Association of Clinical personal choices, etc.; the Health Insurance Endocrinologists guidelines recommend some Portability and Accountability Act rules typi- change to treatment if patients do not achieve cally preclude the database from linking multi- control within 3 months. Our data suggest that ple records from different sources using maintaining a BI regimen over 12 or 24 months personal information. The EMR data collected in a patient who has not yet reached the in the Explorys database provides only a snap- HbA1c \ 7% target provides little additional shot of patients’ medical histories, and cannot benefit towards reaching this goal, and that provide the rigor and completeness of data that prescribers should consider additional medica- is typically expected from a prospective longi- tion options if a patient does not reach glycemic tudinal clinical trial. Laboratory variables are control within 12 months of BI initiation. assessed for clinical practice instead of research Information generated from a large real-world purposes, and are from multiple sources, EMR database offers complementary obtained as per local clinical practices. HbA1c 1356 Diabetes Ther (2018) 9:1347–1358 information to that provided by RCTs, with the received consultancy fees from Lilly, Novo potential for additional insights of use for HCPs, Nordisk, and Sanofi; and speaker bureau fees healthcare systems, payers, and industry. from Novo Nordisk and Sanofi. Compliance with Ethics Guidelines. This article is based on an existing EMR database and ACKNOWLEDGEMENTS does not contain any studies with human par- ticipants or animals performed by any of the The authors would like to acknowledge Sharon authors. Hensley Alford for her contributions to the manuscript. Data Availability. These analyses were conducted on medical records data provided Funding. This study and article processing under a commercial licence, which the authors charges for this journal article were sponsored are unable to share. by Sanofi. Open Access. This article is distributed Medical Writing, Editorial, and Other under the terms of the Creative Commons Assistance. Editorial assistance in the prepara- Attribution-NonCommercial 4.0 International tion of this article was provided by Rob Coover License (http://creativecommons.org/licenses/ of Caudex (New York, NY, USA). Support for by-nc/4.0/), which permits any non- this assistance was funded by Sanofi. commercial use, distribution, and reproduction in any medium, provided you give appropriate Authorship. All named authors meet the credit to the original author(s) and the source, International Committee of Medical Journal provide a link to the Creative Commons license, Editors (ICMJE) criteria for authorship for this and indicate if changes were made. manuscript, take responsibility for the integrity of the work as a whole, and have given final approval for the version to be published. All authors had full access to all of the data in this REFERENCES study and take complete responsibility for the integrity of the data and accuracy of the data 1. Centers for Disease Control. National diabetes analysis. statistics report, 2017. [article online], 2018. https:// www.cdc.gov/diabetes/data/statistics-report/index. html. Accessed 22 Jan 2018. Disclosures. Lawrence Blonde has received grant/research support from AstraZeneca, Jans- 2. Dandona P. Minimizing glycemic fluctuations in sen, Lexicon, Merck, Novo Nordisk, and Sanofi; patients with type 2 diabetes: approaches and speaker honoraria from AstraZeneca, Janssen, importance. Diabetes Technol Ther. 2017;19(9):498–506. and Merck; and consultancy fees from AstraZe- neca, GlaxoSmithKline, Intarcia Therapeutics, 3. Paul SK, Klein K, Thorsted BL, Wolden ML, Khunti Janssen, Merck, and Novo Nordisk. Luigi K. Delay in treatment intensification increases the Meneghini has received advisory board/consul- risks of cardiovascular events in patients with type 2 diabetes. Cardiovasc Diabetol. 2015;14:100. tancy fees from Novo Nordisk, Sanofi, and Intarcia. Anders Boss is an employee of Sanofi. 4. Stokes A, Preston SH. Deaths attributable to dia- Alka Shaunik is an employee of Sanofi. Claire betes in the United States: comparison of data Brulle-Wohlhueter is an employee of Sanofi. sources and estimation approaches. PLoS One. 2017;12(1):e0170219. Xuejun Victor Peng is an employee of Sanofi. Kyu Rhee is an employee of Watson Health, 5. Stratton IM, Adler AI, Neil HA, et al. Association of IBM. Supriya Kumar is an employee of Watson glycaemia with macrovascular and microvascular Health, IBM. Sidhartha Balodi is an employee of complications of type 2 diabetes (UKPDS 35): prospective observational study. BMJ. Watson Health, IBM. Rory J McCrimmon has 2000;321(7258):405–12. Diabetes Ther (2018) 9:1347–1358 1357 6. American Diabetes Association. 6. Glycemic targets: 16. Lin SD, Tsai ST, Tu ST, et al. Glycosylated hemo- standards of medical care in diabetes—2018. Dia- globin level and number of oral antidiabetic drugs betes Care. 2018;41(Suppl 1):S55–64. predict whether or not glycemic target is achieved in insulin-requiring type 2 diabetes. Prim Care 7. Esposito K, Chiodini P, Bellastella G, Maiorino MI, Diabetes. 2015;9(2):135–41. Giugliano D. Proportion of patients at HbA1c tar- get \ 7% with eight classes of antidiabetic drugs in 17. Mauricio D, Meneghini L, Seufert J, et al. Glycaemic type 2 diabetes: systematic review of 218 random- control and hypoglycaemia burden in patients with ized controlled trials with 78 945 patients. Diabetes type 2 diabetes initiating basal insulin in Europe Obes Metab. 2012;14(3):228–33. and the USA. Diabetes Obes Metab. 2017;19(19):1155–64. 8. Hermansen K, Davies M, Derezinski T, Martinez RG, Clauson P, Home P. A 26-week, randomized, par- 18. Wu N, Aagren M, Boulanger L, Friedman M, Wilkey allel, treat-to-target trial comparing insulin detemir K. Assessing achievement and maintenance of gly- with NPH insulin as add-on therapy to oral glucose- cemic control by patients initiating basal insulin. lowering drugs in insulin-naive people with type 2 Curr Med Res Opin. 2012;28(10):1647–56. diabetes. Diabetes Care. 2006;29(6):1269–74. 19. Kostev K, Dippel FW, Rathmann W. Glycemic 9. Garber AJ, Abrahamson MJ, Barzilay JI, et al. Con- control after initiating basal insulin therapy in sensus statement by the American Association of patients with type 2 diabetes: a primary care data- Clinical Endocrinologists and American College of base analysis. Diabetes Metab Syndr Obes. Endocrinology on the comprehensive type 2 dia- 2015;8:45–8. betes management algorithm—2016 executive summary. Endocr Pract. 2016;22(1):84–113. 20. Kim SG, Kim NH, Ku BJ, et al. Delay of insulin ini- tiation in patients with type 2 diabetes mellitus 10. Inzucchi SE, Bergenstal RM, Buse JB, et al. Man- inadequately controlled with oral hypoglycemic agement of hyperglycemia in type 2 diabetes: a agents (analysis of patient- and physician-related patient-centered approach: position statement of factors): a prospective observational DIPP-FACTOR the American Diabetes Association (ADA) and the study in Korea. J Diabetes Investig. European Association for the Study of Diabetes 2017;8(3):346–53. (EASD). Diabetes Care. 2012;35(6):1364–79. 21. Peyrot M, Rubin RR, Lauritzen T, et al. Resistance to 11. Dalal MR, Grabner M, Bonine N, Stephenson JJ, insulin therapy among patients and providers: DiGenio A, Bieszk N. Are patients on basal insulin results of the cross-national Diabetes Attitudes, attaining glycemic targets? Characteristics and goal Wishes, and Needs (DAWN) study. Diabetes Care. achievement of patients with type 2 diabetes mel- 2005;28(11):2673–9. litus treated with basal insulin and physician-per- ceived barriers to achieving glycemic targets. 22. Strain WD, Cos X, Hirst M, et al. Time to do more: Diabetes Res Clin Pract. 2016;121:17–26. addressing clinical inertia in the management of type 2 diabetes mellitus. Diabetes Res Clin Pract. 12. Gordon J, Pockett RD, Tetlow AP, McEwan P, Home 2014;105(3):302–12. PD. A comparison of intermediate and long-acting insulins in people with type 2 diabetes starting 23. Khunti K, Millar-Jones D. Clinical inertia to insulin insulin: an observational database study. Int J Clin initiation and intensification in the UK: a focused Pract. 2010;64(12):1609–18. literature review. Prim Care Diabetes. 2017;11(1):3–12. 13. Harris SB, Kapor J, Lank CN, Willan AR, Houston T. Clinical inertia in patients with T2DM requiring 24. Khunti K, Nikolajsen A, Thorsted BL, Andersen M, insulin in family practice. Can Fam Physician. Davies MJ, Paul SK. Clinical inertia with regard to 2010;56(12):e418–24. intensifying therapy in people with type 2 diabetes treated with basal insulin. Diabetes Obes Metab. 14. Heintjes EM, Thomsen TL, Penning-van Beest FJ, 2016;18(4):401–9. Christensen TE, Herings RM. Glycemic control and long-acting insulin analog utilization in patients 25. Berard L, Bonnemaire M, Mical M, Edelman S. with type 2 diabetes. Adv Ther. 2010;27(4):211–22. Insights into optimal basal insulin titration in type 2 diabetes: results of a quantitative survey. Diabetes 15. Levin PA, Zhou S, Durden E, Farr AM, Gill J, Wei W. Obes Metab. 2018;20(2):301–8. Clinical and economic outcomes associated with the timing of initiation of basal insulin in patients 26. McFarlane SI. Insulin therapy and type 2 diabetes: with type 2 diabetes mellitus previously treated management of weight gain. J Clin Hypertens with oral antidiabetes drugs. Clin Ther. (Greenwich). 2009;11(10):601–7. 2016;38(1):110–21. 1358 Diabetes Ther (2018) 9:1347–1358 27. Polonsky WH, Thompson S, Wei W, et al. Greater glargine/lixisenatide treatment in type 2 diabetes: a fear of hypoglycaemia with premixed insulin than systematic review and meta-analysis. Expert Opin with basal-bolus insulin glargine and glulisine: Pharmacother. 2017;18(17):1789–98. patient-reported outcomes from a 60-week ran- domised study. Diabetes Obes Metab. 30. Steen O, Goldenberg RM. The role of sodium-glu- 2014;16(11):1121–7. cose cotransporter 2 inhibitors in the management of type 2 diabetes. Can J Diabetes. 28. Vijan S, Hayward RA, Ronis DL, Hofer TP. Brief 2017;41(5):517–23. report: the burden of diabetes therapy: implications for the design of effective patient-centered treat- 31. Valentine V, Goldman J, Shubrook JH. Rationale ment regimens. J Gen Intern Med. for, initiation and titration of the basal insulin/ 2005;20(5):479–82. GLP-1RA fixed-ratio combination products, IDe- gLira and iGlarLixi, for the management of type 2 29. Cai X, Gao X, Yang W, Ji L. Comparison between diabetes. Diabetes Ther. 2017;8(4):739–52. insulin degludec/liraglutide treatment and insulin

Journal

Diabetes TherapySpringer Journals

Published: Mar 29, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off