Hypoglycaemia and treatment patterns among insulin‐treated patients with type 2 diabetes who switched to insulin glargine 300 units/mL versus other basal insulin in a real‐world setting

Abstract Introduction Type 2 diabetes (T2D) is characterized by worsening pancreatic β‐cell function often requiring treatment escalation with oral antidiabetic drugs (OADs), glucagon‐like peptide‐1 and eventually insulin. Although there is much evidence available on the initiation of basal insulins, fewer studies have investigated the effects of switching from one basal insulin to another. This study aims to evaluate treatment persistence and hypoglycaemia in adult patients with T2D on prior basal insulin who were switched to insulin glargine 300 units/mL (Gla‐300) or other basal insulins in a real‐world setting. Materials and methods This study is a retrospective cohort analysis of patient‐level data extracted from the Optum® Clinformatics™ database between 1 October 2014 and 30 June 2016. Adult patients (≥18 years) with T2D who were being treated with basal insulin during the 6‐month baseline period, who switched to either Gla‐300 or other basal insulins, were followed up for ≥3 months after switching. Outcomes included treatment persistence, and incidence and number of hypoglycaemic events. Results Of the included patients, 1204 switched to Gla‐300 and 616 switched to other basal insulins. Adjusting for baseline confounders, patients who switched to Gla‐300 were 34% less likely to discontinue their basal insulin than patients who switched to other basal insulins (hazard ratio [HR] 0.66; 95% confidence interval [CI] 0.54‐0.81; P < 0.001). Patients who switched to Gla‐300 were less likely to experience hypoglycaemia at 3‐month follow‐up (odds ratio [OR] 0.56, 95% CI 0.32‐0.97; P = 0.039) and at 6‐month follow‐up (OR 0.58, 95% CI 0.38‐0.87; P = 0.009) compared with patients who switched to other basal insulins. Conclusions Patients with T2D on prior basal insulin in a real‐world setting who switched to Gla‐300 were more persistent with their basal insulin and experienced less hypoglycaemia than patients who switched to other basal insulins.

pharmacy claims data and laboratory results, for members of a large US managed care group covering a geographically diverse population across all 50 states of the USA. The database is updated monthly and includes data with service dates from May 2000, comprising approximately 15 million annual covered lives for a total of approximately 47 million unique entries over a 10-year period.

| Patient selection
Eligible patients were adults (≥18 years of age) with T2D who had ≥1 inpatient/emergency room (ER) visit or ≥2 ambulatory medical claims (≥30 days apart) with a primary or secondary diagnosis of T2D (ICD-9-CM codes: 250.x0 or 250.x2/ICD-10-CM code: E11) during the study period. Patients categorized as long-acting insulin users were defined with ≥1 basal insulin claim for Gla-300, Gla-100, insulin detemir or insulin degludec, during the identification period (between 1 April 2015 and 31 March 2016). The index date was defined as the date of the first Gla-300 or other index basal insulin claim.
Patients were considered eligible, as switching insulin patients, if they had ≥1 claim in the 6-month baseline period prior to the index date for a different long-acting insulin, either neutral protamine Hagedorn (NPH), insulin detemir or Gla-100.
Patients were assigned to cohorts depending on whether they switched to Gla-300 or another basal insulin (Gla-100, detemir, degludec) on the index date. Included patients had continuous enrolment in the database with medical and pharmacy coverage for the 6-month baseline period and ≥3-month follow-up period after the index date; they were followed up until disenrolment, death or study end. All patients had ≥1 measurement of A1C level during the baseline period.

| Outcome assessments
The study endpoints included treatment persistence, and incidence and number of hypoglycaemic events. Patients were defined as "persistent" if they remained on the index insulin during the follow-up period without discontinuation after the index date. Patients were considered to have discontinued basal insulin if no claim was made for a refill within the expected time of medication coverage based on the metric quantity of that prescription, defined as the 90th percentile of the time between the first and second fills among patients with at least 1 refill and same metric quantity for the first fill. If no refill was claimed within that period, the date used for discontinuation was the estimated run-out data of the last claimed refill.
The incidence and number of events (adjusted for baseline confounders) of hypoglycaemia were identified by ICD-9-CM/ICD-10-CM codes for medical claims in inpatient, outpatient or ER settings during 3-and 6-month follow-up.
In a subgroup of patients who had A1C measurements at baseline and during 3-to 6-month follow-up, change in A1C from baseline to follow-up (latest available value during follow-up) was also analysed (Tables S1-S3).

| Patient baseline characteristics
This analysis included 1204 patients who switched to Gla-300 and 616 patients who switched to other basal insulin analogs (Figure 1).

| Treatment persistence
Fewer patients who switched to Gla-300 (20.4%) discontinued their insulin prescription compared with patients who switched to other basal insulins (36.4%) during the 6-month follow-up (Figure 2A,B).
After adjusting for baseline confounders, Gla-300 switchers were found to be 34% less likely to discontinue basal insulin therapy during follow-up compared with other basal insulin switchers (HR 0.66, 95% confidence interval [CI] 0.54-0.81; P < 0.001). The baseline confounders that were associated with persistence or treatment discontinuation included ( Figure 3) female gender, African American race, higher CCI, higher baseline A1C levels, and presence of outpatient hypoglycaemia at baseline (associated with increased risk of treatment discontinuation), use of a GLP-1 RA or sulfonylurea at baseline and baseline comorbidities except for mild liver disease and depression (associated with an increased likelihood of treatment persistence).
After adjusting for baseline confounders, patients who switched to

| Change in A1C
In the subgroup of patients who had A1C measurements at baseline and during 3-to 6-month follow-up, mean reductions in A1C, adjusted for baseline confounders, were modest but comparable between patients who switched to Gla-300 and other basal insulins (−0.35% vs −0.27%, respectively; adjusted mean difference: −0.08%; P = 0.578) ( Figure 4B). Differences in baseline demographics and clinical characteristics may guide physicians' prescribing decisions. Although there was a statistically significant difference in age between the two groups, the actual numerical difference of only 2 years would suggest that age was an unlikely driver of treatment decisions. The lower CCI score, and lower use of inpatient and ER facilities, suggest that patients switching to Gla-300 were healthier. The presence of comorbidities is known to be a major consideration for physicians considering initiating insulin therapy; however, it is unclear how general health might affect switching in patients already using basal insulin. 17 Data from previous studies on insulin switching suggest that a higher burden of illness is a driver for switching between basal insulin analogs. 25 It may be that in our cohort, the lower level of illness in Gla-300 switchers is reflective of switching driven by the perceived advantages of the Gla-300 rather than by problems with the current basal insulin therapy. Additionally, patients switching to Gla-300 had higher baseline insulin doses.

| D ISCUSS I ON
Patients who switched to Gla-300 had experienced fewer hypoglycaemic events at baseline than those who switched to another basal insulin. The substantially higher use of NPH insulin in the other switcher group is a possible explanation for this difference, as NPH insulin is associated with a higher rate of hypoglycaemia than Gla-100. 26,27 It is notable that in a small retrospective study, concern regarding hypoglycaemia was one of the main reasons cited for switching to Gla-300, a result somewhat in contrast to our own data. 10 Patients switching to Gla-300 were found to be more likely to remain persistent during follow-up than those switching to other basal insulins. Although there were twice as many Gla-300 switchers, our selection criteria were not preferential for Gla-300 switchers over other switchers. Therefore, the prevalence of more Gla-300 switchers during this time period may be a function of prescriber pattern or health plan characteristics. However, we did adjust for all patient characteristics.
The previous studies have demonstrated differences in treatment persistence following switching of basal insulin, with patients switching from Gla-100 to insulin detemir being less persistent than those switching from insulin detemir to Gla-100 26,28 ). Additionally, a positive correlation has been shown between specialist visits and There are several other factors associated with Gla-300 treatment that may predispose to better persistence, including flexibility of injection timing, single daily injections and lower injections volumes. 25 Our study reinforces data from the EDITION trial series showing that Gla-300 is associated with less hypoglycaemia than other basal insulins. [6][7][8][9] Patients in this cohort who switched to Gla-300 were less likely to experience hypoglycaemia both at 3-and at 6-month followup compared with those who switched to other basal insulins. Similar results were found in a real-world study that used propensity matching to harmonize baseline characteristics between groups, controlling for baseline confounders of hypoglycaemia. 11 In the subgroup analysis of patients who had A1C measurements at baseline and during 3-to 6-month follow-up, glycaemic control was similar between the two switcher groups, a result which is in line with both the Phase III EDITION trial programme and a similar study comparing patients switching to Gla-300 or other basal insulins. 6-9

| Limitations
While the findings represent actual treatment patterns and out-  The successful early initiation of insulin may therefore be an important factor in the long-term successful use of insulin.

ACK N OWLED G EM ENTS
This study was funded by Sanofi US, Inc. The authors received writing/editorial support in the preparation of this manuscript from Joseph Worrall, PhD, of Excerpta Medica, funded by Sanofi US, Inc.

AUTH O R CO NTR I B UTI O N
LZ, LX and YW designed the study. LX, YW and NV acquired the data. All authors interpreted the data and provided critical revisions to the manuscript.

E TH I C A L A PPROVA L
This study was conducted in compliance with the ethics guidelines for research in humans as recorded in the Helsinki Declaration of 1964.
Given the observational retrospective nature of this study, individual consent was not required after ensuring for anonymization of data.

DATA AVA I L A B I L I T Y
All data are included within the manuscript.