Treatment Patterns, Adherence, and Persistence Associated With Human Regular U-500 Insulin: A Real-World Evidence Study
Abstract
OBJECTIVE | Human regular U-500 insulin (U-500R) is concentrated insulin with basal and prandial activity that can be used as insulin monotherapy. The goal of this study was to better understand treatment patterns (total daily dose [TDD] and concomitant medications), adherence, and persistence in real-world patients treated with U-500R.
DESIGN AND METHODS | We selected patients from the Truven Health MarketScan database who initiated U-500R between 2010 and 2013. We collected data for three periods: pre-index (12 months before initiation), post-index (12 months after initiation or until a gap of ≥60 days in U-500R claims), and follow-up (12 months after post-index). Data were analyzed using descriptive statistics and a regression model as appropriate.
RESULTS | We identified 1,582 patients who met the selection criteria. The median TDD of U-500R during the post-index period was 333 units/day, with 70.0% of patients using 300–400 units/day. During the post-index period, 74.1% of patients had U-500R claims that did not overlap with prescriptions for other insulins, interpreted as U-500R monotherapy. Among patients with ≥1 U-500R fill in the post-index period (n = 1,208), 54.4% had a medication possession ratio (MPR, a measure of adherence) ≥80%. Although 849 patients had a gap of ≥60 days in U-500R claims in the post-index period, 602 of those resumed U-500R in the follow-up period. Of the 733 patients who had no gap in U-500R claims in the post-index period, 286 had a gap of ≥60 days in claims in year 2, and 447 continued with U-500R treatment beyond 2 years.
CONCLUSION | These results demonstrate that U-500R was commonly used as insulin monotherapy, with a median TDD >300 units/day. Compared with published, relevant studies of other insulins, U-500R showed similar or greater adherence and persistence rates. These new data may help guide clinical decision-making when choosing insulin therapy for patients requiring high doses of insulin.
With the prevalence of obesity and diabetes increasing globally (1–3), physicians and other health care providers are faced with making clinical decisions for managing patients who are severely insulin-resistant or who require large doses of insulin to achieve their glycemic goals (4). Patients with type 2 diabetes with severe insulin resistance typically need to inject large volumes of both prandial and basal insulins. Delivery of high doses of U-100 insulin is associated with practical volume-related challenges, including requiring more than five or six daily injections, causing injection site discomfort, exceeding the maximum limit of insulin syringes and pens covered by insurance, and increasing the complexity of the treatment regimen (5). As a result, many patients may have low adherence to and persistence with the medication, leading to a less-than-optimal amount of insulin being used (6–8).
Poor adherence is linked to poor glycemic control, increased health care resource utilization, and higher mortality rates (9–11). In people with high insulin requirements who are not on concentrated insulins, A1C is often >8.5% (6,8,12,13), and comorbidities and diabetes complications are common, with a Charlson comorbidity index (CCI) >2.0 (4,7,13). Low medication possession ratio (MPR), a measure of adherence, is generally associated with higher health care costs (14). A comprehensive literature review found that an increase of ∼10% in MPR is associated with decreases in health care costs of 8.6–28.9%, mostly resulting from a decrease of 4.1–31% in hospitalizations (15).
Recombinant human regular U-500 insulin (U-500R), which is five times more concentrated than its U-100 formulation counterpart, has been available since 1997 for patients with diabetes who require more than 200 units/day of insulin (16). Because of its unique pharmacokinetics, U-500R can cover both prandial and basal insulin needs (17). In a randomized controlled trial of patients switching from 2–10 injections of U-100 insulin to U-500R administered twice or three times daily, patients were able to reduce A1C by >1%, a clinically meaningful improvement, without the need for rescue intensification (6,13). Even in a real-world setting, switching from U-100 insulin to U-500R has resulted in A1C reduction of ∼0.7% in the first year (4).
Besides improved glycemic control, multiple retrospective analyses have shown that patients who initiate U-500R have lower annual pharmacy costs and lower overall health care costs than patients receiving a high dose (>150–200 units/day) of U-100 insulin (4,7,8). These studies showed greater adherence with U-500R compared with U-100 insulin, albeit with a higher incidence of hypoglycemia.
Medication compliance is a multifaceted concept that attempts to quantify “the extent to which a patient acts in accordance with the prescribed interval and dose of a dosing regimen” (18). In the case of insulin therapy, taking the appropriate amount each day with an appropriate regimen is key to glycemic control. Also, both insulin dose and glycemic control may be affected by concomitant antihyperglycemic medications. Therefore, to provide a complete understanding of insulin therapy compliance and its impact, an evaluation of treatment patterns should include insulin daily dose and regimen, concomitant medications, insulin possession, and insulin persistence.
No previous real-world studies have comprehensively examined the treatment patterns among patients taking U-500R. Clinical studies of U-500R did not allow concomitant antihyperglycemic medications other than metformin, thereby providing no guidance on their use with U-500R. Using up to 2 years of data, this study aims to expand our understanding of U-500R treatment patterns and compliance in the real-world setting by comprehensively evaluating total daily dose (TDD), concomitant medication use, adherence, and persistence in patients who initiate U-500R therapy. We also explored the baseline factors associated with interruption of treatment within the first year after U-500R initiation.
Research Design and Methods
Study Design
In this retrospective cohort study, we used claims data from the Truven Health Analytics MarketScan Research Databases for the period between January 2009 and December 2015 to identify patients with type 2 diabetes who were prescribed U-500R treatment between January 2010 and December 2013 (Supplementary Figure 1). The database captured longitudinal person-specific drug and medical claims from large employers, health plans, and government and public organizations. All data were de-identifed to comply with Health Insurance Portability and Accountability Act regulations.
The index date of the study was defined as the date of the first U-500R pharmacy claim between January 2010 and December 2013. We collected data for three periods: 1) the pre-index period: the 12 months before the index date, 2) the post-index period: up to 12 months after the index date or until there was a gap of ≥60 days in U-500R claims within the 12 months after the index date, and 3) the follow-up period: the 12 months after the post-index period. A gap of ≥60 days in U-500R claims was defined as no refill of U-500R within 60 days of the end of the last U-500R fill (i.e., claim date plus days’ supply).
Patients were included if they had at least one U-500R prescription claim during the identification period, had no prescription claim for U-500R 1 year before the index date, and had 2 years of continuous enrollment after the index date. Patients were excluded if, in the pre-index period, there was a diagnosis of secondary diabetes; gestational diabetes; diabetes complicating pregnancy, childbirth, or puerperium; pregnancy; other abnormal glucose; or neonatal diabetes. Patients with type 1 diabetes were excluded because U-500R is prescribed mainly for the treatment of type 2 diabetes, and treatment patterns for type 1 diabetes would be different. Complete inclusion and exclusion criteria are listed in the Supplementary Materials.
The following variables were reviewed for the pre-index period (baseline): demographics, comorbidities, complications, antidiabetic medications, and Deyo-CCI (derived; a commonly used modified CCI) (19,20). Aspects of treatment patterns evaluated included the observed TDD from all insulin claims, concomitant medication use, adherence, and persistence.
Total Daily Dose
The TDDs of insulin from all insulin claims were calculated as the total units of all insulins for claims in the post-index period divided by the total number of days’ supply on these claims. The mean and median TDDs for patients’ first U-500R claim as well as the mean and median TDDs of U-500R for the whole post-index period were also calculated. As a sensitivity analysis, we calculated the average daily dose (ADD) by using the total number of days in the post-index period, instead of the total number of days’ supply on claims, as the denominator in the TDD calculation.
Concomitant Medication Use
For the pre-index period, we specifically examined the percentage of patients who had no claims for U-500R overlapping with continuous claims for other insulins. Overlap with other insulin claims was assessed for ≥1 day, ≥15 days, ≥30 days, ≥45 days, and ≥60 days. In addition, the pre-index, post-index, and follow-up periods were compared for use of concomitant insulins, noninsulin injectable medications, oral antidiabetes medications (OADs), and U-500R insulin monotherapy.
Adherence
Adherence to U-500R during the post-index period was measured using the MPR, calculated as the number of U-500R days’ supply (minus the last post-index refill) divided by the number of days between the first and last refill post-index (21). Five MPR categories (<20%, 20 to <40%, 40 to <60%, 60 to <80%, and ≥80%) were evaluated by age-group (<65 years, ≥65 years). For patients with an MPR >100%, the value was truncated to 100%.
Persistence
To examine treatment continuation and persistence, we first defined an interruption in U-500R treatment as a gap of ≥60 days in insulin claims. During the post-index period, persistence was estimated as time to treatment interruption using the Kaplan-Meier method. Key baseline factors associated with treatment interruption were identified via the gradient boosting method and validated with a Cox regression model. Age-group was included (18–44, 45–54, 55–64, 65–74, 75–84, and ≥85 years), with the age-group 18–44 years as the reference group. Hazard ratios (HRs) for identified significant factors were determined.
The post-index period study sample comprised two cohorts: patients who had an interruption (gap of ≥60 days) within 12 months and those who continued with no gap in claims for 12 months. Both cohorts were followed for another 12 months in the follow-up period, during which insulin-use patterns were examined. For interrupters, this included U-500R resumption and/or switching to other insulins. For continuers, this included percentage of patients who continued U-500R for the second year.
Categorical measures are reported as frequency and percentages, and continuous measures are expressed as mean ± SD. If the distribution of a measure was skewed, the median was also reported. Time-to-event analysis reports the median time to the event, which was the time when 50% of patients had experienced the event. If <50% of patients had the event by the end of the observation period, then the median could not be estimated.
All statistical analyses were performed using SAS v. 9.2 (SAS Institute, Cary, NC) or R 3.3.3 (R Core Team, Vienna, Austria).
Results
Baseline Demographics and Clinical Characteristics
A total of 1,582 patients from the Truven MarketScan Research database met inclusion criteria and were not eliminated based on exclusion criteria. The baseline demographics and clinical characteristics of these patients are given in Table 1. The mean age was 56.8 ± 9.4 years, and 58.9% were men. The majority of patients had hypertension and hyperlipidemia and a mean CCI of 3.1.
Baseline Demographics and Clinical Characteristics
Pattern of U-500R Use After Initiation
Of the 1,582 patients who initiated U-500R therapy, 733 (46.3%) had continuous claims (no gap of ≥60 days) for U-500R for >1 year, and 447 (28.3%) had continuous claims for 2 years (Figure 1). Of the U-500R initiators, 849 (53.7%) had a gap of ≥60 days in U-500R claims in the first year, but 602 (70.9%) of those restarted U-500R in the follow-up period. Of those who restarted U-500R in the follow-up period, 402 (66.8%) restarted without adding a new insulin, whereas 200 (33.2%) restarted and had a new claim for another insulin besides U-500R. (New insulin was defined as insulin not used in the post-index period.) The other 247 (29.1%) patients with a U-500R claims gap in the first year did not restart U-500R; 187 (22%) initiated a different formulation of insulin, and 60 (7.1%) neither restarted U-500R nor initiated a new insulin during the follow-up period (Figure 1).
Persistence; patients who initiated U-500R (top chart) claims in the post-index period who had continued treatment (gray) or had an interruption in treatment (≥60-day gap in claims) (red). The bottom left chart shows the percentages of patients who restarted U-500R and/or initiated a new insulin or who had no new insulin and did not restart U-500R in the follow-up period. The bottom right chart shows the percentages of patients who had continuous treatment or who had a ≥60-day gap in the follow-up period.
U-500R TDD in the Post-Index Period
The TDD for the first U-500R claim was >300 units/day for 1,404 patients (88.7%) (Table 2), and consecutive U-500R fills remained consistent over time (Supplementary Figure 2). The median TDD for U-500R in the post-index period was 333 units/day (mean 375 units/day). The median and mean ADD, which is a more conservative estimate of the daily dose, was very close to the median/mean TDD (median 323 units/day, mean 324 units/day).
U-500R TDD at First U-500R Claim and at 1 Year or Until a Gap of ≥60 Days Occurred
TDD of All Insulins in the Post-Index Period
The dose for all insulins was >300 units/day for 1,390 patients (87.9%) (Table 2). The median TDD of all insulins used during the post-index period, including U-500R and other insulins, was 335 units/day (mean 377 units/day). The median ADD was 331 units/day (mean 342 units/day). The similarities between the median/mean values for ADD and TDD suggest that the average gap between refills was small, adherence was good, the actual median daily dose was likely between 331 and 335 units/day, and the actual mean daily dose was likely to be between 342 and 377 units/day.
Concomitant Medication Use
Before U-500R initiation, most patients were using at least two insulins (mean 1.81 ± 0.71, median 2.00); 76.2% of patients took a long-acting insulin, and 68% took a rapid-acting insulin. In the post-index period, 1,172 patients (74.1%) had claims for U-500R that did not overlap with other insulin claims (Table 3), suggesting U-500R monotherapy. In this post-index period, only 251 (15.9%) and 185 (11.7%) patients, respectively, had claims for long-acting insulin and rapid-acting insulin (Table 4), with 120 patients (7.6%) having claims that had ≥60 days of overlap with a U-500R claim (Table 3). On the other hand, among the 849 patients who had a gap of ≥60 days in U-500R claims in the post-index period, the percentage with long-acting insulin and rapid-acting insulin claims in the follow-up period increased to 32.9 and 27.7%, respectively (Table 4).
Post-Index Overlapping Insulin Use With U-500R
Concomitant Medication Use by Period
Before U-500R initiation, 65.7% of patients had at least one OAD claim during the 12-month pre-index period, with metformin being the most common, followed by sulfonylureas. After U-500R initiation in the post-index period, claims for OADs declined (Table 4), with sodium–glucose cotransporter 2 (SGLT2) inhibitors as an exception, showing a slight increase in claims during this period. For patients who had a gap of ≥60 days in U-500R claims in the post-index period, all OAD claims increased but did not rise to the same level as in the pre-index period (Table 4). Noninsulin injectable claims also declined in the post-index period (Table 4).
Adherence
Among 1,208 patients (76.4%) with more than one U-500R prescription filled during the post-index period, 54.4% had an MPR of ≥80%, and none had an MPR <20% (Figure 2). The overall mean MPR was 78.6 ± 19% (Supplementary Table 1).
Adherence as determined by the percentages of patients by MPR category in the post-index period.
Persistence
Although a little more than half of patients had a gap of ≥60 days in the first year, most of the gaps ≥60 days occurred within 100 days of the index date (Figure 3), including 374 patients (23.6%) who had the gap after the first fill of U-500R. Because about half of the patients had no claim gap in the first year, the median time to a gap of ≥60 days in U-500R claims during the post-index period was 264 days (95% CI 232–285) (Figure 3).
Kaplan-Meier curves for time from U-500R initiation to ≥60-day gap or time from the end of the last-U-500R claim in the post-index period to restart of U-500R by age-groups (<65 and ≥65 years).
Baseline (pre-index) factors associated with a decreased rate of having a gap of ≥60 days in U-500R claims included rapid-acting insulin use at baseline (21.1% decrease in gap event rate [P = 0.001]), more insulins used at baseline (15.3% decrease in gap event rate for each additional insulin [P <0.001]), and being the ages of 45 and 74 years (age-group 45–54, 55–64, or 65–74 years) compared with those in the age-group 18–44 years (25–29% decrease [P <0.05 for all]). The ≥60-day gap event rate was similar between patients >74 years of age and those 18–44 years of age. For those with a claim gap of ≥60 days, the median time from the end of the last U-500R claim to the time of restart of U-500R in the follow-up period was 135 days (95% CI 123–148).
Age-Related Differences
In the age-group ≥65 years of age (n = 289), 160 patients had a gap of ≥60 days compared with the 689 in the cohort <65 years of age (n = 1,293) (Figure 4). Patients who were ≥65 years of age were slightly more adherent with U-500R than those who were <65 years of age (mean MPR 81.8 vs. 78%), but the difference was not statistically significant (P = 0.12). In patients with a 60-day claims gap for U-500R in the post-index period, restart rates in the follow-up period for U-500R were numerically higher in those ≥65 years of age than in those <65 years of age (63.1 vs. 72.7%, P = 0.055). In patients who restarted U-500R after a gap, the time to restart was longer for patients age ≥65 years of age (median 176 days [95% CI 121–242]) than for patients <65 years of age (median 131 days [95% CI 120–144], P = 0.018).
Time to restart U-500R after gap (censoring time: from the end of the last U-500R claim in the post-index period to U-500R restart or end of the follow-up period).
There was a significant difference between age-groups (<65 vs. ≥65 years) in the time to new insulin initiation after a 60-day claim gap (P = 0.02). For patients ≥65 years of age, the median time to initiating a new insulin was 258 days (95% CI 160 to >365). Because fewer than half of the patients <65 years of age reinitiated a new insulin by the end of the follow-up period (365 days), the median time was longer than 1 year and these patients were unable to quantify.
Discussion
We explored the real-world use of U-500R insulin in 1,582 patients from the Truven MarketScan database for up to 2 years. The starting TDD of U-500R in our cohort was >300 units/day for most patients, and the mean TDD for U-500R remained consistent over time for multiple refills. The mean TDD in the post-index period was 375 units/day. The mean ADD (averaged overall days rather than the per day’s supply) was 324 units/day, indicating that the average gap between refills was small. These data are consistent with the doses used in a clinical trial in which patients were randomized to initiation and titration of U-500R using either two or three daily injections (6). In that study at end point, the twice-daily group used 343 units/day and the three-times-daily group used 335 units/day.
In this study, 74% of patients used U-500R as insulin monotherapy in the post-index period compared with 47% of patients in a previous report (4). Although the difference could be a result of different databases or patient populations, it is possible that the understanding of U-500R has evolved over time, and more physicians now are recognizing the basal and prandial properties of U-500R and therefore using it as insulin monotherapy. No OADs or noninsulin injectable therapies (except metformin) were used by ≥7% of patients. However, SGLT2 inhibitors were the exception, with a slight increase in use.
Adherence to U-500R, as measured by mean MPR, was almost 80% among those who had more than one U-500R claim in this study. Compared with other medications for chronic conditions, this adherence rate is quite good (22–24). Adherence rates for other insulins used in type 2 diabetes range from 38 to 77% depending on the methodology used (Supplementary Table 1).
Adherence to U-500R was measured in two earlier studies of patients on high-dose insulin using an adherence measure called Proportion of Days Covered (PDC) (7,8). In those studies, adherence was significantly greater in the U-500R cohorts (∼65%) than in the matched high-dose U-100 cohorts (∼40 and 48%) (Supplementary Table 1). Our data show even greater adherence to U-500R. For comparison, we have listed adherence rates to other insulins that were determined using either MPR or PDC (Supplementary Table 1).
This was the first study examining the persistence of U-500R use. We followed patients for up to 2 years and studied multiple aspects of persistence (as indicated by the absence of a ≥60-day claims gap), including rates of persistence at 1 year and at 2 years, persistence duration, and rates of resumption after a 60-day claims gap. In our study, almost half of patients (46.3%) used U-500R for at least 1 year, and nearly one-third (28.3%) used it for at least 2 years without a gap of ≥60 days in claims. Comparing our results to those of other studies using a claims-gap measure, it appears that the 1-year persistence of 46.3% in our study is better than that reported for other insulins (Supplementary Table 1).
The persistence duration for U-500R (mean/median 202/264 days with no 60-day gap) was similar to that seen for glargine (7.8 months, or 237 days) (25). Other studies, using different measures, have reported mean persistence durations ranging from 253 to 284 days with glargine and from 123 to 262 days with NPH insulin (26–29).
To better understand treatment persistence, we examined claims during the 12 months after a claim gap because the observed claim gap could have resulted from insurance design, pharmacy claim-system entry requirement, or missing inpatient pharmacy data and may not have reflected permanent discontinuation of the treatment. In this study, the majority of patients with a claim gap (70.9%) restarted U-500R, and about one-third had an additional claim for insulin that was not used before the U-500R claim gap. This finding suggests that the U-500R persistence rate might be higher than the 46.3% estimated above. About one-fifth of patients with a claim gap did discontinue U-500R and switch to a different regimen (Supplementary Table 2).
Restarting insulin therapy after a claims gap or initial discontinuation is widely observed in studies of (mostly basal) insulin (30,31). However, patients may or may not restart the same insulin (31). For example, Bonafede et al. (31) found that the majority of patients with a 90-day gap started another insulin after the gap: 86.9% of basal insulin initiators and 81.5% of premixed insulin initiators. Furthermore, 87% of patients with a 90-day gap added a mealtime insulin claim after the gap, suggesting intensification among those who had a treatment gap (32).
A clinical study found improved diabetes-specific patient-reported outcomes after initiating U-500R in a twice- or three-times-daily regimen (33). Treatment simplicity and convenience may be one of the key contributing factors to the adherence and persistence seen with U-500R. This seems to be especially true among the patients who previously had claims for multiple insulins or rapid-acting insulin. These patients may have appreciated the benefit of the simplified regimen, lower volume, and fewer injections with U-500R, making them less likely to have a subsequent U-500R treatment gap.
Patients ≥65 years of age were less likely to restart U-500R and more likely to initiate a new insulin compared with those <65 years of age. This group took longer to restart U-500R if they restarted and were quicker to initiate a new insulin. Those >74 years of age or younger than 44 years of age were more likely to have a U-500R claim gap. Hypoglycemia is a concern for older patients, and future research is needed to determine whether fear of hypoglycemia is associated with these behaviors.
Limitations
A limitation inherent to the use of administrative claims data is that it can only inform about prescriptions that were filled through insurance (1). Claims data may not give a thorough depiction of the actual amount of insulin used by patients because there could be self-paid insulin purchases or the patient could hoard, waste, or discard unused insulin. Patients may not use the amount of insulin prescribed or may not fill their prescriptions on time (2). Furthermore, claims may not reflect the actual prescriptions written by physicians; patients may abandon the prescriptions, or pharmacists may enter the days’ supply in a certain way because of insurance restrictions or computer system restrictions. As a result, days’ supply obtained from claims data may not reflect the actual days that the insulin supply lasted for patients. Furthermore, a gap in prescription claims data may be artificial and not accurately reflect discontinuation (3). Although the actual daily dose for U-500R in the study population cannot be accurately determined, a good estimation could lie between the calculated TDD (a likely overestimation using days’ supply) and the calculated ADD (a likely underestimation including all days between refills) (4).
The persistence rate was likely underestimated in this study (5). The administrative data did not contain socioeconomic factors that could influence claims gap or treatment discontinuation (6). These data also lack clinical characteristics such as A1C, adverse events, or BMI, preventing us from examining the dynamic relationship among adherence, glycemic control, adverse events, and other clinical outcomes.
Future studies should seek data from linked databases that allow comprehensive evaluation of these relationships. The U-500R insulin pen was only approved in 2016, so patients included in this study did not have access to this device. Therefore, results from this study do not apply to and should not be generalized to include U-500R delivered via the pen.
Conclusion
Patients with diabetes requiring high doses of insulin face many challenges regarding adherence and persistence with therapy. This study suggests that the concentrated insulin U-500R may offer some solutions. Patients prescribed U-500R showed similar to better adherence and persistence to their insulin therapy relative to studies of other insulins. The observed adherence and persistence patterns support the concept that U-500R monotherapy regimens are less complex than other treatment regimens and therefore easier to follow.
Evidence suggests that increased adherence and persistence in patients with type 2 diabetes is associated with improved clinical outcomes and reduced health care resource utilization, with reduced or similar overall health care costs despite higher pharmacy costs (34,35). As a better understanding of U-500R treatment patterns emerge, this knowledge can be used to assist clinicians and other health care professionals in decision-making when managing the treatment of patients with type 2 diabetes.
Article Information
Acknowledgments
The authors thank Caryl Antalis of Eli Lilly and Company and Debra A. Sherman of Syneos Health for writing support and Siew Hoong Wong-Jacobson of Eli Lilly and Company for data analysis support.
Funding
This manuscript was funded by Eli Lilly and Company.
Duality of Interest
J.C., C.Y.K., X.H., L.F., and R.J. are employees and stock shareholders of Eli Lilly and Company. J.A.J. is a former employee and stock shareholder of Eli Lilly and Company.
Author Contributions
J.C., X.H., and J.A.J. made substantial contributions to the conception and design of the study. J.C., C.Y.K., X.H., and J.A.J. analyzed the data. J.C. and R.J. drafted the manuscript. All authors interpreted the data, critically revised the manuscript for important intellectual content, and approved the manuscript for publication. J.C. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
J.A.J. is currently affiliated with the Bethesda Health Clinic, Tyler, TX; Center for Hope and Healing, Goshen, IN; and Department of Medicine, Indiana University School of Medicine, Indianapolis, IN.
This article contains supplementary materials online at https://spectrum.diabetesjournals.org/lookup/suppl/doi:10.2337/ds19-0060/-/DC1.
- © 2020 by the American Diabetes Association
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at https://www.diabetesjournals.org/content/license.
References
In this Issue
