The American Journal of Managed Care
March 2021
Volume 27
Issue 3

Costs for Commercially Insured Adults Prescribed Second-line Diabetes Medications

Adults with type 2 diabetes who received nonsulfonylurea medications had relative increases in total costs. Receiving newer medication classes led to relatively decreased medical costs.


Objectives: To examine differences in health care costs associated with choice of second-line antidiabetes medication (ADM) for commercially insured adults with type 2 diabetes.

Study Design: Retrospective cohort study with multiple pretests and posttests.

Methods: Included patients initiated second-line ADM therapy between 2011 and 2015, with variable follow-up through 2017. The 6 index medication classes were sulfonylureas, dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 receptor agonists (GLP-1 RAs), basal insulin, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and thiazolidinediones (TZDs). Multivariable regression models compared between-class changes in adjusted quarterly costs after second-line ADM initiation.

Results: The study cohort included 34,963 adults. Most were prescribed a sulfonylurea (46.0%) or DPP-4 inhibitor (30.4%). Adjusted quarterly index medication costs were significantly higher for all patients receiving nonsulfonylurea medications, ranging from $108 (95% CI, $99-$118) for TZDs to $742 (95% CI, $720-$765) for GLP-1 RAs. Changes in quarterly total health care costs were significantly higher for all nonsulfonylurea classes. Conversely, changes in quarterly nonpharmacy medical costs were significantly lower for patients receiving DPP-4 inhibitors (–$67; 95% CI, –$92 to –$43), GLP-1 RAs (–$43; 95% CI, –$85 to –$1), and SGLT-2 inhibitors (–$46; 95% CI, –$87 to –$6); changes in all other quarterly costs besides the index medication were significantly lower for patients receiving DPP-4 inhibitors (–$60; 95% CI, –$94 to –$26) and SGLT-2 inhibitors (–$113; 95% CI, –$169 to –$57).

Conclusions: The higher cost of nonsulfonylurea medications was the main driver of relative increases in total costs. Relative decreases in nonpharmacy medical costs among patients receiving newer ADM classes reflect these medications’ potential value.

Am J Manag Care. 2021;27(3):e72-e79.


Takeaway Points

Among commercially insured adults starting second-line medications for type 2 diabetes, the higher cost of nonsulfonylurea medications was the main driver of relative increases in total health care costs. However, there were relative decreases in nonpharmacy medical costs among patients receiving dipeptidyl peptidase-4 inhibitors, glucagon-like peptide-1 receptor agonists, and sodium-glucose cotransporter-2 inhibitors.

  • Findings reflect the higher up-front cost, but also the potential benefits, of newer medication classes.
  • Consideration of medications’ impacts on both medical and pharmacy costs may be useful in pharmacy benefit design.
  • More research is needed on medications’ benefits, tolerability, and harms in particular patient subgroups.


Although considerable progress has been made in preventing diabetes complications,1 the annual cost of care for Americans with type 2 diabetes increased more than 70% during a recent 10-year period.2 Increasing costs of diabetes care stem in large part from increased diagnosis and prevalence of diabetes, in addition to increased costs of outpatient services, supplies, and outpatient medications,2,3 especially insulin4 and newer antidiabetes medications (ADMs).5

Metformin is widely recommended as first-line pharmacologic therapy for type 2 diabetes.6 When a second-line ADM is required after metformin, nearly half of patients receive sulfonylureas,7,8 which have demonstrated glycemic-lowering effects similar to those of newer alternatives at much lower cost.9 Despite sulfonylureas’ low price, other second-line ADMs may offer value through improved patient outcomes that generate downstream cost savings. Sulfonylureas may increase the risk of hypoglycemia-related hospital admissions (and related costs),10 whereas other second-line ADMs may reduce downstream costs due to varying glycemic effects,6 safety profiles,11 or potentially preventing cardiovascular events12-15 and renal disease progression.16

Cost-effectiveness analyses can provide important information about the value of different second-line ADMs. However, prior cost-effectiveness analyses have modeled disease progression and medication adherence17 based on randomized trials in high-risk populations18,19 or trials including many participants from outside the United States.20 Real-world evidence in American populations can help health care stakeholders in the United States evaluate whether the higher up-front costs of newer ADMs may be offset by downstream impacts on other health care costs.

In this study, we examined differential changes in quarterly health care costs associated with clinicians’ choice of second-line ADM class in a commercially insured cohort of US adults, following the use of metformin alone. In addition to evaluating between-class changes in second-line ADM costs, total health care costs, nonpharmacy medical costs, and all other costs besides the second-line ADM, we explored between-class changes for several secondary cost measures.


Study Design and Data

As done previously,8,21 our study team extracted and evaluated data from national administrative databases provided by a large commercial health payer. We used health plan enrollment files, medical inpatient and ambulatory claims, and pharmacy claims from the period between January 1, 2011, and December 31, 2017, to evaluate outcomes for a large cohort of commercially insured US adults with type 2 diabetes. Due to our use of coded, nonidentifiable data, Northwestern University’s institutional review board judged this work as non–human subjects research.

This retrospective new-user cohort study used a 6-group pretest-posttest design, with multiple pretests and posttests, and a patient-quarter unit of analysis. We compared costs between classes of patients’ initial second-line ADM (described henceforth as the index medication), following the use of metformin alone. The 6 index medication classes under study were sulfonylureas, dipeptidyl peptidase-4 (DPP-4) inhibitors, glucagon-like peptide-1 receptor agonists (GLP-1 RAs), basal insulin, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, and thiazolidinediones (TZDs). These index medication classes include all 5 major drug classes being tested in an ongoing comparative effectiveness trial of leading second-line ADMs,22 as well as the SGLT-2 inhibitor class that was more recently brought to market. eAppendix Table 1 (eAppendix available at lists included medications within each included class.

Study Cohort

The study cohort included adult commercial health plan enrollees with type 2 diabetes, with metformin treatment augmented by a second-line ADM. Patients met this case definition by fulfilling 4 criteria, where their “index date” was defined as the date of the initial pharmacy claim for their index medication: (1) at least 1 medical claim with an International Classification of Diseases, Ninth Revision (ICD-9) diabetes diagnosis code for type 2 diabetes on or before the index date; (2) at least 1 pharmacy claim for metformin during the 180-day periods before and after the index date; (3) at least 1 pharmacy claim for an included index medication class between January 2011 and September 2015; and (4) no prior claims for any other included index medication class. Patient-quarters preceding the index date were categorized as the baseline period. The eAppendix Figure provides additional details on our methods and identification of study periods.

Included patients were required to have continuous health plan enrollment for at least 2 calendar quarters preceding and following the index date, guaranteeing at least 4 patient-quarters of data. To eliminate bias in baseline period comorbidity measurement attributable to differences between ICD-9 and ICD, Tenth Revision (ICD-10) coding architectures,23,24 we included only patients whose index date preceded the October 2015 transition to ICD-10 coding.

We excluded patients who had ever had any diagnostic codes for type 1 diabetes. Due to expected high costs that were not attributable to the ADMs under study, we excluded patients with a major organ transplant or end-stage renal disease diagnosis during the baseline period. We also excluded patients with major organ transplant during follow-up—other than kidney or heart transplant, which could potentially be related to the index medications under study—and women with pregnancy/childbirth during follow-up or the last 2 baseline quarters. Patients were censored from analysis at the time of commercial health plan disenrollment or transition to Medicare Advantage enrollment.


We calculated cost outcomes for each included patient-quarter. Costs of each claim were allocated to individual service categories using methods previously employed for commercial health plan data.25 All study outcomes represent the total amount of plan-defined costs for relevant covered benefits, including both health plan payments and plan-defined patient responsibilities (ie, deductibles, co-pays, and coinsurance) for paid claims.

We examined 4 primary outcomes: quarterly costs of the index medication, quarterly total health care costs, quarterly nonpharmacy medical costs (ie, total costs excluding all pharmacy claims), and all other quarterly costs besides the index medication (ie, total costs excluding index medication claims). The last 3 outcomes were winsorized at the 95th percentile (ie, across all included patient-quarters, costs above the 95th percentile were transformed to 95th percentile costs) to reduce the influence of outlier observations.26

There were several secondary measures of quarterly costs, including pharmacy claims for other (ie, nonindex) diabetes medications, pharmacy claims for all diabetes medications, total pharmacy claims, and diabetes supplies. Outpatient cost measures included costs of primary care office visits, endocrinology office visits, total office visits, diabetes education, and pathology/laboratory costs. We additionally examined hospital cost data separately for emergency department visits (including observation stays) and inpatient admissions.

We defined several covariate measures for multivariable regression modeling. Patient characteristics included sex, age, race/ethnicity (imputed), year of initiating index medication, US Census region, total health care costs during the last 2 baseline quarters, and most recent glycated hemoglobin (A1C) level. We additionally defined health plan characteristics for plan structure (eg, health maintenance organization or preferred provider organization), self-funded (ie, administrative services only) employer plans, and high-deductible plans. Because provider characteristics can influence the choice of index medication,8 the prescribing clinician’s area of clinical training was also defined. To enable adjustment for variation in costs and care delivery patterns across regions and over time, we collected data on hospital referral region27 and observation year. CMS Hierarchical Condition Categories risk adjustment variables described patients’ overall morbidity burden.28

Statistical Analysis

Statistical analyses were conducted using SAS version 9.4 (SAS Institute Inc) with α = 0.05. Descriptive statistics were computed for all baseline characteristics. We also calculated quarterly measures of total health care costs, by index medication class, adjusted for medical care inflation to 2011 levels.29

We used multivariable regression analyses to adjust for potential confounders of each primary and secondary outcome. Generalized estimating equations accounted for longitudinal patient-level clustering of quarterly observations.30 In our main analysis, models used an identity link and normal distribution to estimate changes in US$ relative to sulfonylureas, an approach shown to provide valid inference on differences in mean costs in large samples.31,32 Each regression model adjusted for all covariates described above.

In each regression model, the key independent variables were interactions between the follow-up period and each of the 5 nonsulfonylurea index medication classes. These interaction terms represent the differential change in costs during follow-up (vs baseline) by index medication class (referent: change in costs for patients receiving sulfonylureas). Key independent variables and baseline patient characteristics remained fixed over the entire follow-up period. This modeling approach focused inference on clinicians’ initial choice of index medication while effectively disregarding any subsequent changes in second-line ADM use (eg, switching from a sulfonylurea to a DPP-4 inhibitor).

We conducted a sensitivity analysis where regression models for our 4 primary outcomes omitted the categorical covariate for most recent A1C level, which had a high rate of missingness due to limited availability of laboratory testing results (although most patients had recent laboratory claims for A1C tests, test results are typically not available from administrative data). We also conducted multiple sensitivity analyses that adjusted for the same covariates as the main adjusted analysis. First, we investigated proportional differences in cost changes under varying distributional assumptions that are often employed in cost analyses (ie, log link, gamma distribution; outlier costs not winsorized). Second, we repeated the analytic approach from our primary analysis, restricted to the following subgroups: (1) subjects whose follow-up period began no earlier than January 2013 (ie, when SGLT-2 inhibitors were introduced to the US market), and (2) “adherent” subjects who, during at least the first 6 months of follow-up, used no other diabetes medications (besides metformin and the index medication) and filled prescriptions for the index medication for at least 80% of included days’ supply. These 2 subgroups comprised 63.8% and 44.6% of the final cohort, respectively.


The final cohort included 34,963 commercially insured adults with type 2 diabetes (eAppendix Table 2 presents numbers of patients meeting individual inclusion/exclusion criteria). Included patients contributed a mean of 8.0 patient-quarters during the baseline period (median, 7; range, 2-18) and 11.1 patient-quarters during follow-up (median, 10; range, 2-26).

More than three-fourths of included patients received either a sulfonylurea (n = 16,096; 46.0%) or DPP-4 inhibitor (n = 10,635; 30.4%) as their index medication (Table 1). Fewer patients were prescribed a GLP-1 RA (n = 2874; 8.2%), basal insulin (n = 1618; 4.6%), SGLT-2 inhibitor (n = 2475; 7.1%), or TZD (n = 1265; 3.6%). A majority of included patients were male (59.3%), were non-Hispanic White (65.8%), and received coverage through a self-funded employer health plan (52.2%). During the 2 quarters preceding the index date, patients whose index medications were either basal insulin or GLP-1 RAs tended to have higher baseline total health care costs: A higher proportion of patients started on basal insulin (9.0%) were in the highest baseline cost category, and a smaller proportion of patients started on a GLP-1 RA (37.6%) were in the lowest baseline cost category. Conversely, baseline total health care costs tended to be lower among patients whose index medication was a sulfonylurea or TZD (P < .001 for all χ2 tests of differences across groups).

The Figure presents inflation-adjusted mean total quarterly health care costs (to 2011 levels), by index medication class, for the last 2 years of the baseline period and first 3 years of follow-up. In all 6 index medication classes, costs were highest during the first follow-up quarter. Later in the first follow-up year, mean total health care costs declined and then stabilized at a level higher than that of the baseline period. Mean costs then increased modestly but consistently for several index medication classes.

Mean total adjusted health care costs for patients prescribed sulfonylurea as their index medication were $1595 per quarter during the baseline period and grew by $252 per quarter during follow-up. In multivariable model estimates of adjusted differential changes in our primary outcomes, higher costs of nonsulfonylurea medications seemed to be the main driver of growth in total costs during follow-up (Table 2). After initiating second-line ADM therapy, per-person per-quarter index medication costs were significantly higher for all patients receiving a nonsulfonylurea index medication, ranging from an additional $108 (95% CI, $99-$118) for TZDs to $742 (95% CI, $720-$765) for GLP-1 RAs. The change in quarterly total costs was similarly higher for all nonsulfonylurea classes, ranging from an additional $107 (95% CI, $32-$182) for TZDs to $664 (95% CI, $601-$726) for GLP-1 RAs.

However, patients receiving some index medication classes experienced relative decreases in other adjusted primary outcomes during follow-up. Compared with patients receiving sulfonylureas, the change in quarterly nonpharmacy medical costs was significantly lower for patients receiving DPP-4 inhibitors (–$67; 95% CI, –$92 to –$43), GLP-1 RAs (–$43; 95% CI, –$85 to –$1), and SGLT-2 inhibitors (–$46; 95% CI, –$87 to –$6). The change in all other quarterly costs besides the index medication (ie, after adding back all other pharmacy expenditures) was significantly lower for patients receiving DPP-4 inhibitors (–$60; 95% CI, –$94 to –$26) and SGLT-2 inhibitors (–$113; 95% CI, –$169 to –$57) but significantly higher for patients receiving basal insulin ($186; 95% CI, $106-$267). In the sensitivity analysis where we did not adjust for most recent A1C level, inference was identical for all primary outcomes (results available from authors upon request). In sensitivity analyses adjusting for the same covariates as the main adjusted analysis, inference on index medication costs and total health care costs was largely consistent with results of the main adjusted analysis (eAppendix Tables 4-6).

In secondary outcome analysis, changes in adjusted office visit costs (primary care and total) were significantly lower for patients receiving a DPP-4 inhibitor, a GLP-1 RA, or an SGLT-2 inhibitor index medication compared with sulfonylureas (Table 3); however, these relative decreases did not exceed $15 per quarter in either the main or sensitivity analyses. There was also some evidence of relative decreases in costs for hospital encounters, but results were not consistent across analyses. For example, in our main analysis, the change in quarterly inpatient costs was lower for patients receiving a DPP-4 inhibitor (–$129; 95% CI, –$230 to –$29) index medication vs sulfonylureas; however, in some sensitivity analyses, significant relative reductions were observed for either inpatient costs or emergency department costs (but not both inpatient and emergency costs). In the main analysis, patients receiving basal insulin as their index medication had a significant relative increase in quarterly costs of nonindex diabetes medications ($105; 95% CI, $72-$137), diabetes supplies ($16; 95% CI, $15-$17), primary care office visits ($8; 95% CI, $4-$12), and total office visits ($19; 95% CI, $11-$27) vs sulfonylureas; inference on these 4 outcomes was consistent across sensitivity analyses for those receiving basal insulin.


In this cohort of commercially insured adults with type 2 diabetes who were prescribed a second-line ADM, patients who initiated a sulfonylurea—the most common and least costly option—experienced a $252 increase in mean total adjusted health care costs per quarter. Patients initiating nonsulfonylurea medications experienced relatively greater increases in total costs, reaching as high as an additional $664 per quarter for patients receiving GLP-1 RAs. However, compared with patients initiating sulfonylureas, those initiating the 3 newest ADM classes (DPP-4 inhibitors, GLP-1 RAs, SGLT-2 inhibitors) had relative decreases in nonpharmacy medical costs. These relative decreases in nonpharmacy medical costs (ie, total costs excluding all pharmacy claims) were at least partially attributable to factors observed in secondary analysis, such as relative decreases in office visit costs among patients initiating these newer ADM classes. Patients initiating basal insulin experienced relative increases in quarterly costs excluding the index medication, which were driven by relative increases in costs of nonindex diabetes medications, diabetes supplies, and office visits.

These results demonstrate the higher up-front cost, but also potential value, of nonsulfonylurea medications. The observed relative decreases in nonpharmacy medical costs may conceivably reflect improvements in outcomes observed in prior studies such as reduced cardiovascular disease risk,12 renal protection,16 and weight loss.33 Relative decreases in medical costs might also be attributable to pleiotropic effects of newer ADM classes such as GLP-1 RAs34 and SGLT-2 inhibitors,35 although a longer follow-up period may be required for such impacts to be observed. Cumulatively or individually, these potential benefits of newer ADM classes could contribute to relative reductions in patient morbidity, mortality, and medical costs or—if desirable impacts such as weight loss occur early in the medication course—possibly even outcomes such as increased medication adherence.36

Current recommendations that clinicians and patients collaborate on personalized prescribing decisions provide a strong rationale for this study of real-world data; however, although we are confident that our regression modeling approach was methodologically appropriate, it provides little context on changes in patients’ medication regimens after the index date. Rather, our results (including findings on index medication costs) reflect real-world patterns of clinical practice and medication use, in which many included patients had incomplete refill adherence or discontinued the index medication during follow-up. To aid prescribing decisions going forward, it is important that real-world data be used to identify which medications have the greatest potential benefits, tolerability, and harms in particular patient subgroups. Furthermore, considering both medical and pharmacy costs may prove useful for payers using or considering a pharmacy benefit carve-out design.

Limitations and Strengths

This study has some notable limitations. Despite the cohort’s large size and the consistency of results across sensitivity analyses, we cannot assume that index ADM choice was the sole reason for observed differences during follow-up. Although multivariable regression models adjusted for covariates, we did not have data on some predictors of costs and diabetes-related complications, including duration of diabetes37 or health behaviors such as physical activity. Even after adjustment for morbidity burden, our analyses may not have fully accounted for potential selection bias in groups such as patients receiving basal insulin or those whose index medication was prescribed by a specialist (ie, an endocrinologist). Among patients who were censored at the time of health plan disenrollment, we did not have information on the reason for disenrollment, which could have included events such as change in employment, loss of employment, switching to another commercial insurer, or death. The study cohort was identified and evaluated at a relatively early point in the diabetes disease course (ie, when starting a second-line ADM), and patients may have been at relatively low risk of medication-related complications that can lead to additional costs. Additional research is needed to understand the extent to which nonsulfonylurea medications may reduce costs in higher-risk, higher-cost diabetes populations.

A notable strength of our study is its use of commercial health plan claims, reflecting actual plan costs and patient responsibilities. Nevertheless, our outcome data on medication costs represent contracted prescription drug prices; these prices do not reflect prescription drug rebates negotiated between pharmaceutical manufacturers and pharmacy benefit managers, which ultimately decrease the net cost of brand-name medications.38 As a result, our findings on index medication costs may effectively overstate the net costs of many nonsulfonylurea ADM classes. We were additionally able to link health care claims data with laboratory results for some—but not all—patients, and multivariable regression models adjusted for patients’ most recent A1C results. A sensitivity analysis found no differences in inference when models did not adjust for A1C.


In a large cohort of commercially insured adults with type 2 diabetes who were prescribed second-line ADM augmenting metformin monotherapy, the higher cost of nonsulfonylurea medications was the main driver of relative increases in total health care costs during follow-up. However, there were relative decreases in nonpharmacy medical costs among patients receiving DPP-4 inhibitors, GLP-1 RAs, and SGLT-2 inhibitors, reflecting the potential value and benefits of these newer ADM classes. Additional research is needed to investigate real-world outcomes of different second-line ADMs in high-risk, high-cost diabetes populations and to enable clinicians and patients to make more informed, patient-centered prescribing decisions.


The authors wish to acknowledge the UnitedHealth Group Enterprise Research and Development team for ensuring access to the data. They also thank Cassandra Aikman, MPH, of Northwestern University, and Sanya Pasricha, BS, of UnitedHealth Group, for project management support.

Author Affiliations: Division of General Internal Medicine and Geriatrics (DTL, NL, MJO, AJC, SAH, RTA) and Division of Endocrinology, Metabolism, and Molecular Medicine (AW, RTA), Department of Medicine, Northwestern University Feinberg School of Medicine, Chicago, IL; Institute of Public Health and Medicine, Northwestern University Feinberg School of Medicine (DTL, RHK, MJO, AW, SAH, RTA), Chicago, IL; Applied Research, UnitedHealth Group (EDP), Minnetonka, MN.

Source of Funding: This work and the evaluation team were supported through a grant to Northwestern University from UnitedHealthcare Services.

Author Disclosures: Dr Liss’ work on the manuscript was supported through a grant from UnitedHealthcare Services, which is the insurer that paid the health care claims of the patients included in this study and, as such, has a financial interest in the manuscript’s subject matter. Dr Wallia has received grants from Novo Nordisk and Eli Lilly. Dr Parker is employed by and owns stock in UnitedHealth Group. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (DTL, RHK, MJO, AW, RTA); acquisition of data (RHK, NL, MJO, AJC, EDP, RTA); analysis and interpretation of data (DTL, RHK, NL, MJO, AW, AJC, EDP, RTA); drafting of the manuscript (DTL, RHK, MJO, AJC, SAH, EDP); critical revision of the manuscript for important intellectual content (DTL, RHK, MJO, AW, SAH, EDP, RTA); statistical analysis (DTL, RHK, NL); obtaining funding (RTA); administrative, technical, or logistic support (SAH); and supervision (DTL).

Address Correspondence to: David T. Liss, PhD, Division of General Internal Medicine and Geriatrics, Department of Medicine, Northwestern University Feinberg School of Medicine, 750 N Lake Shore Dr, 10th Floor, Chicago, IL 60611. Email:


1. Gregg EW, Cheng YJ, Srinivasan M, et al. Trends in cause-specific mortality among adults with and without diagnosed diabetes in the USA: an epidemiological analysis of linked national survey and vital statistics data. Lancet. 2018;391(10138):2430-2440. doi:10.1016/S0140-6736(18)30314-3

2. American Diabetes Association. Economic costs of diabetes in the U.S. in 2017. Diabetes Care. 2018;41(5):917-928. doi:10.2337/dci18-0007

3. Squires E, Duber H, Campbell M, et al. Health care spending on diabetes in the U.S., 1996-2013. Diabetes Care. 2018;41(7):1423-1431. doi:10.2337/dc17-1376

4. Hua X, Carvalho N, Tew M, Huang ES, Herman WH, Clarke P. Expenditures and prices of antihyperglycemic medications in the United States: 2002-2013. JAMA. 2016;315(13):1400-1402. doi:10.1001/jama.2016.0126

5. Zhang Y, McCoy RG, Mason JE, Smith SA, Shah ND, Denton BT. Second-line agents for glycemic control for type 2 diabetes: are newer agents better? Diabetes Care. 2014;37(5):1338-1345. doi:10.2337/dc13-1901

6. Qaseem A, Barry MJ, Humphrey LL, Forciea MA; Clinical Guidelines Committee of the American College of Physicians. Oral pharmacologic treatment of type 2 diabetes mellitus: a clinical practice guideline update from the American College of Physicians. Ann Intern Med. 2017;166(4):279-290. doi:10.7326/M16-1860

7. Montvida O, Shaw J, Atherton JJ, Stringer F, Paul SK. Long-term trends in antidiabetes drug usage in the U.S.: real-world evidence in patients newly diagnosed with type 2 diabetes. Diabetes Care. 2018;41(1):69-78. doi:10.2337/dc17-1414

8. Ackermann RT, Wallia A, O’Brien MJ, et al. Correlates of second-line type 2 diabetes medication selection in the USA. BMJ Open Diabetes Res Care. 2017;5(1):e000421. doi:10.1136/bmjdrc-2017-000421

9. American Diabetes Association. 9. Pharmacologic approaches to glycemic treatment: standards of medical care in diabetes—2019. Diabetes Care. 2019;42(suppl 1):S90-S102. doi:10.2337/dc19-S009

10. Heaton PC, Desai VCA, Kelton CML, Rajpathak SN. Sulfonylurea use and the risk of hospital readmission in patients with type 2 diabetes. BMC Endocr Disord. 2016;16:4. doi:10.1186/s12902-016-0084-z

11. Cefalu WT, Buse JB, Del Prato S, et al. Beyond metformin: safety considerations in the decision-making process for selecting a second medication for type 2 diabetes management: reflections from a Diabetes Care editors’ expert forum. Diabetes Care. 2014;37(9):2647-2659. doi:10.2337/dc14-1395

12. Fitchett D, Inzucchi SE, Cannon CP, et al. Empagliflozin reduced mortality and hospitalization for heart failure across the spectrum of cardiovascular risk in the EMPA-REG OUTCOME trial. Circulation. 2019;139(11):1384-1395. doi:10.1161/CIRCULATIONAHA.118.037778

13. Zinman B, Wanner C, Lachin JM, et al; EMPA-REG OUTCOME Investigators. Empagliflozin, cardiovascular outcomes, and mortality in type 2 diabetes. N Engl J Med. 2015;373(22):2117-2128. doi:10.1056/NEJMoa1504720

14. Marso SP, Bain SC, Consoli A, et al; SUSTAIN-6 Investigators. Semaglutide and cardiovascular outcomes in patients with type 2 diabetes. N Engl J Med. 2016;375(19):1834-1844. doi:10.1056/NEJMoa1607141

15. Verma S, Poulter NR, Bhatt DL, et al. Effects of liraglutide on cardiovascular outcomes in patients with type 2 diabetes mellitus with or without history of myocardial infarction or stroke. Circulation. 2018;138(25):2884-2894. doi:10.1161/CIRCULATIONAHA.118.034516

16. Coppolino G, Leporini C, Rivoli L, et al. Exploring the effects of DPP-4 inhibitors on the kidney from the bench to clinical trials. Pharmacol Res. 2018;129:274-294. doi:10.1016/j.phrs.2017.12.001

17. Hong D, Si L, Jiang M, et al. Cost effectiveness of sodium-glucose cotransporter-2 (SGLT2) inhibitors, glucagon-like peptide-1 (GLP-1) receptor agonists, and dipeptidyl peptidase-4 (DPP-4) inhibitors: a systematic review. Pharmacoeconomics. 2019;37(6):777-818. doi:10.1007/s40273-019-00774-9

18. Buse JB, Bigger JT, Byington RP, et al; ACCORD Study Group. Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial: design and methods. Am J Cardiol. 2007;99(12A):21i-33i. doi:10.1016/j.amjcard.2007.03.003

19. Shao H, Fonseca V, Stoecker C, Liu S, Shi L. Novel risk engine for diabetes progression and mortality in USA: Building, Relating, Assessing, and Validating Outcomes (BRAVO). Pharmacoeconomics. 2018;36(9):1125-1134. doi:10.1007/s40273-018-0662-1

20. Clarke PM, Gray AM, Briggs A, et al; UK Prospective Diabetes Study (UKDPS) Group. A model to estimate the lifetime health outcomes of patients with type 2 diabetes: the United Kingdom Prospective Diabetes Study (UKPDS) Outcomes Model (UKPDS no. 68). Diabetologia. 2004;47(10):1747-1759. doi:10.1007/s00125-004-1527-z

21. O’Brien MJ, Karam SL, Wallia A, et al. Association of second-line antidiabetic medications with cardiovascular events among insured adults with type 2 diabetes. JAMA Netw Open. 2018;1(8):e186125. doi:10.1001/jamanetworkopen.2018.6125

22. Nathan DM, Buse JB, Kahn SE, et al; GRADE Study Research Group. Rationale and design of the glycemia reduction approaches in diabetes: a comparative effectiveness study (GRADE). Diabetes Care. 2013;36(8):2254-2261. doi:10.2337/dc13-0356

23. Yoon J, Chow A. Comparing chronic condition rates using ICD-9 and ICD-10 in VA patients FY2014-2016. BMC Health Serv Res. 2017;17(1):572. doi:10.1186/s12913-017-2504-9

24. Glasheen WP, Renda A, Dong Y. Diabetes Complications Severity Index (DCSI)—update and ICD-10 translation. J Diabetes Complications. 2017;31(6):1007-1013. doi:10.1016/j.jdiacomp.2017.02.018

25. 2016 health care cost and utilization report: analytic methodology, 2016V1.0. Health Care Cost Institute Inc. January 23, 2018. Accessed February 14, 2021.

26. Thomas JW, Ward K. Economic profiling of physician specialists: use of outlier treatment and episode attribution rules. Inquiry. 2006;43(3):271-282. doi:10.5034/inquiryjrnl_43.3.271

27. FAQ. Dartmouth Atlas Project. Accessed February 14, 2021.

28. March 31, 2016, HHS-operated risk adjustment methodology meeting: discussion paper. CMS. March 24, 2016. Accessed October 10, 2019.

29. CPI—all urban consumers (current series)—medical care in U.S. city average, all urban consumers, 2011-2017 [time series]. US Bureau of Labor Statistics. 2019. Accessed May 6, 2019.

30. Diggle PJ, Heagerty P, Liang KY, Zeger SL. Analysis of Longitudinal Data. 2nd ed. Oxford University Press; 2002.

31. Lumley T, Diehr P, Emerson S, Chen L. The importance of the normality assumption in large public health data sets. Annu Rev Public Health. 2002;23:151-169. doi:10.1146/annurev.publhealth.23.100901.140546

32. Song Z, Rose S, Safran DG, Landon BE, Day MP, Chernew ME. Changes in health care spending and quality 4 years into global payment. N Engl J Med. 2014;371(18):1704-1714. doi:10.1056/NEJMsa1404026

33. Pratley RE, Aroda VR, Lingvay I, et al; SUSTAIN 7 Investigators. Semaglutide versus dulaglutide once weekly in patients with type 2 diabetes (SUSTAIN 7): a randomised, open-label, phase 3b trial. Lancet Diabetes Endocrinol. 2018;6(4):275-286. doi:10.1016/S2213-8587(18)30024-X

34. Newsome P, Francque S, Harrison S, et al. Effect of semaglutide on liver enzymes and markers of inflammation in subjects with type 2 diabetes and/or obesity. Aliment Pharmacol Ther. 2019;50(2):193-203. doi:10.1111/apt.15316

35. Kosiborod M, Lam CSP, Kohsaka S, et al; CVD-REAL Investigators and Study Group. Cardiovascular events associated with SGLT-2 inhibitors versus other glucose-lowering drugs: the CVD-REAL 2 study. J Am Coll Cardiol. 2018;71(23):2628-2639. doi:10.1016/j.jacc.2018.03.009

36. Grandy S, Fox KM, Hardy E; SHIELD Study Group. Association of weight loss and medication adherence among adults with type 2 diabetes mellitus: SHIELD (Study to Help Improve Early evaluation and management of risk factors Leading to Diabetes). Curr Ther Res Clin Exp. 2013;75:77-82. doi:10.1016/j.curtheres.2013.06.004

37. Nanayakkara N, Ranasinha S, Gadowski A, et al. Age, age at diagnosis and diabetes duration are all associated with vascular complications in type 2 diabetes. J Diabetes Complications. 2018;32(3):279-290. doi:10.1016/j.jdiacomp.2017.11.009

38. Dieguez G, Alston M, Tomicki S. A primer on prescription drug rebates: insights into why rebates are a target for reducing prices. Milliman. May 21, 2018. Accessed February 14, 2021.

Related Videos
Yael Mauer, MD, MPH
Pregnant Patient | image credit: pressmaster -
Diana Isaacs, PharmD
Beau Raymond, MD
Robert Zimmerman, MD
Beau Raymond, MD
Dr Kevin Mallow, PharmD, BCPS, BC-ADM, CDCES
Ian Neeland, MD
Chase D. Hendrickson, MD, MPH
Steven Coca, MD, MS, Icahn School of Medicine, Mount Sinai
Related Content
CH LogoCenter for Biosimilars Logo