Oral semaglutide versus injectable glucagon-like peptide-1 receptor agonists: a cost of control analysis

Abstract

Aims: The efficacy and safety of oral semaglutide, the first glucagon-like peptide-1 (GLP-1) receptor agonist developed for oral administration for the treatment of type 2 diabetes, was evaluated in the PIONEER clinical trial program, and a recently published network meta-analysis allowed comparison with further injectable GLP-1 receptor agonists. The present study aimed to assess the short-term cost- effectiveness of oral semaglutide 14 mg versus subcutaneous once-weekly dulaglutide 1.5 mg, once-weekly exenatide 2 mg, twice-daily exenatide 10 µg, once-daily liraglutide 1.8 mg, once-daily lixisenatide 20 µg, and once-weekly semaglutide 1 mg, in terms of the cost per patient achieving glycated hemoglobin (HbA1c) targets (cost of control).

Materials and methods: Cost of control was calculated by dividing the annual treatment costs associated with an intervention by the proportion of patients achieving the treatment target with an intervention, with outcomes calculated for targets of HbA1c ≤6.5% and HbA1c <7.0% for all included GLP-1 receptor agonists. Annual treatment costs were accounted in 2019 United States dollars (USD), based on 2019 wholesale acquisition cost.

Results: For the treatment target of HbA1c ≤6.5%, once-weekly semaglutide 1 mg and oral semaglutide 14 mg were associated with the lowest costs of control, at USD 15,430 and USD 17,383 per patient achieving target, respectively. Similarly, the cost of control was lowest with once-weekly semaglutide 1 mg at USD 12,627 per patient achieving target, followed by oral semaglutide 14 mg at USD 13,493 per patient achieving target for the target of HbA1c <7.0%. All other interventions were associated with higher cost of control values for both targets.

Conclusions: Oral semaglutide 14 mg is likely to be cost-effective versus dulaglutide, exenatide (once weekly and twice daily), liraglutide, and lixisenatide in terms of bringing people with type 2 diabetes to glycemic control targets of HbA1c ≤6.5% and HbA1c <7.0% in the US.

Keywords:

• Costs and cost analysis
• cost-effectiveness
• cost of control
• diabetes mellitus
• GLP-1 receptor agonist
• oral semaglutide
• United States

JEL CLASSIFICATION CODES:

• D61
• D60
• I19

Introduction

Clinicians have a range of treatment options available to treat people with type 2 diabetes not achieving glycemic control on metformin monotherapy, including modern interventions such as glucagon-like peptide-1 (GLP-1) receptor agonists, sodium-glucose cotransporter-2 (SGLT-2) inhibitors, dipeptidyl peptidase-4 (DPP-4) inhibitors, and older interventions such as thiazolidinediones and sulfonylureas¹. GLP-1 receptor agonists are recommended when there is a need to minimize the risk of hypoglycemia, and when there is a need to minimize weight gain or promote weight loss¹. Furthermore, GLP-1 receptor agonists with a proven positive impact on the risk of cardiovascular disease are preferred for treatment of patients at high risk of cardiovascular disease¹.

Within the GLP-1 receptor agonists class, options include twice daily (BID), once daily (QD), and once weekly (QW) interventions. Exenatide BID was the first GLP-1 receptor agonist to receive Food and Drug Administration (FDA) approval in 2005. Since then, liraglutide (approved in 2010) and lixisenatide (approved in 2016) have become available as once-daily treatment options, and a long-acting formulation of exenatide (approved in 2012), dulaglutide (approved in 2014), and semaglutide (approved in 2017) have become available as once-weekly treatment options (albiglutide once weekly was also approved in 2014, but has since been withdrawn by the manufacturer). In September 2019, oral semaglutide became the first orally administered GLP-1 receptor agonist to be approved by the FDA². Oral semaglutide uses an absorption enhancer, sodium N-(8-[2-hydroxybenzoyl] amino) caprylate, to facilitate absorption across the gastric mucosa, and aims to provide the benefits of existing GLP-1 receptor agonists without the requirement for daily or weekly injection³. Patients are required to take oral semaglutide at least 30 min before the first food, beverage, or other oral medications of the day with no more than 4 ounces of water only. Waiting less than the full 30 min reduces the effect of oral semaglutide. The dosing conditions of oral semaglutide have been described in more detail in previous publications and the prescribing information^4–12.

The efficacy and safety of oral semaglutide was assessed in the PIONEER clinical trial program, which was comprised of a series of global phase 3 studies, comparing oral semaglutide (3, 7, or 14 mg) with placebo, empagliflozin 25 mg, sitagliptin 100 mg, and liraglutide 1.8 mg, with a further study to establish cardiovascular safety^4–11,13. An additional two phase 3 studies comparing oral semaglutide with liraglutide 0.9 mg and dulaglutide 0.75 mg were conducted in Japan^14,15. To allow comparison of the efficacy of oral semaglutide with other GLP-1 receptor agonists not included in the PIONEER trial program, a network meta-analysis (NMA) has been conducted¹⁶. This found that oral semaglutide 14 mg was associated with significantly greater reductions in glycated hemoglobin (HbA1c) than dulaglutide 0.75 mg QW, exenatide 2 mg QW, exenatide 10 µg BID, liraglutide 1.2 mg QD, and lixisenatide 20 µg QD. Oral semaglutide 14 mg was associated with numerically greater (but not statistically significant) reductions in HbA1c than dulaglutide 1.5 mg QW, liraglutide 1.8 mg QD, and semaglutide 0.5 mg QW. Only semaglutide 1 mg QW was associated with a numerically greater reduction in HbA1c than oral semaglutide 14 mg, with this difference found not to be statistically significant. When change from baseline in bodyweight was analyzed, oral semaglutide was associated with significantly greater reductions than all included GLP-1 receptor agonists, with the exception of semaglutide 0.5 mg QW and semaglutide 1 mg QW, where no significant difference was observed.

Choosing treatment regimens that are both effective and cost-effective is becoming increasingly important, as the prevalence of type 2 diabetes continues to grow, particularly in the United States (US)¹⁷. Improving glycemic control in people with type 2 diabetes has been shown to reduce the incidence of micro- and macrovascular diabetes-related complications, and therefore the American Diabetes Association (ADA) and the American Association of Clinical Endocrinologists (AACE) have proposed HbA1c targets that should be aimed for as part of routine care. There is increasing acknowledgement that the glycemic control target should be individualized depending on the risk of adverse effects of treatment (such as hypoglycemia), disease duration, life expectancy, comorbidities, available support, and patient preferences including consideration of tolerability and dosing conditions, but the ADA recommend an HbA1c target of <7.0% for the majority of people with type 2 diabetes, with a more stringent target of HbA1c <6.5% if this can be achieved without significant hypoglycemia or other adverse effects of treatment^1,18. The AACE suggest an HbA1c target of ≤6.5% for most adults with type 2 diabetes, if this can be achieved safely¹⁹. Data from the National Health and Nutrition Examination Survey (NHANES) in the US show that only 50.9% of people with diabetes had HbA1c levels lower than 7.0% in the 2011–2014 cohort (N = 1,325)²⁰. Failure to intensify treatment despite not achieving glycemic control targets may be due to therapeutic inertia, increasing the time patients spend in sub-optimal glycemic control, and therefore increasing the risk of diabetes-related complications²¹. Therefore, there is a significant opportunity to improve care for people with type 2 diabetes, and thereby reduce the clinical and economic burden of the disease. Achieving improvements in glycemic control in a cost-effective manner is crucial to the future management of type 2 diabetes.

The aim of the present analysis was to compare the short-term cost-effectiveness of oral semaglutide 14 mg versus injectable GLP-1 receptor agonists, in terms of the cost per patient achieving HbA1c targets from a healthcare payer perspective in the US.

Methods

Evaluation of the cost of control

Cost of control evaluates the short-term cost-effectiveness of interventions in a simple and transparent manner, allowing the value for money of each intervention to be assessed. In the present analysis, cost of control was calculated by dividing the annual treatment costs associated with an intervention by the proportion of patients achieving the treatment target with an intervention. The methodology used was consistent with a previously published cost of control analysis of oral semaglutide, as well as a number of other cost of control analyses published in the peer-reviewed literature^22–26. The cost of control calculation was carried out in a cost of control model, programmed in Microsoft Excel (Microsoft Corporation, Redmond, WA, US). Outcomes were not projected beyond a 1-year time horizon, and therefore discounting of cost and clinical outcomes was not required. Cost of control was evaluated for two treatment targets: HbA1c ≤6.5% and HbA1c <7.0%. A total of seven GLP-1 receptor agonists were included in the evaluation: oral semaglutide 14 mg QD, dulaglutide 1.5 mg QW, exenatide 2 mg QW, exenatide 10 µg BID, liraglutide 1.8 mg QD, lixisenatide 20 µg QD, and semaglutide 1 mg QW. Cost of control analyses require data on the clinical efficacy of interventions in terms of the proportion of patients who achieve treatment targets, and data on the costs of interventions, the sources of which are outlined below.

Proportions of patients achieving targets

Data on the proportions of patients achieving the two treatment targets of HbA1c ≤6.5% and HbA1c <7.0% were taken from an NMA (Table 1)¹⁶. The NMA was based on a systematic literature review, which identified 108 publications reporting on 71 trials in people with type 2 diabetes. A total of 25 studies were taken forward into the network used to assess the proportion of patients achieving a target of HbA1c ≤6.5%, and 26 studies were used in the network to evaluate the proportion of patients achieving a target of HbA1c <7.0%. Both outcomes were assessed based on 26-week trial data (±4 weeks). The NMA used WinBUGS software (MRC Biostatistics Unit, Cambridge, UK) and employed a Bayesian framework with the use of uninformative prior distributions. Three Markov Monte Carlo chains were used, starting from different initial values of selected unknown parameters with a burn-in of 50,000 iterations. Convergence for all models was assessed by analyzing history and density plots, and Brooks–Gelman–Rubin diagnostic plots. In addition, autocorrelation plots were assessed to detect the presence of autocorrelation in the chains. Following this, model convergence inferences were made from data obtained by sampling for a further 10,000 iterations on the three chains. Dichotomous outcomes were evaluated using a binomial likelihood, logit link model, with a random-effects model used¹⁶.

Table 1. Estimated proportion of patients achieving treatment targets.

	HbA1c ≤ 6.5%	HbA1c < 7.0%
Oral semaglutide 14 mg QD	54.1% (37.9–69.9%)	69.7% (55.6–81.3%)
Dulaglutide 1.5 mg QW	49.0% (35.2–63.5%)	64.7% (51.9–75.9%)
Exenatide 2 mg QW	42.0% (27.8–57.1%)	60.7% (47.6–72.2%)
Exenatide 10 µg BID	29.2% (18.3–43.7%)	44.7% (31.3–59.2%)
Liraglutide 1.8 mg QD	45.8% (33.9–58.3%)	64.6% (53.5–74.4%)
Lixisenatide 20 µg QD	23.5% (15.9–33.7%)	41.1% (31.0–51.9%)
Semaglutide 1 mg QW	65.3% (50.4–77.5%)	79.8% (69.3–87.4%)

Values are estimated medians (lower and upper 95% credible interval)¹⁶. Abbreviations. BID, twice daily; HbA1c, glycated hemoglobin; QD, once daily; QW, once weekly.

Costs of interventions

The costs of treatment with oral semaglutide 14 mg, dulaglutide 1.5 mg, exenatide 2 mg, exenatide 10 µg, liraglutide 1.8 mg, lixisenatide 20 µg, and semaglutide 1 mg were accounted over a 1-year time horizon, based on wholesale acquisition cost (WAC) (Table 2)²⁷. Needles are not required for administration of oral semaglutide, and are included in the packs of the once-weekly GLP-1 receptor agonists (dulaglutide 1.5 mg, exenatide 2 mg, and semaglutide 1 mg). However, there is a requirement to purchase needles separately for injection of once daily (liraglutide 1.8 mg and lixisenatide 20 µg) and twice daily (exenatide 10 µg) GLP-1 receptor agonists. Costs relating to self-monitoring of blood glucose were not included in the analysis, as GLP-1 receptor agonists are associated with a low risk of hypoglycemia and resource use would be unlikely to differ between interventions.

Table 2. Wholesale acquisition cost applied in the analysis.

Medication	Pack contents	Days per pack	Pack price (USD)
Oral semaglutide 14 mg QD	420.0 mg	30	772.43
Dulaglutide 1.5 mg QW	6.0 mg	28	759.40
Exenatide 2 mg QW	8.0 mg	28	699.77
Exenatide 10 µg BID	600.0 mg	30	729.63
Liraglutide 1.8 mg QD	54.0 mg	30	921.78
Lixisenatide 20 µg QD	560.0 mg	28	620.18
Semaglutide 1 mg QW	4.0 mg	28	772.43
Needles for injection of liraglutide (NovoFine 30G)	100 needles	–	39.41
Needles for injection of lixisenatide 20 µg QD and exenatide 10 µg BID (BD MicroFine+™ 5mm/31G)	500 needles	–	123.25

Costs taken from the IBM Micromedex. RED BOOK in September 2019²⁷. Abbreviations. BID, twice daily; QD, once daily; QW, once weekly; USD, 2019 United States dollars.

Probabilistic sensitivity analysis

To evaluate the uncertainty around the cost of control outcomes, a probabilistic sensitivity analysis (PSA) was performed. The proportion of patients achieving the target endpoints of HbA1c ≤6.5% and HbA1c <7.0% with each intervention was sampled based on the 95% credible intervals identified in the NMA (Table 1). The cost of control with each intervention was calculated based on the sampled values, with this process repeated 1,000 times (as results were stable at this number of model iterations). The mean and standard deviation cost of control across the 1,000 sampled values was then calculated for each intervention.

Results

Annual cost of treatment

Annual pharmacy costs ranged from United States dollars (USD) 8,090 with lixisenatide 20 µg, to USD 11,223 with liraglutide 1.8 g, a difference of USD 3,133 between the most and least costly interventions (Figure 1). Exenatide 10 µg, liraglutide 1.8 mg, and lixisenatide 20 µg were also associated with needle costs, as these are not included with these interventions. However, these costs were relatively low, ranging from USD 90 to USD 180 per year depending on the frequency of injection. In total, annual costs of treatment (the sum of pharmacy and any needles costs) associated with each treatment ranged from USD 8,180 with lixisenatide 20 µg, to USD 11,367 with liraglutide 1.8 mg, with oral semaglutide associated with an annual cost of USD 9,404 (Figure 1).

Figure 1. Annual cost of treatment. Abbreviations. BID, twice daily; QD, once daily; QW, once weekly; USD, 2019 United States dollars.

Number needed to treat

Once-weekly semaglutide 1 mg (subcutaneous) was associated with the lowest number needed to treat to bring one patient to a target of HbA1c ≤6.5%, with 1.53 patients requiring treatment to bring one patient to target (Figure 2). Oral semaglutide 14 mg was the second most effective intervention, with 1.85 patients requiring treatment to bring one patient to target. Lixisenatide 20 µg was the least effective treatment, with 4.26 patients requiring treatment to bring one patient to target.

Figure 2. Number needed to treat to bring one patient to target. Abbreviations. BID, twice daily; HbA1c, glycated hemoglobin; QD, once daily; QW, once weekly.

For the target of HbA1c <7.0%, a similar pattern was observed (Figure 2), with once-weekly semaglutide 1 mg found to be the most effective treatment (1.25 patients requiring treatment to bring one patient to target), followed by oral semaglutide 14 mg (1.43 patients requiring treatment to bring one patient to target). Lixisenatide 20 µg remained the least effective treatment (2.43 patients requiring treatment to bring one patient to target).

Cost of control

For the treatment target of HbA1c ≤6.5%, once-weekly semaglutide 1 mg was associated with the lowest cost of control, at USD 15,430 per patient achieving target, followed by oral semaglutide 14 mg at USD 17,383 per patient achieving target (Figure 3). The cost of control was the highest with lixisenatide 20 µg, with a cost of USD 34,809 per patient achieving a target of HbA1c ≤6.5%. Cost of control values for dulaglutide 1.5 mg, exenatide 2 mg, liraglutide 1.8 mg, and exenatide 10 µg ranged from USD 20,217 to USD 21,039 per patient achieving target.

Figure 3. Cost of control. Abbreviations. BID, twice daily; HbA1c, glycated hemoglobin; QD, once daily; QW, once weekly; USD, 2019 United States dollars.

Cost of control values for the treatment target of HbA1c <7.0% were lower for all interventions than when the treatment target of HbA1c ≤6.5% was considered, driven by the greater proportion of patients achieving the less strict glycemic control target. The cost of control was lowest with once-weekly semaglutide 1 mg at USD 12,627 per patient achieving target, and oral semaglutide 14 mg was associated with the second lowest cost of control, at USD 13,493 per patient achieving target (Figure 3). The intervention with the highest cost of control was found to be exenatide 10 µg, with a cost of USD 20,276 per patient achieving target.

Probabilistic sensitivity analysis

Results produced using a probabilistic approach were consistent with the deterministic approach used in the base case analysis. In the PSA, for the analysis based on a treatment target of HbA1c ≤6.5%, once-weekly semaglutide 1 mg was associated with the lowest cost of control, with a mean (standard deviation) value of USD 15,651 (USD 1,726) per patient achieving target, followed by oral semaglutide 14 mg at USD 17,791 (USD 2,820) per patient achieving target (Table 3). All other interventions were associated with a higher cost of control, with lixisenatide 20 µg associated with the highest cost of control at USD 36,552 (USD 8,293) per patient achieving target.

Table 3. Results of the probabilistic sensitivity analysis.

	Cost of control (USD)
	HbA1c ≤ 6.5%	HbA1c < 7.0%
Oral semaglutide 14 mg QD	17,791 (2,820)	13,587 (1,333)
Dulaglutide 1.5 mg QW	20,727 (3,165)	15,490 (1,466)
Exenatide 2 mg QW	22,621 (4,897)	15,161 (1,624)
Exenatide 10 µg BID	33,065 (10,425)	20,695 (3,590)
Liraglutide 1.8 mg QD	25,468 (3,653)	17,733 (1,455)
Lixisenatide 20 µg QD	36,552 (8,293)	20,232 (2,640)
Semaglutide 1 mg QW	15,651 (1,726)	12,653 (707)

Values are mean (standard deviation). Abbreviations. BID, twice daily; HbA1c, glycated hemoglobin; QD, once daily; QW, once weekly; USD, 2019 United States dollars.

Once-weekly semaglutide 1 mg and oral semaglutide 14 mg were also associated with the lowest cost of control values for the endpoint of HbA1c <7.0%, at USD 12,653 (USD 707) and USD 13,587 (USD 1,333) per patient achieving target, respectively. As in the base case analysis, exenatide 10 µg was found to have the highest cost of control, at USD 20,695 (USD 3,590) per patient achieving target.

Discussion

The present study compared the short-term cost-effectiveness of a number of GLP-1 receptor agonists based on a published NMA, and found that once-weekly semaglutide 1 mg and oral semaglutide 14 mg were associated with the lowest cost per patient achieving treatment targets of HbA1c ≤6.5% and HbA1c <7.0%, and therefore these interventions are likely to be the most cost-effective GLP-1 receptor agonists in the US in terms of bringing patients to glycemic control targets. A key strength of this cost of control study is that these glycemic control targets are recommended in guidelines for the treatment of type 2 diabetes released by the ADA and the AACE, and therefore the analysis is highly applicable to clinical practice in the US^1,18,19.

A further strength of the study is the transparency of the cost-effectiveness analysis, which can be easily replicated by other research groups based on the information provided in Tables 1 and 2. Furthermore, the analysis can be easily updated should new clinical data become available or if costs change. Conventional, long-term modeling of the cost-effectiveness of interventions for type 2 diabetes involves projection of short-term trial data over patient lifetimes based on published risk equations, and there is always uncertainty around how well risk equations represent the cohort of interest²⁸. The present analysis avoids this uncertainty by evaluating cost-effectiveness over a 1-year period in natural units (i.e. cost per patient achieving target). However, a key strength of long-term modeling is that projections of quality-adjusted life expectancy (measured in quality-adjusted life years [QALYs]) and costs are made. Interventions can be compared, with calculation of an incremental cost-effectiveness ratio (ICER) expressed in terms of the cost per QALY gained with the new intervention compared with the existing intervention. ICERs can then be compared against willingness to pay thresholds to evaluate whether an intervention represents good value for money, and can be compared across therapeutic areas with the aim of maximizing healthcare gains across a population within a constrained healthcare budget. The use of natural units in the present analysis limits the comparability of the findings, as no defined willingness to pay threshold per patient achieving glycemic control target have been published and, as such, the present analysis is intended to provide additional, complementary information to conventional, long-term analyses, rather than replace them.

Cost-effectiveness analyses are only as robust as the clinical data used to inform them, and therefore the strengths and weaknesses of the previously published NMA must be considered¹⁶. The NMA was conducted using robust methodology, in line with guidance released by the National Institute for Health and Care Excellence, the International Society of Pharmacoeconomics and Outcomes Research, and the Cochrane Collaboration^29–33. The networks for each endpoint included a large number of trials, across which the risk of bias was considered to be low. However, pooling of data across multiple studies is always associated with some limitations, including heterogeneity in patient populations enrolled in the randomized controlled trials, heterogeneity in study design across the randomized controlled trials, and that publication bias may result in some studies of GLP-1 receptor agonists not being published (and, therefore, these unpublished studies cannot be captured in the systematic literature review). Furthermore, the authors of the NMA did not identify potential effect modifiers for the endpoints used in the present analysis (HbA1c <7% and HbA1c ≤6.5%), and this represents a potential weakness of the analysis. A patient-level meta-analysis would provide a robust and useful data source for future cost of control analyses evaluating oral semaglutide, with similar analyses conducted to assess the impact of liraglutide and insulin glargine on glycemic control and safety outcomes^34–36. However, the patient-level data required for such an analysis are not currently publicly available, and therefore this is not possible at the present time. While the NMA used in the present analysis has a number of limitations, it is the best available data source to inform the present analysis, expanding the number of comparisons that can be made beyond head-to-head trials.

The present analysis focused on glycemic control targets, but treatment guidelines recommend that treatment of type 2 diabetes should take a patient-centered approach, with the impact on other risk factors for diabetes-related complications taken into account when selecting treatments; particularly the risk of hypoglycemia and the impact of medications on bodyweight^1,18,19,37. While all GLP-1 receptor agonists are associated with a low risk of hypoglycemic events, efficacy in terms of weight loss varies. Cost of control analyses with endpoints based on weight loss or composite endpoints capturing both a glycemic control target and a weight loss target would be of further interest, but these were not possible in the present analysis, as these endpoints could not be evaluated in the published NMA¹⁶. The most common side effects with GLP-1 receptor agonists, including oral semaglutide, are gastrointestinal adverse events. The NMA found no statistically significant differences in the odds of experiencing a gastrointestinal adverse event across the GLP-1 receptor agonists. Achieving composite endpoints of glycemic control targets without experiencing a gastrointestinal adverse event could not be calculated in the NMA due to limitations in the published data, but these would be of interest in a future analysis, if such data become available.

A previously published cost of control analysis based on the PIONEER trial program included composite endpoints of ≥1.0%-point HbA1c reduction and weight loss ≥3.0%, and HbA1c <7.0% without hypoglycemia and without weight gain, as well as the glycemic control targets of HbA1c ≤6.5% and HbA1c <7.0% used in the present analysis²². This study compared oral semaglutide 14 mg with empagliflozin 25 mg, sitagliptin 100 mg, and liraglutide 1.8 mg, and found that oral semaglutide 14 mg was associated with a lower cost of control for all endpoints versus all comparators. Therefore the previously published analysis and the present analysis concur that oral semaglutide 14 mg is associated with a lower cost of control for treatment targets of HbA1c ≤6.5% and HbA1c <7.0% than liraglutide 1.8 mg. Expansion of the present analysis to include further endpoints when additional clinical data comparing oral semaglutide with other GLP-1 receptor agonists are available would provide further information for clinicians and healthcare payers.

Previous studies have shown that there is a preference for avoidance of injections (both daily and weekly)^38,39. Oral administration may remove a barrier to treatment intensification, with fear of injection often cited as a reason for therapeutic inertia^40–44. For some patients, oral semaglutide may help overcome therapeutic inertia and therefore be associated with further benefits not captured in the present analysis in real-world clinical practice. While oral semaglutide avoids the requirement for daily or weekly injections, the required dosing conditions also need to be considered. Furthermore, some patients may have a preference for a weekly injectable medication over a daily oral medication. The optimal medication for each patient needs to be determined using a patient-centric, holistic approach as recommended by the ADA, through discussion with a physician, where a number of factors including preferences around dosing schedule and mode of administration (including dosing conditions) are considered¹.

The present economic evaluation used clinical trial data (through an NMA) and unit costs of medications as inputs, and an inherent limitation of this approach is that it does not take into consideration the real-world use of the drugs or the impact of adherence of any of the drugs in this study. As with all medications, it will be important to assess adherence differences between the comparators, as the differences in modes of administration and dosing conditions may introduce variability in adherence and persistence in the real-world. To date, there are no real-world studies assessing adherence rates with oral semaglutide and the injectable GLP-1 receptor agonists, or the impact that nonadherence has on outcomes, and, therefore, it could not be included in the study. It will be important to assess adherence to oral semaglutide outside of a clinical trial setting, and comparing adherence with once-daily oral semaglutide, once-daily injectable GLP-1 receptor agonists and once-weekly GLP-1 receptor agonists represents a key aspect for future studies in order to assess the impact on both effectiveness and cost-effectiveness. Furthermore, capturing adherence within economic evaluations is difficult, as both costs and clinical outcomes are affected, but the extent of this is often unclear. Further real-world studies are required to elucidate adherence patterns and the impact of these on cost and clinical outcomes with GLP-1 receptor agonists, before these can be incorporated into cost-effectiveness analyses.

The use of WAC price for the drug costs may also represent a limitation of the analysis, as these prices may not reflect the costs borne by the healthcare payer due to privately-negotiated discounts, rebates, and patient co-pay. However, these factors differ from medication to medication and from payer to payer, and are confidential. Therefore these could not be included in the analysis, and use of WAC represents the best-available approach.

Conclusions

Oral semaglutide 14 mg is likely to be cost-effective versus dulaglutide, exenatide (once weekly and twice daily), liraglutide, and lixisenatide in terms of bringing people with type 2 diabetes to glycemic control targets of HbA1c ≤6.5% and HbA1c <7.0% in the US.

Transparency

Declaration of funding

This study was supported by funding from Novo Nordisk A/S.

Declaration of financial/other interests

Barnaby Hunt, Samuel Malkin, and William Valentine are employees of Ossian Health Economics and Communications, which received consulting fees from Novo Nordisk A/S to support preparation of the analysis. Brian Bekker Hansen and Solomon Nuhoho are employees of Novo Nordisk A/S. Sarah Naz Ali and Tam Dang-Tan are employees of Novo Nordisk Inc. Brian Bekker Hansen and Tam Dang-Tan are shareholders in Novo Nordisk.

JME peer reviewers on this manuscript have received an honorarium from JME for their review work, but have no other relevant financial relationships to disclose.

Author contributions

The study was conceived by all authors, and conducted by Barnaby Hunt. The paper was drafted by Barnaby Hunt and reviewed by Brian Bekker Hansen, Solomon Nuhoho, Sarah Naz Ali, Tam Dang-Tan T, William Valentine, and Samuel Malkin. All authors agree to be accountable for all aspects of the work, and have given their approval for this version to be published.

Previous presentations

The analysis presented in this manuscript has not been presented previously.

Data availability statement

All data used in the analysis are included in the manuscript.

Acknowledgements

No assistance in the preparation of this article is to be declared.