Target Population and Intervention Measures

This study’s target population consists of Chinese patients with type 2 diabetes who have not achieved control with metformin monotherapy. The study compared once-weekly subcutaneous semaglutide at doses of 0.5 mg and 1.0 mg to dulaglutide at a dose of 1.5 mg, which is administered once a week.

Long-Term Cost-Effectiveness AnalysisEvaluation Model and Setting

Long-term cost-effectiveness analysis is based on the perspective of the Chinese health system. The Institute of Health Economics Diabetes Cohort Model (IHE-DCM) version 4.4.2 is used to simulate the long-term health outcomes and cost outcomes of patients with type 2 diabetes using semaglutide 0.5 mg and 1.0 mg versus dulaglutide 1.5 mg.

The IHE-DCM model is a cohort-based cost-effectiveness model constructed in Microsoft® Excel 2013 with built-in Visual Basic for Applications (VBA). The cycle length is 1 year, and the time horizon can be set up to 40 years. It is frequently used to evaluate the cost-effectiveness of treatment interventions for type 2 diabetes [12]. The model includes multiple Markov health states that captured microvascular and macrovascular complications related to type 2 diabetes. Microvascular complications include eye diseases (such as retinopathy), lower extremity diseases (such as symptomatic neuropathy), and kidney diseases (such as end-stage renal disease), while macrovascular complications include ischemic heart disease, myocardial infarction, heart failure, and stroke. The model will stimulate the progression of type 2 diabetes and predict mortality using risk equations from the United Kingdom Prospective Diabetes Study (UKPDS) [13]. After input parameters, including baseline characteristics of the cohort, treatment efficacy, costs, and utilities, the model can output main outcomes, including mean survival, life expectancy, QALYs, and direct costs.

In this study, the time horizon was set to 40 years, aiming to estimate cost-effectiveness throughout the patient’s life. The risk equation UPKDS 82 was applied in the model for projections. The discount rate was set at 5%, following the recommendation in the Guidelines for Pharmacoeconomic Evaluation in China [11]. The threshold for incremental analysis was 89,358 Chinese yuan (CNY), equivalent to the China GDP per capita in 2023 [14].

Patient Baseline Characteristics and Clinical Efficacy Parameters

The baseline characteristics and clinical efficacy data for the population in this study were derived from the SUSTAIN 7 study [15]. SUSTAIN 7 was a randomized, open-label, active-controlled phase 3b clinical trial that compared semaglutide with dulaglutide in patients with type 2 diabetes over 40 weeks across 194 sites, including Hong Kong, China. A total of 1201 patients with type 2 diabetes uncontrolled on metformin monotherapy were included in the clinical trial. At baseline, the average age of the patients was 55.67 years, with 44.79% being women. The average duration of diabetes was 7.42 years, the baseline hemoglobin A1c (HbA1c) level was 8.22%, and body mass index (BMI) was 33.5 kg/m2 (Table S1 in the supplementary material).

Clinical efficacy and safety parameters include glycemia, blood pressure, blood lipids, BMI, and adverse events, which are also sourced from the SUSTAIN 7. The changes in HbA1c from baseline in the semaglutide 0.5 mg group, semaglutide 1.0 mg group, and dulaglutide 1.5 mg group were − 1.51%, − 1.78%, and − 1.37%, respectively. The changes in BMI from baseline were − 1.63 kg/m2, − 2.33 kg/m2, and − 1.08 kg/m2, respectively. For safety data, the main adverse events considered in the model are severe hypoglycemic events and non-severe hypoglycemic events. The study also considered the cardiovascular protective effects of semaglutide and dulaglutide, and hazard ratios of macrovascular complications are derived from the SUSTAIN 6 [16] and REWIND [17], respectively. Detailed clinical efficacy data are shown in Table 1.

Table 1 Clinical efficacy parameters and treatment costs

Our analysis assumed that the treatment effect observed in the SUSTAIN-7 clinical trial (based on a 40-week duration) is equivalent to a 1-year effect. This assumption was made because the CEA model used in the study operates on a fixed cycle, and the 40-week duration is relatively close to the 52-week (1-year) period. Therefore, we did not adjust or extrapolate the treatment effect further for simplification and consistency with the model’s structure.

At the start of the simulation, patients in the study were assumed to initiate treatment with semaglutide or dulaglutide. After 1 year of treatment with GLP-1 RAs, it was assumed that all patients would transition to intensive treatment with basal insulin (insulin glargine). This transition was based on a real-world study of GLP-1 RA use in China, which reported a mean treatment duration of 7 months with GLP-1 RAs [18]. For simplicity and to align with the structure of the IHE-DCM, this duration was rounded to 1 year, with the treatment transition assumed to occur at the end of the annual cycle. In the model, the effects of once-weekly semaglutide and dulaglutide ceased after 1 year, and the clinical efficacy of insulin glargine was applied from the second year onward. HbA1c change during the first year was based on the efficacy reported in the clinical trial, after which HbA1c was assumed to increase by 0.14% annually [19].

Cost and Utility Parameters

In the model, direct medical costs associated with anti-hyperglycemic treatments, hypoglycemia, and macrovascular and microvascular complication treatment are considered. The costs of GLP-1 RAs, metformin, insulin, and needles were based on the national median bidding prices of drugs in January 2024. The annual treatment costs of semaglutide 0.5 mg, semaglutide 1.0 mg, and dulaglutide 1.5 mg, including drug and needle costs, were CNY 6956.11, CNY 10,210.98, and CNY 7972.12, respectively. For semaglutide, dose titration was considered.

The costs related to macrovascular and microvascular complications and hypoglycemia in the model are sourced from published literature based on the Chinese type 2 diabetes population (Table S2 in the supplementary material) [20,21,22,23,24,25,26,27]. All price and cost data were adjusted to the 2023 price level using the China Healthcare Consumer Price Index [28].

To simplify the analysis and maintain a focused evaluation, we did not include office visits, hospitalization, or indirect costs, such as productivity losses. The rationale behind this decision was to concentrate on the direct costs related to the treatment and management of type 2 diabetes, which are typically the most significant factors influencing healthcare decisions. Including office visits, hospitalization, and indirect costs would have added complexity to the model and could have diverted attention from the primary costs directly associated with diabetes management.

The utility data for diabetes-related health status are also sourced from published literature (Table S2 in the supplementary material) [10, 13, 29,30,31]. The baseline quality of life utility value is 0.091 (0.008), and the negative utility values for proliferative retinopathy, macular edema, end-stage renal disease, and myocardial infarction events are from a study of the Chinese type 2 diabetes population [24].

Sensitivity Analysis

Sensitivity analyses were conducted to assess the robustness of the base-case results and address uncertainty in the projected long-term health and cost results using short-term data. The following one-way sensitivity analyses were conducted to assess the impact of relevant parameters on the results:

(1)

Time horizon: 30 years;

(2)

Discount rate: 0%, 3%, and 8%;

(3)

HbA1c drift: HbA1c drift upwards at 0.2% per year;

(4)

GLP-1 RAs treatment length: assuming that GLP-1 RAs are used for 2 years before switching to insulin treatment;

(5)

Complication costs: increase or decrease by 10%;

(6)

Health utility value: the BMI negative utility value reported in other studies is − 0.0472 [26];

In addition, probabilistic sensitivity analysis (PSA) was conducted using non-parametric bootstrap technology for 1000 sampling simulation calculations.

Short-Term Cost-Effectiveness Analysis

The short-term cost-effectiveness study was also conducted from the perspective of the Chinese health system. Using data from the SUSTAIN 7 clinical trial, a cost-of-control analysis model was used to compare the relative costs of control required for each successful treatment of a patient to achieve the clinical treatment goals for semaglutide 0.5 mg, 1.0 mg, and dulaglutide 1.5 mg. The time horizon was set as 40 weeks, consistent with the SUSTAIN7 trial. Since the study duration was less than 1 year, there was no need to consider the discount rate for costs and clinical outcomes. The cost of control analysis model was constructed in Excel, and the specific calculation process is as follows:

$$\text\left(\text\right)=\frac},$$

$$\text=\frac\times }_} }\times }_}}.$$

The number needed to treat (NNT) is a clinical-specific indicator, representing the number of patients required to achieve a clinical treatment goal for one patient. It can assist in the measurement of clinical effectiveness, and convert clinical trial results into indicators for clinical practice. The relative cost of control refers to the cost of dulaglutide relative to semaglutide to bring a single patient to a treatment goal (the semaglutide group was set as a reference). When the relative cost of control > 1, the cost required for dulaglutide treatment to achieve a goal for one patient is greater than the cost of semaglutide treatment, indicating semaglutide has better short-term cost-effectiveness than dulaglutide; conversely, the relative cost of control < 1 means that dulaglutide has better short-term cost-effectiveness.

According to the prevention and treatment strategies in the T2DM guidelines, glycemic targets should be individualized and the treatment should be comprehensive, targeting weight control in addition to glycemic [5]. Therefore, six clinical targets were considered in the study, including different levels of glycemic control targets, weight management targets, and composite treatment endpoints: (1) HbA1c ≤ 6.5%; (2) HbA1c < 7.0%; (3) HbA1c < 7.0% without hypoglycemia and no weight gain; (4) weight loss ≥ 5%; (5) weight loss ≥ 10%; (6) reduction in HbA1c ≥ 1.0% and weight loss ≥ 3.0%. The proportions of patients achieving clinical targets with semaglutide and dulaglutide were derived from the SUSTAIN 7. The treatment cost included GLP-1 RA and metformin drug costs, needle costs, and treatment costs for adverse events. Adverse events included nausea, diarrhea, vomiting, decreased appetite, headache, nasopharyngitis, upper respiratory tract infection, constipation, and hypoglycemia (Table S3 in the supplementary material). Incidences of adverse events were derived from the SUSTAIN 7 [15], and the treatment costs for adverse reactions were obtained from the previous literature [20, 32] and expressed in the 2023 Chinese yuan.

For the short-term cost control study, we did not conduct a sensitivity analysis for the following reasons. First, study focus: The primary focus of our research was on the long-term cost-effectiveness analysis. Long-term analyses provide a more comprehensive view of the overall impact of treatment on patient health and economic burden. The short-term cost control analysis was considered supplementary, aiming to highlight cost differences over a shorter duration rather than assessing the robustness of these differences. Given the nature of the short-term data, sensitivity analysis may not have been as critical for this part of the study. Second, data availability and quality: The short-term cost control analysis relied on data over a relatively short time frame. These data may not have been sufficiently stable or varied enough to warrant a meaningful sensitivity analysis. The limited variability in the data made the results less prone to significant changes under different assumptions, reducing the need for further robustness testing. Third, assumptions and model limitations: The short-term cost control analysis was based on a set of assumptions that may not hold over the longer term or might need to be adjusted. We felt that adding a sensitivity analysis in the short-term context could introduce excessive uncertainty, potentially complicating the interpretation of results without providing substantial additional insights. Fourth, robustness of findings: The short-term cost control findings were straightforward and demonstrated clear, significant differences. Given the clarity and immediate implications of these results, we felt that a sensitivity analysis would not substantially alter the conclusions or improve the interpretation of the short-term outcomes.

Ethical Approval

This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.