A cost-utility analysis of adding SGLT2 inhibitors for the management of type 2 diabetes with chronic kidney disease in Thailand

Abstract

Chronic kidney disease (CKD) in type 2 diabetes (T2D) patients is associated with end-stage renal disease and significant economic burden. While sodium glucose cotransporter-2 inhibitors (SGLT2i) show renal benefits in randomized controlled trials (RCTs), their cost-effectiveness in Thailand remains unclear. This study evaluates the cost-utility of adding SGLT2i (dapagliflozin, empagliflozin, and canagliflozin) to standard of care therapy (SoCT) for T2D patients with CKD in Thailand. A lifetime Markov model assessed economic and clinical outcomes. Data were derived from Thai studies, RCT subgroup analyses, and patient interviews. Sensitivity analysis was performed. Adding SGLT2i increased life expectancy (0.42-0.52 years) and QALYs (3.83- 3.91 vs. 3.50 with SoCT alone), but also increased lifetime costs ($1,275–$1,903). Empagliflozin was cost-effective at a WTP threshold of $4,336 per QALY ($3,386/QALY), while dapagliflozin ($5,783/QALY) and canagliflozin ($4,591/QALY) required price reductions. SGLT2i showed potential cost savings for dialysis and kidney transplantation compared to SoCT alone. Adding SGLT2i to SoCT for T2D and CKD patients increases costs but provides significant clinical benefits. Empagliflozin is cost-effective at a WTP threshold of $4,336/QALY, while dapagliflozin and canagliflozin require price reductions to be cost-effective. However, the analysis solely focuses on renal benefits, excluding other advantages like cardiovascular and heart failure protection.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-024-81747-7.

Introduction

Up to 40% of diabetes patients develop chronic kidney disease (CKD) and the incidence of CKD in type 2 diabetes (T2D) patients increased by 74% from 1990 to 2017¹. The global prevalence of CKD among T2D patients is estimated to be around 27%². Patients with diabetic kidney disease (DKD) face an elevated risk of severe renal and cardiovascular morbidity and mortality³, and it is a leading cause of end-stage renal disease (ESRD) requiring dialysis or transplantation. In Thailand, diabetes is the most significant cause of dialysis, with more than one-third of renal replacement therapy (RRT) for CKD attributed to diabetes⁴. The prevalence of RRT is expected to more than double, reaching 5.4 million people by 2030, with the greatest increase occurring in Asia⁵. DKD imposes a significant economic burden on healthcare systems because of the high costs associated with dialysis and kidney transplantation (KT). Although they represent a small fraction of the global population, patients undergoing dialysis typically contribute to 2–4% of national healthcare expenditures. This underscores the significant financial strain on healthcare systems caused by DKD⁶.

Sodium glucose cotransporter-2 inhibitors (SGLT2i) have demonstrated a powerful nephroprotective effect in clinical studies, independent of their glucose-lowering, weight-reducing, and blood pressure-lowering effects^7,8. Several clinical trials have shown that adding SGLT2i to standard of care therapy (SoCT) can significantly delay the progression to macroalbuminuria, slow the decline of glomerular filtration rate (GFR), ESRD, and reduce the need for RRT or KT in T2D patients^9–14. As a result, SGLT2i has become a crucial treatment for halting the progression of CKD. However, SGLT2i are costly and have not been included in the reimbursement schemes of many countries, including Thailand. To incorporate new medications into Thailand’s National List of Essential Medicines (NLEM), thorough cost-utility analyses of all accessible medications must be conducted. While SGLT2i share a common mechanism of action, they differ in their selectivity for the SGLT2 transporter relative to SGLT1. Empagliflozin exhibits the highest selectivity for SGLT2, with a selectivity ratio significantly greater than that of dapagliflozin and canagliflozin¹⁵. This superior selectivity for SGLT2 potentially allows empagliflozin to more effectively inhibit renal glucose reabsorption while sparing SGLT1. In contrast, dapagliflozin and canagliflozin, with lower SGLT2 selectivity, may exhibit a broader inhibition profile, potentially influencing their therapeutic effects and safety profiles¹⁵. Therefore, it is imperative to conduct economic evaluations for all available SGLT2i. Currently, no comprehensive economic evaluation has been performed in Thailand for all SGLT2i in the treatment of T2D with CKD (T2D-CKD) patients. This study aims to comprehensively compare the cost-utility of all SGLT2i with CKD-proven benefits available in Thailand (dapagliflozin, empagliflozin, and canagliflozin) when added to the SoCT versus using SoC alone for T2D-CKD patients. The analysis focuses exclusively on the renal benefits of SGLT2i, excluding other potential advantages such as cardiovascular protection and heart failure risk reduction. The findings will provide critical insights to policymakers regarding their value-for-money and their potential inclusion in Thailand’s NLEM for the treatment of T2D-CKD patients. Moreover, the results would be relevant to other middle-income countries facing similar healthcare challenges, providing valuable guidance for their own healthcare policy and economic evaluations.

Methods

This study employed an economic model to compare the cost-utility of adding SGLT2i to SoCT compared with SoCT alone for T2D patients with CKD using societal perspective to comply to Thai health technology assessment (HTA) guideline recommendations¹⁶.

Cohort population

This study included patients diagnosed with CKD according to the KDIGO 2012 guidelines¹⁷. The diagnosis required two abnormal results of a GFR of less than 60 ml/min/1.73 m² and/or albuminuria. Patients with a previous CKD diagnosis were also included. Patients entered the model at aged 60 years and above for all SGLT2i, based on a previous study in Thailand that indicated this is the average age of T2D patients with CKD¹⁸. Additionally, the initial distribution among CKD stages 3–5 was set uniformly across all three SGLT2i, ensuring that each treatment group began with comparable baseline CKD stage proportions.

Intervention and comparator

Based on their demonstrated efficacy in RCTs^9–11, low doses of all available SGLT2i in Thailand, including dapagliflozin 10 mg, empagliflozin 10 mg and canagliflozin 100 mg, were evaluated. These were compared with SoCT alone, which include lifestyle modification, optimum control of blood pressure and lipid as well as optimum dose of angiotensin-converting enzyme inhibitor (ACEI) or an angiotensin receptor blocker (ARB).

Model structure

The life time Markov model was constructed to simulate the progression of CKD based on a previous study^9–11 and clinical practice guideline recommendations^4,19. The model comprises five states: CKD stages 3, 4 (patients with a GFR of 15–59 ml/min/1.73 m²), CKD stage 5 (patients with a GFR of less than 15 ml/min/1.73 m²), kidney transplantation (patients undergoing RRT via KT), dialysis (patients undergoing RRT via peritoneal dialysis or hemodialysis), and death, as depicted in Fig. 1. The arrows represent the annual probabilities of transitioning between these mutually exclusive health states, based on the cycle length. Patients can enter the model in either CKD stage 3, 4 or CKD stage 5. Patients with CKD stage 3,4 may remain in the same stage or progress to CKD stage 5. At CKD stage 5, patients may progress to receive RRT via KT or dialysis. Patients in the KT state who do not respond to the treatment may revert to dialysis, and patients in the dialysis state can also transition to the KT state. Patients can progress to death from any health state in the model.

Fig. 1.

A lifetime Markov model representing the progression of chronic kidney disease (CKD) in patients with type 2 diabetes. CKD stages 3, 4 are defined as a GFR of 15–59 ml/min/1.73 m², and CKD stage 5 is defined as a GFR of less than 15 ml/min/1.73 m². Patients enter the model at either CKD stage 3, 4 or CKD stage 5, with a cycle length of 1 year.

Model input parameters

Transitional probability

Probability of progression from CKD 3,4 to CKD 5

A WebPlotDigitizer (version 4.6, 2020) was used to extract data of the progression of CKD 3, 4 to CKD stage 5 in Thai population from a previous study¹⁸ Based on the result, this can be represented as an equation showing the relationship between the probability of developing kidney failure (CKD stage 5) and the number of years since the onset of CKD as follows: Probability of developing CKD stage 5 = (− 7 × 10^− 5 × years³) + (0.0013 × years²) + (0.0171 × years) + 0.0046. “Year” refers to the duration a patient has lived with CKD stages 3–4 before progressing to stage 5, see Supplementary Fig. 1.

The probability of receiving renal replacement therapy, including KT and dialysis, in patients with CKD stage 5

A WebPlotDigitizer (version 4.6, 2020) was applied to extract data from Thammatacharee et al.²⁰, which presents the rates of dialysis and KT in the Thai population, categorized by age at the time of therapy. The result is shown in Table 1.

Table 1.

Transitional probability and clinical efficacy input parameters in the model.

Parameter	Mean	SE	Distribution	References
Probability of receiving dialysis (by age)
20–24 years	0.392			Thammatacharee et al.20
25–29 years	0.378
30–34 years	0.344
35–39 years	0.319
40–44 years	0.311
45–49 years	0.262
50–54 years	0.236
55–59 years	0.200
60–64 years	0.161
≥ 65 years	0.118
Probability of receiving kidney transplantation (by age)
20–24 years	0.054			Thammatacharee et al.20
25–29 years	0.042
30–34 years	0.030
35–39 years	0.031
40–44 years	0.016
45–49 years	0.012
50–54 years	0.003
55–59 years	0.002
≥ 60 years	0.001
Probability of death from CKD stage 3, 4 and CKD stage 5
CKD stage 3,4	0.06	0.002	Beta	Vejakama et al.18
CKD stage 5	0.30	0.049	Beta	Vejakama et al.18
Hazard ratios for mortality in dialysis and transplantation compared to the general population
50–59 years	8.6	0.31	Log normal	Choi et al.21
60–69 years	4.6	0.20	Log normal
70–79 years	1.9	0.15	Log normal
≥ 80 years	2.2	0.15	Log normal
Age-specific mortality
60–64 years	0.0472	–	Beta	WHO life’s table (Thailand)40
65–69 years	0.0713	–	Beta
70–74 years	0.1073	–	Beta
75–79 years	0.1634	–	Beta
80–84 years	0.2491	–	Beta
> 85 years	1.0000	–	Beta
Clinical efficacy of SGLT2i in reducing the progression from CKD stage 3, 4 to CKD stage 5
Dapagliflozin (RR, 95% CI)	0.78	0.35–1.72	Log normal	Subgroup analysis from DAPA-CKD9, DECLARE-TIMI 5814, DAPA-HF24
Empagliflozin (RR, 95% CI)	0.61	0.39–0.94	Log normal	Subgroup analysis from data on file Boehringer Ingelheim (EMPA-KIDNEY10, EMPA-REG outcome12)
Canagliflozin (RR, 95% CI)	0.66	0.53– 0.83	Log normal	Subgroup analysis from CREDENCE11,22,23
Clinical efficacy of SGLT2i in reducing mortality in CKD stage 3, 4
Dapagliflozin (RR, 95% CI)	0.78	0.58–1.05	Log normal	Subgroup analysis from DAPA-CKD9, DECLARE-TIMI14, DAPA-HF24
Empagliflozin (RR, 95% CI)	0.82	0.63–1.06	Log normal	Subgroup analysis from data on file Boehringer Ingelheim (EMPA-KIDNEY10, EMPA-REG outcome12)
Canagliflozin (RR, 95% CI)	0.76	0.57–1.01	Log normal	Subgroup analysis from CREDENCE11,22,23

Probability of death from CKD stage 3, 4 and CKD stage 5

Mortality data for T2D with CKD patients who were not treated with SGLT2i, with a GFR of ≤59 ml/min/1.73 m² and who had not yet undergone RRT, was obtained from a study by Vejakama et al.¹⁸. This study collected data from T2D Patients with CKD in Ubon Ratchathani province, Thailand. Among the 13,581 patients with a GFR between 15 and 59 ml/min/1.73 m², 900 died within the median follow-up period of 4.5 years, resulting in a 1-year mortality probability of 0.064. For the 384 patients with a GFR < 15 ml/min/1.73 m², 134 died within the median follow-up period of 4.5 years, resulting in a one-year mortality probability of 0.295, Table 1.

Probability of death from dialysis and kidney transplantation

To address the lack of data specific to the Thai population, we utilized hazard ratios for KT and dialysis mortality previously reported in an Asian population²¹. The study in reference provided mortality data for individuals undergoing dialysis and KT compared to the general population in a Korean context. These findings were instrumental in calculating the age-specific standardized mortality rates for dialysis and KT in our model, see Table 1.

Clinical efficacy

Since previous clinical trials on the efficacy and safety of SGLT2i did not focus on T2D patients with CKD and did not report outcomes according to the health states in our model, we were unable to extract efficacy data specific to these health states. To address this limitation, we performed subgroup analyses of relevant landmark RCTs^9,10,22,23 that matched our target population (T2D patient with CKD). However, we recognized that the smaller number of participants in these subgroup analyses could lead to underpowered results. To mitigate this concern, we allowed pharmaceutical companies to include all participants from any landmark RCTs that aligned with our target population and provide outcomes relevant to our model’s health states. These data were obtained from AstraZeneca (dapagliflozin 10 mg), Boehringer Ingelheim (empagliflozin 10 mg), and Merck and Janssen (canagliflozin 100 mg). The subgroup analysis for dapagliflozin primarily drew upon data from DAPA-CKD⁹, DELCARE-TIMI¹⁴, and DAPA-HF trials²⁴; for empagliflozin, from the EMPA-KIDNEY¹⁰ and EMPA-REG OUTCOME trials¹²; and canagliflozin, from the CREDENCE trial^11,22,23. The results of these analyses are shown in Supplementary Table 1 and Table 1.

Cost

Direct medical costs and direct non-medical costs were included for analysis according to Thai HTA guideline that recommends to use a societal perspective for economic evaluation study aiming for policy recommendation¹⁶. These costs have been adjusted to the year 2023 using the Consumer Price Index (CPI) for medical care²⁵. Conversion rate of 36.9 Thai Baht per U.S dollar²⁶.

Direct medical cost

Direct medical costs include medication costs, laboratory test costs, hospital service costs, and treatment costs. These direct medical costs are derived from three sources including previous study in Thailand^27,28, unit costs of medical services from standard cost lists²⁹, and the price of SGLT2i, which are referenced from the median prices provided by the Drug Information Center, Ministry of Public Health³⁰ (Table 2).

Table 2.

Cost and utility input parameters for the model.

Parameter	Mean	SE	Distribution	References
Drug cost (per year)
Dapagliflozin 10 mg	13,786 ($374)	–	–	DMSIC30
Empagliflozin 10 mg	16,403 ($445)	–	–
Canagliflozin 100 mg	15,582 ($422)	–	–
Direct medical cost (per year)
CKD stage 3, 4	31,359 ($850)	3,199	Gamma	Vareesangthip et al.27
CKD stage 5	44,555 ($1,207)	4,546	Gamma
Dialysis	451,508 ($12,236)	67,723	Gamma	Phongphithakchai et al.28
Transplant (first year)	457,229 ($12,391)	50,792	Gamma
Transplant (second year)	641,725 ($17,391)	44,743	Gamma
Transplant (third year onward)	344,467 ($9,335)	14,465	Gamma

Direct non-medical cost (per year)
	Mean	Range
First year
CKD stage 3,4	29,248 ($793)	21,936−36,560	Gamma	Directly interview patients or caregivers at Siriraj hospital
CKD stage 5	108,495 ($2,940)	81,371−135,619	Gamma
Dialysis	296,947 ($8,047)	222,710−371,184	Gamma
Transplant	115,045 ($3,118)	86,284−143,807	Gamma
Second year onward
CKD stage 3,4	22,457 ($609)	16,842−28,070	Gamma
CKD stage 5	87,068 ($2,360)	65,301−108,834	Gamma
Dialysis	283,242 ($7,676)	212,431−354,052	Gamma
Transplant	86,786 ($2,352)	65,090–108,483	Gamma

Utility	Mean	SE
CKD stage 3, 4	0.79	0.04	Beta	Directly interview patients or caregivers at Siriraj hospital
CKD stage 5	0.82	0.05	Beta
Dialysis	0.56	0.06	Beta
Transplant	0.90	0.03	Beta

Costs are presented in Thai Baht (U.S. dollar, $). Conversion rate of 36.9 Thai Baht per U.S dollar, year 2024²⁶.

Direct non-medical cost

Direct non-medical costs include travel, food, and accommodation expenses for patients and caregivers related to treatment. These costs were obtained from patient and caregiver interviews at Siriraj hospital between October 1, 2022, and March 31, 2023, and from the standard cost list for HTA²⁹. The result is shown in Table 2.

Utility

The Thai version of the EQ-5D-5 L questionnaire was employed to obtain utility scores (hybrid model) through direct interviews with CKD patients in various health states who visited Siriraj hospital from October 1, 2022, to March 31, 2023. The characteristics of the included participants are presented in Supplementary Table 2. The mean and standard error (SE) of each health state is detailed in Table 2.

We obtained subgroup analysis results specific to our study population (T2D with CKD) for clinical efficacy and adverse events, including major hypoglycemia, genital infections, urinary infections, fractures, amputations, and volume depletion, as reported in clinical studies^9,11,14. Data indicated no difference in adverse events between groups receiving and not receiving SGLT2i (see Supplementary Table 3). A recent meta-analysis also found no significant difference in drug discontinuation rates for SGLT2i (RR 1.03, 95% CI 0.94–1.13)³¹. Thus, we did not include SGLT2i side effects in the analysis model.

Outcomes

Base case analysis

The results are presented as the incremental cost-effectiveness ratio (ICER). This analysis determines the additional cost per quality-adjusted life year (QALY) gained when using an SGLT2i in conjunction with SoCT compared to SoCT alone. A QALY represents the number of years lived adjusted for utility. The calculation is as follows: ICER = (total cost of SGLT2i group − total cost of SoCT group)/(QALY of SGLT2i group − QALY of SoCT group). If the ICER is lower than Thailand’s willingness-to-pay (WTP) threshold of 160,000 Baht or $4,336 per QALY¹⁶, it is considered cost-effective. Based on Thailand’s HTA guidelines, a discount rate of 3% per year was applied to both costs and outcomes¹⁶. The percentage of individuals in each health state is presented in Supplementary Table 4.

Sensitivity analysis

This study analyzes the uncertainty of all variables using two methods: one-way sensitivity analysis and probabilistic sensitivity analysis.

One-way sensitivity analysis

This method varies one variable of interest at a time while keeping other variables in the model constant. The range of variable variation in this study is calculated using the Bayesian interval method to estimate the 95% confidence interval (95% CI) for each variable, determining the lower bound and upper bound values. This helps identify which variables significantly influence changes in the ICER. Additionally, the discount rate was adjusted to 0% and 6% per year, with the results presented in a tornado diagram^32,33.

Probabilistic sensitivity analysis

This method uses Monte Carlo simulation with Microsoft Office Excel 2010 (Microsoft Corp., Redmond, WA). The simulation involves randomly selecting values for each variable and repeating the process 1,000 times, allowing all model variables to vary simultaneously according to their natural data distributions: Beta (for data between 0 and 1), Gamma (for data greater than 0 to +∞), and Log normal (for data greater than 0, around 1.0, or greater than 1). The results are presented as the average cost, health outcomes, and ICER. The findings are displayed using cost-effectiveness acceptability curves, which show the relationship between the probability that each option is cost-effective and different willingness-to-pay thresholds per additional QALY gained.

Best-case analysis

Since we derived the clinical efficacy of each SGLT2i from subgroup analyses of primary RCTs, the small number of participants might lead to insufficient power to detect the clinical benefits of the medication, especially its efficacy in reducing dialysis and kidney transplant. Therefore, a best-case analysis utilizing clinical efficacy data from previously published studies was performed.

Ethic approval and consent to participate

The study was performed in accordance with the Declaration of Helsinki³⁴. The study protocol was approved by the Institutional Review Board of Siriraj hospital, Mahidol University (MU-MOU CoA 628/2022) to ensure adherence to ethical standards throughout all stages of the research. Informed consent agreements were obtained from all participants.

Results

Base case analysis

From a societal perspective, the lifetime cost of SoCT alone was the lowest, at $14,371. Most of the cost (35% or $5,003) was for treating patients with CKD stage 3,4, followed by dialysis (29% or $4,207), and KT (18% or $2,636). For the dapagliflozin add-on SoCT, the lifetime cost was the highest at $16,273, with the cost of dapagliflozin accounting for 10% ($1,669). However, the use of dapagliflozin reduced the cost for dialysis and KT compared to SoCT alone by $189 and $161, respectively. For the empagliflozin add-on SoCT, the lifetime cost was $15,645, with the cost of empagliflozin accounting for 12% of the total cost. Nevertheless, the use of empagliflozin reduced the cost for treating patients with CKD stage 5, dialysis, and KT compared to SoCT alone by $172, $709, and $503, respectively. For the canagliflozin add-on SoCT, the lifetime cost was $16,248, with the cost of canagliflozin accounting for 13% of the total cost. The use of canagliflozin reduced the cost for dialysis and KT compared to SoCT alone by $478 and $357, respectively, see Table 3.

Table 3.

The result of economic and health outcomes of base case analysis (societal perspective).

	Standard treatment	Dapagliflozin 10 mg	Empagliflozin 10 mg	Canagliflozin 100 mg
Total lifetime cost	530,282 ($14,371)	600,489 ($16,273)	577,313 ($15,645)	599,537 ($16,248)
Drug cost	0	61,586 ($1,669)	71,335 ($1,933)	75,405 ($2,043)
Care cost for CKD stage 3, 4	184,609 ($5,003)	261,987 ($7,100)	282,717 ($7,662)	284,467 ($7,709)
Care cost for CKD stage 5	93,175 ($2,525)	98,899 ($2,680)	86,839 ($2,353)	93,391 ($2,531)
Care cost for dialysis	155,232 ($4,207)	148,276 ($4,018)	129,056 ($3,497)	137,596 ($3,729)
Care cost for kidney transplantation	97,265 ($2,636)	91,326 ($2,475)	78,702 ($2,133)	84,082 ($2,279)
Total life years	4.43	4.85	4.91	4.95
Total QALYs	3.50	3.83	3.88	3.91
Incremental cost		70,207 ($1,903)	47,031 ($1,275)	69,254 ($1,877)
Incremental LYs		0.42	0.48	0.52
Incremental QALYs		0.33	0.38	0.41
ICER/LY gain		158,309 ($4,290)	117,296 ($3,179)	136,685 ($3,704)
ICER/QALY gain		213,389 ($5,783)	124,938 ($3,386)	169,405 ($4,591)

Costs are presented in Thai Baht (U.S. dollar, $). ICER are presented in Thai Baht/QALY ($/QALY). Conversion rate of 36.9 Thai Baht per U.S dollar²⁶.

LYs life years, QALY quality adjusted life year, ICER incremental cost-effectiveness ratio, CKD chronic kidney disease.

In terms of health outcomes, treating with SoCT alone had the lowest LYs at 4.43 years, while add-on SGLT2i to SoCT resulted in higher LYs (4.85–4.95 years). Specifically, canagliflozin increased LYs by 0.52 years, empagliflozin by 0.48 years, and dapagliflozin by 0.42 years. Additionally, treatment with SGLT2i resulted in higher QALYs compared to SoCT alone. canagliflozin resulted in 3.91 QALYs, empagliflozin in 3.88 QALYs, and dapagliflozin in 3.83 QALYs, while SoCT alone resulted in 3.50 QALYs (Table 3).

With the Thailand’s WTP threshold of $4,336 per QALY, the use of empagliflozin add-on SoCT was cost-effective in the context of Thailand, with an ICER of $3,386/QALY. The use of canagliflozin was close to WTP threshold, with an ICER of $4,591/QALY, and the use of dapagliflozin was not cost-effective, with an ICER of $5,783/QALY, see Table 3.

Sensitivity analysis

Univariate analysis

One-way sensitivity analysis adjusted probability variables based on the 95% CI and cost variables by 25%, with discount rates from 0 to 6%. For dapagliflozin with SoCT, the ICER is most influenced by the efficacy in slowing CKD progression, followed by the discount rate and dapagliflozin cost. For empagliflozin and canagliflozin with SoCT, the ICER is similarly influenced by efficacy, discount rate, and drug cost, see Fig. 2.

Fig. 2.

One way sensitivity analysis (a) Dapagliflozin, (b) Empagliflozin, (c) Canagliflozin. KT kidney transplantation, DNM direct non-medical cost, RR relative risk, CKD chronic kidney disease.

Probabilistic sensitivity analysis (PSA)

PSA, which involves varying all model parameters simultaneously, was conducted to compare the probability of treatment options being cost-effective or the likelihood of making the correct decision at different ceiling thresholds or willingness-to-pay values. The results from 1,000 iterations of the PSA indicated that most ICER values fell in the upper right quadrant, suggesting that adding dapagliflozin, empagliflozin, and canagliflozin to SoCT yielded greater clinical benefits but also incurred higher costs compared to SoCT alone (Fig. 3).

Fig. 3.

Cost-effectiveness plane displaying 1,000 iterations of incremental cost (Baht) and incremental QALY (years) for dapagliflozin (a), empagliflozin (b), canagliflozin (c) and all three SGLT-2 inhibitors combined (d) add-on standard of care therapy (SoCT) compared with SoCT alone in the treatment of type 2 diabetes patients with chronic kidney disease.

PSA confirmed the deterministic findings. Considering a ceiling threshold of $4,336 per QALY, it was found that treatment with empagliflozin combined with SoCT has the highest probability of being the most cost-effective option, with a 42% chance of cost-effectiveness. This is followed by dapagliflozin at 23%, canagliflozin at 21%, and SoCT alone at 14% (Fig. 4).

Fig. 4.

The probability of individual treatment options for patients with type 2 diabetes and chronic kidney disease being cost-effective at different willingness-to-pay per quality-adjusted life year (QALY) gain.

Threshold analysis

Threshold analysis found that, at Thailand’s cost-effectiveness threshold of $4,336 per QALY, dapagliflozin 10 mg and canagliflozin 100 mg would be cost-effective if their prices were reduced by 29% and 5%, respectively.

Best-case analysis

The one-way sensitivity analysis results showed that efficacy significantly impacts the ICER. Since the original studies included only a small number of T2D participants with CKD, limiting the sample size and potentially lacking sufficient power to demonstrate clinical efficacy, particularly in reducing dialysis and KT, we conducted a best-case analysis using clinical efficacy data on reducing renal dialysis from the DAPA-CKD study⁹ to evaluate the value-for-money of each SGLT2i-dapagliflozin, empagliflozin and canagliflozin- as this was the only reported efficacy that could be applied to our model (HR 0.66; 95% CI 0.48–0.90)⁹. The results showed that the ICER was reduced from $5,783 to $1,250 and $4,591 to $1,219 per QALY gained for dapagliflozin and canagliflozin, respectively. This demonstrates cost-effectiveness based on the WTP threshold of $4,336/QALY. Additionally, the ICER for empagliflozin showed cost savings (− $131/QALY). The primary reduction in overall costs, attributed to decreased expenses for KT and dialysis, is detailed in Supplementary Table 5.

Model validation

An external validation was conducted through two conferences with multiple stakeholders, including diabetologists, nephrologists, and health economists. The first conference aimed to optimize all input parameters and construct a model that accurately represents disease progression and local practice. The second meeting, focused on face validation, was held to assess the preliminary study results. All suggestions were considered and incorporated into this final report.

Discussion

This study is the first cost-utility analysis that comprehensively evaluate multiple SGLT2i (dapagliflozin, empagliflozin, and canagliflozin) add-on SoCT compared to SoCT alone for treating patients with T2D and CKD under societal perspectives in the context of Thailand, an upper-middle income country. Patients with T2D, particularly when complicated with advanced stage of CKD experience significantly reduced quality of life and increased medical expenses for both patients and their caregivers. Healthcare costs can vary significantly depending on the stage of CKD. Overall costs rise markedly as CKD progresses^35,36, reflecting the increasing complexity and resource intensity required for patient care.

SGLT2i can slow kidney deterioration and reduce mortality in diabetic patients at risk for cardiovascular disease, regardless of their blood sugar-lowering effects^7,8. Our study affirms that patients receiving SGLT2i had life expectancy of 4.85 to 4.95 years, higher than those receiving SoCT alone (4.43 years). Additionally, SGLT2i provided more QALYs compared to SoCT alone (3.83 to 3.91 years vs. 3.50 years, respectively). However, our study also shows that patients receiving all types of SGLT2i incur higher lifetime costs compared to SoCT alone ($1,275–$1,903). Dapagliflozin had the highest lifetime cost ($16,273), followed by canagliflozin ($16,248) and empagliflozin ($15,645), with drug costs accounting for 10%, 13%, and 12% respectively. Nevertheless, SGLT2i can reduce costs for patients undergoing dialysis ($189–$709) and KT (savings of $161 to $503) compared to SoCT alone.

At the WTP threshold of $4,336/QALY, empagliflozin was found to be cost-effective compared to SoCT alone ($3,386/QALY). However, canagliflozin ($4,591/QALY) and dapagliflozin ($5,783/QALY) were not considered cost-effective in the current analysis. Nevertheless, a threshold analysis revealed that dapagliflozin 10 mg should reduce the price by 29% and canagliflozin should reduce the price by 5% to be cost-effective in the present WTP threshold in Thailand with the current clinical efficacies. The most influential variable on the ICER for all SGLT2i is the efficacy in slowing the progression from CKD stage 3, 4 to CKD stage 5, followed by the discount rate for outcomes and the cost of SGLT2i based on the result of one-way sensitivity analysis. The PSA results confirmed the findings of the deterministic analysis, showing similar outcomes to the base case analysis (dapagliflozin at $4,435/QALY, empagliflozin at $3,156/QALY, and canagliflozin at $4,397/QALY). Considering the probability of cost-effectiveness at a WTP of $4,336/QALY, treatment with empagliflozin combined with SoCT has the highest probability of being cost-effective at 42%, followed by dapagliflozin at 23% and canagliflozin at 21%.

Our findings are consistent with the study by Reifsnider et al.³⁷ conducted in the United States, which reported that empagliflozin is cost-effective in treating T2D patients with CKD, with an ICER of $25,974 per QALY gained. This value is considered cost-effective at WTP thresholds ranging from $50,000 to $150,000 per QALY, in the context of a high-income country. Due to differential selectivity among the SGLT2i, empagliflozin has been found to have a high selectivity ratio, meaning it strongly inhibits SGLT2 while sparing SGLT1. This high selectivity may make empagliflozin more effective in targeting renal glucose reabsorption, potentially enhancing its efficacy in managing CKD in patients with T2D. This pharmacological advantage likely underpins the favorable cost-effectiveness of empagliflozin observed in our analysis. Moreover, empagliflozin has the lowest probability of progression from CKD stages 3 and 4 to CKD stage 5 compared to dapagliflozin and canagliflozin (progression rates: dapagliflozin > canagliflozin > empagliflozin). This slower progression allows more patients in the empagliflozin group to remain in CKD stages 3 and 4 for longer, which not only delays the onset of advanced CKD stages but also helps avoid the significantly higher medical costs associated with CKD stage 5, dialysis, and KT. By reducing the economic burden of managing advanced CKD, empagliflozin leads to overall cost-effectiveness.

In contrast to a previous study in Egypt³⁸ and Thailand²⁷ that evaluated the cost-utility of add-on dapagliflozin to SoCT in patients with CKD, both with and without diabetes, our findings show several key differences. The earlier study found that dapagliflozin was economically dominant, meaning it was cost-saving or incurred lower costs while providing higher quality of life compared to SoCT alone. Specifically, it reduced costs associated with dialysis or KT. Our findings contrast with this previous research for several reasons. First, the earlier study utilized transitional probabilities solely from the DAPA-CKD trial, which included patients with CKD regardless of diabetes status. In contrast, our study obtained transitional probabilities from a study on the Thai population and focused exclusively on diabetic patients, leading to different probability calculations. Second, our study population focused on patients with CKD entering the model at GFR < 60 ml/min/1.73 m², while the previous study included all CKD stages (CKD stages 1–5) according to the baseline characteristics of the DAPA-CKD trial. Third, the earlier study applied a 1 and 3-month cycle length, whereas our study used a one-year cycle length to represent the progression of CKD.

The one-way sensitivity analysis results indicate that the efficacy of SGLT2i in reducing kidney disease progression has the most significant impact on the ICER among all medications. Additionally, the subgroup analysis of the primary studies did not show a benefit in delaying dialysis and KT as previously reported in the original study^9,10,13. This discrepancy may be attributed to the small sample size of T2D patients with CKD in the previous study, which lacked sufficient power to demonstrate a definitive benefit, resulting in a wide 95% confidence interval. Therefore, if the efficacy of SGLT2i, particularly in reducing the need for KT and dialysis, were to change, the ICER results could differ from those presented in this study, as we demonstrate in best-case analysis and Supplementary Table 5. Further study is necessary to assess the clinical efficacy of SGLT2i in CKD among T2D patients, especially in Thailand, where the prevalence of CKD among T2D patients is notably high at 35%³⁹ compared to the global average of 27%². The benefits of SGLT2i would be particularly significant in countries with high prevalence rates of CKD.

This study has several strengths. First, this is the first study that comprehensively evaluates several SGLT2i (dapagliflozin, empagliflozin, and canagliflozin) comparing their combination with SoCT against SoCT alone. The results will provide clear insights into medication choices for health care personals and policy makers. Second, the model variables, such as transitional probabilities, direct medical costs, direct non-medical cost and utilities, were primarily derived from studies conducted on the Thai population and adding more primary data collection in Thailand. Consequently, the results would accurately represent the cost-utility in the Thai context.

We also acknowledge some limitations. First, due to a lack of information on the probabilities of transitioning from CKD stage 5 to RRT (dialysis or KT) in T2D patients in Thailand. We had to apply the data from CKD patients, regardless of diabetes status. However, univariate analysis revealed that these variables did not significantly affect the ICER. Second, the utility data used in the study were derived solely from patient interviews at single center (Siriraj hospital), a tertiary care hospital in Thailand. Patients treated at Siriraj hospital typically have multiple comorbidities and often present with more severe symptoms compared to patients at general hospitals. This may result in lower utility values compared to those from other hospitals, potentially not fully representing the overall diabetic population with CKD across the country. In addition, the sample size was limited for certain health states, such as CKD stage 5 without dialysis, where only 10 participants were interviewed. However, the utility applied in this study relatively close to previous studies in Thailand²⁷. Third, there is no clinical efficacy data for SGLT2i in Thai T2D patients with CKD and the availability of data that align perfectly with our model’s health states. Nevertheless, we requested subgroup analyses of clinical efficacy from RCTs for each medication from primary studies, this ensures the efficacy data aligns with the health states and focuses exclusively on studied population (T2D patients with CKD) in our model. However, it is important to note that subgroup analyses of RCTs may result in smaller sample sizes, which could limit the robustness of the clinical efficacy data. Additionally, if more comprehensive clinical efficacy data for each SGLT2i on T2D patient with CKD were available, it could potentially lead to different outcomes in our model.

Conclusion

This study demonstrates that while adding SGLT2i (dapagliflozin, empagliflozin, and canagliflozin) to SoCT for patients with T2D and CKD results in higher lifetime costs, it also provides significant clinical benefits, including increased life expectancy and improved QALYs. Among the SGLT2i analyzed, empagliflozin was found to be cost-effective at a WTP threshold of $4,336/QALY in Thai context, whereas dapagliflozin and canagliflozin were not, unless their prices were reduced by 29% and 5%, respectively. This underscores the potential value of SGLT2i in enhancing patient outcomes, despite their higher costs. However, the analysis focuses solely on the renal benefits of SGLT2i, excluding other advantages like cardiovascular and heart failure protection.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Author contributions

TK, PL, VS conceptualized the original idea and design of the study methodology. NC, PL, JP, and TK participated in primary and secondary data collection. SI participated in review literature. NC, PL, VS and TK performed data analysis and interpretation. TK wrote the first draft of the manuscript. TK and VS critically revised the manuscript.

Funding

This work was supported by the Medicine Regulations Division of the Food and Drug Administration, Ministry of Public Health, Thailand and conducted at the request of the National List of Essential Medicine (NLEM) subcommittee. This study is part of the overarching “Economic evaluation of add-on SGLT-2 inhibitor and GLP-1 agonist in type 2 diabetes with established cardiovascular disease and/or diabetes kidney disease in Thailand” project, which was used to support the policymaking process under the Subcommittee for the Development of the NLEM in Thailand through the Health Economic Working Group (HEWG), although the HEWG is not responsible for the study findings or the dissemination of research funds. The funders had no role in study design, data collection, management, analysis, interpretation of the data, preparation of the manuscript or decision to submit the article for publication.

Data availability

The datasets utilized and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.