Cost-effectiveness of empagliflozin as add-on to standard of care for chronic kidney disease management in the United Kingdom

ABSTRACT

Objective

The sodium-glucose co-transporter-2 inhibitor empagliflozin was approved for treatment of adults with chronic kidney disease (CKD) on the basis of its demonstrated ability to slow CKD progression and reduce the risk of cardiovascular death. This analysis was performed to assess the cost effectiveness of empagliflozin plus standard of care (SoC) versus SoC alone in the treatment of CKD in the UK.

Methods

A comprehensive, patient-level CKD progression model that simulates the evolution of risk factors for disease progression based on CKD-specific equations and clinical data was used to project a broad range of CKD-related complications. Patient baseline characteristics, distribution across Kidney Disease Improving Global Outcomes (KDIGO) health states, and changes in estimated glomerular filtration rate (eGFR), urine albumin-creatinine ratio (uACR), and other parameters while on treatment, were derived from the EMPA-KIDNEY trial. UK cost and utilities/disutilities were sourced from the literature. Univariate and probabilistic sensitivity analyses were conducted. Annual discounting of 3.5% was applied on costs and outcomes.

Results

Over a 50-year horizon, SoC resulted in per-patient costs, life years, and QALYs of £95,930, 8.55, and 6.28, respectively. Empagliflozin plus SoC resulted in an incremental gain in life years (+1.04) and QALYs (+0.84), while decreasing per-patient costs by £6,019. Empagliflozin was more effective and less costly (dominant) with a net monetary benefit of £22,849 at the willingness-to-pay threshold of £20,000. Although treatment cost was higher for empagliflozin, this was more than offset by savings in kidney replacement therapy. Empagliflozin remained highly cost-effective in patients with and without diabetes, and across scenario and sensitivity analyses.

Limitations

This analysis is limited by reliance on short-term clinical trial data and by uncertainties in modelling CKD progression.

Conclusions

Empagliflozin as an add-on to SoC for treatment of adults with CKD represents cost-effective use of UK National Health Service (NHS) resources.

KEYWORDS:

• Health economics
• chronic kidney disease
• cost effective
• empagliflozin

JEL codes:

• I19
• I1
• I
• I15

Disclaimer

As a service to authors and researchers we are providing this version of an accepted manuscript (AM). Copyediting, typesetting, and review of the resulting proofs will be undertaken on this manuscript before final publication of the Version of Record (VoR). During production and pre-press, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal relate to these versions also.

INTRODUCTION

Chronic kidney disease (CKD) is a general term for a range of disorders affecting kidney function and structure, present for more than 3 months and with implications for poor health [1]. Kidney function is measured by estimated glomerular filtration rate (eGFR); albuminuria, a marker of kidney damage, is often assessed in terms of the urine albumin-creatinine ratio (uACR) [1]. CKD is often progressive and can lead to severe outcomes such as end-stage kidney disease (ESKD) and cardiovascular disease, both of which affect the burden of morbidity and mortality [2,3]. The public health burden of CKD is substantial, with an estimated global prevalence of 9.1% in 2017; this equates to nearly 700 million cases of all-stage CKD [4]. In the same year, 1.2 million people died globally from CKD and a further 1.4 million died from cardiovascular disease that was attributable to impaired kidney function. Global all-age CKD prevalence and mortality rates increased by approximately 30% and 40%, respectively, from 1990 to 2017, and are expected to continue to rise [4]. In the UK, in 2023, it is estimated that approximately 7.2 million people are living with CKD, which equates to more than 10% of the entire population; 3.3 million of these have stage 3–5 CKD, i.e., mild-to-moderate loss of function through to kidney failure [5]. The number living with stage 3–5 CKD is expected to rise to 3.9 million in the next 10 years.

The financial burden of CKD is also substantial, both in developing and wealthy countries [3,6]. In the UK in 2023, the annual economic burden of kidney disease to the National Health Service (NHS) was estimated at £6.4 billion, accounting for approximately 3% of the NHS budget [5]. Dialysis costs the NHS an estimated £34,000 per patient per year, and with 30,000 adults and children in the United Kingdom (UK) receiving dialysis due to kidney failure, it is a key driver of the economic burden [5]. Dialysis is also time-consuming for the patient, and contributes to the dramatic impact of CKD on patients’ and caregivers’ lives, and subsequent loss of productivity [5].

Prevention and measures to slow progression of CKD are key in reducing the burden of disease and associated costs [6]. The sodium-glucose co-transporter-2 (SGLT2) inhibitor empagliflozin was shown in the EMPA-KIDNEY trial to slow disease progression and reduce cardiovascular death in CKD [7]. Empagliflozin is approved in the UK and European Union (EU) for the treatment of adults with insufficiently controlled type 2 diabetes mellitus (T2DM), symptomatic chronic heart failure, and CKD [8,9]. However, its cost effectiveness remains to be evaluated. A comprehensive CKD progression model has been developed that simulates the evolution of risk factors for CKD progression and projects a broad range of CKD-related complications based on CKD-specific equations and clinical data.[10] The objective of this study was to use the CKD progression model to assess the cost effectiveness of empagliflozin used as an adjunct to individually optimised standard of care (SoC) versus SoC alone in the treatment of CKD in the UK.

METHODS

CKD progression model

The development and validation of the CKD progression model have been previously reported.[10] Briefly, the CKD progression model is a patient-level (microsimulation) state-transition model consisting of 18 mutually exclusive health states based on the Kidney Disease Improving Global Outcomes (KDIGO) categories, which are defined by discrete eGFR and uACR thresholds [1]. A model diagram is presented in Supplementary Figure 1. Transition between health states is determined by projected changes in eGFR and uACR over time. The model links the evolution of disease markers (eGFR and uACR) and risk factors (e.g., age, body mass index, glycated hemoglobin, serum cholesterol, systolic blood pressure, and comorbidities such as diabetes and hypertension) to project CKD progression between health states and the occurrence of a broad range of complications. Model outcomes include time (years) in each health state, time to ESKD, life expectancy, and incidences of a broad range of CKD complications, comprising cardiovascular disease, acute kidney injury (AKI), bone and mineral disorders (BMD), anemia, hyperkalemia, gout, metabolic acidosis, renal and urothelial cancer in non-functioning kidneys, infections, all-cause hospitalization, and mortality (cardiovascular, kidney-related and non-specific).

A health economic module was integrated into the CKD progression model to assess the cost-effectiveness of potential new therapeutic interventions compared with current CKD treatments. Costs were assigned for specific health states, complications, or events. The health economic module reports total costs, total life years (LYs), quality-adjusted life years (QALYs), and breakdown of costs per clinical outcome. The analysis was performed according to the National Institute for Health and Care Excellence (NICE) guidelines from the perspective of the UK NHS and personal social services (PSS).

Clinical parameters and variables

Cost-effectiveness analyses were performed using data from the EMPA-KIDNEY trial [7]. This was a multicenter, double-blind, phase 3 clinical trial conducted in 6,609 patients with CKD who received empagliflozin 10 mg oral once daily (OD) in addition to standard of care (SoC) or placebo plus SoC (SoC alone). Median follow-up was 2.0 years. For the cost-effectiveness analyses, patient baseline characteristics, distribution across KDIGO health states, and changes in eGFR, uACR, and other clinical parameters that affect disease progression were derived from the trial.

Patient population

The intention to treat (ITT) population of the EMPA-KIDNEY trial was used [7]. These were adults with evidence of CKD and at risk of disease progression (eGFR ≥20 to <45 mL/min/1.73 m², or eGFR ≥45 to <90 mL/min/1.73 m² with uACR ≥200 mg/g). Subgroup analyses were conducted for those with and without T2DM. Baseline characteristics and clinical risk factors for the study population were sourced from the EMPA-KIDNEY trial, and, where necessary, from the relevant published literature (Supplementary Table 1) [11-13].

Disease progression and risk of complications

While alive and on treatment, patients’ disease progression through the KDIGO health states was modelled according to projected treatment-specific changes in eGFR and uACR values; these were derived from observed total annual eGFR slopes and changes at 18 months in uACR from the EMPA-KIDNEY trial (Table 1). Projected changes in eGFR and uACR were KDIGO class-specific but not diabetes-specific, since the EMPA-KIDNEY trial data did not provide this level of detail. Along with eGFR and uACR treatment effects measured in the EMPA-KIDNEY trial, changes in other risk factors (HbA1c, bodyweight, BMI, and systolic BP) that impact the occurrence of complications and death were programmed into the model (Table 2). Following treatment discontinuation, natural disease progression was assumed, with no residual treatment effect. Changes in eGFR (KDIGO class- and diabetes-specific) and uACR (non-specific) were informed by the literature [11,14,15].

Risk of complications was determined by patients’ baseline characteristics and the model’s projection of the evolution of clinical risk factors over a lifetime horizon. The probability of experiencing complications or events per cycle was predicted using either clinical data sourced from the published literature (probabilities or incidence rates) or validated, established predictive risk equations. Risk of all-cause mortality was defined as the sum of cardiovascular mortality, kidney-related death, and death due to other causes, as previously described.[10]

Treatment effects, adverse events, and treatment discontinuation

Treatment effects of empagliflozin plus SoC on AKI (hazard ratio [HR] of 0.78 [95% confidence interval 0.60, 1.00] versus SoC alone) and hospitalization for heart failure (HR of 0.80 [95% confidence interval 0.60, 1.06] versus SoC alone) were derived from the EMPA-KIDNEY trial and applied for the duration of treatment. Off-treatment rates of AKI were predicted from eGFR and uACR values, according to a meta-analysis of cohort studies [16]. Off-treatment rates of hospitalization for heart failure were predicted from eGFR and uACR values and diabetes status, according to the Chronic Renal Insufficiency Cohort (CRIC) study [11]. Rates of lower limb amputation and discontinuation from the EMPA-KIDNEY trial were included in the model. Lower limb amputation rates were 0.43 and 0.29 per 100 patient-years with empagliflozin and placebo, respectively. Annual discontinuation rates were 12.56 and 14.16 per 100 patient-years with empagliflozin and placebo, respectively. In addition, treatment was assumed to be discontinued upon initiation of kidney replacement therapy (KRT).

Utilities

UK health state utilities and event disutilities were based on published sources (Supplementary Table 2). Relevant health state utilities from the EMPA-KIDNEY trial were used in the scenario analysis (detailed below). Disutilities were applied once in the cycle of occurrence of each specific complication. Time spent on dialysis and during renal transplantation was associated with a specific health state utility [17,18].

Costs

Annual CKD health state costs and event-driven costs for complications were incorporated into the model. Costs were compiled from a UK healthcare payer perspective (NHS and PSS) and informed by published sources and expert clinical opinion. Empagliflozin cost was £476.98, which was applied annually until discontinuation, death, or KRT initiation [19]. For SoC, an annual weighted average cost for the most frequently prescribed medications was used (£34.64), based on a previous cost analysis [18] and updated using the British National Formulary (BNF) 2022 [19]. Costs involved in the management of CKD and associated complications over time were obtained primarily through structured literature search (Supplementary Table 3). Annual maintenance healthcare costs per KDIGO health state, excluding hospitalization and critical care costs, ranged from £1,187 to £4,604 [20]. For CVD complications, an acute cost was applied in the year during which the complication occurred, then follow-up costs were applied in successive years while alive. For other complications, costs were either applied annually, per event, or during the first year of occurrence, as appropriate.

Analytical approach

A base-case analysis was conducted to evaluate the cost-effectiveness of empagliflozin 10 mg oral OD in addition to individually optimized SoC in adult patients with CKD, using the model settings detailed in Table 3; the comparator was individually optimized SoC alone. An annual discount rate of 3.5% was applied to costs and outcomes in line with NICE guidelines [21]. The time horizon was 50 years to reflect lifetime analysis in the EMPA-KIDNEY population, which had a mean (SD) age of 63.8 (13.9) years at baseline. The health economic module was used to calculate incremental QALYs, Lys gained, incremental net monetary benefit (iNMB), incremental net health benefit (iNHB), and incremental cost-effectiveness ratios (ICERs) (incremental cost per LY or QALY). Willingness-to-pay (WTP) thresholds of £20,000 and £30,000 per QALY gained were used, in line with NICE guidelines [21]. Cost effectiveness was described in terms of the ICER, iNMB and iNHB. The ICER is the cost per additional unit of effectiveness gained for empagliflozin versus SoC (incremental total costs/incremental QALY). The iNMB indicates the value of an intervention ([incremental QALY × WTP threshold] – incremental total cost), while the iNHB represents the impact on population health of moving funding to pay for a new intervention (incremental QALYs – [incremental cost/opportunity cost threshold]).[22]

Scenario analysis

Four scenario analyses, involving alternative choices of risk equations, and variations in model parameters and assumptions applied in the base-case analysis, were performed to assess the robustness of the cost-effectiveness findings: (i) the 6-variable Kidney Failure Risk Equation (KFRE) [23] used in the base-case analysis was replaced with a 4-variable KFRE (which excludes diabetes and hypertension as risk factors) generated from a UK primary care population [24]; (ii) literature-derived health state utilities were replaced with utilities from the EMPA-KIDNEY trial; (iii) hospitalization costs for acute cardiovascular events (myocardial infarction, stroke, transient ischemic attack) in the first year, cancer in the first year, and infections, AKI, and fractures were replaced with all-cause hospitalization rates and costs; and (iv) the eGFR threshold of 15 Ml/min/1.73 m² at which risk equations for KRT were applied was replaced with an eGFR threshold of 20 ml/min/1.73 m².

Univariate sensitivity analysis

Univariate sensitivity analysis assessed how varying individual parameters sequentially, while holding all other parameters constant, impacted model-predicted costs, outcomes, and ICER. An exhaustive list of parameters was analyzed, including complication probabilities, treatment effects (except adverse events), costs, utilities, and disutilities. The majority of upper and lower values were defined by the observed 95% confidence interval (CI). If the 95% CI was not available, a 20% change from the point estimate was assumed for cost parameters and a 10% change was used for clinical parameters to ensure upper values ≤ 1 for utility, probabilities, and proportions.

Probabilistic sensitivity analysis

A probabilistic sensitivity analysis was run with 1,000 patients and 500 iterations to assess the effect of parameter uncertainties. Observed standard error (SE) was used to determine the probabilistic distribution of all parameters, except for costs where the SE was assumed equal to 20% of the mean. Parameters were simultaneously sampled from their respective distributions. Clinical data for risk of events and complications and treatment effects were typically assigned lognormal or beta distributions. All cost parameters were assigned a gamma distribution. Health state utilities, CKD complications, and adverse event (AE) disutilities were assigned a beta distribution.

RESULTS

Base case analysis

In the base case, time on treatment for empagliflozin plus SoC was 5.9 years, compared with 4.2 years for SoC alone. Undiscounted life expectancy for empagliflozin plus SoC was 12.6 years, compared with 11.0 years for SoC alone (1.6 undiscounted incremental life years), and time to progression to ESKD or KRT was 9.7 and 7.0 years, respectively. Mortality rates and AKI event rates were lower with empagliflozin plus SoC compared with SoC alone (Supplementary Table 4). Over a lifetime horizon, empagliflozin plus SoC compared with SoC alone resulted in an incremental gain in discounted life years (+1.04) and QALYs (+0.84) while decreasing costs by £6,019 per person (Table 4). Empagliflozin dominated SoC alone and, with a net monetary benefit of £22,849, was cost-effective at £20,000/QALY. Although the cost of treatment was higher for empagliflozin, this was more than offset by savings in kidney replacement therapy; some costs, such as monitoring and cardiovascular or CKD complications, were slightly higher in the empagliflozin group owing to better survival (Supplementary Table 5).

In patients with T2DM, life expectancy was shorter than the overall population (11.2 years for empagliflozin plus SoC and 9.5 years for SoC alone); rates of cardiovascular events were higher, but ESKD and KRT event rates were lower (data not shown) owing to the shorter life expectancy. In patients without T2DM, life expectancy was longer than that of the overall population (13.9 years for empagliflozin plus SoC and 12.3 years for SoC alone). Cardiovascular event rates were lower compared with those in patients with T2DM, but rates of ESKD and KRT events were higher owing to the longer life expectancy. Empagliflozin was cost effective in patients with and without T2DM; cost savings and QALYs were greater in the non-T2DM group compared with the overall population (Table 4), owing to longer life expectancy. The breakdown of costs showed similar trends to the overall population (data not shown).

Scenario analysis

Empagliflozin remained highly cost effective in the scenario analyses (Table 5). Scenarios involving the use of a 4-variable KFRE [24] to predict the risk of KRT, and the use of all-cause hospitalization costs, resulted in a net monetary benefit within 4% of the base-case value.

Univariate sensitivity analysis

The 20 model parameters that had the most impact on net monetary benefit in the deterministic univariate sensitivity analysis are shown in Figure 1. Key drivers were health state utilities and management costs for patients with KDIGO classification G4 * A2, and treatment effects for Uacr progression for patients with KDIGO classification G4 * A3.

Probabilistic sensitivity analyses

There was a consistent reduction in cost associated with an increase in QALYs (Figure 2A). Empagliflozin was dominant in 94% of simulations, and the probability of cost-effectiveness compared with SoC alone at WTP thresholds of £20,000/QALY and £30,000/QALY was 100% (Figure 2B).

DISCUSSION

A de novo microsimulation cost-effectiveness model, developed to capture the heterogeneity of the population with CKD and to project a complete disease outcome, demonstrated that empagliflozin is a cost-effective treatment option for CKD in the UK. In the base-case analysis, empagliflozin 10 mg OD plus SoC compared with SoC alone was highly dominant, with an incremental gain in life years of 1.04 and a gain in QALYs of 0.84. This was attributable to improved survival due to slowing of disease progression and delay in the need to use KRT. Costs fell by £6,019 per person, compared with SoC alone, with a net monetary benefit of £22,849. Univariate and probabilistic sensitivity analyses identified health state utilities and management costs for patients in the KDIGO G4*A3 category as key determinants of net monetary benefit. In all scenario analyses, empagliflozin remained highly cost effective, and dominant versus SoC alone at a WTP threshold of £20,000 per QALY. The patient population used in this analysis, which was derived from the EMPA-KIDNEY trial, represents a broad and heterogenous CKD population with a wide range of Egfr and Uacr levels at baseline, including patients without T2DM, for whom cost-effectiveness was also demonstrated. The comparator, defined as ‘individually optimized SoC’, was mapped to the SoC arm of EMPA-KIDNEY trial. Each trial participant received appropriate management of their cardiovascular risk factors and other existing comorbidities (e.g. hypertension, diabetes) which aligned with the established clinical practice in the NHS and the recommendations for CKD management.[25]

These findings are in line with increases in QALYs observed with dapagliflozin, another SGLT2 inhibitor that has been evaluated and approved for treatment of CKD in the UK [18,26]. A cost-effectiveness analysis conducted by the UK National Institute for Health and Care Excellence (NICE), based on clinical data from the Dapagliflozin and Prevention of Adverse outcomes in Chronic Kidney Disease (DAPA-CKD) trial [27], found that dapagliflozin in addition to SoC was associated with 9.26 LYs and 6.80 QALYs, while SoC was associated with 8.25 LYs and 6.03 QALYs [18]. However, the DAPA-CKD trial population had a higher proportion of patients with T2DM and was at higher risk for progression of CKD and cardiovascular disease, since patients were in a higher baseline albuminuria range than those in the EMPA-KIDNEY trial. Costs of disease management in both treatment arms were higher in the empagliflozin analysis compared with the dapagliflozin analysis (£89,911 for empagliflozin and £95,930 for SoC, compared with £56,526 for dapagliflozin and £51,408 for SoC), owing to more complications, longer life expectancy, and increased costs driven by inflation.

Another cost-effectiveness analysis conducted from a UK perspective, likewise using data from the DAPA-CKD trial [27], reported that dapagliflozin was associated with an increase in QALYs of 8.68, versus 7.86 for SoC in patients with CKD [28]; ICERs were comparable between subgroups of patients with and without diabetes. As in the present analysis, while dapagliflozin was associated with increased total costs versus SoC, this was due to increased life expectancy; delays in CKD progression and kidney failure offset this additional cost. The undiscounted incremental gain in life years in that analysis was 1.7 years, compared with 1.6 in the present analysis.

Empagliflozin is approved in the UK for the treatment of adults with CKD, adults with insufficiently controlled T2DM, and adults with symptomatic heart failure [8]. Cost effectiveness of empagliflozin in the treatment of diabetes has been demonstrated [29-31], along with cost effectiveness in patients with diabetes and cardiovascular disease or heart failure [32,33]. Clinically, empagliflozin has demonstrated consistent effects on renal outcomes across a range of diabetic kidney disease phenotypes [34]. Further, the EMPA-KIDNEY trial showed that risk of kidney disease progression or cardiovascular death was reduced with empagliflozin plus SoC, among a broad population of patients (with or without diabetes and with a range of albuminuria levels) at risk of CKD progression [7]. A health economic analysis in the United States concluded that empagliflozin plus SoC may be a cost-effective treatment for patients with diabetic kidney disease [35]. The results of the present analysis build on the body of evidence for use of empagliflozin in patients with CKD, demonstrating cost-effectiveness when added to SoC in the overall population, and also separately in patients with and without T2DM.

Access to additional options for the treatment of CKD would have a substantial impact in reducing or avoiding downstream healthcare consumption for many patients. This has economic implications for the NHS, as demonstrated in this cost-effectiveness analysis, but would also have broader effects at a societal level, e.g., by reducing the impact on patient and caregiver productivity.

Limitations

As in all long-term health economic analyses, a limitation to this approach is the reliance on short-term clinical trial data. In mitigation, longer-term studies indicate that the decline in kidney function seen over the short clinical trial period results in significant progression of CKD complications. Other limitations include those related to the CKD-progression model and the methodology used for the cost-effectiveness analysis. Risk prediction provided by the CKD progression model is a composite of several published risk equations from different sources, and consistency across data inputs may not be optimal. Outlier observations (e.g., patients with >1 renal transplant) were not excluded from the analysis. However, these phenomena affect both treatment arms with similar frequencies, and are also encountered in real-world practice. Moreover, when generating baseline patient characteristics, it was not feasible to account for possible correlation between certain parameters—for example, age was sampled independently from eGFR. While this may affect the face validity of specific outcomes (e.g., age at KRT initiation), both treatment arms would be anticipated to be equally affected in this regard. In addition, for estimation of net monetary benefit, model convergence was achieved with 1,000 patients; however, it is possible that incremental cost and incremental QALY may not achieve the same degree of convergence with a sample of this size.

CONCLUSIONS

This cost-effectiveness analysis demonstrates that empagliflozin represents a cost-effective use of UK NHS resources as an add-on to SoC for treatment in adult patients with CKD.

TRANSPARENCY

Declaration of funding

This analysis was funded by Boehringer Ingelheim & Eli Lilly and Company Diabetes Alliance.

Declaration of financial/other interests

LG is an employee of IQVIA. MR and ML were employees of IQVIA at the time of the study but are now employed at Th(is)²Modeling, Asse, Belgium. IQVIA received consulting fees from Boehringer Ingelheim. AU and LM are employees of Boehringer Ingelheim. AF has received research grants and advisory board fees from Boehringer Ingelheim, Lilly, Astra Zeneca, Menarini, and Bayer.

The authors meet criteria for authorship as recommended by the International Committee of Medical Journal Editors (ICMJE). The authors received no payment related to the development of the manuscript.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

MR contributed to study conception and design, acquisition of data, and data analysis and interpretation. AU and AHF contributed to study conception and design, and data interpretation. ML contributed to study conception and data interpretation. LG contributed to data validation, analysis, and interpretation. LM contributed to data interpretation.

All authors were involved in critically revising the manuscript for intellectual content and approve the final version to be published. All authors agree to be accountable for all aspects of the work.

Acknowledgments

The EMPA-KIDNEY trial was initiated, designed, and conducted by the University of Oxford in collaboration with a Steering Committee of experts and Boehringer Ingelheim. The presented analyses were initiated and conducted by Boehringer Ingelheim independently from the EMPA-KIDNEY Collaborative Group.

Editorial support was provided by Andrew Fitton, on behalf of Envision Pharma Group, which was contracted and funded by Boehringer Ingelheim. Boehringer Ingelheim were given the opportunity to review the manuscript for medical and scientific accuracy as well as intellectual property considerations.

Data sharing statement

The data underlying this article will be shared upon reasonable request to the corresponding author.

Previous presentation

Poster presentation at: The Professional Society for Health Economics and Outcomes Research (ISPOR) Europe Conference 2023; 12–15 November 2023; Copenhagen, Denmark.

Table 1. Projected changes in eGFR and uACR values over 1 year in patients receiving empagliflozin as add on to SoC and SoC alone.

KDIGO classification	Empagliflozin 10 mg OD plus SoC					Placebo plus SoC
	A1	A2		A3		A1	A2		A3
	Mean (95% CI) change in eGFR (mL/min/1.73 m^2)
G2	NA	–2.2 (–3.26, –1.14)		–3.39 (–3.96, –2.81)		NA	–2.76 (–3.92, –1.59)		–5.14 (–5.7, –4.58)
G3a	NA	–1.60 (–2.32, –0.89)		–3.45 (–3.91, –2.98)		NA	–2.29 (–3.04, –1.55)		–4.66 (–5.14, –4.19)
G3b	–0.58 (–0.96, –0.19)	–1.041 (–1.40, –0.67)		–2.90 (–3.20, –2.60)		–0.83 (–1.2, –0.46)	–1.56 (–1.92, –1.20)		–4.11 (–4.42, –3.80)
G4	–0.32 (–0.87, 0.22)	–0.62 (–1.04, –0.19)		–2.76 (–3.08, –2.45)		–0.15 (–0.71, 0.40)	–0.85 (–1.27, –0.43)		–3.76 (–4.09, –3.44)
All	–1.96 (–2.11, –1.82)					–2.68 (–2.82, –2.53)
	Mean (95% CI) change in uACR
G2	NA		1.26 (0.93, 1.70)		0.67 (0.56, 0.79)	NA		0.93 (0.67, 1.29)		0.75 (0.63, 0.89)
G3a	NA		0.87 (0.71, 1.06)		0.53 (0.46, 0.61)	NA		0.99 (0.80, 1.23)		0.71 (0.62, 0.81)
G3b	1.62 (1.44, 1.82)		0.84 (0.76, 0.94)		0.62 (0.56, 0.67)	1.65 (1.47, 1.85)		1.09 (0.98, 1.22)		0.81 (0.74, 0.88)
G4	2.08 (1.76, 2.44)		1.03 (0.91, 1.18)		0.68 (0.62, 0.74)	2.44 (2.07, 2.87)		1.51 (1.33, 1.71)		0.95 (0.86, 1.04)
All	0.86 (0.83, 0.90)					1.08 (1.04, 1.13)

Abbreviations: CI, confidence interval; eGFR, estimated glomerular filtration rate; NA, not available; OD, once daily; SoC, standard of care; uACR, urine albumin-to-creatinine ratio.

Table 2. Incremental effects of empagliflozin and placebo on other risk factors across health states.*

Risk factor	Empagliflozin 10 mg OD plus SoC	Placebo plus SoC
HbA1c, mmol/mol	–0.5585 (–0.8253, –0.2917)	–0.1455 (–0.4142, 0.1232)
Weight, kg	–1.5525 (–1.7375, –1.3676)	–0.679 (–0.8648, –0.4933)
BMI (calculated), kg/m²	–0.55	–0.24
SBP (mmHg)	–3.915 (–4.3245, –3.5056)	–1.2913 (–1.7019, –0.8808)

*The impact on lipids was not reported

^†Impact on serum hemoglobin was not programmed into the model as it is not a driver of any of the risk equations in the model.

Abbreviations: BMI, body mass index; HbA1c, glycated hemoglobin; OD, once daily; SBP, systolic blood pressure; SoC, standard of care.

Table 3. Base case model settings,

Model parameter	Option selected
Key model parameters
Time horizon (years)	50 (lifetime)
Number of patients	1,000
Number of iterations (PSA)	500 iterations of 1000 patients
Discounting rate – effects	3.5%
Discounting rate – costs	3.5%
Half cycle correction	Yes
Cost approach	Option 1: Includes 1st and follow-up years
Utility approach	Option 1: Health state utilities plus disutilities per event/complication
Population baseline risk factors
Patient population	Option 1 – EMPA-KIDNEY full population
eGFR	Distribution over eGFR classes according to EMPA-KIDNEY
eGFR midpoints	As per EMPA-KIDNEY
uACR	Distribution over uACR classes according to EMPA-KIDNEY
uACR midpoints	As per EMPA-KIDNEY
Treatment effect options
Comparators	Empagliflozin + SoC vs. SoC alone
eGFR	Annual eGFR change per KDIGO
uACR	Annual uACR change per KDIGO
Adverse events	Rate per 100-patient year
Discontinuation	Rate per 100-patient year
HHF	Hazard ratio vs. SoC
Acute kidney injury	Hazard ratio vs. SoC
All-cause hospitalization	Unadjusted rate per 100 person-years
Duration of treatment (max)	50 years
Risk factor progression/disease progression after treatment discontinuation
eGFR progression	Fixed changes per eGFR and uACR classes [11] (Grams 2021; Supplementary Table 3)
uACR progression	Lognormal distribution uACR fold [14] (Coresh 2019; Figure 1)
Engine options
Risk of death	Three mutually exclusive causes of death were modelled: CVD death with CKD patch [36] (Matsushita 2020; Web Table 13) plus non-specific cause death (UK National life tables; excluding mortality from CVD and renal causes as per death counts statistics) plus renal death during KRT (UK Renal Registry, Annual Report 2022; Table 2.8)
Risk prediction of CVD mortality	CKD patch, low-risk countries [36] (Matsushita 2020; Web Table 13)
Risk of KRT	Tangri risk equations, 5 years, pooled with six variables (selected because DM is a parameter in the equation) [23] (Tangri 2016; eAppendix 2)
Risk of KRT is applied when eGFR is equal to or below	15 mL/min/1.73 m^2
Age threshold after which KT is not performed	80 years
When in conservative therapy which cost to apply? (ESKD health state costs and/or conservative therapy costs)	Conservative therapy cost only
Risk prediction of CVD	[36] (Matsushita 2020; Web Table 13)

Abbreviations: CVD, cardiovascular disease; DM, diabetes mellitus; eGFR: estimated glomerular filtration rate; ESKD, end-stage kidney disease; HHF, hospitalization for heart failure; KDIGO, Kidney Disease Improving Global Outcomes; KRT, kidney replacement therapy; KT, kidney transplantation; PSA, probabilistic sensitivity analysis; SoC, standard of care; uACR, urine albumin-to-creatinine ratio.

Table 4. Base case cost-effectiveness analysis over a lifetime (50 years).

	Overall population			Patients with T2DM			Patients without T2DM
	Empagliflozin 10 mg OD plus SoC	SoC	Incremental	Empagliflozin 10 mg OD plus SoC	SoC	Incremental	Empagliflozin 10 mg OD plus SoC	SoC	Incremental
Total discounted costs (£)	89,911	95,930	–6,019	80,264	81,792	–1,528	90,447	98,510	–8,063
Total discounted Lys	9.59	8.55	1.04	8.73	7.63	1.099	10.35	9.33	1.017
Total discounted QALYs	7.12	6.28	0.84	6.29	5.36	0.925	7.53	6.61	0.921
ICER (£/QALY)	–	–	Dominant	–	–	Dominant			Dominant
iNMB (£) at £20,000/QALY	22,849		Cost-effective at £20,000/QALY	20,036	–	Cost-effective at £20,000/QALY	26,473		Cost-effective at £20,000/QALY
iNHB (QALYs) at £20,000/QALY	1.14	–	–	1.00			1.32

Abbreviations: ICER, incremental cost-effectiveness ratio; iNHB, incremental net health benefit; iNMB, incremental net monetary benefit; LY, life years; OD, once daily; QALY, quality-adjusted life year; SoC, standard of care; T2DM, type 2 diabetes mellitus.

Table 5. Scenario analyses: Incremental cost-effectiveness ratio and net monetary benefit for empagliflozin in addition to SoC compared to SoC alone.

Scenario	ICER (£/QALY)	NMB (£)*	Change (%) relative to base-case NMB
Base case	Dominant	22,849	–
Major, et al. risk equation to predict risk of KRT	Dominant	22,573	–1.21
EMPA-KIDNEY derived trial utilities to predict QALYs	Dominant	24,596	7.64
All-cause hospitalizations	Dominant	22,020	–3.63
eGFR threshold at 20 mL/min/1.73 m^2 to estimate the risk of KRT	Dominant	20,218	–11.51
Annual discount rate 2% (costs; health effects)	Dominant	25,707	+12.51
Annual discount rate 5% (costs; health effects)	Dominant	20,385	−10.78

*At a willingness-to-pay threshold of £20,000.

Abbreviations: eGFR, estimated glomerular filtration rate; ICER, incremental cost-effectiveness ratio; KRT, kidney replacement therapy; NMB, net monetary benefit; QALY, quality-adjusted life year; SoC, standard of care.

Figure 1. Deterministic univariate sensitivity analysis: tornado diagram of the 20 parameters with most impact on net monetary benefit, when comparing empagliflozin in addition to SoC against SoC alone.

Abbreviations: eGFR, estimated glomerular filtration rate; ESKD, end-stage kidney disease; HD, hemodialysis; HS, health state; ITT, intention to treat; KRT, kidney replacement therapy; PD, peritoneal dialysis; RFP, risk factor progression, RRT, renal replacement therapy; SoC, standard of care; uACR, urine albumin creatinine ratio.

Figure 2. Probabilistic sensitivity analysis: cost-effectiveness plane (A) and acceptability curve (B), for empagliflozin in addition to SoC.

Abbreviations: BC, base case; PSA, probabilistic sensitivity analysis; QALY, quality-adjusted life years; SoC, standard of care.