Cost-effectiveness of Dapagliflozin for Treatment of Patients With Heart Failure With Reduced Ejection Fraction

Key Points

Question  Is dapagliflozin cost-effective when added to standard of care for patients who have heart failure with reduced ejection fraction (HFrEF) with and without diabetes?

Findings  This economic evaluation used a Markov model based on the Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure population and found dapagliflozin had an incremental cost-effectiveness ratio of $83 650 per quality-adjusted life-year gained, indicating intermediate value according to American College of Cardiology/American Heart Association benchmarks. The cost-effectiveness of dapagliflozin was similar for subgroups with and without diabetes and among those with more or less impaired health status due to heart failure.

Meaning  These findings suggest that dapagliflozin is cost-effective for patients with HFrEF; therefore, given evidence of clinical benefit and economic value, focus should be on increasing therapy rates among patients who have HFrEF with and without diabetes.

Importance  In the Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure (DAPA-HF) trial, dapagliflozin was shown to reduce cardiovascular mortality and hospitalizations due to heart failure while improving patient-reported health status. However, the cost-effectiveness of adding dapagliflozin therapy to standard of care (SOC) is unknown.

Objective  To estimate the cost-effectiveness of dapagliflozin therapy among patients with chronic heart failure with reduced ejection fraction (HFrEF).

Design, Setting, and Participants  This Markov cohort cost-effectiveness model used estimates of therapy effectiveness, transition probabilities, and utilities from the DAPA-HF trial and other published literature. Costs were derived from published sources. Patients with HFrEF included subgroups based on diabetes status and health status impairment due to heart failure. We compiled parameters from the literature including DAPA-HF, on which our model is based, and many other sources from December 2019 to February 27, 2021. We performed our analysis in February 2021.

Exposures  Dapagliflozin or SOC.

Main Outcomes and Measures  Hospitalizations for heart failure, life-years, quality-adjusted life-years (QALYs), costs, and the cost per QALY gained (incremental cost-effectiveness ratio).

Results  In the model, dapagliflozin therapy yielded a mean of 0.78 additional life-years and 0.46 additional QALYs compared with SOC at an incremental cost of $38 212, resulting in a cost per QALY gained of $83 650. The cost per QALY was similar for patients with or without diabetes and for patients with mild or moderate impairment of health status due to heart failure. The cost-effectiveness was most sensitive to estimates of the effect on mortality and duration of therapy effectiveness. If the cost of dapagliflozin decreased from $474 to $270 (43% decline), the cost per QALY gained would drop below $50 000.

Conclusions and Relevance  These findings suggest that dapagliflozin provides intermediate value compared with SOC, based on American College of Cardiology/American Heart Association benchmarks. Additional data regarding the magnitude of mortality reduction would improve the precision of cost-effectiveness estimates.

Heart failure is projected to affect 9.6 million persons in the US with a cost of $69.7 billion annually by 2030.¹ It contributes to 1 of 8 deaths in the US.^2,3 In recent years, several promising new HF treatment options have been approved. However, given concerns regarding health care costs, a cost-effectiveness evaluation of these therapies is imperative.

Dapagliflozin, a sodium-glucose cotransport 2 (SGLT2) inhibitor, is the newest medication approved by the US Food and Drug Administration for heart failure with reduced ejection fraction (HFrEF). In multiple diabetes treatment trials, SGLT2 inhibitors reduced the incidence of heart failure and hospitalizations for heart failure.^4-7 As a result, the Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure (DAPA-HF) trial⁸ evaluated dapagliflozin in patients with symptomatic HFrEF, with and without diabetes, compared with standard of care (SOC)—that is, standard heart failure device and drug therapy, which included appropriate treatment with angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, or sacubitril-valsartan plus a β-blocker. In that study,⁸ 4744 patients (2761 [58.2%] without type 2 diabetes) were randomized to dapagliflozin or placebo and were followed up for a median of 18.2 (range, 0-27.8) months. Dapagliflozin significantly reduced the primary composite end point of cardiovascular death, hospitalization for heart failure, or urgent visits for heart failure; reduced the secondary outcome of cardiovascular death; and improved patient-reported health status. Dapagliflozin had similar effectiveness for patients with and without diabetes.

At present, dapagliflozin’s price is approximately $470 per month.⁹ It remains unclear whether it is cost-effective at this price point. We performed an independent cost-effectiveness analysis of dapagliflozin in patients with HFrEF, with subgroup analyses by diabetes status and health status impairment due to heart failure.

We developed a state-transition Markov cohort model that compares dapagliflozin and SOC without dapagliflozin across the DAPA-HF population (eFigure 1 in the Supplement).⁸ The DAPA-HF trial included patients older than 18 years with left ventricular ejection fraction of 40% or less, New York Heart Association (NYHA) class II to IV heart failure, and N-terminal pro–B-type natriuretic peptide (NT-proBNP) levels of at least 600 pg/mL (see DAPA-HF methods for complete inclusion criteria).⁸ Patients with type 1 diabetes, hypotension (systolic blood pressure <95 mm Hg), or estimated glomerular filtration rate of less than 30 mL/min/1.73 m² were excluded. This economic simulation analysis was exempt from institutional review board approval and informed consent. DAPA-HF collected data on average for 18 months after enrollment/randomization, which occurred from February 15, 2017, to August 17, 2018. We developed our model and performed our analysis from December 2019 to February 27, 2021.

We performed 2 secondary subgroup analyses in which we separately evaluated (1) groups with and without diabetes within each arm and (2) 2 different levels of health status impairment due to heart failure within each arm. Both diabetes and heart failure health status affect the baseline morbidity rates.^8,10-13 Although relative treatment effects appeared similar across diabetes status, the DAPA-HF trial suggested possible effectiveness differences between patients with more and less severe heart failure. The effect of dapagliflozin was significantly larger among patients with NYHA class II vs III or IV disease. In addition, although not significant, the effectiveness estimates were greater for those with better health status (based on the Kansas City Cardiomyopathy Questionnaire Total Symptoms Score [KCCQ-TSS]; cardiovascular mortality: KCCQ-TSS ≤65.6 [hazard ratio (HR), 0.84], KCCQ-TSS 65.7-87.5 [HR, 0.78], and KCCQ-TSS >87.5 [HR, 0.72]; P = .82) and lower NT-proBNP levels.¹⁰ However, the data are not entirely clear. Effectiveness point estimates were nonsignificantly smaller for patients with higher left ventricular ejection fraction and those without past hospitalization for heart failure. We stratified patients into mild and moderate impairment of health status due to heart failure using the KCCQ-TSS.

We used the KCCQ-TSS–based stratification owing to its superior reproducibility and prognostic significance compared with NYHA class as well as its patient-centered approach.^14-16 In DAPA-HF, baseline KCCQ-TSS tertiles correlated with probabilities of cardiovascular death (16.8% in the low [most severe impairment] tertile, 10.2% in the middle tertile, and 7.7% in the high tertile of KCCQ-TSS scores).¹⁰ We created 2 strata: moderate (KCCQ-TSS <65.7) and mild (KCCQ-TSS ≥65.7) health status impairment due to heart failure.¹⁰

Transition probabilities were based on DAPA-HF trial outcomes and the natural history of HFrEF.⁸ Patients had a monthly risk of hospitalization for heart failure, hospitalization for other causes, left ventricular assist device implantation, heart transplantation, cardiovascular or noncardiovascular death, and treatment intolerance (Table 1).

We developed our model from a health care payor perspective with a lifetime horizon (eMethods 1 in the Supplement).²³ Future costs and utilities were discounted to present value at 3% annually. The primary outcome was the incremental cost-effectiveness ratio (ICER), which we calculated as incremental cost divided by incremental quality-adjusted life-years (QALYs). We used the American College of Cardiology/American Heart Association value threshold for high value (ICER <$50 000), intermediate value (ICER $50 000-$150 000), and low value (ICER >$150 000).²⁴ Below we provide an overview of model parameters (Table 1, with additional details in eMethods 2 to 13 and eTables 1 to 9 in the Supplement).

The monthly risk of hospitalization for heart failure, cardiovascular mortality, and noncardiovascular mortality for SOC were estimated from the DAPA-HF control arm (eMethods 2 in the Supplement).⁸ We assumed risk increased with age based on prior literature (eTables 2 and 3 in the Supplement).^25-27 For patients treated with dapagliflozin, the probability of hospitalization for heart failure and cardiovascular mortality were calculated based on the SOC rates multiplied by the DAPA-HF HR (eMethods 3 in the Supplement). The SOC arm’s noncardiovascular mortality rate was assumed for the dapagliflozin arm. We modeled an adverse event rate based on DAPA-HF; we assumed patients discontinued dapagliflozin therapy after an adverse event, incurred no further drug costs, and received no further risk reduction relative to the SOC arm (eMethods 4 in the Supplement).

The DAPA-HF trial did not report rates of left ventricular assist device implantation and heart transplantation. We estimated these rates from the literature, assuming no dapagliflozin effect on advanced therapy rates or outcomes (eMethods 5 in the Supplement).^1,28-31

For subgroup analyses, we estimated subgroup-specific event rates using DAPA-HF data when available and otherwise from the literature (eMethods 6 in the Supplement). For the diabetes subgroup analysis, we assumed a similar dapagliflozin effect in both arms.¹² For the subgroups with health status impairment due to heart failure, we used different treatment effects based on baseline KCCQ-TSS. We calibrated each analysis to match the overall event rates and treatment effects observed in DAPA-HF (eMethods 7 and eTables 4 and 5 in the Supplement).

We estimated a 2020 monthly cost of dapagliflozin of $473.64 based on the Medicare Part D drug payments from calendar year 2018 (eMethods 8 and eTable 6 in the Supplement).⁹ This cost comprised the total drug cost, including the dispensing fee, the drug plan payment, and the beneficiary copayment. Medicare Part D cost was chosen as opposed to wholesale acquisition cost because Medicare Part D data are consistently and accurately reported and most patients with heart failure have Medicare coverage.³² In a sensitivity analysis, we evaluated ±50% of this cost because the pricing may vary across payors or change over time. We also evaluated several cost scenarios, including entry of multiple generics, wholesale acquisition cost, retail pharmacy price, and uninsured cost.

We divided other health care costs into ambulatory and hospitalization costs (eMethods 9 and 10 and eTable 6 in the Supplement). We assumed outpatient heart failure care costs increased with age by 0.88% annually (eTable 7 in the Supplement).³³ Based on prior literature, we assumed similar outpatient costs among patients who had heart failure with and without diabetes (eMethods 9 in the Supplement).³³ Higher costs for moderate impairment of health status due to heart failure were based on estimates of cost differences between patients with NYHA class III or IV and class II.³⁴ The costs of hospitalization for heart failure and hospitalizations for other causes were derived from previous literature estimates.^21,35 Based on DAPA-HF proportions, we assumed the cost of a hospitalization for causes other than heart failure for severe adverse events (eg, diabetic ketoacidosis) and an outpatient clinic appointment for mild adverse events.^35,36 The cost of cardiovascular and noncardiovascular death were taken from literature estimates and applied as a one-time cost at the time of death (eMethods 10 in the Supplement).²² Costs are inflation adjusted to 2020 using annual US Bureau of Economic Analysis Personal Consumption Expenditure–Health Index deflators.³⁷

We estimated the mean baseline utility of the DAPA-HF population using the KCCQ-TSS (eMethods 11 and eTable 8 in the Supplement).¹⁸ We used an algorithm that maps KCCQ-TSS scores to utility estimates.¹⁷ The utility estimates are based on a US time trade-off valuation of EQ-5D-3L scores. We estimated a baseline utility of 0.78. To account for aging and disease progression, we modeled an annual utility decrease of 0.7%, which we varied in sensitivity analysis.³⁸ We estimated subgroup-specific differences for patients with and without diabetes.¹⁹ For the heart failure health status subgroups, we estimated overall KCCQ scores based on subgroup KCCQ-TSS values and mapped these scores to estimated utilities.^10,17 We estimated the effect of dapagliflozin on utility based on the change in KCCQ observed in DAPA-HF (annual utility change of 0.008) (eMethods 12 in the Supplement).

We used literature estimates for disutilities associated with adverse events, as well as adjustment of long-term utility after advanced therapy (eMethods 13 and eTable 8 in the Supplement).^20,39-42 A disutility of 29% of baseline utility was used for the month of a hospitalization for heart failure.²⁰ The disutility for adverse events was again weighted to DAPA-HF proportions of mild (one-third of disutility of a hospitalization for causes other than heart failure) and severe (disutility of a hospitalization) adverse events.

As a sensitivity analysis, we tested an alternative model with generally higher utility estimates (eMethods 12 and eTable 9 in the Supplement). This model used prior literature estimates of utility based on populations similar to those in the DAPA-HF trial. We also removed the age-related decline in utility and estimated the dapagliflozin effect on utility based on the ratio of KCCQ-TSS change in DAPA-HF to PARADIGM-HF (Prospective Comparison of ARNI [Angiotensin Receptor–Neprilysin Inhibitor] with ACEI [Angiotensin-Converting–Enzyme Inhibitor] to Determine Impact on Global Mortality and Morbidity in Heart Failure Trial) (annual utility change of 0.017).⁴³

We performed 1-way sensitivity analyses across all model parameters (Table 1 and eTables 1, 6, and 9 in the Supplement). Parameters were varied across 95% CIs where available. Conservatively wide distributions were estimated for other parameters. We performed threshold analyses of key parameters (eg, dapagliflozin mortality effect and cost) to determine values at which dapagliflozin became high-value (ICER <$50 000) or low-value (ICER >$150 000) treatment. Two-way sensitivity analyses were performed for parameter combinations hypothesized to have significant interactions.

We performed several clinically relevant scenario analyses (eMethods 14 in the Supplement). First, in 2 alternate analyses, we assumed dapagliflozin was only effective for 18 months (median DAPA-HF follow-up) or had decreasing effectiveness after 18 months. Second, we assumed the risk of advanced therapies increased after hospitalization for heart failure to incorporate a dapagliflozin effect on advanced therapy rates.^44,45 Third, we modeled a pill disutility of 0.5% of baseline utility among patients receiving dapagliflozin.^46,47 Fourth, we modeled an equivalent effect of dapagliflozin in the subgroups with mild and moderate impairment of health status due to heart failure to evaluate whether timing of initiation alone affects the value of dapagliflozin. Finally, based on evidence from DAPA-HF and other SGLT2 inhibitor trials, we modeled a dapagliflozin effect on noncardiovascular death using the rate ratio in DAPA-HF of 0.89.^4,5,8

We performed a probabilistic sensitivity analysis to evaluate the overall uncertainty of our model’s results (eMethods 15 in the Supplement). We simultaneously resampled all parameters across their uncertainty distributions for 10 000 simulations (distributions in eTables 1, 6, and 9 in the Supplement).

We compared our model outcomes with DAPA-HF trial event rates during the median trial follow-up period of 18.2 (range, 0-27.8) months (eTables 4 and 5 in the Supplement).⁸ Our model produced an SOC cardiovascular mortality rate of 7.9 events per 100 person-years compared with 7.9 events per 100 person-years in the DAPA-HF trial. The model’s cardiovascular mortality rate ratio with dapagliflozin was 0.82 compared with 0.82 (95% CI, 0.69-0.98) in the DAPA-HF trial.

In our model, patients in the SOC arm survived a mean of 8.5 life-years and had a mean of 2.0 hospitalizations for heart failure. Patients in the dapagliflozin arm lived a mean of 0.78 more life-years with a mean of 0.13 fewer hospitalizations for heart failure. On a discounted basis, the SOC arm had a mean of 5.2 QALYs at a lifetime cost of $145 371, whereas the dapagliflozin arm experienced an additional 0.46 QALYs at an additional cost of $38 212, which included $35 708 in dapagliflozin costs. Dapagliflozin therapy had an intermediate value compared with SOC, with an ICER of $83 650 per QALY gained (Table 2).

We found that dapagliflozin had similar cost-effectiveness among patients with and without diabetes (Table 2). Among patients with diabetes, dapagliflozin led to fewer incremental QALYs (0.43) and costs ($34 367) compared with patients without diabetes (0.49 QALYs and $42 191, respectively). The ICERs were similar at $79 726 and $85 420 per QALY gained for patients with and without diabetes, respectively.

We found similar results among patients with mild and moderate impairment of health status due to heart failure. In the subgroup with mild impairment, patients in the SOC arm lived a mean of 10.0 years with 2.4 hospitalizations for heart failure. Patients in the dapagliflozin arm lived an additional 1.0 year with 0.08 fewer hospitalizations for heart failure at an incremental cost of $44 813, resulting in a cost per QALY gained of $78 483. In the moderate impairment subgroup, patients in the SOC arm lived a mean of 6.5 years with 1.6 hospitalizations for heart failure. Patients in the dapagliflozin arm lived an additional 0.5 years with 0.18 fewer hospitalizations for heart failure at an incremental cost of $30 262, resulting in a cost per QALY gained of $97 608.

In 1-way sensitivity analyses (Figure 1), the decision was most sensitive to the HR for cardiovascular mortality with dapagliflozin, the duration of dapagliflozin effectiveness, and dapagliflozin cost. Across the 95% CI (0.69-0.98) for HR of cardiovascular mortality, the ICER spanned from $58 747 to $361 739 per QALY gained (Figure 1). For an ICER of $50 000 or less, the dapagliflozin HR for cardiovascular mortality would need to be less than 0.59; for an ICER $100 000 or less, less than 0.86; and for an ICER of $150 000 or less, less than 0.92. In the case where dapagliflozin is only effective for 18 months (trial duration), it costs $242 096 per QALY gained (Figure 1). To meet willingness-to-pay thresholds of $100 000 and $150 000 per QALY, dapagliflozin would need to be effective for at least 111 and 44 months, respectively. Dapagliflozin’s effectiveness would need to decrease by 0.5% or less and by 3.0% or less monthly in the posttrial period to cost less than $100 000 and $150 000 per QALY gained, respectively. The results were notably insensitive to the risk of dapagliflozin intolerance or the effect on hospitalizations for heart failure (Figure 1).

The cost of dapagliflozin may vary across payors and over time. In our main analysis, we used the Medicare Part D cost of $473.64. Varying this cost by ±50% led to a cost per QALY gained from $44 568 to $122 731 (Figure 1 and Figure 2). The dapagliflozin cost would need to decrease by 43% (from $474 to $270) to cost less than $50 000 per QALY gained. In Figure 2, we display dapagliflozin’s ICER across different subgroups and estimated prices including uninsured patients, wholesale acquisition cost, Medicare Part D cost with estimated rebate, and a scenario with multiple available generic formulations (eTable 9 in the Supplement). The results were insensitive to changes in other cost parameters.

The baseline utility may vary substantially across different populations with heart failure. However, our results were relatively insensitive to baseline utility estimate, assumptions regarding utility decline, addition of a pill disutility, or other utility parameters (eTables 10 and 11 in the Supplement). In an alternate high-utility scenario, dapagliflozin’s ICER remained similar ($73 581 per QALY gained) (eTable 12 in the Supplement).

In our alternate model linking the HR of advanced therapies to recent hospitalizations for heart failure, dapagliflozin’s ICER changed minimally owing to the relative rarity of advanced therapies (eTable 11 in the Supplement). When modeling an equal effect of dapagliflozin across heart failure health status subgroups, the incremental life-year increase from dapagliflozin was 0.07 years greater in the subgroups with mild vs moderate impairment of health status due to heart failure, and the ICER was similar ($103 778 vs $91 820 per QALY gained, respectively). Incorporating effects on noncardiovascular death decreased the ICER to $74 800 per QALY gained. Finally, Figure 3A shows a threshold analysis demonstrating the interaction of the probability of cardiovascular mortality in the SOC arm (displayed as mean survival) and the effect of dapagliflozin on cardiovascular mortality. Several other 2-way sensitivity analyses of parameters hypothesized to have interactions did not demonstrate notable interactions (eTable 13 in the Supplement),

In the probabilistic sensitivity analysis, dapagliflozin had an ICER below $50 000 per QALY gained in 8% of simulations, below $100 000 in 65%, and below $150 000 in 89%. Details are given in Figure 3B and eFigure 2 in the Supplement.

We have demonstrated that dapagliflozin provides intermediate value, with a cost per QALY gained of $83 650 among patients with HFrEF based on the DAPA-HF trial. Dapagliflozin had similar value among patients with and without diabetes and patients with mild and moderate impairment of health status due to heart failure. The results were most sensitive to dapagliflozin’s effect on cardiovascular mortality and its cost. In addition, dapagliflozin would need to remain effective for at least 44 months to have a cost per QALY gained of less than $150 000.

Although dapagliflozin decreases both hospitalization for heart failure and cardiovascular mortality, its intermediate value largely depends on the effect estimate on cardiovascular mortality. We found that dapagliflozin must reduce cardiovascular mortality by at least 8% and 14% for the cost per QALY gained to remain below $150 000 and $100 000, respectively. The DAPA-HF 95% CI ranged from a reduction of 2% to 31%. This finding emphasizes the importance of additional data to better estimate the extent to which dapagliflozin can reduce cardiovascular mortality. Pooling results with empagliflozin from the EMPEROR-Reduced (Empagliflozin Outcome Trial in Patients With Chronic Heart Failure With Reduced Ejection Fraction) trial led to fairly similar estimates of cardiovascular mortality (HR, 0.86; 95% CI, 0.77-0.98).⁴⁸

The DAPA-HF trial suggested potential heterogeneity of treatment effect across disease severity subgroups. The treatment effect point estimate for the primary composite outcome was greater for patients with milder disease severity based on NT-proBNP levels and KCCQ-TSS score and was significantly greater for NYHA class. We found that dapagliflozin provided similar value across heart failure health status strata, although in the subgroup with mild health status impairment due to heart failure, the cost per QALY gained was lower, and there was a larger absolute life-year and QALY benefit despite lower baseline event rates. This finding emphasizes the potential benefit of starting dapagliflozin therapy early.

Although SGLT2 inhibitors have demonstrated a broad range of benefits, we conservatively limited our analysis to the effect on heart failure outcomes. In DECLARE-TIMI 58 (Dapagliflozin Effect on Cardiovascular Events Trial) and CREDENCE (Canagliflozin and Renal Events in Diabetes and Nephropathy Clinical Evaluation Trial), dapagliflozin and canagliflozin, respectively, significantly reduced renal outcomes, including kidney failure and death.^6,7 In DAPA-HF, there were fewer noncardiovascular deaths among patients in the dapagliflozin arm. Including a reduction of noncardiovascular death reduced the ICER further. Other non–heart failure effects may further improve the value among patients who are at high risk for those outcomes.

The cost of dapagliflozin likely varies substantially across payors. In addition, publicly available drug cost estimates often fail to account for manufacturer rebates. We modeled the cost-effectiveness of multiple different prices; with our cost estimate incorporating manufacturer rebate, we found dapagliflozin has intermediate value, with an ICER of $61 139 per QALY gained (Figure 2 and eTable 9 in the Supplement). The cost of dapagliflozin would need to exceed $875 (85% higher than the estimated Medicare Part D paid price to pharmacies) to be low value. Most notably, the US patent for dapagliflozin expired in October 2020, and multiple generic options would be expected to enter the market in the coming years, driving price competition that could significantly lower the cost of dapagliflozin. If the introduction of generic drugs drops the price below $270, dapagliflozin would have high value.

Our use of the KCCQ to stratify our cohort on heart failure health status has advantages over traditional NYHA stratification. Being patient-reported, the KCCQ estimates quality of life directly from patients, which is often inconsistent with clinician assessment.⁴⁹ The KCCQ is more prognostic and reproducible as a standardized measurement tool, whereas the NYHA class has significant interoperator variability.^50,51 Using the KCCQ will help clinicians compare their patients with a given published analysis. Therefore, we believe using patient-reported health status as an estimate of illness severity stratification is an appealing alternative to traditional approaches in clinical trials and economic analyses.

Previous cost-effectiveness analyses, including those by McEwan et al,²⁷ Savira et al,⁵² and Yao et al,⁵³ have found that dapagliflozin is cost-effective among patients with HFrEF. These analyses focused on non-US settings, including the UK, with an ICER of £5822 (US$7681); Germany, with an ICER of €5379 (US$6044); Spain, with an ICER of €9406 (US$10 569); Australia, with an ICER of $12 482 (US$8759); and China, with an ICER of $3828. McEwan et al²⁷ used primary data from the DAPA-HF trial and estimated 0.48 QALYs gained with dapagliflozin at an incremental cost of £2780 in the UK compared with 0.46 QALYs gained and incremental cost of $38 212 in our analysis. Their markedly lower ICER largely stems from differences in the annual cost of dapagliflozin ($5684 in the US vs £477 in the UK [US$629]) as well as higher health care costs in the US. In both analyses, exploratory subgroup analyses by diabetes and heart failure health status showed similar ICERs.

It is important to note that fewer than 1% of the trial population had NYHA class IV. The effectiveness, and thereby cost-effectiveness, in this population remains unclear.⁸ In addition, only 11% of patients were receiving sacubitril-valsartan. Although a post hoc analysis indicated the effects of dapagliflozin were independent of sacubitril-valsartan use, further evaluation of the interaction between dapagliflozin and sacubitril-valsartan will be important.

Understanding therapy cost-effectiveness is critical for decision-making for health care systems and payors. However, affordability to patients remains an equally if not more important issue with regard to therapy access and adoption. The out-of-pocket costs will likely play a major role in driving suboptimal use of dapagliflozin. For therapies of high or intermediate value, payors and health care systems must reduce financial barriers that contribute to decreased use of health care services and worse outcomes.

This study has some limitations. Our effects and transition probabilities were largely based on a single trial; however, this was a large trial that is consistent with the literature with regard to the use of dapagliflozin in patients who have diabetes.^6,54 Also, limited evidence is available on the duration of dapagliflozin’s effectiveness. Given the sensitivity of the model to this parameter, this is an important question for postmarket monitoring. In addition, although we believe the KCCQ is an appealing tool for reproducible and prognostic stratification of patients with heart failure, it does not provide an exact estimate of utilities. In addition, although several SGLT2 inhibitors are being studied for HFrEF, this analysis does not indicate broad cost-effectiveness across this class. Finally, we did not model long-term transitions in heart failure severity owing to limited data and focus on the DAPA-HF trial.

This economic evaluation study demonstrates that dapagliflozin provides intermediate value among patients with HFrEF regardless of diabetes or heart failure–related health status. This effect is largely driven by a decrease in cardiovascular mortality. These results were insensitive to variation of model parameters except for the HR of cardiovascular mortality with dapagliflozin treatment, the cost of dapagliflozin, and the duration of effectiveness. Additional data regarding the mortality effect and long-term effectiveness of dapagliflozin are critical to a better understanding of its cost-effectiveness.

Accepted for Publication: March 5, 2021.

Published Online: May 26, 2021. doi:10.1001/jamacardio.2021.1437

Corresponding Author: Justin T. Parizo, MD, Division of Cardiovascular Medicine, Stanford University School of Medicine, 870 Quarry Rd, Falk Building, Cardiovascular Medicine, Stanford, CA 94305 (jparizo@stanford.edu).

Author Contributions: Dr Sandhu was the principal investigator. Drs Parizo and Sandhu had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Parizo, Goldhaber-Fiebert, Salomon, Khush, Spertus, Sandhu.

Acquisition, analysis, or interpretation of data: Parizo, Heidenreich, Sandhu.

Drafting of the manuscript: Parizo.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Parizo, Salomon, Sandhu.

Obtained funding: Parizo.

Administrative, technical, or material support: Parizo, Sandhu.

Supervision: Goldhaber-Fiebert, Salomon, Khush, Sandhu.

Conflict of Interest Disclosures: Dr Khush reported receiving personal fees from CareDx, Inc, outside the submitted work. Dr Spertus reported consulting for Janssen Pharmaceutica NV, Novartis International AG, Amgen Inc, Bayer AG, Myokardia, and Merck & Co, Inc, on patient-reported outcomes; serving as study principal investigator for Janssen; receiving grants from Abbott Vascular for collection of patient-reported outcome data; serving on the scientific advisory board of United Healthcare; serving as board member for Blue Cross Blue Shield of Kansas City outside the submitted work; and holding a patent for copyright to the Seattle Angina Questionnaire, Kansas City Cardiomyopathy Questionnaire, and Peripheral Artery Questionnaire with royalties paid. No other disclosures were reported.

Funding/Support: The research reported in this publication was supported by awards 1F32HL149253-01 (Dr Parizo) and 1K23HL151672-01 (Dr Sandhu) from the National Heart, Lung, and Blood Institute of the National Institutes of Health.

Role of the Funder/Sponsor: The sponsor had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.