Reply: Cost-effectiveness of sacubitril/valsartan versus enalapril in patients with heart failure and reduced ejection fraction

This article responds to:

RE: Cost-effectiveness of sacubitril/valsartan versus enalapril in patients with heart failure and reduced ejection fraction

Reply to: https://www.tandfonline.com/doi/full/10.1080/13696998.2018.1503597 Original Article: https://www.tandfonline.com/doi/full/10.1080/13696998.2017.1387119 Cost-effectiveness analyses (CEAs) and the resulting incremental cost-effectiveness ratio (ICERs), defined as cost per quality-adjusted life-year (QALY), are commonly used to guide decisions on resource allocation and to compare the efficiencies of alternative health interventions.

The letter states

The World Health Organization’s Choosing Interventions that are Cost-Effective project (WHO-CHOICE) program¹ defines interventions for which the cost per QALY is less than one gross domestic product (GDP) per capita as very cost-effective, between one and three times GDP per capita as cost-effective and over three times GDP per capita as not cost-effective.

We would like to highlight a 2016 publication by the WHO that WHO-CHOICE has not recommended the indiscriminate sole use of generic GDP-based thresholds in national funding decisions or for setting the price or reimbursement value of a new drug or other intervention². Thresholds should not be misused for the following reasons:

1. Many factors influence the results of CEA, for example, the data used to estimate costs and effects, the choice of comparator, and whether or not sub-groups of the target population are analyzed. Variations in the inputs can have substantial effects on the ICER. If the analyses do not reflect the policy context accurately, over-reliance on ICERs and a fixed cost-effectiveness threshold to guide decision-making may result in the wrong decisions being made.

2. At a technical level, it is important to note that ICERs derived from economic modeling are simply estimates—generally based on several assumptions—produced to indicate the potential value for money of one or more interventions. The construction of economic models is prone to problems and errors, but such models can still be a valuable input for decision-making if well-constructed and validated. However, even well-constructed models can produce a range of estimates, depending on the assumptions adopted and the formulation of the policy question being evaluated. Use of a rigid cost-effectiveness threshold to determine funding decisions may simply encourage the interested parties to tailor their estimates so that they trigger funding.

3. Even if estimated accurately, generic GDP-based cost-effectiveness ratios—or other estimates of willingness to pay—do not provide information on affordability, budget impact, or the feasibility.

Consistent with the above, there is no fixed threshold to determine cost-effectiveness in Singapore. As part of the deliberative process, our decision-makers make reimbursement decisions after considering a range of factors, such as the clinical need of the patients, comparative clinical effectiveness and safety of the intervention, affordability, and the cost impact to the system, rather than in isolation based on a single threshold value.

We, therefore, do not agree with the letter that sacubitril/valsartan can be considered cost-effective in Singapore at current price, based on an ICER of between SGD 35,578–38,819/QALY gained (using the model published by McMurray et al.³) that fall below the GDP per capita of Singapore, which approximates SGD 79,000–80,000.

We believe our cost-effectiveness model is based on inputs and assumptions that are most relevant and reflective of the local Singapore context.

1. Our model structure was adapted from King et al.⁴, who included the New York Heart Association (NYHA) classes as health states, to reflect real-world practice whereby patients have different disease severity stages. In line with King et al., utility values by NYHA classes were obtained from SENIORS trial, and transition probabilities between NYHA classes derived from CARE-HF trial.

2. The base-case time horizon of 10 years reflected local clinical experts’ inputs of the average life expectancy for patients with heart failure in Singapore. It also took into consideration that the long-term safety profile of sacubitril/valsartan remains unknown. Given concerns regarding potential increased risk of dementia from chronic neprilysin inhibition with sacubitril/valsartan⁵, this aspect should be modelled if the proposed time horizon of up to 120 years of age is applied.

3. The prices used in the model were obtained from pharmacy databases across the public healthcare institutions in Singapore.

4. All model inputs, including utilities, time horizon, and prices, were tested in sensitivity analyses.

5. Additional sensitivity analyses were also conducted for various willingness-to-pay thresholds of between SGD 20,000 and SGD 100,000/QALY gained.

Overall, our study suggests a base-case ICER for sacubitril/valsartan vs enalapril of SGD 74,592/QALY gained, with the ICERs ranging from SGD 16,325–1,447,103/QALY gained in sensitivity analyses⁶. The wide range of ICERs reflects high uncertainty in its cost-effectiveness, driven by the uncertainty in its treatment effect to reduce cardiovascular death in the Asian sub-group⁷, and the extrapolation of benefit beyond the follow-up duration of PARADIGM-HF (27 months). The model was not sensitive to hospitalization for HF or utility values. These results need to be considered, alongside other factors, to inform subsidy decision-making in the Singapore healthcare context.

In Australia, the Pharmaceutical Benefits Advisory Committee (PBAC)—that recommends medicines for subsidy listing—also expressed concerns over the uncertainty of the cost-effectiveness of sacubitril/valsartan. Key model drivers were similar: risk of cardiovascular death, time horizon (10 years’ base case), and continuing treatment effect, all of which were considered to favour sacubitril/valsartan. The PBAC eventually recommended its listing, after price reduction and risk sharing arrangements addressed concerns around the cost-effectiveness of treatment being unknown^8,9.

Our findings are also consistent with several published CEAs of sacubitril/valsartan. King et al.⁴ reported the cost-effectiveness of sacubtril/valsartan to range from USD 249,411 per QALY at 3 years to USD 50,959 per QALY gained over a lifetime. Gaziano et al.¹⁰ reported similar findings, with ICER ranging from USD 35,357 per QALY to USD 75,301 per QALY gained when the mortality hazard was varied over the 95% CI. The ICER increased further to USD 135,964 per QALY gained when the benefit only lasting 27 months was modeled. Sandhu et al.¹¹ also reported ICERs of up to USD 120,623 per QALY gained if the treatment was effective only for 27 months.

On the other hand, the CEA by McMurray et al.³ cited in the letter did not test the time horizon in sensitivity analyses, when this is commonly done in other published studies. We note that the CEA by McMurray et al.³ was funded by the company marketing sacubitril/valsartan, and most of the authors have received financial interests from the company. This is of some concern, given it has been shown that the majority of CEAs with financial conflict of interest directly or indirectly support the company’s product^12,13.

Transparency

Declaration of funding

No sponsorship/funding requires declaration.

Declaration of financial/other relationships

The authors have no financial relationships to declare.

Acknowledgments

None reported.