Cost-effectiveness of de-escalation from micafungin versus escalation from fluconazole for invasive candidiasis in China

Abstract

Aims: Guidelines on treating invasive candidiasis recommend initial treatment with a broad-spectrum echinocandin (e.g. micafungin), then switching to fluconazole if isolates prove sensitive (de-escalation strategy). This study aimed to evaluate the cost-effectiveness of de-escalation from micafungin vs escalation from fluconazole from a Chinese public payers perspective.

Materials and methods: Cost-effectiveness was estimated using a decision analytic model, in which patients begin treatment with fluconazole 400 mg/day (escalation) or micafungin 100 mg/day (de-escalation). From Day 3, when susceptibility results are available, patients are treated with either fluconazole (if isolates are fluconazole-sensitive/dose-dependent) or micafungin (if isolates are resistant). The total duration of (appropriate) treatment is 14 days. Model inputs are early (Day 3) and end-of-treatment mortality rates, treatment success rates, and health resource utilization. Model outputs are costs of health resource utilization over 42 days, incremental cost per life-year, and incremental cost per quality-adjusted life-year (QALY) over a lifetime horizon.

Results: In the base-case analysis, the de-escalation strategy was associated with longer survival and higher treatment success rates compared with escalation, at a lower overall cost (–¥1,154; –175 United States Dollars). Life-years and QALYs were also better with de-escalation. Thus, this strategy dominated the escalation strategy for all outcomes. In a probabilistic sensitivity analysis, 99% of 10,000 simulations were below the very cost-effective threshold (1 × gross domestic product).

Limitations: The main limitation of the study was the lack of real-world input data for clinical outcomes on treatment with micafungin in China; data from other countries were included in the model.

Conclusion: A de-escalation strategy is cost-saving from the Chinese public health payer perspective compared with escalation. It improves outcomes and reduces costs to the health system by reducing hospitalization, due to an increase in the proportion of patients receiving appropriate treatment.

Keywords:

• Echinocandin
• micafungin
• fluconazole
• invasive candidiasis
• cost-effectiveness

Background

Invasive candidiasis is a major cause of morbidity and mortality in hospitalized patients in many countries worldwide¹. Those in intensive care units (ICUs) are particularly vulnerable, as they have multiple risk factors for developing invasive candidiasis, including a prolonged hospital stay, use of broad-spectrum antibiotics, and use of intravascular catheters². Furthermore, invasive candidiasis-associated mortality in this setting is particularly high¹, with attributable mortality estimated at up to 47%; however, crude rates vary depending on the study design and patient population². In a prospective, observational study of ICUs at 10 university hospitals in China, mortality rates associated with invasive infection with C. albicans or C. tropicalis were 68.5% and 74.1%, respectively³.

Following its approval in the 1990s, fluconazole became one of the most widely used antifungal agents to treat and prevent invasive candidiasis⁴. Although C. albicans has retained sensitivity to fluconazole, there has been an increase in other Candida species with reduced susceptibility or resistance to fluconazole^4,5. According to a recently published systematic review, the relative frequency of C. albicans and non-albicans species amongst in-patients with invasive candidiasis varies in different parts of the world⁶. For example, the proportion of C. albicans was highest in North and Central Europe and the US, while non-albicans species were more common in South America, Asia, and Southern Europe⁶. In China, a prospective study of 67 ICUs revealed that 59.9% of invasive candidiasis cases were caused by non-albicans species⁷. Further analyses within this Chinese population found that mortality rates were slightly higher in patients with non-albicans species (38.4% vs 29.6%), although the difference was non-significant⁷. It has also been shown that fluconazole-resistant isolates in cancer patients with invasive candidiasis are associated with an increased risk of mortality compared with those who have fluconazole-susceptible isolates⁸.

Micafungin is an antifungal agent in the echinocandin group, which works by inhibiting beta-(1,3)-D-glucan synthase, an enzyme necessary for the synthesis of an essential component of the fungal cell wall⁹. It has potent fungicidal, broad-spectrum activity against Candida species, including azole-resistant Candida pathogens^10,11. Both European and United States (US) guidelines strongly recommend first-line treatment of invasive candidiasis using echinocandins in non-neutropenic patients^12,13. The US guidelines advocate transition from an echinocandin to fluconazole, within 5–7 days, in patients shown to have infection with fluconazole-sensitive isolates¹³; this approach is referred to as “de-escalation”¹. Compared with “escalation” (the transition from fluconazole to a broad-spectrum antifungal agent if isolates are fluconazole-resistant), de-escalation avoids the delay in initiation of effective treatment for those with fluconazole-resistant invasive candidiasis¹⁴. Nevertheless, fluconazole continues to be used as first-line therapy, partly because of its efficacy against most Candida species, and partly because of its low acquisition costs¹⁴. Consequently, patients with fluconazole-resistant isolates are exposed to an inappropriate treatment until the results of susceptibility testing are available; this, in turn, can increase the duration of hospitalization and increase the associated management costs^15,16. It has been shown that de-escalation from micafungin (to fluconazole) vs escalation from fluconazole (to micafungin) in patients with invasive candidiasis is a cost-effective option in the United Kingdom (UK)¹⁴.

The aim of this study was to evaluate the cost-effectiveness of de-escalation from micafungin vs escalation from fluconazole from the public–payer perspective in China.

Methods

Purpose of the study

This study evaluated the cost-effectiveness of de-escalation from micafungin vs escalation from fluconazole, using a decision analytic model. Patients treated according to the de-escalation strategy start treatment with micafungin and are switched to fluconazole if isolates prove fluconazole-sensitive. For the escalation strategy, patients start treatment with fluconazole and switch to micafungin if isolates are fluconazole-resistant.

The economic analysis was performed from a public-payer perspective in China, and all costs considered in the model were based on 2015 pricing, or adjusted accordingly.

Model design

The decision analytic model is based on that used in a UK-based study, to assess the cost-effectiveness of de-escalation¹⁴. According to the model, treatment begins as soon as a sample is taken for culture. Patients start treatment with fluconazole 400 mg/day (escalation approach) or micafungin 100 mg/day (de-escalation approach) (Figure 1)¹⁷. On Day 3, when results of the susceptibility testing are available, patients still alive in the escalation arm can: continue on fluconazole (if isolates are sensitive), have their fluconazole dose increased to 800 mg/day (if isolates are dose-dependent), or switch to micafungin (if isolates are resistant). Those in the de-escalation arm can continue micafungin (fluconazole-resistant isolates) or switch to fluconazole 400 mg (fluconazole-sensitive isolates) or 800 mg (dose-dependent isolates) (Figure 1). To simplify the model, it was assumed that all patients are non-neutropenic and that invasive candidiasis is identified in all patients.

Figure 1. Decision analytic model structure. Fluconazole-sensitive: MIC <8 μg/mL; fluconazole dose-dependent: MIC =16–32 μg/mL; fluconazole-resistant: MIC ≥64 μg/mL^14,17. Abbreviation. MIC, minimum inhibitory concentration.

The total duration of appropriate treatment used in the model was 14 days, i.e. in patients who escalate to micafungin on Day 3, the overall treatment period was 17 days. Treatment success, defined as clearance of Candida from the bloodstream and resolution of symptoms attributable to candidemia, was evaluated at the end of treatment (Day 14/Day 17). If treatment was not successful, patients on fluconazole were switched to micafungin and those on micafungin were titrated to a dose of 200 mg/day.

Epidemiology (base-case analysis)

Data relating to Candida species distribution and fluconazole susceptibility were obtained from the China-SCAN study, in which the prevalence of Candida species and their antifungal susceptibility were determined separately¹⁸. The study, which was conducted in ICUs across China, analyzed 389 Candida isolates from 244 patients. The most frequently identified species and their susceptibility to fluconazole are summarized in Supplementary Table 1.

Model inputs (base-case analysis)

Inputs used in the model were early (Day 3) and late (Day 14/17) mortality, treatment success rates, and health resource utilization (Table 1)^8,16,19–26. Early mortality was based on rates reported on Day 5 for patients with invasive candidiasis treated with appropriate and inappropriate therapy⁸. Late mortality for patients with fluconazole-sensitive or dose-dependent isolates was based on data from the same study⁸. Late mortality for those with fluconazole-resistant isolates was derived from the weighted average from three micafungin studies^19–21. Clinical success rates were based on data from published studies on fluconazole^22–25 and micafungin^19–21,26. Health resource utilization comprised treatment costs and excess hospitalization incurred by patients on inappropriate treatment only, based on data from a published study¹⁶.

Table 1. Model input parameters.

	Fluconazole-sensitive/dose-dependent	Fluconazole-resistant	Reference
Mortality
Early (Day 3) (%)
Escalation strategy	9.8 (6.1; 14.4)	14.5 (7.3; 17.1)	Slavin et al.^8 (2010)
De-escalation strategy	9.8 (6.1; 14.4)	9.8 (6.1; 14.4)	Slavin et al.^8 (2010)
Late (end-of-treatment) (%)
Escalation strategy	21.3 (15.5; 27.6)	50.8 (38.2; 63.4)	Slavin et al.^8 (2010)
De-escalation strategy	21.3 (15.5; 27.6)	19.9 (24.3; 31.3)	Kuse et al.^19 (2007); Pappas et al.^20 (2007); Kubiak et al.^21 (2010)
Treatment success
Fluconazole	65.6 (59.8; 71.1)		Rex et al.^22 (1994); Anaissie et al.^23 (1996); Phillips et al.^24 (1997); Sipsas et al.^25 (2009)
Micafungin	76.9 (73.6; 80.1)	76.9 (73.6; 80.1)	Kuse et al.^19 (2007); Pappas et al.^20 (2007); Marfo and Guo^26 (2009); Kubiak et al.^21 (2010)
Health resource utilization
Excess LOS (general ward, days)
Escalation strategy	0	7.7 (0.6; 13.5)	Zilberberg et al.^16 (2010)
De-escalation strategy	0	0
Excess LOS (ICU, days)
Escalation strategy	0	0	Zilberberg et al.^16 (2010)
De-escalation strategy	0	0

All data are mean (95% CI).

Abbreviations. CI, confidence interval; ICU, intensive care unit; LOS, length of stay.

Costs (base-case analysis)

The costs for fluconazole and micafungin (Table 2)^14,27–30 were obtained from a pricing document issued by the Price Bureau in Shanghai and published on the official website of Shanghai Pharmaceutical Trade Association²⁷. The costs of hospitalization (Table 2) were based on data from a published study conducted in China²⁸. Daily hospitalization costs were converted from US dollars (USD) to Chinese Yuan (¥), using an exchange rate of 8.192, and inflated to 2015 prices using government data³¹.

Table 2. Costs, life expectancy values, and utilities used in the model.

	Mean^a	Reference
Medication costs, ¥ (USD)^a		Drugs’ Price Information in Shanghai^27 (2015)
Micafungin (100 mg)	1,032 (157)
Fluconazole (400 mg)	439 (67)
Fluconazole (800 mg)	877 (133)
Daily hospitalization costs^b^,^c
¥	4,374 (2,187–6,560)	Xie et al.^28 (2008)
USD	664 (332–995)
Life expectancy, years	7.11 (3.95–17.51)	Masterton et al.^14 (2013)
Utilities
Invasive candidiasis	0.50 (0.37–0.63)	Masterton et al.^14 (2013)
Underlying disease	0.60 (0.50–0.87)	Diekema et al.^29 (2002); Health Protection Authority^30 (2008); Masterton et al.^14 (2013)

^a For consistency throughout the manuscript, a 2017 currency conversion rate was applied.

^b 95% CIs where applicable.

^c On general ward.

Abbreviations. CI, confidence interval; USD, United States dollars.

Only reimbursed costs were included, and a discount rate of 4% was applied. Based on data from the Chinese Government’s regulation on basic medical insurance^32,33, it was assumed that 85% of the total costs would be reimbursed.

Outcomes

The cost of health resource utilization was evaluated over a time horizon of 42 days. This time point was selected because, on average, mortality was assessed in published micafungin studies at Day 42^19–21.

Results from the model were extrapolated over a lifetime horizon, using the average life expectancy for an ICU survivor and the corresponding utilities (Table 2) to calculate the incremental cost per life-year and incremental cost per quality-adjusted life-year (QALY). Costs are presented in ¥ and in 2017 USD³⁴.

Probabilistic sensitivity analysis

Probabilistic sensitivity analysis was conducted to test the uncertainty of the base-case results for the incremental cost-effectiveness ratio (ICER) per QALY. Measures of distribution were obtained from the literature for all parameters, except the cost of hospitalization, which was conservatively estimated as 60% of the mean. In total, 10,000 Monte Carlo simulations were conducted to obtain 95% confidence ranges for the costs and QALYs for the escalation and de-escalation strategies, and for the ICER. A cost-acceptability curve was then generated to determine the proportion of simulations that were below cost-effectiveness thresholds recommended by the World Health Organization (WHO)³⁵: less than 1 × gross domestic product (GDP) (¥47,457 in 2013; 7,201 USD) was defined as “very cost-effective”; less than 3 × GDP (¥142,371; 21,604 USD) as “cost-effective”.

One-way sensitivity analysis

A one-way sensitivity analysis was conducted to assess the impact of varying the following parameters on the ICER: utility of invasive candidiasis; success rates for micafungin and fluconazole; resistance rates for fluconazole; life expectancy; cost of a stay on a general ward; and level of reimbursement. Upper and lower 95% confidence intervals (CIs) were used for the invasive candidiasis utility and treatment success rates. The minimum and maximum values for utilities and life expectancy associated with underlying conditions were used to vary the effect of life expectancy on the results. The cost of a stay on a general ward was varied by 50% and the level of reimbursement was varied between 50% and 100% in a sensitivity analysis.

Results

Costs and outcomes results

It was estimated (based on results of the model) that 26.5% of patients with invasive candidiasis in China have dose-dependent isolates and 9.5% have fluconazole-resistant isolates. Thus, more than one-third of patients would receive inappropriate first-line treatment via the escalation strategy.

In the base-case analysis, the de-escalation strategy was associated with longer survival and higher treatment success rates compared with escalation, equivalent to incremental benefits of 2.9% and 2.2%, respectively (Table 3). In those with fluconazole-resistant isolates, the de-escalation strategy was also associated with fewer deaths (27.7% vs 58.0% for the escalation strategy) and more successfully treated patients (55.6% and 32.3%, respectively) and the incremental benefits vs the escalation strategy were considerably greater.

Table 3. Clinical outcomes associated with de-escalation and escalation strategies (base-case analysis).

	De-escalation	Escalation	Incremental effect (de-escalation vs escalation)	ICER
Clinical outcomes (over 42 days), %
Survival rate	71.1	68.2	2.9%	40,150
Treatment success rate	47.4	45.2	2.2%	52,203
Lifetime effects
Life years	4.60	4.42	0.18	6,327
QALYs	2.77	2.66	0.11	10,465

Abbreviations. ICER, incremental cost-effectiveness ratio; QALYs, quality-adjusted life-years.

Life-years were 4.60 for the de-escalation strategy and 4.42 for the escalation strategy; corresponding QALYs were 2.77 and 2.66, respectively. As costs were lower for the de-escalation strategy, the ICER was dominant for both measures (Table 3).

Overall, the mean cost of the de-escalation strategy was lower than that of the escalation strategy. Although medication costs were higher for the de-escalation strategy, it was not associated with excess hospitalization costs, resulting in an overall net saving of ¥1,154 (175 USD; Table 4). The de-escalation strategy was, therefore, dominant, as it generated better outcomes (survival and treatment success) at a lower cost.

Table 4. Costs associated with de-escalation and escalation strategies (base-case analysis).

	De-escalation	Escalation	Incremental effect (de-escalation vs escalation)
Cost, ¥ (USD)^a
Medication	13,063 (1,982)	11,478 (1,742)	1,585 (241)
Excess hospitalization	0 (0)	2,739 (416)	−2,739 (−416)
Total	13,063 (1,982)	14,218 (2,158)	−1,154 (−175)

^a For consistency throughout the manuscript, a 2017 currency conversion rate was applied.

Abbreviation. USD, United States dollars.

Probabilistic sensitivity analysis

Results of the probabilistic sensitivity analysis are shown in Supplementary Figure 1, and mean (95% CI) values for costs and QALYs are summarized in Supplementary Table 2. The corresponding ICER (95% CI) was –¥9,774 (–¥84,077; ¥16,553) or −1,483 USD (–12,758 USD; 2,511 USD). Overall, 99% of simulations were below the very cost-effective threshold (1 × GDP), and the de-escalation strategy dominated the escalation strategy in 57.5% of simulations.

One-way sensitivity analysis

Results of the one-way sensitivity analysis are shown in Figure 2. Variables with the greatest effect on the ICER were excess hospitalization on a general ward, general ward costs, and fluconazole resistance.

Figure 2. ICERs for the de-escalation strategy vs the escalation strategy (one-way sensitivity analysis). ^aOn general ward. ^bFor consistency throughout the manuscript, a 2017 currency conversion rate was applied. Abbreviations. ICER, incremental cost-effectiveness ratio; USD, United States dollar.

Discussion

The results of the current study demonstrate that, in China, de-escalation from micafungin is associated with improved clinical outcomes and lower overall costs compared with escalation from fluconazole. The higher overall costs with the escalation strategy are likely to reflect the delay in initiation of effective treatment for those with fluconazole-resistant systemic Candida infection; this has been shown to increase the duration of hospitalization and, consequently, the management costs^15,16. De-escalation also increased both life-years and QALYs. In other words, it dominated the escalation strategy for all outcomes. Results of the probabilistic sensitivity analysis emphasize the robustness of the results, as 99% of the 10,000 simulations were below the WHO very cost-effective threshold (1 × GDP).

The epidemiology of invasive candidiasis varies in different parts of the world^36,37. For example, the incidence of C. albicans appears to be higher in Europe than in China³⁷. C. glabrata is the second most common species after C. albicans in North America and some parts of Europe, while C. parapsilosis or C. tropicalis are relatively more common in Asia, Australia, Latin America, and other parts of Europe^37,38. Fluconazole resistance rates also vary between countries. For example, in China, resistance rates for C. albicans and C. glabrata were estimated to be 9.6% and 4.0%, respectively¹⁸, while the corresponding rates in the UK were 0.6% and 11.6%, respectively¹⁴. These data highlight the importance of using country-specific epidemiological information when evaluating the cost-effectiveness of antifungal treatments for invasive candidiasis. Indeed, the results of the one-way sensitivity analysis in the current study indicate that fluconazole resistance rates are one of the main drivers of cost-effectiveness of the de-escalation strategy.

A previous analysis, conducted using a similar model, demonstrated that the micafungin de-escalation strategy was more cost-effective than the fluconazole escalation strategy from a UK National Health Service perspective¹⁴. De-escalation was associated with greater medication costs, but was more effective overall, generating an ICER (£25,673) below the National Institute for Health and Care Excellence threshold for cost-effectiveness (£30,000). In the current analysis, the de-escalation strategy was cost saving. This likely reflects the higher incidence of non-albicans species and greater resistance of C. albicans to fluconazole in China, leading to a higher proportion of patients avoiding inappropriate treatment, thus reducing excess hospitalization and subsequent costs. These results are consistent with data from the one-way sensitivity analysis in the current study, which showed that the duration of excess hospitalization and the cost of a hospital stay are two of the main drivers of cost-effectiveness of the de-escalation strategy.

We do not know how the differential effect of echinocandins and fluconazole against biofilms may have contributed to our findings. However, infective endocarditis occurs frequently in patients with invasive Candida infections, so micafungin may also have reduced medical costs in affected patients in this study.

One of the limitations of economic modeling is that its application generally requires simplification of strategies used in everyday clinical practice. In the current model, it was assumed that all patients develop invasive candidiasis; however, in a study of patients with severe sepsis in the ICU setting in China, only 25.5% developed candidemia³. When this value, reflective of real-world practice, was used in the model (and the costs for Candida-negative patients comprised only treatment costs for the first 3 days), the ICER was ¥25,061 (3,803 USD), which is substantially lower than the very cost-effective threshold (¥47,457; 7,201 USD). Zilderberg et al.³⁹ also reported a low proportion of patients (14%; range = 5–28) with suspected invasive candidiasis in various ICU settings actually having confirmed candidemia; together, these results suggest that confirming Candida isolates in the current study would not have had a significant impact on the results. For these reasons, a “no yeasts isolated” group was not included in the model.

For the purposes of the model, it was assumed the results of susceptibility testing would be available by Day 3. In reality, this process may take longer, particularly if samples are sent to an external laboratory for testing. In this case, it is likely that the outcomes for the escalation strategy may be poorer, as those with fluconazole-resistant infection would be treated with inappropriate treatment for longer.

Another assumption of the model was that follow-up costs in invasive candidiasis survivors would not differ between the two strategies and, for this reason, long-term follow-up costs were not included. This is consistent with the theory put forward by Gold et al.⁴⁰, that future health costs unrelated to the intervention under consideration can be excluded, as the range of interventions used is expected to be the same. The WHO cost-effectiveness guidelines recommend that a consistent approach is used, but do not make specific recommendations regarding the inclusion or exclusion of health costs associated with extended life-years⁴¹.

One limitation of the model is the lack of real-world input data from China on clinical outcomes. Data for micafungin were obtained from studies conducted across a number of countries (Europe, North America, Brazil, India, Thailand, and South Africa)^19–21,26. For fluconazole, treatment success rates were based on studies carried out in North America and Canada^22–24, while early and late mortality rates were obtained from a single study conducted in Australia⁸. Nevertheless, the higher mortality rate in those with isolates resistant to fluconazole has been substantiated elsewhere⁴². Furthermore, varying the mortality rates associated with fluconazole-susceptible and fluconazole-resistant strains did not have a substantial effect on the ICER. Another limitation was the lack of an interrupted time-series analysis to proof the results, due to limitations of the model design and the data collected.

Conclusions

In summary, this analysis shows that a de-escalation strategy is cost-saving from the Chinese public health payer perspective. De-escalation in this setting improves outcomes, in particular life-years and QALYs, and reduces cost to the health system by reducing hospitalization due to an increase in the proportion of patients who receive the appropriate treatment. These results are particularly pertinent given the difficulty in differentiating non-albicans and albicans infections on the basis of clinical symptoms alone and the increasing prevalence of fluconazole-resistant strains. They also support guideline recommendations on the first-line use of echinocandins for invasive candidiasis.

Transparency

Declaration of funding

This study was funded by Astellas Pharma China Inc.

Declaration of financial/other relationships

The authors report the following disclosures: CC and EK are employees of QuintilesIMS; they were formerly employed by IMS Health—who were contracted by Astellas China to conduct the analysis and provide writing support—during the conduct of the study and preparation of the manuscript. LW is an employee of Astellas Pharma China Inc. The remaining authors have no conflicts to report.

Acknowledgements

Medical writing support, which was funded by Astellas Pharma Inc, was provided by Nicola French, PhD, of Bioscript Medical.