Introduction

Invasive fungal disease is associated with increased mortality, morbidity and use of critical care and hospital beds.^6,7 About half of IFD occurs in non-neutropenic patients in critical care units,¹ and most of these infections are due to the Candida species.^2,3 In the USA, candidaemia has been estimated to lead to excess costs of US$44,000 per episode.⁹³ RCTs have established that antifungal prophylaxis with either fluconazole or ketoconazole is effective in reducing mortality in non-neutropenic, critically ill patients.¹⁹ However, patients included in the RCTs were at high risk of IFD, with baseline risk in the control arm typically exceeding 10%.

The previous chapters have highlighted that the incidence of invasive Candida infection in unselected, non-neutropenic adult patients admitted to NHS critical care units is relatively low (see Chapter 4). Given this low incidence, the costs of prophylaxis and concerns about resistance, it is unclear at what levels of baseline risk prophylaxis is cost-effective for critically ill patients. Indeed, current usual practice in the NHS is not to provide prophylaxis, irrespective of baseline risk. The relative gains and costs of administering antifungal prophylaxis are also anticipated to differ according to the time at which the prophylaxis is administered. In particular, providing prophylaxis to those patients who are judged to be high risk on admission to critical care would entail providing prophylaxis to more patients than waiting and only providing prophylaxis to those patients who were still in critical care, and high risk, after 3 days. Furthermore, it is unclear whether or not it is more cost-effective to consider prophylaxis at single or at multiple time points. Although current usual practice in the NHS is not to provide prophylaxis, this standpoint has not been informed by a careful assessment of the relative cost-effectiveness of alternative prophylaxis strategies for invasive Candida infection. Indeed, no previous study has assessed the cost-effectiveness of using a risk model to define a risk threshold above which to initiate antifungal prophylaxis for preventing invasive Candida infection in non-neutropenic, critically ill adult patients.⁴⁶

This chapter therefore presents an economic evaluation with the aim to report the relative cost-effectiveness of alternative strategies to prevent invasive Candida infection for non-neutropenic, critically ill adult patients admitted to NHS critical care units. The objectives of the economic evaluation were to establish the relative cost-effectiveness of risk assessment using the FIRE Study risk models, followed by initiation of prophylaxis at different thresholds of baseline risk and at different time points; and to assess the relative value of further research to reduce uncertainty about the optimum strategy to adopt.

Methods: overview

The economic evaluation assessed the cost-effectiveness of alternative strategies to risk assessment followed by prophylaxis using the risk models developed for invasive Candida infection. The study compared alternative treatment protocols for providing antifungal prophylaxis to patients identified as high risk (‘interventions’) with providing no prophylaxis (‘current practice’). The prophylaxis treatment regimen evaluated followed current recommendations and is for 400 mg of fluconazole per day.^4,42 There is no specific guideline on the duration of prophylaxis with fluconazole in critical care. A previous systematic review suggested that, across studies, prophylaxis was generally administered until discharge from critical care.¹⁹ For the FIRE Study economic evaluation, prophylaxis was assumed to be applied for 10 days, which is the mean LOS in critical care for the study population. The economic evaluation used a decision-analytical approach to project lifetime cost-effectiveness. The decision model was populated with estimates of PPV (the proportion of those identified as high risk that subsequently developed invasive Candida infection) and NPV (the proportion of those identified as low risk that did not subsequently develop invasive Candida infection) from the FIRE Study risk models at each time point (see Chapter 5), and estimates of the effectiveness of antifungal prophylaxis from systematic reviews of published RCTs.¹⁹ The input parameters were all estimated for patients aged 60 years, the mean age of patients who met the inclusion criteria for the FIRE Study. A probabilistic sensitivity analysis was undertaken to recognise the sampling uncertainty surrounding the input parameters, following general recommendations in the choice of distribution for each parameter.⁹⁴ The main structural assumptions were subjected to sensitivity analyses. Finally, the value of further research was established both overall and for specific parameters.

Methods: base case

Population of interest

The population of interest, represented by the development and validation samples described in Chapter 5, was defined as non-neutropenic adult patients admitted to NHS critical care units, and excluded those with IFD prior to a decision time point and those receiving systemic antifungal therapy as part of routine clinical practice prior to a decision time point. This last criterion excluded a number of patients that received systemic antifungal therapy in the absence of IFD, and may therefore have represented use of antifungal prophylaxis. Figure 14 reports the proportion of patients that were excluded at each time point due to receiving systemic antifungal therapy that would otherwise have met the inclusion criteria for the models. Overall, 1635 patients (3.0%) otherwise eligible for the models were excluded for this reason at any of the three time points. There were 26 cases of invasive Candida infection among these patients, and assuming a RR of invasive Candida infection associated with antifungal prophylaxis of approximately 0.519 we may anticipate that a further 26 cases may have been prevented.

FIGURE 14

Current use of antifungal prophylaxis and potential impact of antifungal prophylaxis on observed outcome. a, Assuming a RR of invasive Candida infection of 0.5 associated with antifungal prophylaxis; b, prior to 24 hours; c, prior to end of calendar day (more...)

Strategies under comparison

The economic evaluation defined three decision time points at which to consider assessment of the risk of invasive Candida infection. These were:

• on admission
• at the end of the first 24 hours, and
• at the end of calendar day 3.

The alternative strategies recognised these time points (Table 22). Each time point defined a decision node, with two possibilities: either assessment of the risk of invasive Candida infection was not undertaken (no risk assessment) or risk assessment was undertaken (risk assessment). For example, under risk assessment on admission, the patients' risk was assessed according to the predicted risk from the on admission FIRE Study risk model. Those patients whose predicted risk of invasive Candida infection during the critical care unit stay exceeded the specified threshold (P_T) were designated ‘high risk’. These high-risk patients were then assumed to receive a single course of prophylaxis with fluconazole.

TABLE 22

Alternative treatment strategies for non-neutropenic, critically ill adult patients

Strategy 1 assumed that there was no assessment of the risk of invasive Candida infection, as is usual practice in UK critical care units for the majority of patients (see Chapter 4). Strategies 2–4 assumed that risk assessment was performed using a single FIRE Study risk model at a single point in time – either on admission, at the end of 24 hours or at the end of calendar day 3. So, for example, strategy 4 assumed that patients' risk was only assessed at the end of calendar day 3 with prophylaxis initiated for those patients defined by the end of calendar day 3 FIRE Study risk model as having high risk of invasive Candida infection. Strategies 5–8 allowed for risk assessment at multiple time points. At any time point, risk assessment was only considered for the subgroup who were still in critical care, not already receiving systemic antifungal therapy, and without IFD. It was assumed that prophylaxis was initiated for those newly defined as high risk at the particular time point.

The risk thresholds (P_T) defined a priori according to the literature and expert opinion were 0.5%, 1%, 2%, 5% and 10%. However, the low incidence of Candida infection in the FIRE Study meant that under some of the strategies there were no infections in those designated high risk, and so 2% was the highest risk threshold considered in the analysis. In strategies involving risk assessment at multiple time points, the same risk threshold was applied at all time points.

Model structure

The model included a hypothetical cohort of 1000 cases with characteristics defined by the patients who met the FIRE Study inclusion criteria. The model structure (Figure 15) recognised the alternative strategies and time points described in Table 22.

FIGURE 15

Structure of the model comparing alternative strategies for assessing risk and initiating prophylaxis for non-neutropenic, critically ill adult patients.

If, on admission, no risk assessment was undertaken (strategies 1, 3, 4 and 6), patients faced the risk of either having (R₁) or not having (1 − R₁) invasive Candida infection within the first 24 hours in the critical care unit. From the ‘no invasive Candida infection’ health state, patients faced a baseline risk of all cause death. For patients predicted to develop invasive Candida infection, an excess risk of death was applied. Under the strategies where risk assessment was undertaken on admission (strategies 2, 5, 7 and 8), the predicted probability of invasive Candida infection at any time during the critical care stay was estimated from the on-admission FIRE Study risk model. The proportion of patients (P₁) whose predicted risk of infection was higher than the risk threshold (e.g. 2%) were judged ‘high risk’ and assumed to receive prophylaxis. For these patients, the probability of developing invasive Candida infection at any time during critical care was estimated by multiplying the PPV from the ‘on-admission’ FIRE model (PPV₁) by the RR of invasive Candida infection associated with receiving antifungal prophylaxis versus no prophylaxis.¹⁹ The proportion of patients (1 − P₁) whose predicted risk of infection was lower than the risk threshold (e.g. 2%) were judged ‘low risk’ and assumed not to receive prophylaxis. [For these patients, the probability of developing invasive Candida infection prior to the next decision time point in the strategy under consideration (or at any time during critical care, if admission was the only decision time point, i.e. strategy 2) was estimated as one minus the corresponding NPV from the ‘on admission’ FIRE model (1 − NPV₁). Note that the NPVs were calculated for each time point because the patients that did not receive antifungal prophylaxis may be reconsidered for risk assessment and prophylaxis at subsequent time points. By contrast, as antifungal prophylaxis was a ‘one-off’ treatment, those that received antifungal prophylaxis had a subsequent risk of invasive Candida infection at any time during the critical care stay.]

The risk of death, conditional on presence or absence of invasive Candida infection, was assumed the same whether or not patients received antifungal prophylaxis.

At the second assessment time point, the model considered those patients still on the critical care unit, not receiving systemic antifungal therapy and without IFD at the end of 24 hours. The strategies with no assessment at this time point (strategies 1, 2, 4 and 7) allowed for patients to face a baseline risk of invasive Candida infection (R₂) before the end of calendar day 3. For strategies with risk assessment at this time point (strategies 3, 5, 6 and 8), prophylaxis was initiated for the proportion of patients (P₂) whose predicted risk from the 24-hour FIRE Study risk model exceeded the risk threshold. The subsequent probability of invasive Candida infection was then calculated by multiplying the PPV of invasive Candida infection from the 24-hour FIRE Study risk model (PPV₂) by the RR of invasive Candida infection. For the proportion of patients (1 − P₂) whose predicted risk from the 24-hour FIRE Study risk model was below the risk threshold, the subsequent probability of invasive Candida infection prior to the next decision time point was estimated as one minus the corresponding NPV from the 24-hour FIRE Study risk model (1 − NPV₂).

At the third assessment time point, the model considered those patients still on the critical care unit, not receiving systemic antifungal therapy and without IFD at the end of the third calendar day. The strategies with no assessment at this time point (strategies 1, 2, 3 and 5), allowed for patients to face a baseline risk of invasive Candida infection (R₃) over the remaining critical care stay. For strategies with risk assessment at this time point (strategies 4, 6, 7 and 8), prophylaxis was initiated for the proportion of patients (P₃) whose predicted risk from the end of calendar day 3 FIRE Study risk model exceeded the risk threshold. The subsequent probability of invasive Candida infection was then calculated by multiplying the PPV of invasive Candida infection from the end of calendar day 3 FIRE Study risk model (PPV₃) by the RR of invasive Candida infection. For the proportion of patients (1 − P₃) whose predicted risk from the end of calendar day 3 FIRE Study risk model was below the risk threshold, the subsequent probability of invasive Candida infection at any subsequent time during critical care was estimated as one minus the corresponding NPV from the end of calendar day 3 FIRE Study risk model (1 − NPV₃).

At each decision node, a proportion of patients left the model because they died or were discharged from critical care. After the last decision node (end of calendar day 3), remaining patients were assigned the mean LOS and cost of those remaining in critical care in the FIRE Study for > 3 days, according to whether or not they had invasive Candida infection. These patients were then assigned lifetime quality-adjusted life-years (QALYs) according to the assumptions detailed below.

Transition probabilities

Risk of invasive Candida infection

The risk of invasive Candida infection (R₁, R₂, R₃) for each decision node without prophylaxis was predicted within each time period (admission to 24 hours, 24 hours to day 3, and after day 3). These baseline risks of infection were estimated from the combined FIRE Study development and validation samples. The probabilistic sensitivity analysis assumed that each probability was drawn from a beta distribution.⁹⁴Table 23 presents the risks of invasive Candida infection for each time period.

TABLE 23

Predicted probabilities of invasive Candida infection in critical care without risk assessment

Positive and negative predictive value The decision model required PPV and NPV for each strategy and for each risk threshold. Estimates of PPV were required to predict the risk of invasive Candida infection at any subsequent time in critical care (as once prophylaxis was initiated, it was assumed that no further risk assessments took place), whereas estimates of NPV only considered the risk of invasive Candida infection prior to the next time point at which risk assessment would take place. For the strategies where multiple risk assessments were undertaken, PPV and NPV were calculated depending on the decision at the previous time points, i.e. they were conditional on the previous predicted risk and assumed risk threshold. To avoid concerns about overfitting from use of the FIRE Study development data set, PPVs and NPVs were estimated from the validation sample only. Table 24 presents PPVs and NPVs for each strategy and risk threshold. The PPVs were low for all strategies; even at the 2% risk threshold and with prophylaxis at each time point, the PPVs remained below 2%. By contrast, the NPVs all exceeded 99%. To recognise the uncertainties in estimating the PPVs and NPVs given the small number of infections, we assumed vague priors with uniform distributions (range from 0 to 1).⁹⁵ Parameter values for the PPVs and NPVs were sampled from the resultant posterior distributions.

TABLE 24

Positive predictive value and NPV according to strategy and threshold

Probabilities of death and relative risk of invasive Candida infection

For patients who did not have invasive Candida infection, the probabilities of death were estimated for the three time periods (admission to 24 hours, 24 hours to day 3, and after day 3) using the combined FIRE Study development and validation samples (Table 25). For patients with invasive Candida infection, the excess risk of death was estimated from the combined data set. The same excess risk of death for patients was applied for each time point and irrespective of whether or not patients had received prophylaxis. The effectiveness of prophylaxis was recognised by taking the RR of invasive Candida infection after prophylaxis from the Cochrane systematic review by Playford et al.¹⁹ The systematic review reported similar RRs across different levels of baseline risk. Hence, we applied the same RR for all time points and all risk thresholds.

TABLE 25

Baseline probabilities of death, RR of death with invasive Candida infection, and RR of invasive Candida infection after prophylaxis

Costs

Risk assessment was assumed (based on expert opinion) to require 10 minutes of nurse time, giving a cost of £8.67.⁹⁶ Prophylaxis costs were calculated assuming a standard regimen of 400 mg for 10 days, with unit costs taken from the British National Formulary (BNF)⁹⁷ (Table 26). This unit cost recognises that, according to the BNF, non-proprietary fluconazole intravenous infusion was available from September 2011. The recommended dose of 400 mg per day would be given as two 100 ml (200 mg) infusions. This would be an appropriate regimen to use for non-neutropenic, critically ill patients when the source of infection is unknown and there is a very high suspicion of invasive Candida infection. Once the patient can absorb, they may, even within critical care, be switched to the lower-cost oral formulation and this, together with the possibility of local discounts, is considered in the sensitivity analysis. The resultant unit cost for intravenous fluconazole of £7.78 per day replaces the unit cost of £45.74 per day used in a previous analysis of the FIRE Study, which was taken from a 2006 HPA report,⁴ inflated to 2010–2011 prices. The previous unit cost did not reflect the price reductions for the generic indication. Note also that, unlike the previous version, the base case assumes a more realistic treatment duration of 10 days rather than 14 days. The net effect is that in the base case the unit cost of antifungal prophylaxis is £77.80 per day, not the £640 assumed previously. Morbidity costs were included from critical care admission until ultimate discharge from acute hospital. These costs were calculated by estimating LOS both within and after critical care. Each day of critical care was classified according to Healthcare Resource Group 4 (HRG4) category derived from organ support data in the CCMDS, which forms part of the routine CMP data collection. Each bed-day was costed with the corresponding cost per bed-day from the UK ‘Payment by Results’ database.⁹⁸ Costs per bed-day in hospital after discharge from critical care were taken from the literature.⁹⁹ No costs after the initial hospital episode were considered. All costs were adjusted to 2010–11 price levels.⁹⁶

TABLE 26

Resource use and cost input parameters

Results: base case

The model predicted that following risk assessment with the threshold set to 0.5%, the proportion of patients receiving prophylaxis ranged from 17% to 30% (Table 27). The strategies that had risk assessment at multiple time points were predicted to result in a higher proportion of patients receiving antifungal prophylaxis; if, for example, risk assessment was only undertaken at the end of calendar day 3, then around 23% of patients were predicted to have prophylaxis versus 30% if risk assessment was performed at admission and the end of calendar day 3 or at all three time points. At the higher risk thresholds of 1% and 2%, the predicted proportions receiving prophylaxis ranged from 4% to 14% and from 1% to 5%, respectively (Tables 28 and 29). The current practice of no risk assessment and prophylaxis was predicted to have an incidence of invasive Candida infection during the critical care stay of 0.57%. The prophylaxis strategies were predicted to somewhat reduce the incidence of infection, for example to 0.47% if prophylaxis was provided for patients whose risk at the end of day 3 exceeded 2%. The lowest incidences of invasive Candida infection were following the strategies which required risk assessment at all three time points. The proportion who died after invasive Candida infection was slightly lower following prophylaxis than under current practice, but the reductions in overall mortality during the critical care stay were small (see Tables 27–29).

TABLE 27

Decision model outputs, by strategy, for a risk threshold of 0.5%

TABLE 28

Decision model outputs, by strategy, for a risk threshold of 1%

TABLE 29

Decision model outputs, by strategy, for a risk threshold of 2%

Prophylaxis was predicted to slightly reduce mean hospitalisation costs. For example, at all three risk thresholds, a strategy of providing prophylaxis at the end of calendar day 3 was predicted to reduce mean hospitalisation costs by around £25 (see Tables 27–29). For some risk assessment and prophylaxis strategies, for example risk assessment at admission or the end of 24 hours, the reduction in hospitalisation costs was offset by higher assessment and prophylaxis costs, leading to higher total costs than current practice (Tables 30–32). For other risk assessment strategies, for example risk assessment at the end of day 3, the costs of assessment and prophylaxis were exceeded by the reduction in hospitalisation costs, leading to lower total costs than that of current practice at each risk threshold.

TABLE 30

Mean total costs, life-years, QALY and net monetary benefits at risk threshold of 0.5%

TABLE 32

Mean total costs (£), life-years, QALY and net monetary benefits at risk threshold of 2%

TABLE 31

Mean total costs, life-years, QALY and net monetary benefits at risk threshold of 1%

The incremental analysis compared each prophylaxis strategy with current practice (Tables 33–35). These results showed that, irrespective of the risk threshold, the incremental QALYs of the prophylaxis strategies compared with current practice were positive, but small. The incremental costs of the risk assessment strategies were negative for strategies including risk assessment at the end of day 3, whether at single or at multiple time points. The INB at a risk threshold of 0.5% was highest when assessment and prophylaxis were administered at all time points. For risk threshold of 1% and 2%, the highest INB was associated with risk assessment and prophylaxis at the end of calendar day 3.

TABLE 33

Incremental cost, QALYs and INBs for each prophylaxis strategy vs no risk assessment (at risk threshold of 0.5%)

TABLE 35

Incremental cost, QALYs and INBs for each prophylaxis strategy vs no risk assessment (at risk threshold of 2%)

TABLE 34

Incremental cost, QALYs and INBs for each prophylaxis strategy vs no risk assessment (at risk threshold of 1%)

The CEACs are plotted for each risk threshold in Figures 16–18. They show that at the 1% and 2% risk thresholds, risk assessment and prophylaxis at the end of calendar day 3 was the strategy most likely to be cost-effective at the recommended cost-effectiveness threshold of £20,000 per QALY gain. For the lower risk threshold (0.5%), the strategy with the highest probability of being cost-effective at £20,000 per QALY was to assess risk at all three time points. At each risk threshold there was considerable uncertainty surrounding the relative cost-effectiveness of the alternative strategies, and at the £20,000 per QALY threshold the probability that any particular strategy would in fact be the most cost-effective did not exceed 30%.

FIGURE 16

Cost-effectiveness acceptability curves at risk threshold of 0.5%. Note: the strategies for ‘admission’ and ‘admission and 24 hours’ are indistinguishable from ‘24 hours’.

FIGURE 18

Cost-effectiveness acceptability curves at risk threshold of 2%. Note: the strategies for ‘admission’ and ‘admission and 24 hours’ are indistinguishable from ‘24 hours’.

FIGURE 17

Cost-effectiveness acceptability curves at risk threshold of 1%. Note: the strategies for ‘admission’ and ‘admission and 24 hours’ are indistinguishable from ‘24 hours’.

Methods: scenario analyses

The main assumptions made in the base-case analysis were challenged in the following scenario analyses, which repeated the above analyses but with alternative assumptions.

1.

The base-case analysis assumed the cost of risk assessment was based on 10 minutes of nursing time. The sensitivity analysis considered alternative scenarios assuming 5 or 20 minutes of nursing time to undertake the risk assessment.

2.

The base-case analysis assumed each patient had the recommended dose of fluconazole (400 mg per day), administered in an intravenous form. However, the systematic review reported that some RCTs have reported similar levels of effectiveness, with doses as low as 100 mg.¹⁹ We therefore undertook a sensitivity analysis in which the dose of fluconazole was reduced to 100 mg, and the costs lowered accordingly by 75%, while the RR of invasive Candida infection remained the same as in the base case. Another rationale for considering a 75% reduction in the unit costs of fluconazole is that even if the dose is maintained at 400 mg, local discounts in the range 50–70% may be available. Also, once the patient is able to absorb orally, some hospitals may administer the lower-cost oral formulation of fluconazole. Therefore, it is relevant to consider a reduction in the unit cost of 75% even for the same dose.

3.

The base case analysis assumed a duration of prophylaxis of 10 days based on the average LOS in critical care. The sensitivity analysis considered alternative scenarios assuming 5 days and 14 days of prophylaxis as informed by systematic review.¹⁹

4.

The base-case analysis assumed a duration of excess mortality for critical care survivors of 4 years after discharge from critical care, but excess mortality could continue for up to 25 years.¹⁰⁴ The sensitivity analysis assumed the magnitude of excess death rates for an extended period of time (25 years) taken from a previous study.¹⁰⁴ These excess death rates, relative to age- and sex-matched mortality in the general population, were applied beyond 4 years for up to 25 years.

5.

The base-case analysis assumed the HRQOL of critical care survivors was 80% of that of the general population. The sensitivity analysis assumed a lower figure of 70%.

The scenario analysis also examined best-case and worst-case scenarios as defined below:

6.

Best-case scenario: 10 minutes of nursing time, low dose of prophylaxis/75% discount on cost of prophylaxis, prophylaxis administered for 5 days, excess mortality for four years and HRQOL of survivors 80% that of general population.

7.

Worst-case scenario: 20 minutes of nursing time, standard dose of prophylaxis/no discount on cost of prophylaxis, prophylaxis administered for 14 days, excess mortality for 25 years and HRQOL of survivors 70% that of general population.

Results: scenario analyses

The results of the scenario analyses showed that the base-case results were generally robust to each of the alternative assumptions (Tables 36–38). Across risk thresholds and prophylaxis strategies, the scenarios that considered increased costs of assessment and prophylaxis (e.g. higher nursing time, increased duration of prophylaxis) led to lower INB for the risk assessment strategies than for the base case, assuming lower costs of assessment and prophylaxis (e.g. lower nursing time, shorter duration of prophylaxis) led to higher INB. Excess mortality for 25 years and decrement in HRQOL weights showed small effects on INB. The general conclusion that the strategies that included risk assessment at the end of calendar day 3 were relatively cost-effective was robust to the alternative best-case and worst-case scenarios considered.

TABLE 36

Sensitivity analyses on INBs (£) at risk threshold of 0.5%

TABLE 38

Sensitivity analyses on INBs (£) at risk threshold of 2%

TABLE 37

Sensitivity analyses on INBs (£) at risk threshold of 1%

Methods: value of information analysis

The decision as to which prophylaxis strategy to adopt based on the current evidence available remains uncertain. There is always the possibility that the strategy that appears the most cost-effective from current evidence would not be the optimal approach if perfect information was available. The expected costs of choosing the wrong strategy can be considered in terms of lost resources, but also health gain forgone. The expected costs of this decision uncertainty can be quantified according to the expected value of perfect information (EVPI).¹⁰⁷ EVPI can inform whether or not further research is worthwhile, by reporting whether or not the EVPI exceeds the anticipated research costs.^108,109 EVPI was calculated for the total population anticipated to benefit from the strategies considered, assuming that the eligible population of interest was 100,000 admissions to critical care each year, and that the relevant life cycle for the technology was 5 years.

To establish where further research might be best targeted, EVPI can also be reported for groups of parameters termed EVPI for parameters or expected value of partial perfect information (EVPPI).¹¹⁰ This approach can direct research towards those parameters and research designs that have most value. The groups of parameters considered were baseline probability of invasive Candida infection, mortality with and without invasive Candida infection, PPV and NPV, RRs of infection after prophylaxis, morbidity costs, long-term HRQOL, and survival.

Results: value of information analysis

Figures 19 and 20 summarise the population EVPI for the three threshold levels of risk. At a threshold of £20,000 per QALY the EVPI estimates for the population ranged between £12M (0.5% risk) and £14M (1% risk) at £20,000 per QALY. The corresponding EVPI per patient ranged from around £120 to £140.

FIGURE 19

Expected value of perfect information for the population.

FIGURE 20

Expected value of perfect information per patient.

These results indicate that across all parameters in the decision model, the value of further research for the whole population of interest is high relative to the likely research costs, and that the value is similar across risk thresholds.

Figure 21 reports for the overall population EVPPI estimates for each group of parameters according to risk threshold. The results highlight that the value of information for each group of parameters is similar across risk thresholds. The results also suggest that even after the FIRE Study, given the large population of interest for this decision problem (100,000 per year), there is still high value in acquiring more information on parameters such as PPV and NPV.

FIGURE 21

Expected value of partial perfect information for the population at £20,000 per QALY, according to alternative risk thresholds.

The EVPPI per patient (Figure 22) suggests the value of further research for most parameters is from £80 to £127 per patient. The decison whether or not to fund further research must then be weighed against the additional costs.

FIGURE 22

Expected value of partial perfect information per patient, at £20,000 per QALY, for alternative risk thresholds.

Discussion

This economic evaluation assessed the relative cost-effectiveness of alternative prophylaxis strategies for preventing invasive Candida infection for patients admitted to critical care who do not currently receive antifungal prophylaxis. The main finding was that, at a threshold risk for invasive Candida infection of 1% or 2%, the most cost-effective strategy was risk assessment and prophylaxis at the end of calendar day 3, which would require approximately 5–12% of eligible patients to receive antifungal prophylaxis. With a lower threshold risk for invasive Candida infection (0.5%), risk assessment and prophylaxis at all time points was the most cost-effective prophylaxis strategy, but would require around 30% of eligible patients to receive antifungal prophylaxis, which raises concerns about the impact on resistance.

In this general population of non-neutropenic, critically ill adult patients, the incidence of invasive Candida infection was low. As a result, the cost-effectiveness model predicted that prophylaxis prevented relatively few invasive Candida infections leading to small reductions in mortality and gains in QALYs. However, the costs of risk assessment and prophylaxis were also low relative to other hospitalisation costs, and for some risk assessment strategies this led to a net reduction in hospitalisation costs. The risk assessment strategies only lead to small gains in net monetary benefits compared with current practice, however, and so, given the large uncertainties, the probability that any particular risk assessment strategy was most cost-effective did not exceed 30% at the recommended threshold willingness to pay for a QALY gain.

This is the first cost-effectiveness analysis (CEA) to compare alternative strategies for preventing invasive Candida infection. The study has several important strengths. First, the decision model is populated mainly by parameters (e.g. PPV, NPV, baseline risk of infection and death, cost with and without infection) estimated from patient-level data collected prospectively in the FIRE Study. The PPVs and NPVs were taken from the FIRE Study validation sample, rather than from the development sample, to avoid any concerns about overfitting the models. Second, the RR of invasive Candida infection after prophylaxis was taken from a systematic review of published RCTs. The review suggested that the RRs were similar irrespective of the level of baseline risk, and were applicable to the low-risk population considered here. Third, the CEA followed methodological recommendations and fully considered both parameter uncertainty and structural uncertainty emanating from the assumptions made in constructing and populating the model. Fourth, the initial proposal was for a CEA limited to comparing current practice with risk assessment and prophylaxis at a single time point, but instead we followed expert clinical advice and broadened the range of strategies to allow risk assessment at several time points. Fifth, the CEA quantified the expected value of further research both overall and for specific groups of parameters.

Cost-effectiveness analyses that estimate relative costs and outcomes over the lifetime inevitably make assumptions.¹¹¹ Here we made plausible assumptions, for example about life expectancy for critical care survivors by drawing on the previous literature.^101,102 The base-case finding that a strategy of risk assessment and prophylaxis was relatively cost-effective was robust to the base-case assumptions concerning the costs of risk assessment and the long-term prognosis for critical care survivors. However, the results were relatively more sensitive to the worst- and best-case scenarios in which a number of parameters such as duration of prophylaxis, nursing time, discounted price, dose of prophylaxis, duration of excess mortality and quality of life were varied jointly.

The decision model also reported the expected value of further research. The EVPI for the entire population relevant to this decision problem (100,000 critical care admissions per year) may be as high as £14M. The reasons why further research may have high overall value are that the differences in costs and outcomes across strategies were small, the parameter uncertainty was relatively high and the overall target population is large. It should be recognised, however, that these estimates of the upper bound on the value of research in this area are lower than previous estimates for other interventions in critical care, although these interventions would affect substantially smaller populations.¹⁰⁴ The estimates of EVPPI provide an upper bound on the relative value of further research on each group of parameters such as the RRs after prophylaxis, baseline risks of infection and mortality, PPVs and NPVs, costs and lifetime QALYs. For RRs, further research would imply additional RCTs, but it could be argued that the anticipated research would be justified by the value (£11M). For other parameters such as baseline risk of infection and death, where further research is of similar value, such parameters could be collected at relatively low cost, for example alongside existing national clinical audit through the CMP. Parameters such as PPV and NPV also have high value but they would require a new prospective study, and so would be costly to estimate. Such parameters are specific to this decision problem. Instead, it might be more worthwhile to invest in further research on estimating parameters, such as lifetime QALYs after critical care survival, which would be useful for all decision problems in this area and would inform subsequent economic evaluations both for ongoing and future research studies in critical care.

This CEA does have some limitations. First, the model ignored any impact on resistance from increased use of prophylaxis. Including the effects of resistance on the costs and health outcomes of future patients would reduce the relative cost-effectiveness of the risk assessment strategies compared with current practice. Second, no consideration is given to prevention of onward transmission. Hence, the gain from prophylaxis could have been understated. That said, given the model predicted that prophylaxis prevented few invasive Candida infections, it can be anticipated that moving to a dynamic model structure and including the effect on onward transmission would be unlikely to change the conclusions. When an additional variable was added to the FIRE Study risk models indicating the presence of another patient with invasive Candida infection in the critical care unit this was not significantly associated with increased risk of invasive Candida infection, which suggests that onwards transmission was not a major factor. Third, the CEA took a narrow perspective to costing, and included hospitalisation costs only for the initial hospital episode. Hence, any additional costs attributable to invasive Candida infection that fell on community health services or came from hospital readmissions were excluded. It should be noted that the model did incorporate a relatively large increase in average hospitalisation costs following invasive Candida infection but, given the low incidence, it would seem unlikely that including a broader range of costs would alter the conclusions. Finally, it should be noted that the results of this CEA do not apply to those patients (approximately 3% of those otherwise eligible for the decision problem) who are currently prescribed prophylaxis according to clinical judgement.

The study suggests that a strategy of risk assessment and prophylaxis within three calendar days of admission to critical care may be cost-effective. However, it should be recognised that this could increase the risk of resistance, leading to higher costs and increased morbidity for future patients. Emergence of resistance has only been directly linked to fluconazole usage in cases of prolonged treatment in HIV-associated candidosis or in patients with chronic mucocutaneous disease and, as such, is relatively unlikely in this patient population. ¹¹² However, fluconazole usage has been linked to the pathogen shifts away from Candida albicans towards fluconazole-resistant species such as Candida glabrata.¹¹³ The possible consequences of resistance to antifungal prophylaxis could include increased lengths of stay in critical care and in hospital, the additional diagnostic tests and treatment costs for a patient infected with a resistant organism. ¹¹⁴ Studies to date have not fully assessed the cost of resistance to antifungal prophylaxis. In a related context, Smith and Coast¹¹⁵ highlighted that published studies underestimated the true costs of antibacterial resistance. Further research is required to consider the full costs of antifungal prophylaxis in terms of the additional burden to future patients whose treatment with antifungal agents becomes inappropriate due to increased resistance, and the consequent increased use of newer, and more costly, next generation antifungals. ¹¹⁶ Incorporating these effects of resistance in decision analytic modelling is challenging as it requires estimates of additional parameters, such as the resistance rate, the ensuing effect on morbidity and mortality, and a broader model structure to consider future populations who may be affected by increased resistance.

In conclusion, this CEA found that, for non-neutropenic, critically ill adult patients who met the inclusion criteria for the FIRE Study, the most cost-effective strategy at a 1% or 2% risk threshold was to assess the risk of invasive Candida infection at the end of calendar day 3, which would lead to 5–12% of patients receiving antifungal prophylaxis. Although risk assessment at all three time points was the most cost-effective strategy at a 0.5% risk threshold, this would lead to antifungal prophylaxis for around 30% of patients and thus raise concerns about resistance. The incremental costs and QALYs for each of the prophylaxis strategies compared with current practice were relatively small, and there is considerable uncertainty surrounding the cost effectiveness of the alternative strategies. Hence, even after the FIRE study the expected value of further research for this population appears large, but any further research recommendations pertaining to this decision problem should recognise the cost of further research and also consider whether or not the value of such research can be transferred to other decision problems of high clinical relevance. In particular, further research could take the approach followed in this study to assess whether or not a diagnostic test, such as a quantitative polymerase chain reaction (PCR) test¹¹⁷ or (1→3)-β-d-glucan assay,¹¹⁸ would be worthwhile for those patients who according to the FIRE Study risk models are at high risk of invasive Candida infection.