|
Empirical anti-candida therapy among selected patients in the intensive care unit: a cost-effectiveness analysis |
Golan Y, Wolf M P, Pauker S G, Wong J B, Hadley S |
|
|
Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology The health interventions examined in the study were 4 empirical anti-Candida strategies and 4 culture-based anti-Candida strategies for patients admitted to the intensive care unit (ICU). Empirical strategies included therapy with amphotericin B, lipid formulations of amphotericin B (L-amphotericin), caspofungin, or fluconazole. All empirical therapies were started before culture results were available, were continued even if fungal culture results were negative, and were given to all patients in the ICU with fever, hypothermia, or unexplained hypotension that persisted after 3 days of antibacterial therapy. Culture-based strategies included therapy with amphotericin B, L-amphotericin, caspofungin, or fluconazole only when cultures are reported to grow Candida. The dose and duration of therapy were based on treatment guidelines recommended by the Infectious Diseases Society of America (IDSA), which advises a 14-day course of an anti-Candida agent after the last positive culture result for invasive candidiasis (IC). Fluconazole was given as an intravenous dose of 800 mg on the first day of therapy followed by 400 mg intravenously daily for 6 days, followed by 400 mg orally daily for 7 days. Caspofungin was administered intravenously with 70 mg given on the first day of therapy followed by 50 mg daily for 13 days. L-amphotericin was given as 3mg/kg of body weight per day for 14 days.
Economic study type Cost-effectiveness analysis.
Study population The study population comprised a hypothetical cohort of patients in the ICU who, despite receiving antibacterial therapy in the ICU for 3 days, had fever, hypothermia, or unexplained hypotension. Patients who responded to antibacterial treatment (who became normothermic and normotensive within 3 days of antibacterial therapy) were excluded.
Setting The setting was a hospital ICU. The economic study was carried out in the USA.
Dates to which data relate Effectiveness and some resource use data were derived from studies published between 1988 and 2004. The price year was not stated.
Source of effectiveness data The effectiveness evidence came from a synthesis of published studies and the authors' assumptions.
Modelling A decision tree model was constructed to assess the cost-effectiveness of the 4 empirical anti-Candida strategies, the 4 culture-based anti-Candida strategies, and the no treatment strategy for ICU patients with suspected IC. The time horizon of the model was lifetime. Culture specimens were obtained from all patients and culture results could be positive or negative. Positive results could represent an infection or culture contamination. Negative results could represent no infection or a false-negative result. In the 4 branches of empirical therapy, all patients received therapy regardless of culture result, while in the 4 branches of culture-based therapy, only patients with Candida-positive cultures received therapy. In the branch representing no therapy, no patient received therapy regardless of culture results. For patients with true-positive or false-negative culture results, Candida infection could lead to death or survival. Patients who died of their infection could die early (3 days after culture specimens were obtained) or late (7 days after culture specimens were obtained) due to fulminant infection. Those who survived or who were not infected could have anti-Candida drug-related toxicity (renal failure requiring dialysis), which might result in toxicity-related death. Surviving patients could still die of other causes unrelated to Candida infection or drug toxicity, such as an underlying comorbid condition.
Outcomes assessed in the review The outcomes assessed in the review were IC prevalence, culture sensitivity, age of ICU survivors, duration and efficacy of anti-Candida therapy, probability of drug resistance, rates of severe toxicity, mortality rates, excess mortality in ICU survivors and length of hospital stay.
Study designs and other criteria for inclusion in the review All clinical data were derived from a review of the literature as well as from ICU databases of 3 US tertiary care academic medical centres (Tufts-New England Medical Center and Beth Israel Deaconess Medical Center, both in Boston, Massachusetts, and Long Island Jewish Medical Center, New Hyde Park, New York). A separate systematic review of the randomized clinical trials was carried out to assess treatment efficacy. Extensive details on the estimation of some clinical inputs from the literature were provided in the technical appendix.
Sources searched to identify primary studies MEDLINE, the Cochrane Library, reference lists of articles, and selected conference proceedings were searched to identify relevant articles reporting treatment efficacy rates.
Criteria used to ensure the validity of primary studies For the systematic review of treatment efficacy, the inclusion of clinical trials ensures a high internal validity.
Methods used to judge relevance and validity, and for extracting data Two independent reviewers decided which trials to include in the review of treatment efficacy.
Number of primary studies included Approximately 67 primary studies were included in the review. Some data were estimated from 3 hospital databases.
Methods of combining primary studies A meta-analysis using a random-effects model was used to combine primary estimates of treatment efficacy. A narrative approach appears to have been used to combine other clinical data.
Investigation of differences between primary studies Results of the review IC prevalence was set at 10%.
Culture sensitivity (ratio of IC diagnosis in patients with IC) was 0.7 (range: 0.5 - 0.9).
The average age of ICU survivors was 60.7 years (range: 20 - 80 years).
The duration of anti-Candida therapy was 14 days (range: 7 - 21 years).
The efficacy of anti-Candida therapy (decrease in mortality from IC) was 0.6 (range: 0.6 - 1.0) with amphotericin B, L-amphotericin, fluconazole, and caspofungin.
The probability of fluconazole resistance (decrease in efficacy) was 0.05 (range: 0.01 - 0.50).
The probability of caspofungin resistance (decrease in efficacy) was 0 (range: 0.01 - 0.50).
The increased treatment efficacy if administered early was 0.3 (range: 0.0 - 0.5).
The rate of severe toxicity related to anti-Candida therapy was 0.12 (range: 0.05 - 0.20) with amphotericin B (kidney toxicity and anaphylaxis), 0.08 (range: 0.03 - 0.20) with L-amphotericin (kidney toxicity and anaphylaxis), and 0.0001 (range: 0.0001 - 0.01) for both fluconazole (anaphylaxis) and caspofungin (anaphylaxis).
The probability of mortality from untreated IC was 0.4 (range: 0.20 - 0.80), with 0.8 probability (range: 0.5 - 1.0) of death occurring in the ICU rather than later, in the ward.
The probability of mortality from reasons unrelated to IC was 0.2 (range: 0.10 - 0.40), with 0.5 probability (range: 0.2 - 1.0) of death occurring in the ICU rather than later, in the ward.
The probability of mortality from severe drug-related nephrotoxicity was 0.3 (range: 0.05 - 0.50), with probability of 1.0 (range: 0.4 - 1.0) for this occurring in the ICU, rather than later, in the ward.
The annual excess probability of mortality in ICU survivors was 0.049 in the first year after ICU discharge, 0.036 in the second year after ICU discharge, 0.019 in the third year, 0.018 in the fourth year, 0.006 in the fifth year, and 0 after 5 years.
The length of hospital stay was 14 days (range: 7 - 21) for ICU survivors with no IC or drug toxicity as well as for survivors of IC, 20 days (range: 10 - 30) for survivors of drug-related severe toxicity, 9 days(range: 3 - 15) for early death from IC, 44 days (range: 20 - 60) for late death from IC, 17 days (range: 5 - 30) for death from drug-related nephrotoxicity, 3 days (range: 1 - 10) for early death unrelated to IC or drug toxicity, and 17 days (range: 5 - 30) for late death unrelated to IC or drug toxicity.
Methods used to derive estimates of effectiveness The authors made assumptions to derive some clinical estimates.
Estimates of effectiveness and key assumptions The rate of culture contamination by Candida (false-positive result) was estimated to be 0.05 (range: 0.01 - 0.20). In general, when data were unavailable, the most conservative estimate was used, biasing the results against the empirical therapies.
Measure of benefits used in the economic analysis The summary benefit measure used in the economic analysis was the number of life years (LYs) saved (life-expectancy), which were estimated using the modelling approach, combining the hospital survival rate and the life-expectancy at discharge. Discounting was applied and a 3% annual rate was used.
Direct costs Although the authors stated that a societal perspective was adopted, the perspective adopted in fact appears to have been that of the third-party payer. The following items were considered: cost of an ICU day, cost of an ICU day if the patient died of drug-related nephrotoxicity, non-ICU (floor) day, and drugs (amphotericin B, fluconazole, L-amphotericin, and caspofungin). The costs associated with Candida culture were not considered since culture was performed on all patients, regardless of treatment strategy. Unit costs and the quantities of resources used were reported for most items. The estimation of resource use was mainly based on published data. Costs were estimated using average wholesale prices or from hospital billing departments. Discounting was not relevant as all costs were incurred over a short-time frame. The price year was not reported.
Statistical analysis of costs Statistical analyses of costs were not performed.
Indirect Costs Indirect costs were not considered.
Sensitivity analysis Univariate sensitivity analyses were carried out on all model inputs to assess the robustness of the base case cost-effectiveness ratios. The ranges used in the sensitivity analysis were based on values obtained from the literature or set by the authors. A justification for ranges of costs was not provided. Two- and three-way sensitivity analyses were also used to assess the impact of the inputs with the greatest effect on the base case results.
Estimated benefits used in the economic analysis The discounted LYs were:
10.972 with no anti-Candida treatment;
11.121 with culture-based amphotericin B;
11.154 with culture-based fluconazole;
11.164 with culture-based caspofungin;
11.135 with culture-based L-amphotericin;
11.261 with empirical fluconazole therapy;
10.868 with empirical amphotericin B therapy;
11.274 with empirical caspofungin therapy;
11.003 with empirical L-amphotericin B therapy.
Cost results The expected cost per patient was $21,722 with no anti-Candida treatment, $21,901 with culture-based amphotericin B, $21,926 with culture-based fluconazole, $22,373 with culture-based caspofungin, $22,574 with culture-based L-amphotericin, $23,272 with empirical fluconazole therapy, $23,554 with empirical amphotericin B therapy, $27,080 with empirical caspofungin therapy, and $29,125 with empirical L-amphotericin B therapy.
Synthesis of costs and benefits Incremental cost-effectiveness ratios were calculated to combine the costs and benefits of the alternative strategies. Each strategy was compared with the previous most effective strategy. Dominated or extended dominated alternatives, namely culture-based amphotericin B, culture-based caspofungin, culture-based L-amphotericin, empirical amphotericin B, and empirical L-amphotericin B, were eliminated.
The incremental cost per discounted LY (DLY) gained was $1,122 with culture-based fluconazole over no anti-Candida treatment, $12,593 with empirical fluconazole therapy over culture-based fluconazole, and $295,115 with empirical caspofungin therapy over empirical fluconazole.
The sensitivity analysis led to some interesting results. If the prevalence of IC decreased to less than 2.5%, empirical fluconazole exceeded the willingness-to-pay threshold of $50,000 per DLY saved and culture-based fluconazole was the preferred option. Only at an unrealistic prevalence of 60% did empirical caspofungin become the preferred option. Given no resistance to caspofungin, a 28% threshold of fluconazole resistance was required for empirical caspofungin to have a marginal cost-effectiveness ratio less than $50,000. The cost of caspofungin needed to be reduced by 58% for the cost-effectiveness ratio to be below the threshold. For rates of severe nephrotoxicity lower than 5.0%, culture-based standard and L-amphotericin therapies were acceptable strategies, but empirical fluconazole therapy remained the preferred option. Changes in other model inputs did not alter the results of the base case analysis.
Authors' conclusions The authors concluded that, for patients in the ICU with suspected IC who have not responded to antibiotic treatment, empiric therapy with fluconazole was the most cost-effective strategy. However, empirical therapy was not cost-effective in low-risk patients.
CRD COMMENTARY - Selection of comparators The selection of comparators was intended to cover all possible strategies for the management of patients with suspected IC. The option of no intervention was also considered. The authors stated that empirical therapy was implemented in several institutions. Dosages reflected those recommended by international societies. You should decide whether they are valid comparators in your own setting.
Validity of estimate of measure of effectiveness The effectiveness data were estimated from published studies. A systematic review of the literature was undertaken to identify primary studies. However, details of the review were reported only for the studies that were used to assess treatment effect. For these studies, the method used to combine primary estimates was reported and appears to have been appropriate. Furthermore, only clinical trials were included, which strengthens the validity of the primary estimates. The issue of homogeneity was addressed using random-effect models. Some data were estimated from the administrative databases of three hospitals. However, limited information on the design and characteristics of the methods and data sources used to estimate other clinical inputs was provided. Furthermore, the authors made some assumptions due to the lack of published evidence or because of the uncertainty in some data. The issue of variability in the data was addressed in the sensitivity analysis. The authors noted that, when ranges of values were available from the literature, the model was biased against empirical therapy.
Validity of estimate of measure of benefit The use of expected survival as the summary benefit measure was appropriate as this is the most relevant dimension of health for ICU patients suspected of having IC. The authors noted that quality of life adjustments were not used since it was assumed that survivors of IC or nephrotoxicity related to anti-Candida medications were unlikely to have long-term consequences related to these conditions. Discounting was applied, and the impact of different discount rates was investigated.
Validity of estimate of costs The perspective adopted in the study was not clear. The authors stated that a societal perspective was used, but indirect costs were not considered. Thus, costs relevant to the third-party payer appear to have been included. A justification for the exclusion of culture costs was provided. Extensive information on unit costs and the quantities of resources used was provided, which will assist in reproducing the cost analysis in other settings. The source of all economic data was provided. However, the price year was not given, which limits the possibility of making reflation exercises in other settings. Costs were treated deterministically, but some key cost estimates were varied in the sensitivity analysis.
Other issues The authors did not compare their findings with those from other studies. The authors pointed out that the clinical data used in the model might not apply to all ICUs but that the probability estimates might reflect the reality of ICUs in large referral hospitals. However, the extensive use of sensitivity analysis, especially for uncertain parameters, enhances the robustness of the study conclusions and the external validity of the analysis. The analysis refers to high-risk patients admitted to the ICU and this reflects the authors' conclusions.
Implications of the study The study results support the use of an empiric therapy with fluconazole to manage ICU patients suspected of having IC. The authors state that research to develop a clinical prediction rule for IC and survival is ongoing and could help refine the group of patients who would benefit from prophylactic treatment of IC.
Source of funding This research was supported by Pfizer Inc and by the National Library of Medicine.
Bibliographic details Golan Y, Wolf M P, Pauker S G, Wong J B, Hadley S. Empirical anti-candida therapy among selected patients in the intensive care unit: a cost-effectiveness analysis. Annals of Internal Medicine 2005; 143: 857-869 Other publications of related interest Rex J H, Bennett J E, Sugar A M et al. A randomized trial comparing fluconazole with amphotericin B for the treatment of candidemia in patients without neutropenia. Candidemia Study Group and the National Institute. New England Journal of Medicine 1994;331:1325-30.
Phillips P, Shafran S, Garber G, et al. Multicenter randomized trial of fluconazole versus amphotericin B for treatment of candidemia in non-neutropenic patients. Canadian Candidemia Study Group. European Journal of Clinical Microbiology and Infectious Diseases 1997;16:337-45.
Rex J H, Pappas P G, Karchmer A W, et al. A randomized and blinded multicenter trial of high-dose fluconazole plus placebo versus fluconazole plus amphotericin B as therapy for candidemia and its consequences in nonneutropenic subjects. Clinical Infectious Diseases 2003;36:1221-8.
Indexing Status Subject indexing assigned by NLM MeSH Amphotericin B /economics /therapeutic use; Anti-Bacterial Agents /therapeutic use; Antifungal Agents /economics /therapeutic use; Candidiasis /drug therapy /epidemiology; Cost-Benefit Analysis; Cross Infection /drug therapy /epidemiology; Decision Support Techniques; Echinocandins; Fluconazole /economics /therapeutic use; Humans; Intensive Care Units; Microbiological Techniques; Peptides, Cyclic /economics /therapeutic use; Prevalence; Risk Factors; Sensitivity and Specificity; Treatment Outcome AccessionNumber 22006008051 Date bibliographic record published 31/05/2006 Date abstract record published 31/05/2006 |
|
|
|