Journal of Infection (I99I) z3, I7-3I

Cost implications of alternative treatments for AIDS patients with cryptococcal meningitis. Comparison of fluconazole and amphotericin B-based therapies M. J. Buxton,* [[, D. J. Dubois,t R. R. Turner,~ M. J. Sculpher,* P. A. Robinson,~ and C. Searcy~ * Health Economics Research Group, Brunel University, Uxbridge, U.K., t Pfizer Central Research, Brussels, Belgium, :~Pfizer Central Research, Groton, Connecticut and ~Pfizer International, New York, U.S.A. Accepted for publication 4 February I99I Summary The extra demands placed upon health care resources by management of AIDS patients have increased the focus on cost implications of therapeutic alternatives. Cryptococcal meningitis is a common life-threatening infection in AIDS patients, usually treated with amphotericin B, often in combination with flucytosine. Administered intravenously, this therapy is associated With frequent and often severe side effects. Fluconazole is a new alternative which can be given orally once daily and has fewer such side effects. The purpose of this study was to examine the cost implications of these different therapies for both primary and maintenance treatment of cryptococcal meningitis. Comparison of these two therapies in recent clinical trials has indicated that fluconazole is at least as effective as amphotericin B, and therefore cost-minimisation analysis is an appropriate method to study the economic consequences of the alternative treatments. Patient management and resource-use information for both treatments was obtained using a modified Delphi technique with a panel of European physicians experienced in the treatment of this disease, and three models were developed to reflect the variability of practice evident among the panel members. U.K. health care costs were used to value these resources. The results indicated that, despite the higher cost of the drug itself, the costs associated with fluconazole were likely to be markedly less than those for amphotericin B for primary treatment, and similar or slightly cheaper for maintenance treatment. Over I year of treatment, the saving from the use of fluconazole would be in the range of £4ooo-I 4 ooo.

Introduction G i v e n the considerable d e m a n d on health care resources imposed by the m a n a g e m e n t of A I D S , it is essential to examine the cost implications of alternative treatments. T h e costs of new drugs need to be justified, particularly w h e n they are used not only for p r i m a r y b u t also for maintenance t r e a t m e n t for the lifetime of the patient. Cryptococcal meningitis is one of the most c o m m o n life-threatening infections in A I D S , with reported incidences outside Africa ranging from 2-5 % up to IO-I3 % o f patients. 1-3 T h e high risk of cryptococcal infection [r Address correspondence to: Professor M. J. Buxton, Health Economics Research Group, Brunel University, Uxbridge, Middlesex UB8 3PH, U.K.

o[63-4453/9I/O4OOI7+ I5 $03.00/0

© I99I The British Society for the Study of Infection

18

M. J. B U X T O N

ET AL.

relapse in A I D S has led to the recommendation of lifelong maintenance treatment,4.5 t h e benefits of which have been demonstrated in a recent study." Previously, treatment of cryptococcal meningitis has required therapy with amphotericin B, often in combination with flucytosine. Amphotericin B must be given intravenously, and is associated with frequent side effects that have been well-documented in n o n - A I D S patients. Azotaemia has been reported in approximately 26 % of patients, and bone marrow suppression related to flucytosine co-therapy in zz %.7. s Infusion-related side effects include fever, rigors, nausea and vomiting. Fluconazole, a new triazole antifungal c o m p o u n d , can be given orally or intravenously once a day and is effective against a variety of serious fungal infections including cryptococcal meningitis. 9 It is well-absorbed orally and is widely distributed in the body, reaching high concentrations in the serum and the cerebrospinal fluid, a°, 11 Results from non-comparative trials in patients with acute cryptococcal meningitis and A I D S have demonstrated the effectiveness of fluconazole and its low incidence of side effects in the treatment of such infections. It has also been shown to be effective for maintenance therapy. 12'1~ Recently, the interim results of a clinical trial comparing amphotericin B-based therapy and fluconazole have been reported, which indicated no statistically significant difference in efficacy between the two drugs in the treatment of acute cryptococcal meningitis in A I D S patients. 1~ Further, results from a clinical trial comparing fluconazole v s . arnphotericin B as maintenance therapy for the prevention of relapse of cryptococcal meningitis found fluconazole to be at least as effective and less toxic.* Because of the differing routes of administration and side effects of both these drugs, their use might be expected to lead to quite different clinical management practices, varying in the nature and extend of medical resource utilisation. T h e purpose of the present study, therefore, was to identify these differences by defining the resources required for these two approaches to the treatment of cryptococcal meningitis and, from these differences, to estimate their cost consequences. Material and methods

Economic evaluation of alternative treatments requires assessments of a n d / o r assumptions about clinical effectiveness, side effects, medical resource utilisation and associated unit costs. Information for this study was determined directly from a controlled clinical trial, but since the economic evaluation was initiated during the final stages of the trial, it was not feasible to collect resource utilisation prospectively. Rather, estimates of the resources needed for the management of cryptococcal meningitis were obtained from a panel of European infectious disease specialists using a structured consensus method. This approach, referred to as the modified Delphi method, used a * Medical communication. Protocol No. o56-i57 and o56-I58 (ACTG no. o26), Multicenter comparison of fluconazole (U.K.-49 858) and amphotericin B as maintenance treatment for the prevention of relapse of cryptococcal meningitis in patients with acquired immunodeficiency syndrome. Drug Safety Monitoring Board: April 23, I99o.

Cost implications for alternative care in A I D S

I9

questionnaire and group discussion to describe the clinical management of these patients. Costs to the health care system were determined for one setting (U.K.). T h e consequences of variation in unit costs and resource utilisation on total treatment costs were explored through sensitivity analysis. Selection of economic evaluation method

Evidence of the efficacy and side effects of the two compounds was available from controlled multicentre clinical trials conducted in the U.S.A. For primary therapy, patients were randomly assigned using a 2 : I schedule to fluconazole (n = 78) or amphotericin B (n = 4o) for IO weeks of therapy. Fluconazole patients received 2oo-4oo rag/day while amphotericin B patients received at least ~5 m g / k g over 6-IO weeks. 14 Survival rates and culture- and clinically-defined cure rates served as the primary endpoints. Side effects were tabulated from standardised reports which included clinical symptoms, severity, presumed relationship to study drug, onset, duration, and any action taken in response. Both therapies were compared in a multicentre, randomised comparative trial undertaken to assess efficacy for the prevention or relapse of cryptococcal meningitis. Patients were randomly assigned to zoo rag/day of fluconazole (n = IO6) or I m g / k g / d a y amphotericin B (n = 77). Relapse and survival rates served as the primary endpoints. Interim results from the primary treatment trial showed no statistically significant difference in efficacy between the two therapies. Results from the maintenance trial revealed different relapse rates (I6"8 % for amphotericin B vs. 1.8 % for fluconazole) and similar overall mortality rates. Taken together, these results indicate that clinically these two drugs are at least comparable for primary and maintenance therapy. Therefore cost minimisation would be the appropriate economic method for the analysis. 15 Cost-minimisation analysis focuses on cost differences only, as opposed to cost-effectiveness and cost-benefit analyses, which explore differences in effect as well as costs. 15 Identification of medical resource utilisation

Formal methods are often used to establish expert opinion in situations where empirical data are not available. ~6 For this study, those methods used in health-related decision-making were reviewed 17-~9 and the modified Delphi method, modelled after that reported by Park et al., ~° was selected. This method involved an expert panel, a structured questionnaire that described clinical management practices, and the conduct of a panel meeting during which the responses to this questionnaire were reviewed and discussed. Six European panellists were selected on the basis of their experience in treating cryptococcal meningitis in A I D S patients with both amphotericin B and fluconazole. T h e panellists and their institutional affiliation are recorded in the acknowledgement section. A preliminary evaluation by one of the authors (M. J. B.) of the likely costs of the treatments undertaken served as a basis for developing a structured questionnaire for describing patient management practices. After checking it for clarity and completeness with a clinical expert, adjustments were made and the revised questionnaire was then sent to the six panellists, who were asked

M. J. BUXTON E T A L .

20

Table I Expert opinion on patient management patterns

summary of questionnaire categories (a) (b) (c) (d) (e) (f) (a) (b) (c) (d) (e)

Acute primary therapy Dosage regimen (route, dose, duration) Iv line placement (surgical, percutaneous, peripheral) Hospitalisation (in-patient, out-patient) Prophylaxis for side effects Monitoring for side effects Treatment of side effects Maintenance therapy Dosage regimen (route, dose, frequency, duration) Hospitalisation (out-patient, day-patient, home-patient) Prophylaxis for side effects Monitoring for side effects Hospitalisation (in-patient) rate during maintenance (for any reason)

to complete it and return it by post. T h e major elements covered in the questionnaire are shown in Table I. T h e questions covered acute treatment and follow-up, up to a total period of I year. An anonymised summary of the panel's responses was circulated to the members prior to the group meeting. At the meeting, each component of the questionnaire was discussed, the reasons for differing responses considered, and issues of definition and interpretation clarified. It was stressed that a consensus was not specifically being sought. Rather, the panellists were instructed that their responses should reflect those patient management practices which they would implement with each therapy. At the end of the meeting, the panellists were asked to confirm or revise their initial responses. Construction of composite management

models

After the meeting, the panel's responses were summarised into composite patient management models for both alternative therapies, and included medication, hospitalisation, monitoring, and side-effect management. Summary techniques, including the use of mean or modal responses, were employed. T h e clinical practice for one panellist involved the use of fluconazole in combination with amphotericin B for part of the acute treatment period. This precluded the use of that panellist's responses for those questionnaire components that were treatment-specific. T o determine whether cost differences were consistent across variations of the 'base case' composite model, 'resource-sparing' and 'resource-intensive' models were constructed, reflecting the lowest and the highest resource-use response from the panel. Additional procedures were implemented for describing the resources needed to treat side effects. Panel deliberations revealed side-effect management to be complex and varied. Rather than develop decision-analysis models around specific side effects, an alternative approach was adopted which used expected additional hospitalisation as a proxy variable for the resources required to manage side effects. Those side effects which did not require additional hospitalisations were considered to require comparatively few

Cost implications for alternative care in A I D S

21

T a b l e I I Patient management resources unit cost estimates for the U.K. (£, January 199o prices) Resource Drugs Fluconazole (Diflucan)* Amphotericin (Fungizone)* Flucytosine (Ancoban)t Hydrocortisone (Solu-cortef)* Heparin (Hep-flush)* Domperidone (Motilium)* IV administration Giving sets~ Cannulae~ Infusates~: Surgical implant of IV line~ Monitoring Complete blood count Liver function tests Renal function tests Electrolytes Hospitalisation In-patient** Out-patienttt Day-patient~ ~; Home-patient~

Unit

Low cost

zoo rag/oral 200 mg/Iv 5o mg/Iv 50o mg/oral 25o ml/Iv ioo mg/Iv z ml/Iv io mg/oral

Cost

High cost

IO'9I 33'67 4"25 o'54 I9"O4 I'17 0"30 o'o9

Per Per Per Per

item item item item

0"93 o'75 o'75 47"20

3"I2 I'75 o'93 7o'8o

Per Per Per Per

test test test test

I'O6 [[ I 1.51 [[ 4'08 ]] v27 []

I I'OO4 1I'5O4 IO.OO4 9'5o4

Day Visit Day Visit

I92.oo 45'oo 67"00 45'oo

288"oo 67"oo 192'00 67'00

* Price (including VAT) from MIMS (Monthly Index of Medical Specialities), January 199o. t Price (including VAT) from BNF (British National Formulary) no. 18, September 1989. Price ranges (including VAT) obtained for major suppliers to hospital purchasing authorities, updated to January 199o. Based on Department of Health model charge for private patients for use of operating theatre facilities for lO-3o min (£59.oo) _ 20 %. [] Based on estimated costs for a specific NHS hospital, uprated to January 199o. 4 Based on prices from commercial pathology laboratory, London. ** Ranges of _ 20 % around best estimate of £240 based on mean of estimates for in-patient costs of AIDS patients from Cunningham and Griffiths (1987) 21 for St Mary's Hospital, London, and unpublished data from M. Rees for St Stephen's Hospital, London, adjusted to exclude costs of AZT (which would be unaffected by choice of antifungal therapy) and uprated to January 199o prices. t t Range of 4- 20 % around best estimates of £56, based on mean of average cost per outpatient visit for St Mary's and St Stephen's Hospitals uprated for AIDS patients at the same ratio of costs as implied by in-patient estimates above, and uprated to January 199o prices. $$ In the absence of any reliable published estimates for day-patients of this type, range assumed to lie between upper out-patient cost and lower in-patient cost limits. ~ In the absence of any reliable published estimates of costs of home care for such patients, costed as out-patient visit. r e s o u r c e s for t h e i r m a n a g e m e n t . T h i s a p p r o a c h i n v o l v e d t h e c o m p u t a t i o n o f a loading factor, w h i c h c o m b i n e d i n c i d e n c e o f side effects w h i c h w o u l d r e q u i r e additional hospitalisation with estimates of the duration of that additional h o s p i t a l stay.

M. J. B U X T O N

22

ET AL.

Table I I I Composite patient management models resource use Acute primary therapy Base

Resource-intensive

Resource-sparing

IV 400 m g / d a y x 2 weeks Oral 4oo m g / d a y x 4 weeks IV 0'3 m g / k g / d a y x 4 weeks Flucytosine IV 15o m g / k g x 4 weeks Flucytosine oral 15o m g / k g x 2 weeks further

N o IV

5% o% 95 % 50 % 50 % o %

N o IV

(a) Dosage regimen Fluco

Ampho B

IV 400 m g / d a y x I week Oral 4oo m g / d a y × 5 weeks IV 0"6 m g / k g / d a y x 6 weeks

(b) IV line placement Fluco 1% 8% 9I % Ampho B I9 % 48 % 33 %

surgical percutaneous peripheral surgical percutaneous peripheral

surgical percutaneous peripheral surgical percutaneous peripheral

(c) Hospitalisation (days, IP = in-patient, O P = out-patient) Fluco I8"2 I P 30.0 IP 3'4 O P _I'70P Ampho B 44"6 I P 52 I P o OP o OP(d) Prophylaxis for side effects Nil Fluco Nil Hydrocortisone Ampho B H y d r o c o r t i s o n e : 5o m g Saline IV bolus Heparin Saline Antihistamine : 2o mg x 3/day (e) M o n i t o r i n g for side effects (tests/week x weeks) Fluco CBC o'4 (tests/week) As for base × 6 weeks = 2"4 Electro 0'25 × 6 w e e k s = 1' 5 LFT 0"9 x 6 weeks -- 5"4 Renal o'6 x 6 weeks = 3"6 Ampho B CBC I'2 × 6 weeks = 7"2 As for base Electro I'7 x 6 weeks = Io-2 LFT r 2 x 6 weeks = 7"2 Renal 1.6 x 6 weeks = 9"6

Oral 2o0 m g / d a y x 6 weeks IV o'6 m g / k g x 4 weeks

o % surgical 5o % percutaneous 5o % peripheral IP 4"40P 42 I P o OP I I'O

Nil Saline

As for base

r2 × I'7 × I'2 x 1"6 x

4 4 4 4

weeks weeks weeks weeks

T r e a t m e n t of side effects (additional hospitalisation only) I P day loadings [increase in baseline hospitalisation days, see (c)] Fluco Ampho B

5"6 I5"8

Fluco Ampho B

2"6 22'5

9"9 22"9 Percentage of patients affected 5"o 3o-o

9"0 7"I o-o I5"o

= = = =

4"8 6"8 4"8 6"4

Cost implications for alternative care in A I D S

23

T a b l e I I I (cont.) Acute primary therapy Base

Resource-intensive

Resource-sparing

Maintenance therapy (a) Dosage regimen Fluco Oral 2o0 mg/day × 46 As for base As for base weeks Ampho B IV I'O mg/kg/week × 46 IV 0"7 mg/kg twice IV o-6 mg/kg weeks weekly × 46 weeks week x 48 weeks (b) Additional hospitalisation (out-patient, day-patient, home-patient) per 46 weeks Proportion of patients hospitalised for other reasons 16% 28% 10% Fluco Monthly OP visits = Monthly OP visits = Monthly OP visits = 9"7 visits 8"3 visits lO'3 visits Ampho B Weekly visits Weekly visits Weekly visits Home Home Home 12 % = 4"6 visits o 25 % = lO-8 visits Day Day Day 88 % = 34"0 visits IOO% = 33'1 visits 75 % = 32"4 visits (c) Prophylaxis for side effects Fluco Nil Nil Nil Ampho B Saline Saline Saline (d) Monitoring for side effects (tests/week × week) Fluco o-12 (tests/week) As for base As for base CBC x 46 = 5"5 tests Electro o LFT 0"29 × 46 = 13"3 Renal o'o5 × 46 = 2"3 Ampho B CBC o'44 × 46 = 20-2 As for base o'44 x 48 = 21.1 Electro o o LFT o-75 x 46 = 34"5 0.75 x 48 = 36 Renal o-8 × 46 = 36"8 o.8 x48 = 38'4 Abbreviations: Fluco, fluconazole-based therapy; Ampho B, amphotericin B-based therapy (with/without flucytosine); IV, intravenous; IP, in-patient; OP, out-patient; Home, homepatient therapy; Day, day-patient therapy; CBC, complete blood count; Electro, electrolytes; EFT, liver function tests; Renal, renal function tests.

T h e i n c i d e n c e o f side effects t h a t w o u l d p r o l o n g h o s p i t a l i s a t i o n was e s t i m a t e d f o r each d r u g , after r e v i e w o f i n d i v i d u a l p a t i e n t profiles f r o m t h e c o m p a r a t i v e trial data. T h o s e cases w i t h side effects classified as ' p r o b a b l y ' o r ' p o s s i b l y ' r e l a t e d to s t u d y d r u g a n d t h o s e o f ' u n k n o w n ' r e l a t i o n s h i p w e r e p r e s e n t e d b l i n d e d to t h e t w o p h y s i c i a n a u t h o r s (D. J . D . , P. A . R . ) . B a s e d o n t h e s e v e r i t y a n d t h e d u r a t i o n o f side-effect s y m p t o m s , t h e r a p y for s y m p t o m s (including interruptions of study d r u g therapy), and the time course, the p h y s i c i a n s i n d e p e n d e n t l y j u d g e d w h e t h e r t h e side effects f o r each i n d i v i d u a l p a t i e n t w e r e likely to p r o l o n g h o s p i t a l i s a t i o n . T h e rate o f side effects likely to p r o l o n g h o s p i t a l i s a t i o n , d e t e r m i n e d e m p i r i c a l l y f r o m t h e clinical trial, was

24

M.J. BUXTON ET AL.

similar to that estimated by the expert panel, based on their clinical experience. T h e duration of additional hospitalisation required for severe side effects was estimated by the panel. T h e empirically determined incidence rate was then multiplied by the duration to provide a length-of-stay loading factor, to be applied as an estimate of the cost implications of treating side effects. Application of unit costs

T o translate the resource-use information into cost figures, best available estimates of current unit costs in the U.K. were obtained and applied to the resource-use estimates. T h e unit costs were calculated in January z99o U.K. prices and are shown in Table II. Notes on the sources and calculations are appended to the table. For those costs for which no single market price was available, low and high estimates were used in the sensitivity analysis. T h u s , for each major c o m p o n e n t of resource utilisation (medication, hospitalisation, monitoring, and side effects) six cost estimates were calculated: low and high costs for the 'base case', as well as for the 'resource-sparing' and the 'resource-intensive' patient management models. Results Medical resource utilisation

T h e patterns of medical resources used during both primary and maintenance treatment of A I D S patients with cryptococcal meningitis that were derived from the modified Delphi m e t h o d are summarised in Table III. It shows the resources involved for the 'base case' models and for the 'resource sparing' and 'resource intensive' alternatives. Base case patient management

models

T h e 'base case' clinical management models produced by the panel for cryptococcal meningitis may be described as follows. Primary therapy normally lasted for up to 6 weeks for a patient who responded to therapy. For the purposes of the model, maintenance therapy was analysed on the basis of it lasting for the remainder of I year (46 weeks); in practice, it was the view of the expert panel that I year was the typical life expectancy of such patients. T h e definitions of primary and maintenance therapy are not clinically precise. Rather, the distinction is usually based on a change in drug administration and monitoring intensity as a result of clinical and mycological response. For primary therapy, fluconazole-treated patients would be expected to receive 4o0 mg IV daily for I week followed by 4o0 m g orally, daily for 5 weeks. T h e IV route would be expected to be peripheral in over 90 % of the cases. No prophylactic treatment for side effects would be expected. T h e patient would require in-patient hospitalisation for approximately I8 days followed by three to four out-patient visits. Monitoring tests would be expected to include complete blood count (CBC) and renal function tests approximately twice weekly, weekly liver function tests, and electrolyte tests about every 3 weeks. Approximately 3 % of the patients would be expected to develop side effects severe enough to require about 6 additional days of hospitalisation. For the 46-week maintenance period, fluconazole-treated patients would

Cost implications for alternative care in A I D S

25

Table IV Costs per patent (£), composite base case Primary

Fluconazole Medication Hospitalisation Monitoring Side effects Total Amphotericin B Medication Hospitalisation Monitoring Side

effects

Total Cost differences (AMB-FLUCO)

Maintenance

Total

Low

High

Low

High

Low

High

I26o 3647 8I 28 5o16

128I 5469 I39 42 6931

3514 435 I69 -4118

3514 647 237 4398

4774 4082 250 28 9134

4795 6116 376 42 1I 329

381 8563 143

515 i2 845 355

557 2487 569

7 I2 6839 987

938 i 1050 712

I227 19 684 1342

682

lO23

9769 4753

14738 78o7

--

3613 (5o5)

--

--

8538 414o

682

13382 4248

lO23

23276 i 1947

receive zoo mg by mouth daily. An additional IO out-patient visits would be expected (i.e. approximately monthly). Monitoring tests include monthly liver function tests, twice-monthly CBC tests, and renal function tests every 5 months during treatment. No electrolyte test would be anticipated. In contrast, for primary therapy, amphotericin B-treated patients would be expected to receive o'6 mg/kg zv daily for 6 weeks. Intravenous administration of amphotericin B via a central catheter would be expected in 67 % of patients. In 19 % of patients the central line would be surgically placed and in 48 ~/o the central line would be percutaneously placed. The remaining 33 % of patients would have a peripheral catheter. Prophylactic treatment would be expected to include a 5o mg bolus of hydrocortisone, and saline. T h e patient would require in-patient hospitalisation for acute therapy for approximately 45 days. Monitoring tests would be expected to include CBC and liver function tests slightly more frequently than once a week, and electrolyte and renal function tests every 3 or 4 days. Approximately 23 % of the patients would be expected to develop side effects, which would require an additional 16 days of hospitalisation. For maintenance therapy, amphotericin B-treated patients would receive z-o m g / k g weekly Iv through a peripheral line along with saline therapy. In addition to therapy delivered during hospitalisation for other reasons, about IZ % of these patients might be expected to receive therapy at home while 88 % would receive therapy at a day-bed clinic. To monitor the patient, twiceweekly CBC and weekly liver function and renal function tests would be expected. Hospitalisation for other reasons would be expected in about I6 % of the patients on both therapies, during which monitoring for cryptococcal meningitis therapy would be conducted, and therefore is subtracted from the cost of out-patient maintenance therapy.

26

M.

J.

BUXTON

ET AL.

T a b l e V Costs per patent (£), resource-sparing Primary

Fluconazole Medication Hospitalisation Monitoring Side effects Total Amphotericin B Medication Hospitalisation Monitoring Side effects Total

Cost differences (AMB-FLUCO)

Maintenance

Total

Low

High

Low

High

Low

High

458 2311 8I o 285o

458 3465 I39 o 4062

3514 466 I69 -4149

3514 693 238 -4445

3972 2777 25o o 6999

3972 4158 377 o 85o7

215 8064 95

3oi 12096 237

321 2657 594

482 6944 lO3O

536 10721 689

783 19040 1267

206 858o

308 12942

3572

-8456

206 I2 I52

308 2I 398

573o

888o

(577)

4oII

5153

I2891

E c o n o m i c consequences o f therapy

T h e unit cost estimates shown in T a b l e II were applied to the resources used in each patient m a n a g e m e n t model. T h e costs of the primary and the maintenance treatment were then s u m m e d to p r o d u c e the costs for the first year of treatment. F o r s u b s e q u e n t periods, if any, pro rata maintenance costs w o u l d apply. T h e results for the ' b a s e case' models are shown in T a b l e IV. T h e costs associated with each c o m p o n e n t of the primary and any maintenance treatment periods were estimated using b o t h low and high cost assumptions. Similar analyses were c o n d u c t e d for the ' r e s o u r c e - s p a r i n g ' and ' r e s o u r c e - i n t e n s i v e ' alternative models, the results of which are presented in Tables V and VI respectively. T h e medication cost calculations take account of drug costs and, w h e r e appropriate, the costs of disposable items used in administration of drugs through intravenous infusion. Considerable differences were noted in the IV line placement for fluconazole and amphotericin B, the former being given via a peripheral line in 9o % of cases and the latter being given via central catheter in 67 % of patients. Whilst only point cost estimates are used for the drugs themselves, ranges are used for the costs of intravenous infusion reflecting different assumptions regarding infusion practice and the costs of disposable items. A n estimated p r o c e d u r e cost for surgical implantation of intravenous lines was included, b u t no allowance was made for the additional time cost o f medical or nursing staff associated with other forms o f intravenous administration. F o r all models, the medication cost of fluconazole is higher than that of amphotericin B for b o t h p r i m a r y and maintenance therapy. It is hard to obtain strictly appropriate estimates for the hospitalisation costs for this group of patients, and in-patient costs vary considerably from hospital to hospital. Strictly, a total cost per day or per visit, minus the costs of

Cost implications for alternative care in A I D S

27

Table VI Costs per patent (£), resource-intensive Primary

Fluconazole Medication Hospitalisation Monitoring Side effects Total Amphotericin B Medication Hospitalisation Monitoring Side effects Total Cost differences (AMB-FLUCO)

Maintenance

Low

High

Low

High

I6O4 5837 8I 95 76I 7

I649 8755 I39 I43 1o686

3514 373 169 -4o56

3515 555 238

3639 9984 143 I318 I5o84 7467

3737 14976 355 I977 21045 IO359

723 2219

569 -351I (545)

Total Low

5II8

43o7

621o 25o 95 II 673

lO33 6359 987 -8379 4072

4362 I22O3 712 I318 I8595 6922

--

High

5163 931o 377 I43 14993 477o 21335 I342 I977

29424 ~443x

pharmacy, medical and surgical supplies is required. In practice, available cost estimates are not sufficiently accurate to make such precise distinctions. T h i s analysis showed the expected differences in costs of hospitalisation between the two treatments during primary therapy. T h e costs associated with amphotericin B were approximately two to three times those for fluconazole, refecting the difference in hospital stay (I8 vs. 45 days) expected for responders to each drug. Unlike oral fluconazole, each administration of IV amphotericin B during maintenance therapy requires either a day-care bed or overnight hospital stay, in addition to the regular pattern of out-patient monitoring c o m m o n to both therapies. Further, estimates were derived for the proportion of amphotericin B patients who received therapy either at home (I2 %) or in a day-care bed (88 %). Hospitalisation costs generated by fluconazole treatment, £,435-647, reflect out-patient visits, while the substantially higher amphotericin B costs, £2487-6839, refect the costs of Iv delivery as an out-patient, day-patient or home-patient (see Table IV). T h e monitoring costs for fluconazole are relatively low and constant across primary and maintenance treatment periods, despite different lengths of time. This result reflects the low intensity of the monitoring requirements for fluconazole due to a favourable side-effect profile and less frequent monitoring during the maintenance treatment period. Amphotericin B monitoring costs are two to three times higher than for fluconazole during primary therapy and three to four times higher during maintenance therapy. It was clear from both the clinical trial 1~ and the experience reported by the panel, that the patients treated with fluconazole experienced fewer and less severe side effects. T h e consequences of the differences in side effects were reflected in the costing of the length-of-stay loading factor. T h e loading for fluconazole was 5"6 additional days of hospitalisation x 2.6 incidence of severe

28

M.J. BUXTON

ET AL.

25

~'Z7"/'/2~,

NN 20

O_

~

xN\\\\\\\\•

N\\\\\\\~

io-

\\\\\\\\\\N ~\\\\\\\\\\, \\\\\\\\\\\

5-

I

~3333333333

~\\\\\\\\\\~

~!!!!~

......

\\\\\\\\\\\NI \\\\\\\\\\\M ~xx\xx~ x\\\\\\\\\\~

~NN!NN "dddddd ......

"\\\\\\37~?~l

Fluconazole Amphotericin B Low unit cost

Fluconazole

Amphotericin B

High unit cost

Fig. z. Cost per patient of z year of treatment~ fluconazole vs. amphotericin B. (m), Medication; ([~), hospitalisation; (E), monitoring; ([~), side effects. \

side effects multiplied by the daily cost of hospitalisation. Similarly, the loading for amphotericin B was I5"8 additional days of hospitalisation x 22"5 % incidence of severe side effects multiplied by the daily cost of hospitalisation. Similar differences apply for each of the patient management models across both unit cost assumptions. T h e final column in Table IV sums the primary and maintenance treatment costs of each resource component for the 'base case' management model, using both low and high unit cost assumptions. T h e bottom line of Table IV represents the difference in costs between amphotericin B- and fluconazolebased therapies for the 'base case' model. Differences are shown for primary, maintenance and total treatment costs for both unit cost assumptions. For low unit cost assumptions, the cost advantage for fluconazole is £4248. This difference rises to £ I r 947 when applied to the high-unit-cost scenario. Similar patterns emerged for the 'resource-sparing' models, where the low and high cost differences were £5153 and £ I 2 8 9 I respectively (Table V), and the 'resource-intensive' models (Table VI), where those differences were £6922 and £ I 4 4 3 I . T h e cost implications of the choice between fluconazole and amphotericin B

Cost implications for alternative care in A I D S

29

are summarised in Fig. I. T h e total costs of treating cryptococcal meningitis in A I D S patients for I year on each therapy are presented under both low and high unit cost assumptions for the composite 'base case'. T h e relative contributions of the major components of patient management are indicated within each bar. T h e cost advantage for fluconazole is apparent under both low and high cost assumptions. Hospital costs are clearly dominant for amphotericin B, accounting for over 80 % of the total costs under each unit cost assumption. Hospitalisation costs and medication costs are approximately equal contributors to the total cost of fluconazole therapy and represent approximately 96 % of total costs. Hospitalisation costs for amphotericin B exceed those for fluconazole by an amount larger than that by which direct drug cost of fluconazole exceeds amphotericin B, resulting in the overall cost advantage for fluconazole. Discussion

D e m a n d placed upon health care resources for the management of A I D S continues to grow, ~1 and new drug therapy needs to be assessed with respect to the use of resources as well as efficacy and safety. T h e purpose of this research was to examine the economic consequences of alternative methods of treating cryptococcal meningitis in A I D S patients. Evidence for efficacy of fluconazole and amphotericin B was obtained from multicentre randomised comparative trials for both primary and maintenance treatment. T h e assumption of comparable efficacy for I year of treatment is, in fact, conservative. T h e results from the maintenance trial suggest superior efficacy for fluconazole in the prevention of relapse. However, since these results were based on U.S.A. studies, the more conservative approach was selected. And while these data were obtained from controlled clinical trials, no such information concerning the use of medical resources for the management of these patients was available. Controlled clinical trials are often seen as the preferred method for assembling information for an economic evaluation. However, they may be problematic, because the requirements of the trial may lead to atypical patient management and reduce differences between the arms of the trial. 22 These considerations led to the use of a modified Delphi method as a means for describing, in a structured and quantifiable manner, the clinical management practices needed to treat patients who respond to each therapy. This approach proved to be an effective method of assembling expert opinion in a situation where the opportunity to establish the information empirically was not available. Whilst the classic Delphi approach involves a series of anonymous rounds to avoid the undue influence of any particular respondent,17.19 this panel benefited from the face-to-face discussions and debates concerning clinical practice. As in Park et al., 2° some adjustments were made following the discussion, but these appeared to be more the result of clarification and of context, rather than substantive changes in their views concerning clinical practice. T h e composite patient-management models that resulted from this exercise provide an important systematic description of treatment practices for these patients. It was apparent from the descriptions of patient management that both the

30

M. J. BUXTON E T A L .

patterns and the intensity of resources used to treat cryptococcal meningitis differed substantially between the two therapies. T h e key difference was due to the a m o u n t of hospitalisation. A m p h o t e r i c i n B therapy used more t h a n twice as m a n y days of fluconazole (45 vs. i8) in the composite view of the panel, due not only to the IV route of administration but also to the m o n i t o r i n g for, and t r e a t m e n t of, side effects associated with this drug. Hospitalisation accounted for substantial proportions of total amphotericin B costs u n d e r both low and high unit cost assumptions. Side-effect t r e a t m e n t requiring additional hospitalisation added to the total costs. T h e differences between fluconazole and amphotericin B were essentially the same for ' r e s o u r c e - s p a r i n g ' and ' r e s o u r c e - i n t e n s i v e ' variations of the model. Based on the results of this study, direct comparison of single components of patient m a n a g e m e n t (e.g. medication or monitoring) would not be sufficient to make i n f o r m e d decisions about the cost consequences of alternative therapies. Rather, total costs based u p o n a comprehensive analysis of all resources used for the m a n a g e m e n t of the patient need to be considered. A direct comparison simply of cost of medication would lead to an incorrect conclusion on cost differences. As the requirement to justify clinical m a n a g e m e n t decisions intensifies, a fuller assessment of the cost consequences of alternative therapies will be essential. (This work was supported by Pfizer Inc. We would like to thank Professor B. Dupont, Institut Pasteur, Paris; Dr G. Just, Medizinische Universitatsklinik, Frankfurt; Professor R. Esposito, Institute of Infectious Diseases, University of Milan; Dr J. van't Wout, Department of Infectious Diseases, Leiden University; Dr J. Arrizabalaga, Hospital Nuestra Senora de Arranza, San Sebastian and Dr V. Kitchen, St Mary's Hospital, London. This group served as the expert Cryptococcal Meningitis Treatment Panel. Dr A. J. Pinching, St Mary's Hospital Medical School, London, provided advice for the preliminary evaluation prior to the setting up of the Cryptococcal Meningitis Treatment Panel. Professor M. Boogaerts, Catholic University of Leuven, reviewed the draft questionnaire for clarity and completeness. M. Rees, St Stephen's Hospital, London, gave advice on U.K. costs. Without their input, this report would not have been possible.) References

i. Eng RHK, Bishburg E, Smith KM, Kapila R. Cryptococcal infections in patients with acquired immune deficiency syndrome. Am J Med 1986; 8I: 19-28. 2. Zuger A, Luie E, Holzman RS, SimberkoffMS, Rahal JJ. Cryptococcal disease in patients with the acquired immunodeficiency syndrome: diagnostic features and outcome of treatment. Ann Intern Med 1986; IO4 (2): 234-24o. 3. Chuck SL, Sande MA. Infections with Cryptococcus neoformans in the acquired immunodeficiency syndrome. New Engl J Med I989; 32I: 794-799. 4. Dismukes WH. Cryptococcal meningitis in patients with AIDS. J Infect Dis I988; I57 (4): 624-628. 5. Kovacs JA, Kovacs AA, Polis Met al. Cryptococcosis in the acquired immunodeficiency syndrome. Ann lntern Med I985 ; IO3: 533-5386. Bozzette SA, Larsen R, Chiu J et al. Successful secondary prophylaxis of cryptococcal meningitis with fiuconazole: a placebo controlled study. (Abstr) The VI International Conference on AIDS. San Francisco, CA, June, 199o. 7. Masur H. Fungal infections in AIDS patients : part 2. Optimal Treatment Regimens. Hosp Ther 1987; I2: 47-64.

Cost implications for alternative care in A I D S

3I

8. Stamm AW, Diasio RH, Dismukes WE et al. Toxicity of amphotericin B plus flucytosine in I94 patients with cryptococcal meningitis. A m J Med I987; 83 (2): 236-242. 9. Sugar AM, Saunders O. Oral fluconazole as suppressive therapy of disseminated cryptococcosis in patients with acquired immunodeficiency syndrome. A m J Med I988 ; 85 (4): 81-89. Io. Tucker TM, Williams PL, Arathoon EG et al. Pharmacokinetics of fluconazole in cerebrospinal fluid and serum in human coccidioidal meningitis. Antimicrob Agents Chemother I988; 32: 369-373. I I. Rinaldi MG, Robinson PA, Graybill JR et al. Fluconazole concentrations in patients undergoing antifungal therapy. (Abstr) 28th Interscience Conference on Antimicrobial Agents and Chemotherapy, October 23-26, I988 (Los Angeles, CA), p. IZI. I2. Stern JJ, Hartman BJ, Sharkey P et al. Oral fluconazole therapy for patients with acquired immunodeficiency syndrome and cryptococcosis : experience with 22 patients. A m J Med I988; 85 (4): 477-480. I3. Robinson PA, Knirsch AK, Joseph JA. Fluconazole for life-threatening fungal infections in patients who cannot be treated with conventional antifungal agents. Rev Infect Dis I99o; I2 : $349-363. I4. Dismukes O, Cloud S, Thompson Aet al. Fluconazole versus amphotericin B therapy of acute cryptococcal meningitis. (Abstr) 29th Interscience Conference on Antimicrobial Agents and Chemotherapy, September i7-2o, I989 (Houston, TX), p. 282. 15. Drummond MF, Stoddart GL, Torrance GW. Methods for the economic evaluation of health care programmes. London : Oxford University Press, I98 ~. I6. Institute of Medicine. Assessing Medical Technologies. Washington, D.C.: National Academy Press, r985. 17. Fink A, Kosecoff J, Chassin M, Brook RH. Consensus methods: characteristics and guidelines for use. A m J Public Health I984; 74 (9) : 979-983. I8. Duffield C. The Delphi technique. Aust J Adv Nurs I988; 6 (2): 41-45. 19. Linstone HA, Turoff M. The Delphi method: techniques and applications. Reading, MA: Addison Wesley, I975. 20. Park RE, Fink A, Brook RH et al. Physician ratings of appropriate indications for six medical and surgical procedures. A m J Public Health I986; 76 (7): 766-772. 2I. Cunningham D, Griffiths SF. AIDS : counting the cost. Br M e d J I987; 295 (II): 921-922. 22. Drummond MF, Stoddart GL. Economic analysis and clinical trials. Cont Clin Trials I984; 5 : I I 5 -I28-