Professional Documents
Culture Documents
Transfusion - 2002 - Pereira - A Model of The Health and Economic Impact of Posttransfusion Hepatitis C Application To
Transfusion - 2002 - Pereira - A Model of The Health and Economic Impact of Posttransfusion Hepatitis C Application To
T
he risk of acquiring HCV infection through the
BACKGROUND: Cost-effectiveness analyses are transfusion of screened blood has decreased dra-
needed to decide the value of further expansion of the matically over the last decade, mainly as a result
screening protocols for HCV in blood donors. However, of the availability and increasing sensitivity of
such analyses are hampered by imperfect knowledge of HCV antibody tests. For instance, in the United States, the
the health and economic repercussions of posttransfu- risk declined from 1 per 526 blood units in 19891 to 1 per
sion hepatitis C (PTHC). 103,000 units in 1996.2 Despite this remarkable achieve-
STUDY DESIGN AND METHODS: A Monte Carlo simu- ment, public concern about HCV infection seems to be ris-
lation of a Markov model representing the outcomes of ing rather than abating, and the public urges the blood bank
patients transfused with HCV-infective blood was used community and regulatory agencies to further increase the
to estimate the health and economic impact of PTHC safety of the blood supply.
and to calculate the cost-effectiveness ratio of various HCV NAT utilizing PCR techniques, and the recently
HCV screening methods. described assay for HCV antigen, shortens by several days
RESULTS: Median survival for hypothetical patients with the window period of antibody testing and can further re-
PTHC and for controls without hepatitis was 11.25 and duce the risk of transfusion transmission of HCV infec-
11.75 years, respectively. Overall, 12.3 percent of pa- tion.3,4 HCV NAT is already required by European regula-
tients receiving HCV-infective blood will develop chronic tions for all plasma pools intended to be fractionated in
hepatitis, 9.3 percent will progress to liver failure, and products for clinical use. In Germany, it also constitutes a
9.25 percent will eventually die of liver disease after a requirement for release of cellular blood components.5
median time of 20.75 years (range, 6-70). Ninety-one Other European countries and the major US blood collec-
percent of the infected blood recipients had no reduction tion agencies are adopting a similar approach. Not surpris-
in life expectancy due to PTHC, and the average loss ingly, expanding the screening protocols for HCV will make
per patient was 0.754 years. The present value of the blood transfusion more expensive, whereas the potential
lifetime health costs incurred by patients with PTHC is benefits remain uncertain, given the already low risk of HCV
$6330 per case. HCV antibody testing increases the pa- transmission. Cost-effectiveness studies and other decision
tients’ life expectancy by 20.4 hours per blood collection
tested, and it results in net savings by decreasing the
ABBREVIATIONS: ASLE = age- and sex-adjusted life expectancy;
number of patients that will require treatment for liver
PTHC = posttransfusion hepatitis C; QALY = quality-adjusted life
disease in the future. Adding HCV NAT increases the
years.
patients’ life expectancy by 0.08 hours per blood collec-
tion tested, at a cost-effectiveness ratio of $1,829,611 From the Hemotherapy and Hemostasis Service, and the Blood
per QALY gained. Bank, Hospital Clínic and Augusto Pi-Sunyer Memorial Institute
CONCLUSION: PTHC has low health benefits because for Biomedical Research, Barcelona, Spain.
of the advanced age of many blood recipients. Testing Address reprint requests to: Arturo Pereira, MD, PhD, Ser-
donors for HCV antibodies results in net savings for the vice of Hemotherapy, Hospital Clínic, Villarroel 170, 08060
health care system, despite low health benefits. Adding Barcelona, Spain; e-mail: apereira@clinic.ub.es.
HCV NAT would produce little additional gain at a very Supported in part by Grant FIS 99/0464 from the Ministry
high cost. of Health of the Government of Spain and Grant 1999ACES
00033 from the Autonomous Government of Catalonia.
Received for publication September 9, 1999; revision re-
ceived November 29, 1999, and accepted December 9, 1999.
TRANSFUSION 2000;40:1182-1191.
analyses are therefore pertinent, as they can help patients, how this method allows representation of the patients’
physicians, policy makers, and the public in deciding on the course of disease have been published elsewhere.15 In the
efficacy of these technologies. present model, the posttransfusion course of each patient
However, decision making in regard to further expan- with HCV infection was compared to what would had been
sion of the screening for HCV in blood donors is severely his or her state of health if he or she had not acquired the
constrained by our limited knowledge of the impact that infection, so that each case acted as its own control. Survival
posttransfusion hepatitis C (PTHC) has on both the health curves were derived from the simulation of 100,000 cases.
of patients and the resources for health care. Studies on the For calculating aggregated health repercussions and costs,
follow-up of patients with PTHC represent the disease as a 10 million cases were simulated for every scenario of the
slowly progressive one, in which liver failure is a rare and baseline and sensitivity analysis.
late complication, and which has a minimal repercussion The model was written and compiled in programming
on the patients’ life expectancy.6-8 Studies on the more se- software (PowerBASIC 3.2, PowerBASIC Inc., Carmel, CA).
vere outcomes of hepatitis C reveal, however, that it can be Outputs were analyzed on a spreadsheet (Excel 5.0, Microsoft
associated with significant morbidity and mortality.9 Unfor- Ibérica, Madrid, Spain) and with a statistical package for
tunately, none of the above-mentioned studies provide in- social sciences (SPSS 6.1.3, SPSS Inc., Chicago, IL). Graphi-
formation that is useful for decision making. Prospective cal plots were drawn with analysis software (Prism,
studies on persons with PTHC usually lack control groups, GraphPad Software Inc., San Diego, CA).
exclude patients who die soon after transfusion, and some-
times have short follow-up periods. On the other hand, in Characteristics of blood transfusion recipients
studies on the more severe outcomes of hepatitis C, the size Probability distributions for age and sex of blood recipients
of the population from which the reported cases were ex- were derived from a population of 9661 patients who re-
tracted is not mentioned, and that precludes any risk cal- ceived transfusion(s) in our institution from July 1996 to
culation because of a lack of denominators. Furthermore, December 1998. The age of these patients ranged from new-
most recent investigations on the evolution of HCV infec- born to 104 years, with first, second, and third quartiles
tion focus on histologic progression, which often does not being 52, 67, and 76 years, respectively; 54 percent of the
correlate with clinical progression or with the need for ther- patients were males. Age- and sex-adjusted life expectancy
apeutic interventions.10-13 Finally, there is a lack of studies (ASLE) values were obtained from mortality tables for the
analyzing the lifetime costs of treating patients with PTHC. population of Catalonia, in northeast Spain.16 These pa-
Ideally, a study suitable for ascertaining the health and tients’ characteristics were comparable to those reported
economic consequences of PTHC would entail the long- for blood recipients in the United States17 and other west-
term follow-up of a large and carefully surveyed cohort of ern countries.18
patients and appropriate controls. However, such a study The probability of short-term mortality due to under-
would be too expensive, and its conclusions would be de- lying diseases was stratified by the age of blood recipients,
layed for several decades, by which time the natural history according to the study by Vamvakas and Taswell19 (Table 1).
of HCV infection will have changed because of forthcom- We considered, conservatively, that the impact of underly-
ing and more effective treatments. As decisions on the ap- ing diseases on mortality does not extend longer than 2
propriateness of new preventive interventions should be years after transfusion.
made now, mathematical modeling emerges as the most
suitable tool, albeit an imperfect one, with which to assess Modeling posttransfusion HCV infection
the health and economic repercussions of PTHC. Health states in the model were defined on the basis of clini-
In the present study, the course of patients with PTHC cal repercussion and need for therapeutic intervention, so
and controls without hepatitis has been simulated, focus- that they do not necessarily correspond to histologic stages.
ing on clinical outcomes and associated health care costs. Accordingly, mild chronic hepatitis was included in the as-
The main outputs of the simulation were used to estimate ymptomatic chronic carrier state, and compensated cirrho-
the benefits derived from different HCV screening methods sis, the chronic hepatitis state. This is consistent with recent
and to calculate the cost-effectiveness of further expand- investigations that show no difference in perceived quality
ing the current HCV screening protocols for blood donors. of life in patients with chronic hepatitis and those with
compensated cirrhosis.20,21
We considered four possible health states in the pro-
MATERIALS AND METHODS
gression of HCV infection. Baseline health is the health sta-
Basic model description tus of controls without HCV infection and of patients from
A Monte Carlo simulation of a Markov model14 was built to whom the virus cleared spontaneously after the acute in-
represent the more relevant health states and outcomes of fection. Asymptomatic chronic carriers are patients with no
patients transfused with HCV-infective blood. Details on symptoms and only occasionally mild elevations of ALT;
study by Kimm et al.29 Some of the above cost figures were range was used in sensitivity analysis, including the “zero
adjusted to fit them to our model’s 3-month cycle. Quality risk” that would derived from a hypothetical, perfect test.
of life repercussions assigned to health states are also shown We assumed that each infective blood collection that passes
in Table 1 and represent the most common estimates that undetected through blood bank screening will transmit
can decrease life expectancy, as referenced in the pertinent HCV to 1.656 blood recipients. This figure is the average
medical bibliography.28-30 Future costs and impact on qual- number of components that are transfused from each blood
ity-adjusted life years (QALY) were discounted at a 3-per- collection, according to data from our blood bank and
cent annual rate. transfusion service in 1997.
Costs were assessed from the perspective of a national Marginal cost-effectiveness was calculated per blood
health service, and they included only direct health costs. collection tested according to a ratio:
Indirect costs, such as those related to the loss of produc- (incremental cost of the HCV screening method – costs
tivity and other social expenses incurred by patients with savings derived from treating fewer cases of PTHC in
PTHC, were not accounted for. Life expectancy was ex- the future)/QALY gained by the population of blood re-
pressed in QALY, and also in unadjusted and undiscounted cipients because of fewer PTHC cases.
survival years.
The incremental cost for HCV antibody screening was
RESULTS
estimated at $5 per blood collection tested, according to
prior studies.30 There is much uncertainty about the incre- Health and economic impact of PTHC
mental cost of adding HCV NAT to the donor screening pro- The predicted median survival for patients with PTHC and
tocols. Factors such as pool size, the number of pools tested for controls without hepatitis was 11.25 and 11.75 years,
per run, the algorithms used to retest pools that are HCV respectively, with a 30-year projected survival rate of 18
positive until the individual HCV-positive donor is identi- percent and 19 percent, respectively. As shown in Fig. 1,
fied and removed before component release, the financial early mortality due to underlying diseases was the main
charges due to acquisition or leasing of expensive equip- factor in shortening the life expectancy of patients and con-
ment, and the wages for technician overtime may greatly trols. After exclusion of patients who died early as a result
influence the actual cost of NAT. In addition, economies of of their underlying disease(s), median survival for the re-
scale secondary to centralization of NAT and market com- maining patients and controls increased to 17 and 17.75
petition between firms may further influence costs in the years, respectively. The model predicts that 12.3 percent of
near future. In the baseline analysis, we estimated the in- patients given HCV-infective blood will develop chronic
cremental cost of routine HCV NAT as $10 per blood col- hepatitis and 9.3 percent will progress to liver failure. Inter-
lection tested. However, a wide range, from $5 to $25, was feron will be used in 4.9 percent of patients, and 0.5 percent
used in most of our analyses. This spans from the expected of patients will undergo liver transplantation. Overall, 9.25
cost of HCV antigen testing (Pereira M, oral communica- percent of patients in the simulated cohort will eventually
tion, September 1999), to the cost of HCV NAT under the die of HCV-related liver disease(s), which will occur at a
worst conditions. median of 20.75 years (range, 6-70) after transfusion.
100 –
Calculation of the cost:benefit ratio for HCV
90 –
screening methods
80 –
The main outputs of the model and the residual risk of HCV
70 –
transmission through screened blood were used to calcu- —— theoretical life expectancy
Probability (%)
the baseline analysis, residual risk after HCV NAT was esti- Fig. 1. Projected survival after transfusion-acquired infection,
mated as 2.7 per million blood collections tested, accord- compared with survival of controls without infection, and the
ing to Kleinman et al.3 However, because there is some un- sex- and age-adjusted theoretical life expectancy. Also shown
certainty as to the actual magnitude of this risk, a wide is the Kaplan-Meier estimate of the risk of liver failure.
The loss of life expectancy due to PTHC varies consid- PTHC-induced loss of life expectancy and the associated
erably among particular patients, being null in 90.7 percent lifetime costs, but to a much lesser extent (Figs. 3 and 4).
and longer than 10 years in 3 percent (Fig. 2), with an aver- Growth of health care costs above current average
age loss of 0.754 years per infected blood recipient. With prices can also influence our baseline estimates of the eco-
regard to QALY, 67.9 percent of patients will suffer no health nomic repercussions of PTHC. Fig. 5 shows that lifetime
impairment due to HCV infection, and the average reduc- costs incurred by patients with PTHC will grow exponen-
tion per infected blood recipient is 0.468 QALY. tially as the differential inflation rate of medical costs in-
Table 2 shows the predicted lifetime costs of treating creases, mainly because of the higher costs of treating pa-
PTHC-related liver diseases. Liver failure is the health state tients with liver failure. The long period covered by the
with the largest requirement for health care resources. Liver analysis also makes the expected QALY and costs change
transplant, despite its high unitary cost, represents less than with the discount rate used in the analysis. In sensitivity
16 percent of the lifetime costs, which is mainly due to the analysis varying the discount rate from 0 to 10 percent, the
small number of patients in whom the procedure will be average number of QALY lost per patient decreased from
performed. 1.043 to 0.125. Over the same range of discount rates, the
The PTHC-induced loss of life expectancy was very sensi- lifetime cost per patient varied from $11,511 to $2,324.
tive to the patients’ age. As shown in Fig. 3, the average loss
per patient grew exponentially as the patients became younger. Health and economic gains from HCV screening
The lifetime costs associated with treatment of PTHC-related protocols in blood donors
complications also rose markedly as patient age decreased Table 3 shows the impact that routine HCV RNA testing of
(Fig. 4). Reducing the early mortality caused by the patients’ blood donors would have on the health and economic bur-
underlying disease(s) also brought an increase in both the den of PTHC, as compared with the gains provided by sur-
rogate markers or HCV antibody testing and those to be
expected from a hypothetical perfect test that completely
eliminated PTHC. As it can be seen, each blood collection
Percentage of patients
Short-term mortality due to underlying diseases Short-term mortality rate due to underlying diseases
(in multiples of the baseline rate) (in multiples of the baseline value)
,
Loss of life expectancy (years/patient)
Cost-effectiveness (thousands of
,
dollars/QALY gained)
,
Age at transfusion
Age at transfusion
Fig. 5. Cost-effectiveness of NAT in relation to patients’ age
Fig. 3. Loss of life expectancy of simulated patients with trans- (•—•) and their short-term rate of mortality due to underlying
fusion-acquired HCV infection according to age at the time of disease(s) ({—{). Arrows denote the baseline values.
transfusion (•—•) and to short-term rate of mortality due to
NAT also had little sensitivity to inflation of the medical
their underlying disease(s) ({—{).
costs of treating liver diseases. At a hypothetical annual
Short-term mortality due to underlying diseases inflation rate of 12.5 percent, the cost-effectiveness of NAT
(in multiples of the baseline rate)
declined to $1.529 million, a decrease of only 13 percent
from the baseline estimate.
,
DISCUSSION
Lifetime costs ($/patient)
merase chain reaction: increase of the virus in advanced 40. Iglehart JK. The American health care system—expendi-
liver disease. Hepatology 1993;18:16-20. tures. N Engl J Med 1999;340:70-6.
34. Gretch D, Corey L, Wilson J, et al. Assessment of hepatitis C 41. Casarett DJ, Lantos JD. Have we treated AIDS too well? Ra-
virus RNA levels by quantitative competitive RNA poly- tioning and the future of the AIDS exceptionalism. Ann In-
merase chain reaction: high-titer viremia correlates with tern Med 1998;128:756-9.
advanced stage of disease. J Infect Dis 1994;169:1219-25. 42. Magliocchetti N, Torchio P, Corrao G, et al. Prognostic fac-
35. Niederau C, Lange S, Heintges T, et al. Prognosis of chronic tors for long-term survival in cirrhotic patients after the first
hepatitis C: results of a large prospective cohort study. episode of liver decompensation. Ital J Gastroenterol
Hepatology 1998;28:1687-95. Hepatol 1997;29:38-46.
36. Benvegnu L, Fattovich G, Noventa F, et al. Concurrent 43. Wright JC, Weinstein MC. Gains in life expectancy from
hepatitis B and C virus infection and risk of hepatocellular medical interventions—standardizing data on outcomes. N
carcinoma in cirrhosis: a prospective study. Cancer Engl J Med 1998;339:380-6.
1994;74:2442-8. 44. Hlatky MA, Rogers WJ, Johnstone Y, et al. Medical care
37. Vento S, Garofano T, Renzini C, et al. Fulminant hepatitis costs and quality of life after randomization to coronary
associated with hepatitis A virus superinfection in patients angioplasty or coronary bypass surgery. N Engl J Med
with chronic hepatitis C. N Engl J Med 1998;338:286-90. 1997;336:92-9.
38. Kiyosawa K, Sodeyama T, Tanaka E, et al. Interrrelationship 45. AuBuchon JP, Birkmeyer JD, Busch MP. Safety of the blood
of blood transfusion, non-A, non-B hepatitis and hepato- supply in the United States: opportunities and controver-
cellular carcinoma: analysis by detection of antibody to sies. Ann Intern Med 1997;127:904-9. 6
hepatitis C virus. Hepatology 1990;12:671-5.
39. Fattovich G, Giustina G, Degos F, et al. Morbidity and mor-
tality in compensated cirrhosis type C: a retrospective fol-
low-up of 384 patients. Gastroenterology 1997;112:463-72.