|
Hepatitis C screening strategies in haemodialysis patients |
Saab S, Brezina M, Gitnick G, Martin P, Yee H F |
|
|
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 technologies studied were screening strategies for hepatitis C virus (HCV). These were:
ALT values with confirmatory testing (biochemical);
serial enzyme-linked immunosorbent and strip immunoblot assay testing (serological); and
polymerase chain reaction (viral).
Economic study type Cost-effectiveness analysis.
Study population The study population comprised haemodialysis patients at risk of HCV infection.
Setting The setting was an institution. The economic study was conducted at the Departments of Medicine and Physiology, Division of Digestive Diseases, University of California at Los Angeles, USA.
Dates to which data relate Effectiveness evidence and resource data were derived from studies published between 1993 and 2001. The price year was not reported.
Source of effectiveness data The effectiveness evidence was derived from a review of published studies.
Modelling A decision analytic model was constructed to determine the expected outcomes and costs of each of the three screening options. A hypothetical cohort of 5000 haemodialysis patients was followed for 5 years. In the biochemical strategy, all patients underwent baseline screening with antibody to HCV serological tests and those without HCV underwent monthly ALT testing, whereas those with elevated ALT values underwent serological testing. In the serological screening, patients underwent testing twice yearly by ELISA and positive results were confirmed by SIA. In the viral-based strategy, PCR testing was performed twice yearly in the field. Viral-based screening was considered the gold standard method in the decision model.
Outcomes assessed in the review The outcomes assessed in the review were the probability of infection (prevalence and incidence) and the sensitivity and specificity of ALT and ELISA/SIA. In the literature, SIA was the gold standard for ALT and PCR for ELISA/SIA.
Study designs and other criteria for inclusion in the review Sources searched to identify primary studies The MEDLINE database was searched from 1992 to 2000 to find primary studies using different combinations of the following keywords: hepatitis C, haemodialysis, and screening. Further primary studies were found from the reference checks of initial studies.
Criteria used to ensure the validity of primary studies Methods used to judge relevance and validity, and for extracting data Number of primary studies included Seventeen primary studies were used as the source of the effectiveness evidence.
Methods of combining primary studies Investigation of differences between primary studies Results of the review The prevalence and incidence of HCV infection were 0.09 (range: 0.015 - 0.15) and 0.002 (range: 0.0005 - 0.005), respectively. The sensitivity and specificity values were 0.83 and 0.90 for ALT and 0.93 and 0.84 for ELISA/SIA.
Measure of benefits used in the economic analysis A specific efficacy measure was used in the economic analysis: the likelihood ratio. This was defined by the authors as the odds that a diagnosis of de novo HCV infection is made in patients with HCV infection versus those without HCV infection. This measure was stated to embody both sensitivity and the false-positive rate and previous studies were reported to show that it remained stable for changes in prevalence. In fact the likelihood ratio is the probability of the result given the disease divided by the probability of the result given no disease. More specifically, if the result is labelled as positive, then it is the sensitivity divided by (1 - specificity) or the probability of obtaining a true positive divided by the probability of obtaining a false positive. As such, it is not correlated with prevalence unless the properties of the test itself change with prevalence.
Direct costs Discounting was relevant because the study lasted 5 years, but a discount rate was not reported. Resource quantities were not reported separately and the resource/cost boundary reflected the perspective of the analysis. Only the direct costs associated with the performance of the screening tests were considered in the analysis. The estimation of costs was based on actual charges used at the authors' institution. Charges were then converted into true unit costs using an approximate cost-charge ratio of 0.53, with unit costs of each test being given. The price year was not reported. The resource data were derived from studies published between 1993 and 2001. The test costs were $6.31 for ALT (range: $1.58 - $15.78), $20.30 (range: $5.08 - $50.75) for ELISA, $58.30 (range: $14.58 - $145.75) for SIA, and $74.20 (range: $18.55 - $185.50) for HCV RNA.
Statistical analysis of costs No statistical analysis of costs was reported.
Indirect Costs No indirect costs were included.
Sensitivity analysis One-way and two-way sensitivity analyses were stated to have been conducted on test costs, HCV incidence and prevalence, to assess the robustness of the model and how far uncertainty around the data estimates would affect cost per patient screened. In fact, the authors did not conduct any two-way analyses. The analyses described as two-way were plots of the cost per patient screened versus multiple of baseline unit costs for each strategy.
Estimated benefits used in the economic analysis The estimated likelihood ratios of the biochemical and serologic approach were 44 and 88, respectively. In particular, biochemical-based screening would have misclassified 45 haemodialysis patients as having no evidence of HCV infection (false negative) and 93 patients as having infection (false positive). Serologic-based screening, on the other side, would have misclassified 45 haemodialysis patients as not being infected and 48 patients as being HCV infected. The serologic-based screening was therefore, more effective than the biochemical-based approach.
Cost results The expected cost of screening per patient was $382 for biochemical-based strategy, $195 for serological-based screening, and $696 for viral-based screening. The expected total costs for screening the hypothetical cohort for 5 years were $1,908,524 for the biochemical strategy, $979,965 for the serologic screening, and $3,480,506 for the viral strategy.
Synthesis of costs and benefits The authors computed the cost per true HCV infection detected: $4,257 for biochemical-based screening, $2,210 for serological-based screening, and $6,922 for viral-based screening. Sensitivity analyses indicated that the rankings of costs remained the same.
Authors' conclusions The authors concluded that a serological-based screening strategy is less costly and more effective than biochemical-based screening in the diagnosis of HCV, and that it should be considered for HCV screening in haemodialysis populations.
CRD COMMENTARY - Selection of comparators The technologies were chosen with reference to established practice, with biochemical being recommended by the Centers for Disease Control and Prevention. Clearly there are, potentially, an infinite variety of strategies when accounting for variability by technique and frequency. Perhaps, given the use of a model, more variability could have been tested in order to optimise the strategy. You should consider whether the comparators represent widely used technologies in your own setting.
Validity of estimate of measure of effectiveness The effectiveness estimates were derived from a review of the literature. However, the contribution of each single analysis was not reported clearly and the method of combination of primary studies was not stated. As a consequence, differences in sample size, study designs, and baseline characteristics of patient populations among the studies were not accounted for in the review. Potentially, a major limitation was that accuracy data were derived from studies with different gold standards, the effect of which, without applying the same gold standard, is impossible to determine. A further limitation of the model, as acknowledged by the authors, was that patients with indeterminate SIA results were supposed not to be HCV infected.
Validity of estimate of measure of benefit The benefit measure was based on the likelihood ratio, which incorporated both the sensitivity and the positive rate of the test. However, the authors stated that the likelihood ratio was employed only to overcome the lack of data about long-term consequences of HCV infections needed to compute a more appropriate benefit measure. The authors' definition of likelihood ratio was also incorrect and the description of its usefulness unclear. In fact, it can be shown that tests should be compared on the basis of their resource and effectiveness consequences by the number of each of false positives, false negatives, true positives and true negatives, which depend on sensitivity, specificity and prevalence.
Validity of estimate of costs The estimation of costs was based on actual costs, which appeared to be quite specific to the authors' institution. Although charges were used as the source of cost data, a mechanism based on the cost-charge ratio was use to approximate the true resource consumption. According to the perspective adopted in the study, only the direct costs of the tests were accounted for in the analysis. The sensitivity analysis was, in fact, largely redundant since it did not show anything that could not have been established more simply. The only basis for differentiating between the strategies investigated by the authors was by cost. The only way in which a change in unit costs could alter the choice of strategy would be if those for the more effective options were to increase sufficiently to make total costs greater than those for the biochemical strategy. Even then, the correct approach would be an incremental analysis with which one could weigh the increase in effectiveness against the increase in costs. An increase in prevalence could never affect virology since it is assumed to be perfect. The ranking of the other two, again, would not alter since they only differed in the rate of false positives, the effect of which would diminish with increased prevalence but not reverse. Sensitivity analysis is only worth performing on those parameters whose variability might alter the choice of technology. This should have included some alteration of test accuracy and/or prevalence and/or the effect on the subsequent health consequences and none for the consequences of HCV infection or unnecessary treatment of false positives. Discounting was relevant, but none was reported. Furthermore, costs and quantities were not reported separately.
Other issues The issue of the generalisability of the study results to other settings was addressed, sensitivity analyses being performed on some of the model inputs (although this was largely unnecessary). The authors made few comparisons of their study findings with those from other studies. Finally, the authors highlighted a possible limitation of the decision model structure, which implied that, on entry into the biochemical strategy, patients underwent baseline serological testing. Therefore, the total costs of the biochemical-based screening included also the costs of the serological approach. This was the screening strategy recommended in the USA.
Implications of the study The main implication of the study stated by the authors was that serological screening should be adopted as the best strategy for HCV surveillance in chronic haemodialysis patients. However, as has been shown above, the measures of effectiveness were limited, and the consequences, in terms of test accuracy and prevalence, were not investigated. Finally, it is not clear why virology was not considered for recommendation since, although it was more costly to administer, it was more effective and would reduce future resource consumption.
Source of funding Supported in part by award no. DK 02450 from the National Institute of Diabetes and Digestive and Kidney Diseases (H.F.Y.); American Association for the Study of Liver Diseases/Schering Award (S.S.); and National Service Research Award no. DK07180-2 from the National Institutes of Health (S.S.).
Bibliographic details Saab S, Brezina M, Gitnick G, Martin P, Yee H F. Hepatitis C screening strategies in haemodialysis patients. American Journal of Kidney Diseases 2001; 38(1): 91-97 Other publications of related interest Comment in: American Journal of Kidney Diseases 2001;38(1):186-8.
Indexing Status Subject indexing assigned by NLM MeSH Alanine Transaminase /blood; Cost-Benefit Analysis; Enzyme-Linked Immunosorbent Assay /economics; Hepacivirus /genetics /immunology; Hepatitis C /complications /diagnosis /virology; Humans; Immunoblotting /economics; Kidney Failure, Chronic /complications /therapy; Mass Screening /economics /methods; RNA, Viral /blood; Renal Dialysis; Sensitivity and Specificity AccessionNumber 22001001430 Date bibliographic record published 31/05/2002 Date abstract record published 31/05/2002 |
|
|
|