METHODS

General approach

We used a Markov model to simulate progression through the various states of ill health involved in progressive HCV disease, including death (from hepatic and other causes), using DecisionMaker 7.07 (Pratt Medical Group, Boston, USA). In this approach a cohort of hypothetical patients move between specified health states on an annual basis according to probabilities for progression based on best available evidence. For each annual cycle, the model tracks how many patients are in each state so that, given estimates for the quality of life and resource use associated with the health states, the model yields an average lifetime cost utility for each strategy and allows estimation of the marginal cost per quality adjusted life year (QALY) gained. This approach to modelling HCV disease progression performs well when compared with existing cohort data.^13–15 The model was originally developed by one of us (JW) to investigate aspects of management of HCV disease in a North American setting. In this study we adapted the model for the UK. The analysis is undertaken primarily from the perspective of UK commissioning authorities (primary care groups/trusts and district health authorities). Where possible, information is included that allows a wider societal perspective to be taken.

Parameters used in the analysis

Patient population

At the start of the model, patients are in one of several states of compensated chronic HCV infection defined by sustained abnormal liver function and histological status. These states also summarise the populations studied in the two clinical trials of combination and monotherapy.^10,11 Three variables have been consistently identified as predicting improved sustained response to antiviral therapy^16–22 and these are considered when calculating the cost effectiveness of targeted treatment strategies: histological type (mild hepatitis, moderate hepatitis, or cirrhosis); level of viraemia; and viral genotype (1 and 4, or other).

Following the inclusion criteria of the clinical trials of interferon (alone or in combination), and current practice in the UK, the model does not consider treatment in patients with persistent chaotic intravenous drug use, excessive alcohol consumption, psychiatric disorders including depression, ischaemic heart disease, severe respiratory disease, or diabetic retinopathy.

Natural history

The model has been described in more detail previously.¹³ For progression of non-compensated disease, we used pooled estimates from three retrospective observational studies of non-A non-B hepatitis in which subsequent serology showed HCV as the causative agent,^23–25 and a published cohort study of compensated cirrhosis.¹⁴ Prognosis following decompensation varies according to the predominant clinical feature at presentation—that is, ascites (diuretic sensitive or refractory), variceal haemorrhage, and hepatic encephalopathy.^26–28 We assumed that recovery from decompensated to compensated states would only occur in the event of transplantation. Risk of development of hepatocellular carcinoma was estimated as 1–3% per year in patients with cirrhosis.^14,29,30 Probability of death from other causes was calculated from UK age specific tables.

Transplantation

It is difficult to estimate the transplantation rate in the UK population. In England and Wales in 1998, there were 632 liver transplantations (UKTSSA, personal communication, September 1999). This gives an overall population rate for transplantation of 12 per million, which is 80% of that in the USA where the annual probability for transplantation in the eligible HCV infected population is 3.1%. We therefore estimated the annual UK transplantation probability as 2.5% (80% of 3.1%).

Antiviral treatment

Response rates were based on the results of combined US and international trials of interferon and ribavirin combination therapy.^10,11 Individual patient data were combined to examine response rates in several subgroups.

• Genotype 1 versus genotype 2 or 3; 50% of UK HCV sufferers have genotype 1.

• Viral load, defined as (a) >2 million copies/ml or (b) ≤2 million copies/ml.

• Sex.

• Histological grade: mild or moderate hepatitis.

• Age: 40 years and under; 41–50 years; and over 50 years of age.

As a bias against therapy, we assumed that a few treatment responders could experience ongoing progression after sustained viral clearance,³¹ despite the observation of regression of fibrosis following response to interferon monotherapy.³²

Treatment duration was 24 weeks for genotype 2 or 3, and 48 weeks otherwise. If the patient remained viral positive, we assumed monotherapy would be stopped at 12 weeks and combination therapy would be stopped at 24 weeks. Where response was shown at these points, it was assumed treatment would continue for 12 months except in the subgroup of genotype 2 or 3 where treatment would be limited to 24 weeks. We calculated average costs of treatment for the treated cohorts based on these stopping rules.

Drug costs were calculated on a per diem basis with prices taken from the British National Formulary.³³ Drug dosage is related to weight. Based on the experience of the randomised controlled trials of combination therapy, dosage was as received in the combined trials. In a European population, this likely overestimates the dose because US patients were heavier.

Resource use and valuation

Resource use is based on typical UK management of HCV disease in each of the health states concerned (see table 1). Management patterns were determined using a questionnaire survey of nine UK liver units, including five transplant centres. Net ingredient costs for drug treatments are from the British National Formulary, March 1999.³³ Cost of transplantation was estimated from the average contract price of liver transplantation in the UK for 1997/8.³⁴ This includes costs for transplantation and follow up. Costs have been estimated for 1998/9 using, where appropriate, the Hospital and Community Health Services Index to adjust values.³⁵

Cost estimates for procedures and diagnostic tests (at 1998 prices) were provided by the costings units at Ninewells Hospital, Dundee (Dr D Carson, personal communication, September 1999), from a small purposive sample of NHS trust financial managers surveyed in September 1999 (listed at the end of the paper) and by reference to published tariffs from hospitals in the South and West regions of the UK.³⁶ Unit costs of items associated with drug treatment and outpatient management of HCV disease are shown in table 2. Costs of inpatient admissions for complications of chronic HCV infection were taken from NHS Healthcare Resource Group (HRG) reference costs. HRGs used in relation to health states involved are shown in table 3. HRGs include all hotel and treatment costs associated with admission, including drugs and investigations. However, as combination therapy was not a standard part of management at the time HRG costs were estimated by trusts, we have assumed that HRGs exclude this element.

Utility of health states

We used the utility values obtained by Wong et al from American hepatologists using the standard gamble and time trade off techniques in their study of pretreatment evaluation strategies in HCV disease³¹ to weight the time spent in each health state for quality of life (see table 4). These values were checked with UK hepatologists.

Sensitivity analysis

We carried out one way sensitivity analyses to investigate the effect on the model of varying key cost and disease progression parameters by 50% upwards and downwards. Costs and benefits were discounted at 6%.

RESULTS

Main analysis

The results of the main analysis are shown in table 5. Average cost of drug treatment per patient was £2738 for interferon monotherapy and £7014 for combination therapy.

Subgroup analyses

The results of the subgroup analyses are shown in table 6. We present only the results with benefits and costs discounted at 6%. These therefore represent the highest estimates of cost effectiveness to current commissioners.

Sensitivity analyses

The results of sensitivity analyses are shown in table 7. The conclusions of the main analysis are not markedly sensitive to variation in key parameters, although drug cost appears to be a more important determinant of cost effectiveness than assumptions about costs of treatment or disease progression.

DISCUSSION

Our results suggest that interferon and ribavirin are quite cost effective. Without considering the timing of costs and benefits, the additional cost of combination therapy is almost completely offset by future savings. Because the benefits are delayed, discounting is important. Although the appropriate rate is contentious, using a discount rate of 6% for both costs and benefits, we showed that combination therapy was likely to produce an additional QALY for less than £3000. This is considered by some as below a decision threshold for obtaining strong support from NHS commissioners.³⁷ In some groups of patients combination therapy appears to be more cost effective, with discounted cost effectiveness below £1000 per QALY. Our results show that even in groups least likely to respond, the cost effectiveness of combination over monotherapy falls below £10 000. Compared with accepted monotherapy, our results suggest that changing to combination therapy is cost effective in most groups, with cost per QALY below £5000. The results do not appear to be sensitive to major changes in assumptions regarding key parameters although drug costs emerge as the most important source of variation in the sensitivity analyses.

A number of limitations in our study should be acknowledged. Cost estimates are inevitably imprecise, being based on reports from a small number of trusts or on reported costs for HRGs. In the absence of sound bottom up costings of the consequences of HCV disease, this is difficult to remedy although the sensitivity analysis demonstrates that the findings of our analysis are insensitive to 50% upward and downward variation in cost assumptions. HRGs were used to estimate inpatient costs and there are likely to be variations in what has been included in these costs. For example, inclusion of terlipressin or octreotide in the management of varices in HRG costs (which we believe is unlikely) would bias our results against therapy for HCV as these elements are costed separately. Our survey of hepatologists revealed a move towards nurse led clinics for the management of patients on combination therapy. This will reduce the outpatient management costs of treatment although such changes are very unlikely to be captured in standard NHS accounting practices, further emphasising the importance of better costing studies.

We have assumed that viral clearance is lifelong for almost all patients and that life expectancy is nearly normal. However, with reduction in HCV spread through blood transfusions, the proportion of the seroprevalent population made up of intravenous drug users is likely to increase. Reinfection of those treated because of recurrent injection drug use may be a significant problem, and resumed drug use would also reduce their life expectancy. These factors suggest that our results may be optimistic for this population although it is impossible to predict to what extent.

The transplantation rates we have imputed are lower than those calculated for the USA and possibly lower than current UK practice. It is likely that improvements in transplant technology and organ availability with consequent reduction in the clinical threshold for transplantation could increase the value of avoiding this costly procedure through early treatment.

Our results are necessarily based on the results of clinical trials carried out to date. But exclusion criteria imply that these included only a small proportion of those who might now be considered for combination therapy in the UK. Our survey of hepatologists revealed considerable variation in the application of criteria for combination treatment, reflecting the genuine uncertainties of clinicians. A better understanding of this variation will be important to inform further pragmatic research into the value of combination therapy and to ensure equity of access to this potentially valuable treatment. While such variation and uncertainty exists, we consider it important that treatment should be confined to specialist centres, as recommended by the British Society of Gastroenterology and the European Society for the Study of the Liver.