The clinical effectiveness and cost-effectiveness of peginterferon alfa and ribavirin for the treatment of chronic hepatitis C in children and young people: a systematic review and economic evaluation

Aetiology

Hepatitis C virus is acquired primarily through exposure to contaminated blood. In adults in the UK, the most common source of infection is through the sharing of injecting equipment in intravenous drug misuse. This accounts for around 90% of cases. 3 Other sources of HCV infection include needle stick injury, tattooing and body piercing, and treatment with contaminated blood products (prior to the introduction of blood screening in the mid-1980s). The risk of sexual transmission is thought to be low. 3

In children, mother-to-child (‘vertical’) transmission is the primary reason for HCV infection, with perinatal transmission being the most important route, and to a lesser extent, intrauterine transmission. 4,5 The rate of perinatal transmission from a HCV-infected mother to her child ranges from 2% to 5%. 6,7 Breast feeding does not appear to increase the risk of HCV transmission, even though HCV RNA may be detected in breast milk and colostrum. A number of factors may change the risk of mother-to-child transmission. There is an increased risk of transmission depending on the level of maternal viral load and whether or not the mother is also co-infected with human immunodeficiency virus (HIV). 8 A systematic review of 77 studies published in 2001 showed that the rate of mother-to-child transmission was in the region of 5% from women without HIV infection and 22.1% from women with HIV infection. 9

Epidemiology

Estimates based on laboratory surveillance by the Health Protection Agency (HPA)10 in the UK suggest that around 216,000 individuals were chronically infected with HCV in 2011. The prevalence of HCV in children of all ages is unclear and difficult to establish. The HPA report estimated that 26 children aged 1 year or below, 21 young people aged 1–14 years, and 439 people between the ages of 15 and 24 years were newly diagnosed with HCV in England in 2010. Among those aged 15–24 years, many cases of HCV will be acquired through injecting drug use, which often begins in late adolescence and early adulthood. 10

Published population-based studies range in their estimates, in part owing to many studies having small, and in some cases, unrepresentative samples (e.g. antenatal screening can be selective), and thus vertical transmission may be undetected in some. Estimates generally suggest that the prevalence of HCV in children in developed countries is between 0.1% and 0.4%. 4,7,11 In some populations this may exceed 10% (e.g. in some regions of Saudi Arabia and Africa). 7 Estimates of regional prevalence rates in pregnant women in the UK range from 0.19% to 0.43%. 4 The prevalence of HCV genotypes in children is thought to be similar to that in adults given that the majority are infected by vertical transmission. Studies have shown that genotypes 1, 2 and 3 are the most clinically relevant groups in children with HCV, while genotype 4 is less prevalent. 7

Progression and prognosis

The natural history of HCV acquired during childhood is not completely understood, although the age at onset of HCV infection is thought to be an important factor in the long-term outcome. In children, spontaneous viral clearance tends to occur early in the history of an infection and is more likely before the age of 4 years. 12 Once established, chronic HCV infection tends to persist into adult life, although the associated liver disease may be asymptomatic. 12 In vertical transmission, estimates suggest that somewhere between 2.4% and 55% of children will spontaneously clear the infection, with the cumulative probability of progression to chronic HCV being approximately 80%. 4,7,12 Caution is required in the interpretation of these data, however, as most of the studies that these estimates come from have small numbers of children, with different ages at acquisition of HCV and different comorbidities. 4 Spontaneous viral clearance is thought to be dependent on genotype, with children infected with genotype 3 having a higher likelihood of clearance than those with genotype 1. 7,12

Chronic HCV is a slowly progressing disease that usually develops over a number of years, often decades. The severity of chronic HCV relates to the duration of infection, meaning that progression to advanced disease is less likely in children than in adults. 11 A recent systematic review6 evaluated the outcomes of untreated HCV in children from population-based screening studies. Results from 25 studies including 733 people infected with HCV as children showed that of the 180 (25%) who underwent a liver biopsy as adults, only 4% had developed liver cirrhosis, with no other individuals developing any severe adverse outcomes. The authors concluded that the majority of people with disease acquired during childhood have a mild degree of hepatitis and fibrosis during childhood. No clear risk factors for severe adverse outcomes were identified in the studies reviewed. The review conclusions were limited by the relatively short follow-up periods in most of the studies included. 6 Other studies suggest that the rate of advanced liver fibrosis or cirrhosis seen on liver biopsy in children with chronic HCV infection is also relatively low, in the range of 2–6%. 7,11–15 According to clinical experts, no children have undergone liver transplantation because of chronic HCV infection in the UK. Despite the relatively innocuous nature of chronic HCV in children, clinicians believe that treatment during childhood is beneficial, as a definitive resolution of disease may be achieved in a subgroup of patients, treatment may reduce children’s social burden and factors may be more favourable for a response (e.g. a low viral load, less advanced disease). 7,16 In addition, clinical opinion suggests that children experience fewer side effects of treatment than adults.

Some differences in outcomes between vertically infected and parenterally infected children have been found. For example, in young children vertical transmission may be associated with higher levels of alanine aminotransferase (ALT), an enzyme that may be elevated in concentration if damage to liver cells has occurred. Overall, however, the mode of infection appears to have a relatively limited impact on outcomes, which reflect a slowly progressing disease. 17,18

Diagnosis and staging

The need for diagnostic testing in children is established by assessment of potential risk factors, such as HCV infection or drug use in the mother, or exposure to contaminated blood products or organ transplants. Diagnosis is undertaken using blood tests to detect HCV antibodies and HCV RNA. 4 Identification of HCV antibodies uses enzyme-linked immunosorbent assays or enhanced chemiluminescence tests, where test accuracy indices have been shown to be excellent. 19 In cases of suspected vertical transmission, this testing procedure should ideally be undertaken after the child is older than 18 months, because maternal antibodies can cross the placenta and persist for up to 18 months, leading to potentially unreliable test results. 16

A positive antibody test will be followed up with a test for the presence of HCV RNA in serum in order to determine active infection. 16,19 This is typically undertaken using a real-time polymerase chain reaction (PCR)-based test as these yield both sensitive and quantitative detection ranges. Recent clinical guidelines suggest that if undertaken in early infancy, a positive serum HCV RNA should be rechecked after 12 months of age to establish the presence of chronic HCV. 16 At this point, the determination of positive HCV RNA may indicate acute or chronic infection, and the clinician will use patient history of the timing between the test and the likely exposure to aid diagnosis. If a test for HCV antibodies is positive but a test for HCV RNA is negative, this could indicate a resolved infection, and testing would be repeated after 6 months for confirmation. 4

If chronic HCV infection is established, testing may be undertaken to establish the HCV genotype using a further PCR assay. In adults, evidence has shown that HCV infections with viruses of genotypes 2 or 3 are the most likely to resolve with therapy, whereas infections with viruses of genotypes 1a or 1b are less likely to respond. The determination of the genotype is therefore an important and useful means to establish treatment options including the timing and duration of treatments, and once this is known it may be followed up with a liver biopsy to determine the extent of any liver fibrosis. 16,19 In children infected with the more responsive genotypes 2 or 3, treatments may commence without the need to test for the extent of liver fibrosis because the benefits of treatment are likely to outweigh the risks. In those with genotypes 1a and 1b, however, the extent of liver fibrosis will be used to weigh up the benefits and risks of treating immediately versus waiting. 16 For some children who have been vertically infected the biopsy may be delayed until age 8–10 years, as evidence of the natural history shows that fibrosis is unlikely to occur until at least this age. 19

In children who have undergone a liver biopsy, chronic HCV infection may be classified as mild, moderate or severe based on histological appearance. To determine the severity, two components of the liver biopsy sample are assessed: fibrosis (scarring) and necroinflammation. 20 The extent of fibrosis is expressed as a ‘stage’ ranging from no fibrosis to cirrhosis in its severe form. Cirrhosis can progress from a compensated state, where the liver is still functioning despite the fibrosis, to a decompensated state where the functioning of the liver is seriously impaired. The extent of necroinflammation of the liver is expressed as a ‘grade’ of disease activity which relates to the rate at which the disease stage is changing. A weakness of the histological classification is that it does not differentiate the clinical process of decompensation (compensated or decompensated liver function can occur at the same stage of fibrosis or cirrhosis), so the fibrosis stage score may not necessarily increase as decompensation occurs. Inflammatory activity in the liver can increase or decrease, or remain constant, during the disease process. 20

There are a number of commonly used systems for classifying liver biopsy samples. The three most commonly used are the Knodell histological activity index (HAI), the Ishak revised HAI and the METAVIR system. Knodell and colleagues’ system21 comprises four components; one classifies the amount of fibrosis [scored from 0 (no fibrosis) to 4 (cirrhosis)], and three the extent of necroinflammation (periportal and/or bridging necrosis; intralobular degeneration; and portal inflammation, with a combined maximum score of 18). The maximum combined score is therefore 22, where higher scores reflect more severe disease.

A revision of the HAI, primarily for use as a research tool, was published in 1995 by Ishak and colleagues. 22 The revised system applies five components. Four of these measure components of necroinflammation grading – periportal or periseptal interface hepatitis, confluent necrosis, focal (spotty) lytic necrosis, apoptosis and focal inflammation, and portal inflammation – with a maximum score of 18. The fifth relates to fibrosis staging with a maximum score of 6. The maximum score is therefore 24.

The METAVIR system was developed specifically for use in HCV and again scores the fibrosis stage and the necroinflammation grade. 23 Fibrosis is scored from 0 to 4, from no fibrosis to cirrhosis or advanced scarring. Necroinflammation is scored on a scale of 0 (no histological activity) to 3 (severe activity). The maximum score is therefore 7. The METAVIR system is the most widely used system in clinical trials of antiviral treatment in chronic HCV and is considered to be the most validated instrument currently available. 20

Impact of disease

Many children infected with HCV appear to be clinically asymptomatic or show only mild, non-specific symptoms (e.g. fatigue, flu-like symptoms, nausea). 6,7,16 As mentioned above (see Progression and prognosis), a small proportion of patients with chronic HCV will develop significant liver disease during their childhood. A retrospective study of 246 patients on the UK HCV National Register Database13 found that when patients who were infected with HCV before the age of 16 years reached their late teens, some had started to show signs and symptoms of liver disease, including enlarged liver (43 patients), enlarged spleen (20 patients), visible blood vessel abnormalities (spider nevi) (four patients), abdominal fluid retention (ascites) (three patients), jaundice (three patients), bleeding oesophageal varicose veins (varices) (one patient) and itching (one patient). Many of the patients on the database had comorbidities and, overall, those who developed signs and symptoms of liver disease were found to be statistically significantly more likely to have had underlying medical conditions in addition to HCV infection. 13

Another study, based on medical records submitted by 12 paediatric and infectious diseases centres in Italy, investigated outcomes for 504 children who were infected with HCV before the age of 16 years but who did not have comorbid viral, autoimmune, metabolic or haematological disorders or malignancy. 12 The majority of children had non-specific, transient and mild symptoms at the time of diagnosis. However, six (1.8%) went on to develop signs and symptoms of advanced liver disease, including weakness (asthenia), nosebleed, itching, ascites and gastrointestinal bleeding, with a mean duration of HCV exposure from putative time of exposure to diagnosis of cirrhosis of 9.9 years. 12 However, these data should be interpreted with caution, as they are from retrospective studies where the methods of population selection and data capture are unclear. Transplant would be offered for children with end-stage disease with significant complications of cirrhosis, including variceal bleeding and refractory ascites, those with decompensated liver function (coagulopathy and encephalopathy) or those who develop hepatocellular carcinoma. However, these are rare in children with HCV infection without any other comorbidity and, as mentioned above, no children in the UK are thought to have undergone liver transplantation as a result of chronic HCV.

Evidence from adult populations suggests that chronic HCV infection eventually leads to impairment in quality of life (QoL), even in the absence of liver inflammation, with patients feeling unwell in terms of both their physical and mental health. 24,25 However, information on the impact of infection with chronic HCV on children’s QoL is very limited and it is difficult to draw clear conclusions from the available evidence. A small study on 19 HCV-infected children in Australia concluded that physical and psychosocial summary scores from validated self-reported and parent-reported questionnaires were significantly lower in infected than in non-infected children. Children reported reduced physical functioning but were otherwise less concerned than their parents about future health. 26 Another study on 114 treatment-naive HCV-infected children used validated questionnaires to elucidate the behavioural, emotional and cognitive functioning of the children and their caregivers. 27 Children with HCV had significantly lower cognitive functioning scores than a normative sample, and some caregivers were found to be highly distressed by their children’s medical circumstances, which limited family activities. However, the authors concluded that overall QoL was not impaired in children with chronic HCV infection. 27,28

In adults, chronic infection with HCV is recognised as a social stigma29 and it has been suggested that children chronically infected with HCV, and their families, experience the burden of social stigma,8 although to date this does not appear to have been analysed quantitatively. According to clinical experts consulted during this technology assessment, parents often carry immense guilt, especially mothers, if they have transmitted a HCV infection, and disclosing the diagnosis to their child can also be a huge burden. Children with HCV may experience stigma as a result of carrying an infection that they may later transmit; inappropriate segregation that can arise because of ignorance; and having a virus that may be perceived as related to negative social factors such as drug use and HIV.

Childhood infection with HCV has been estimated to increase the risk of liver-related death 26-fold. 30 There is a significant economic impact of paediatric HCV infection; projected 10-year costs associated with paediatric HCV infection in the USA (arising from the costs of screening, monitoring and treatment) have been estimated at $199–335M. 31 We are not aware of any cost data for the UK.

Population

Children and young people aged 3–17 years (peginterferon alfa-2b) or 5–17 years (peginterferon alfa-2a) with chronic HCV, without liver decompensation and who were positive for HCV RNA. All groups were considered, including:

• people with HIV co-infection

• people with all grades of severity of chronic hepatitis C (mild, moderate and severe)

• people who were treatment naive or, if appropriate, those who had been previously treated but who relapsed or did not respond.

Interventions

• Peginterferon alfa-2a in combination with RBV.

• Peginterferon alfa-2b in combination with RBV.

Comparators

• Best supportive care (e.g. symptomatic treatment, monitoring, treatment without any form of interferon therapy).

• The interventions compared with each other within their licensed indications, i.e. peginterferon alfa-2a and RBV versus peginterferon alfa-2b and RBV.

Outcomes

Studies had to report SVR (defined as undetectable HCV RNA at least 6 months after treatment cessation). Studies could also include one or more of the following:

• virological response to treatment (e.g. during treatment, end of treatment)

• biochemical response (e.g. ALT)

• liver inflammation and fibrosis

• mortality

• adverse effects of treatment, including effects on growth

• HRQoL.

Study design

• Randomised controlled trials were included if available. If no RCTs of relevance were identified, non-RCTs were considered for inclusion. Studies without a control group were only considered for inclusion in the absence of any controlled studies.

• Studies published since 2007 as abstracts or conference presentations were only included if sufficient details were presented to allow an appraisal of the methodology and an assessment of results to be undertaken.

• For the systematic review of cost-effectiveness, studies were only included if they reported the results of full economic evaluations [cost–utility analyses, cost-effectiveness analyses (reporting cost per life-year gained), cost–benefit analyses or cost–consequence analyses].

• Systematic reviews were only used as a source of references.

• Case series, case studies, narrative reviews, editorials and opinions were not included.

• Only studies published in the English language were included.

Overview of the included studies

The key characteristics of the included studies are shown in Table 2. Two studies evaluated peginterferon alfa-2a and RBV,56,57 and five studies evaluated peginterferon alfa-2b and RBV. 46–48,51,59 The dose of peginterferon was administered subcutaneously once per week in all the studies, and was largely similar within peginterferon type. Peginterferon alfa-2a, given according to BSA, was similar in the two studies [180 µg/1.73m²/week56 (= 104 µg/m²/week) and 100 µg/m²/week57], both with a maximum dosage of 180 µg. The peginterferon alfa-2b dosage, given according to body weight, was 1.5 µg/kg/week in three studies,46,47,51 and Wirth and colleagues59 reported that the dose of 60 µg/m²/week used in their study was equivalent to the licensed dose for adults of 1.5 µg/kg/week. The study by Jara and colleagues48 used a lower dose of 1.0 µg/kg/week. RBV was administered orally at a dose of 15 mg/kg/day in all the studies, with the two peginterferon alfa-2a studies stating a maximum dosage of 1200 mg,57 or 1200 mg for body weight ≥ 75 kg and 1000 mg for body weight < 75 kg. 56 RBV is usually administered in two divided doses although this was explicitly stated in only three studies (one peginterferon alfa-2a,56 two peginterferon alfa-2b47,48).

The duration of treatment was 48 weeks46,56 or 52 weeks47 in three studies (one peginterferon alfa-2a,56 two peginterferon alfa-2b46,47), whereas two studies (one peginterferon alfa-2a,57 one peginterferon alfa-2b48) treated participants for different durations according to genotype, which was generally 24 weeks for genotype 2 or 3 and 48 weeks for genotypes 1, 4, 5 or 6. The information provided by Pawlowska and colleagues51 on treatment duration was not clear. They reported a duration of 48 weeks for all participants while also reporting that participants received 24 or 48 weeks of treatment according to genotype (2 and 3 or 1 and 4, respectively). Wirth and colleagues59 also treated participants for different durations according to genotype but further divided those with genotype 3 according to baseline viral load, so that those with genotype 2 and those with genotype 3 and a low viral load received 24 weeks of therapy, and those with genotype 1 or 4 and genotype 3 with a high viral load received 48 weeks of therapy.

All of the included studies were relatively small. The trial by Wirth and colleagues (peginterferon alfa-2b)59 was the largest, recruiting 107 participants. The two peginterferon alfa-2a studies56,57 and one peginterferon alfa-2b study51 were similar in size, with 53–65 participants (although Schwarz and colleagues56 had n = 55 for the peginterferon and RBV arm, with a total study size of n = 114). The numbers of participants in the three smaller studies ranged from 7 to 30. 46–48

Five of the studies (both peginterferon alfa-2a,56,57 three peginterferon alfa-2b48,51,59) included participants with a mix of genotypes, although all included a higher proportion of participants with genotype 1 (range 50–87%), or genotypes 1 or 4 (range 71–96%) than the other genotype subgroups. Participants with genotypes 2 or 3 accounted for only 3–25% of the included populations across the studies. The remaining two studies, both evaluating peginterferon alfa-2b, included only participants with genotype 4 (Al Ali and colleagues46), or largely genotype 4 where six of seven participants had genotype 4 and one was unknown (not tested) (Ghaffar and colleagues47). Over half of the studies included treatment-naive populations, with four (both peginterferon alfa-2a,56,57 two peginterferon alfa-2b46,59) having 100% of participants not previously treated, and a fifth study (of peginterferon alfa-2b48) consisting largely of treatment-naive participants (80%). Ghaffar and colleagues47 stipulated that one of seven children was previously treated whereas the treatment history of the other six children was not reported (so it is unclear whether they were treatment naive or this was unknown). The study by Pawlowska and colleagues51 (peginterferon alfa-2b) was conducted in a mixed population of treatment-naive and previously treated (with non-pegylated interferon alfa-2b and RBV) participants in roughly equal proportions (54% and 46%, respectively).

The age ranges of children included in the peginterferon alfa-2a studies (5–18 years) and peginterferon alfa-2b studies (3–17 years) are within those of the anticipated licence for peginterferon alfa-2a and the existing licence for peginterferon alfa-2b. Three of the peginterferon alfa-2b studies focused on a narrower age range, 8–17 years, which excluded young children. 46,47,51 The mean age was approximately 10–11 years in four studies (both peginterferon alfa-2a,56,57 two peginterferon alfa-2b48,59) and older (14–16 years) in two peginterferon alfa-2b studies;46,51 one peginterferon alfa-2b study47 did not report mean age but the median was 10 years. The proportion of male participants was approximately half to two-thirds of the total population in all studies except one (peginterferon alfa-2b48) (not reported). Vertical transmission was the most common mode of infection in four studies (both peginterferon alfa-2a,56,57 two peginterferon alfa-2b48,59), accounting for nearly half the included population in one study57 and around 70% in the other three. 48,56,59 Parenteral transmission was the most common route in the other three peginterferon alfa-2b studies,46,47,51 ranging from 42% to 100%, and included infection via intravenous drug use, transfusion and other medical procedures. In four studies (both peginterferon alfa-2a,56,57 two peginterferon alfa-2b46,59) the mode of infection was unknown in 14–22% of participants.

Baseline HCV RNA levels across the seven included studies varied. In the two peginterferon alfa-2a studies, approximately two-thirds of participants had relatively high baseline HCV RNA levels [> 500,000 international units (IU)/ml57 or ≥ 600,000 IU/ml56], and one study of peginterferon alfa-2b46 also reported high baseline HCV RNA, with a mean of 780,000 IU/ml. Two other peginterferon alfa-2b studies51,59 reported similar proportions (range 40–55%) of participants with either high (> 500,000 or > 600,000 IU/ml) or low (< 500,000 or < 600,000 IU/ml) HCV RNA levels. In the study by Ghaffar and colleagues47 (peginterferon alfa-2b), most of the participants had low HCV RNA levels at baseline (median 145,000 IU/ml). The seventh study (peginterferon alfa-2b48) reported a mean HCV RNA of 5 log₁₀ IU/ml, stating that 67% (20/30) of participants had a viral load of > 10⁵ × IU/ml with only one patient having log₁₀ viral load < 4.5. Fibrosis levels indicated mild liver disease in most or all of the population across the seven studies, although fibrosis was not reported for all the participants in the study by Ghaffar and colleagues. 47 According to clinical opinion, there is generally no single agreed definition of what constitutes a ‘high’ or ‘low’ viral load and the cut-off value is different between the two peginterferons.

Studies differed in the numbers of centres and countries that they included. Four (all peginterferon alfa-2b) were single-centre studies (Egypt,47 Kuwait,46 Poland51 and Spain48) while three (both peginterferon alfa-2a,56,57 one peginterferon alfa-2b59) were multicentre studies (ranging from 6 to 22 centres). Schwarz and colleagues recruited patients from 11 centres, all located in the USA,56 and the remaining two multicentre studies involved multiple centres in different countries. 57,59 Sokal and colleagues57 included six centres located in Brazil, the UK (Birmingham Children’s Hospital), Belgium, Latvia and Sweden, and Wirth and colleagues59 included 22 centres located in the USA, South America and Europe. For two studies (one peginterferon alfa-2a, one peginterferon alfa-2b), funding was either received from the drug companies involved,57 or the majority of the authors had received funding from, or were employed by, the drug manufacturer. 59 One study (peginterferon alfa-2b) did not state the funding source but did report that the drug manufacturer provided the interventions and ‘assistance for designing the study’,48 while a fourth (peginterferon alfa-2a) reported unspecified ‘additional support’ from the drug manufacturer as well as stating other sources. 56 The three remaining peginterferon alfa-2b studies either received no financial support46 or reported vaguely that their funding was from ‘donations’47 or ‘other departmental sources’. 51

All seven studies specified the patients’ age as an inclusion criterion and this ranged from 3 to 18 years (although the maximum age of included participants was 17 years). All studies required patients to have chronic HCV infection46–48,51,56,59 and/or detectable plasma HCV RNA,46–48,51,56,57 although only one study (peginterferon alfa-2b) specified a detection threshold of HCV RNA for inclusion (> 50 IU/ml48). Inflammation and/or fibrosis from liver biopsy or ultrasound investigations was specified as supporting evidence of liver disease in four studies (one peginterferon alfa-2a,56 three peginterferon alfa-2b46,51,59) and two studies (one peginterferon alfa-2a,56 one peginterferon alfa-2b47) explicitly stated that the liver disease should be compensated. Four studies specified either that treatment-naive patients were included (one peginterferon alfa-2a, two peginterferon alfa-2b)46,57,59 or that patients previously treated with interferon or RBV were excluded,56 whereas one study (peginterferon alfa-2b) permitted non-responders to interferon alfa monotherapy provided that they accounted for < 25% of the population. 48

Six studies excluded participants who were co-infected with HIV or hepatitis B (not reported in one peginterferon alfa-2b study47), and two studies47,48 also excluded those who were co-infected with any other non-HCV liver disease. Five studies (one peginterferon alfa-2a, four peginterferon alfa-2b) excluded those with thrombocytopaenia, anaemia and neutropenia – conditions that are consistent with decompensated liver disease or are made worse by taking peginterferon + RBV – by stipulating certain laboratory readings46,48,56,59 or specifying ‘normal levels’47 in their inclusion/exclusion criteria. Details of other inclusion/exclusion criteria specified by the studies can be found in the data extractions forms in Appendix 3.

Six of the seven studies specified that SVR was the primary outcome. 46,48,51,56,57,59 SVR may also have been the primary outcome in the remaining peginterferon alfa-2b study (Ghaffar and colleagues47), although this was not stated explicitly. In terms of secondary outcomes, EVR was reported by all seven studies46–48,51,56,57,59 (though not specifically stated as secondary by Ghaffar and colleagues47), and five studies (both peginterferon alfa-2a,56,57 three peginterferon alfa-2b46,47,51) reported an end-of-treatment virological response (EOT, abbreviated in some studies as ETR; the former is used hereafter for consistency). Other secondary virological outcomes reported were rapid virological response (RVR),48,56,59 predictors of viral response,48,51,56,57,59 virological response at 24 weeks,48 virological breakthrough51 and relapse. 51,56,59 Biochemical response was reported by two peginterferon alfa-2b studies. 47,59 Adverse events were reported by all seven studies. Four studies (one peginterferon alfa-2a57 and three peginterferon alfa-2b48,51,59) reported growth and only one study (peginterferon alfa-2a) reported QoL. 56

A summary of the included studies in terms of patient population, in line with the NICE scope, is given below:

• Peginterferon alfa-2a/-2b Peginterferon alfa-2a, n = 2; peginterferon alfa-2b, n = 5.

• Treatment naive/previously treated Treatment naive (100% of population), n = 4; mixed treatment (naive and previously treated), n = 2; unclear, n = 1.

• Severity of chronic HCV Mild fibrosis (most or all of the population), n = 6; unclear, n = 1.

• HIV co-infection n = 0.

Quality assessment of included studies

The methodological quality of reporting in the included studies was assessed using criteria based on guidance by the CRD at the University of York,43 and is shown in Table 3. The quality assessment criteria relate to various aspects of study design which may help to gauge the relative strengths and weaknesses of the individual studies. On the whole, the cohort studies were of generally poor quality, although the study by Schwarz and colleagues56 (peginterferon alfa-2a) fared better in its reporting of methodological details. This was a RCT, although as detailed previously, it is treated as a single-cohort study in this assessment.

All the studies specified their criteria for patient selection a priori, stating their inclusion and exclusion criteria to varying degrees of detail (although Ghaffar and colleagues47 did not specify any exclusion criteria). The lack of randomisation procedures (resulting from the single-arm study design of most studies) may mean there is a higher risk of bias. Given that six of the seven studies were uncontrolled cohort studies with only one intervention arm, the blinding of participants was not applicable. The seventh study, by Schwarz and colleagues,56 reported that participants (in both arms) and investigators were blinded as to whether they were receiving placebo or RBV in combination with peginterferon alfa-2a, with placebo/RBV tablets being supplied in the same dosing regimen.

For most of the studies it was unclear whether or not the authors measured more outcomes than they reported. Schwarz and colleagues56 (peginterferon alfa-2a) was the only study to clearly describe measuring more outcomes than were reported (in either the main paper56 or related publications28,49,50,53–55), stating that assessments of body composition and growth would be reported separately. (A recent publication62 reporting these outcomes did not meet the inclusion criteria for the review as combined results were reported for the peginterferon alfa-2a combination and monotherapy groups together). Pawlowska and colleagues51 reported that there were plans to assess growth 5 years after treatment cessation but there were no further details. Most of the studies provided adequate details of participant withdrawals and losses to follow-up, with the exception of two peginterferon alfa-2b studies47,51 where this information was not reported. However, four studies (one peginterferon alfa-2a,57 three peginterferon alfa-2b46,47,59) reported very little or no methodology relating to data analysis, and all seven studies were either not clear or did not report whether or not the statistical analysis accounted for any missing data.

Assessment of the generalisability of the studies is difficult owing to the single-cohort study design, poor methodological quality, and variation in the participant inclusion criteria and countries included, as well as other uncertainties.

Sustained virological response

Sustained virological response was reported to be the primary outcome in six of the included studies, but not specifically stated as such in the seventh (Ghaffar and colleagues47). Results are reported in Table 4.

Sustained virological response was defined as undetectable serum HCV RNA 24 weeks after the end of treatment in six studies (both peginterferon alfa-2a studies56,57 and four peginterferon alfa-2b studies46,48,51,59) and 12 months after the end of treatment in one study (Ghaffar and colleagues47). Three studies specifically defined the lower limit of detection for attainment of SVR as 50 IU/ml,57 < 50 IU/ml46 or < 10 IU/ml56 (although the last is reported to be < 100 IU/ml in two related publications54,55). Quantitative and qualitative lower limits of detection of HCV RNA are reported by most of the other studies,48,51,59 ranging from 25 to 600 IU/ml, but it is not always clear which virological outcome they relate to (i.e. EVR, EOT or SVR). Details for individual studies can be seen in the data extraction forms in Appendix 3.

Sustained virological response according to prognostic factors

It should be noted that there were some differences between studies in the SVR subgroups in terms of how different categories were defined (e.g. low/high viral load, abnormal ALT), and there were also inconsistencies in grouping different categories (e.g. genotypes, histology). These differences should be borne in mind when interpreting the results. Furthermore, numbers in some of the SVR subgroups were very small and are unlikely to be statistically powered, so results should be interpreted with caution.

Virological response during treatment

All seven included studies reported virological response at various time points during treatment, including RVR, EVR and EOT. RVR is defined as a viral load that does not exceed a specified (although not standardised) limit after 4 weeks of therapy. EVR can be defined as complete EVR, which means HCV RNA is undetectable after 12 weeks, or partial EVR, which means virus is still detectable but there has been at least a 2 log₁₀ drop compared with the baseline value.

Rapid virological response was reported by two studies (one peginterferon alfa-2a56 and one peginterferon alfa-2b48), although in the PEDS-C study56 it was defined as a lack of detectable HCV RNA at week 5 of treatment (rather than at week 4). Jara and colleagues48 reported the proportion of children with negative HCV RNA at week 4 of treatment, which we infer to be RVR, although this was not explicitly defined as such by the authors. All seven studies reported EVR. Three peginterferon alfa-2b studies did not specifically define EVR but reported undetectable47,59 or negative48 HCV RNA at week 12 of treatment, which we infer to be EVR. A fourth peginterferon alfa-2b study46 defined EVR as a HCV RNA level < 50 IU/ml at week 12 compared with baseline, while the two peginterferon alfa-2a studies defined EVR as a decrease of ≥ 2 log₁₀ at week 12 compared with baseline. 56,57 Pawlowska and colleagues51 described EVR as ‘levels of HCV RNA viral load at week 12 of treatment’ but did not specify if levels had to be undetectable or reach a lower limit (they did, however, define these for the subcategories of complete and partial EVR – see the data extraction form in Appendix 3). Six of the included studies reported EOT, defined as undetectable HCV RNA at the end of treatment (48 weeks,56 52 weeks,47 or 24 or 48 weeks57)51 or HCV RNA < 50 IU/ml at week 48. 46 The sixth study (Wirth and colleagues59) reported EOT but did not provide a definition. Results can be seen in Table 10.

Non-response and relapse

Five studies (both the peginterferon alfa-2a studies56,57 and three peginterferon alfa-2b studies46,48,51) reported the proportion of participants who did not respond to treatment, although a specific definition of non-response was not given by any of the studies. Five studies (one peginterferon alfa-2a study56 and four peginterferon alfa-2b studies46,48,51,59) reported data for participants who relapsed. Relapse was defined by three studies as the reappearance of HCV RNA (detectable HCV RNA at week 72,51 at last follow-up59 or after stopping therapy56) after previously having undetectable HCV RNA at the end of treatment. Two of the peginterferon alfa-2b studies reported data but did not specifically define relapse. 46,48 Results can be seen in Table 11.

Biochemical response

Histological response

Quality of life

Growth

Four studies (one peginterferon alfa-2a57 and three peginterferon alfa-2b48,51,59) reported whether or not their participants’ height and weight changed during treatment. These studies presented results for the overall study population, not separately for the subgroups of participants who received treatment for 24 or 48 weeks according to HCV genotype. Changes in growth were often presented only in a brief narrative in the publication text, without quantitative data, and relate to short-term follow-up.

Adverse events

Summary of clinical effectiveness

• Seven studies (two peginterferon alfa-2a and five peginterferon alfa-2b) were included; six were single-arm, uncontrolled cohort studies and one was a RCT for which only data for a single arm met the inclusion criteria. No studies were identified that compared peginterferon alfa with BSC, nor peginterferon alfa-2a with peginterferon alfa-2b.

• The studies were relatively small (range 7–107 participants) and of generally poor quality, with a potentially high risk of bias (owing to the study design) and little reporting of data/statistical analysis. Therefore, caution is advised in the interpretation of results. The generalisability of the studies to a UK population of children and young people is uncertain.

• Sustained virological response rates ranged from 53% to 66% for peginterferon alfa-2a and 29% to 75% for peginterferon alfa-2b. Excluding two studies with very small participant numbers resulted in a range of 49–65% for peginterferon alfa-2b.

• For both peginterferon alfa-2a and peginterferon alfa-2b, children with genotype 2 or 3 appeared to have higher SVR rates than those with genotype 1 (two peginterferon alfa-2a and three peginterferon alfa-2b studies), and children with low viral load at baseline achieved higher SVR rates than those with high viral load in three studies (two peginterferon alfa-2a and one peginterferon alfa-2b). Where participants were of mixed treatment history (two peginterferon alfa-2b studies), children who were treatment naive were more likely to achieve a SVR than those who had been previously treated. The rate of SVR appeared higher in those with normal than in those with abnormal ALT levels at baseline in the two peginterferon alfa-2a studies, whereas in one peginterferon alfa-2b study SVR rates were very similar irrespective of ALT levels. There did not appear to be any impact of the degree of liver fibrosis on SVR rates in the two peginterferon alfa-2a studies that reported it. It should be noted that numbers of children in some of these subgroups were very small and none of the studies was powered for subgroup analysis, therefore results should be interpreted with caution. In five studies, children with genotype 2 or 3 appeared to have higher SVR rates than those with genotype 1.

• Rates of non-response were variable, ranging from 12% to 25% (two peginterferon alfa-2a studies) and 17% to 51% (three peginterferon alfa-2b studies). A relapse rate of 17% was reported by one peginterferon alfa-2a study and a range of 3–17% across four peginterferon alfa-2b studies.

• No conclusions can be drawn on the effect of treatment on biochemical response (normalisation of ALT levels) (three studies), or histological response (one study), as these were poorly and inconsistently reported.

• In one peginterferon alfa-2a study, a clinically significant decline was reported in physical health (15% of children) and in the QoL depression score (5% of children) 24 weeks after starting treatment, but most children showed no clinical changes in any of the measures of QoL, behaviour, depression or executive function at 24 weeks. For children who completed 48 weeks of treatment, there were no statistical differences from baseline for any of the QoL outcome measures after 1 or 2 years of follow-up.

• For one peginterferon alfa-2a study, there were no statistically significant changes in height nor weight from baseline to follow-up. For peginterferon alfa-2b (three studies), there was either no impact on height and weight, or rates decreased during treatment but recovered at the end of treatment or follow-up. The impact on growth was often presented only in a brief narrative so results are not reliable.

• Although not consistently reported, the most frequently occurring adverse events were largely similar across all the studies and were typical of those associated with peginterferon and RBV. These included flu-like symptoms, headache, myalgia and/or arthralgia, gastrointestinal symptoms, injection site reactions, anaemia, leukopenia and neutropenia. Serious adverse events occurred at relatively low incidence rates of 4–6% in the two peginterferon alfa-2a studies that reported them.

• The incidence of dose discontinuation due to adverse events was relatively low and ranged from 3% to 7% (two peginterferon alfa-2a studies) and 1% to 10% (two peginterferon alfa-2b studies). Dose modifications occurred at a rate of 23–51% in two peginterferon alfa-2a studies, while one small peginterferon alfa-2b study reported no modifications and one other was unclear because of inconsistent reporting. Adverse events leading to dose modification were usually anaemia and neutropenia.

Merck Sharp & Dohme: peginterferon alfa-2b and ribavirin

The MSD MS reported a systematic review of clinical effectiveness evidence that was conducted by an independent academic group. The bibliographic databases and search strategies were specified and the searches appear to be reproducible; the study selection, data extraction and quality assessment steps were reported. The MS included eight studies but only presented study characteristics for five of these. The studies included were five non-RCTs of peginterferon alfa-2b and RBV, as well as one RCT and two non-RCT studies of peginterferon alfa-2a and RBV. Of these eight studies in the MS, six met the inclusion criteria for the Assessment Group (AG) report (four on peginterferon alfa-2b46,48,51,59 and two on peginterferon alfa-2a and RBV56,57). One study of peginterferon alfa-2b and RBV63 included in the MS was excluded from our AG appraisal because the population age range exceeded the upper limit specified in the scope. The other study (peginterferon alfa-2a and RBV64) was excluded from our AG appraisal because of the intervention (participants received non-pegylated interferon before peginterferon). Conversely, the AG report includes a non-RCT study of peginterferon alfa-2b and RBV47 that was not included in the MS.

Sustained virological response rates in the MS are comparable with those seen in the clinical effectiveness section here, with only minor discrepancies noted in other virological outcomes (in two studies48,59). The MS reported briefly on growth inhibition and adverse events and presented results of meta-analyses which pooled data for SVR, EVR, relapse, discontinuation of treatment and selected adverse events. In addition to the five studies of peginterferon alfa-2b and RBV, some of the meta-analyses also included three studies of peginterferon alfa-2a and RBV. There appears to be moderate to substantial heterogeneity in these meta-analyses, according to the reported I²-values, but the MS does not provide guidance on interpretation.

Overall, the MSD MS analysis appears reasonably well conducted but the methods of meta-analysis were not reported and interpretation of the meta-analysis results, in light of the apparent study heterogeneity, is unclear. The MS seems to focus on comparing peginterferon alfa-2b and RBV against peginterferon alfa-2a and RBV, but it is unclear how the RCT and non-RCT evidence was combined in the meta-analysis. The MS concludes that both forms of peginterferon and RBV are clinically effective compared with BSC, with no clear differences indicated between the two forms.

Roche: peginterferon alfa-2a and ribavirin

Roche did not conduct a systematic review of clinical effectiveness evidence (bibliographic databases and search strategies were not specified and insufficient detail was given for the search for evidence to be reproducible). The MS provided results primarily from the PEDS-C trial, plus three other non-RCTs. The processes used for inclusion/exclusion screening, data extraction and quality assessment were not reported, nor were the study details or patient characteristics of the non-RCTs. Two of the four studies, including PEDS-C, are included in the present AG report. 56,57 The remaining two studies do not meet the inclusion criteria for the AG assessment because the population age range exceeded the upper limit specified in the scope,65 or the trial was retrospective with no details of peginterferon dose or treatment duration. 66

The MS reported comparative data for both arms of the PEDS-C trial, even though PEG monotherapy is outside the licence and scope. SVR rates in the MS are comparable with those seen in the clinical effectiveness section of the present report. Virological outcomes during treatment (RVR, EVR, EOT) and QoL were not reported in the MS, while data on body composition and growth were reported only from those studies that did not meet the inclusion criteria for the AG report. Subgroup analyses for HIV co-infected and retreated patients were included in the MS using an extrapolation study, carried out using data from four studies for which details are extremely limited. In addition, the numbers of participants are very small.

Overall, the Roche MS appears uncritical and does not provide an explicit interpretation of the clinical evidence. The MS concludes that the PEDS-C trial demonstrates efficacy of peginterferon alfa-2a and RBV over monotherapy (although this comparison is outside the scope of the appraisal). The MS states that there is no safety concern with regard to adverse events; however, only adverse event data from the PEDS-C trial were considered.

Merck Sharp & Dohme’s submission: cost-effectiveness analysis

Cost-effectiveness results

The MS reports results by age group and genotype, in terms of total costs, life-years and QALYs. Table 19 shows the base-case results for all patients (aged 5–17 years). Patients receiving peginterferon alfa-2a, peginterferon alfa-2b and BSC accrued a total of 19.16, 19.24 and 16.77 discounted QALYs at a cost of £17,798, £17,526 and £22,750, respectively. Both combinations of peginterferon alfa and RBV dominated BSC in all patients (aged 5–17 years) and in age and genotype subgroup analyses. Peginterferon alfa-2b dominated peginterferon alfa-2a in all patients (aged 5–17 years) and in all subgroup analyses, except in patients aged 9–13 years and those with HCV of genotypes 1 and 4.

TABLE 19 - Base-case results from MSD cost-effectiveness analysis

LYG, life-years gained; NA, not applicable; PEG α-2a, peginterferon alfa-2a; PEG α-2b, peginterferon alfa-2b.

Treatment	Vs. supportive care
	Cost (£)	Life-years	QALYs	Incremental costs (£)	Incremental LYG	Incremental QALYs	ICER (£/QALY)
Supportive care	22,750	56.15	16.77	NA	NA	NA	NA
PEG α-2a	17,798	63.84	19.16	–4952	7.69	2.39	Dominates
PEG α-2b	17,526	64.09	19.24	–5224	7.94	2.47	Dominates

The MS conducted deterministic sensitivity analyses (DSAs) around structural assumptions (time horizon, discount rates) and the model parameter values. The DSA results showed that peginterferon alfa-2b dominated BSC in nearly all analyses, except for time horizon and discount rates. The ICERs for peginterferon alfa-2b versus peginterferon alfa-2a were robust to variation in the model parameters, i.e. peginterferon alfa-2b dominated peginterferon alfa-2a for all analyses. The MS probabilistic sensitivity analyses (PSAs) showed that there is a 100% probability that peginterferon alfa-2a and peginterferon alfa-2b in combination with RBV are cost-effective. The cost-effectiveness plane for cost and QALYs for the treatments for the PSA are shown in Figure 4. The cost-effectiveness acceptability curve (CEAC) is shown in Figure 5.

FIGURE 4.

Merck Sharp & Dohme’s cost-effectiveness plane for all patients aged 5–17 years.

FIGURE 5.

Merck Sharp & Dohme’s CEAC for all patients aged 5–17 years.

Roche’s submission to the National Institute for Health and Care Excellence: cost-effectiveness analysis

Cost-effectiveness results

The MS reports results by genotype, in terms of total costs, life-years and QALYs. Table 21 shows the base-case results for children aged 11 years. Treating genotype 1, 4 and 5 patients with peginterferon alfa-2a and RBV improved outcomes by 1.01 QALYs compared with BSC and cost an additional £3971, which gives an ICER of £3915. For genotypes 2 and 3, treatment for 24 weeks improved QALYs by 1.57 compared with BSC and cost £1864 less. For this group, peginterferon alfa-2a dominates no treatment.

TABLE 21 - Base-case results from Roche’s cost-effectiveness analysis

LYG, life-years gained; NA, not applicable; PEG α-2a, peginterferon alfa-2a.

Treatment	Cost (£)	Life-years	QALYs	Incremental costs (£)	Incremental LYG	Incremental QALYs	ICER (£/QALY)
Genotypes 1, 4 and 5
No treatment	8199	18.47	14.20	NA	NA	NA	NA
PEG α-2a	12,170	18.56	15.21	3971	0.09	1.01	3915
Genotypes 2 and 3
No treatment	8199	18.47	14.20	NA	NA	NA	NA
PEG α-2a, 24 weeks	6336	18.61	15.77	–1864	0.14	1.57	Dominates no treatment

The MS performed one-way and two-way DSAs for the likelihood of SVR, time horizon, discounting, baseline cohort characteristics, rate of progression to cirrhosis, probability of SVR with treatment, health state costs, health state utilities and timing of treatment. The cost-effectiveness of peginterferon alfa-2a compared with BSC remains below £13,000 per QALY for all analyses. Model results are most sensitive to time horizon, rate of disease progression, probability of SVR with treatment, liver disease at entry and annual cost of achieving SVR.

In the PSA, for patients with genotypes 2 and 3 there is a 97.2% probability of 24 weeks of combination therapy being cost-effective compared with no treatment, at a willingness-to-pay (WTP) threshold of £20,000 per QALY. In patients with genotypes 1, 4 and 5 there is a 91.6% probability of this being cost-effective at a £20,000 per QALY threshold. The cost-effectiveness planes for cost and QALYs by genotype group are shown in Figure 6. The CEAC is shown in Figure 7.

FIGURE 6.

Roche’s scatterplots for (a) genotypes 1, 4 and 5; and (b) genotypes 2 and 3.

FIGURE 7.

Roche’s CEACs for (a) genotypes 1, 4 and 5; and (b) genotypes 2 and 3. CE, cost-effectiveness.

Critique of the manufacturers’ submissions for Merck Sharp & Dohme and Roche

Merck Sharp & Dohme and Roche used the state transition model applied in previous HTAs of peginterferon alfa treatments in adult populations. 20,41 The model structure was considered to be appropriate and was also used in the SHTAC analysis with minor changes to the classification of the health states ‘mild chronic HCV’ and ‘moderate chronic HCV’ (using the METAVIR fibrosis scale F0–F3 according to more recently published evidence).

In the absence of child-specific data, the MSD and Roche MSs have used adult data relating to transition probabilities, costs and utility values. It should be noted that within a long-term model, a large proportion of the time spent in the model would represent the patient’s life as an adult rather than a child, and that these values would therefore be appropriate for most of the duration of the model. In addition, these data were based upon previous HTAs which used a systematic approach. 20,41

As the majority of the patients started in mild chronic HCV, few of these would have been expected to progress to the more severe health states before they reach adulthood. Both the MSD and Roche MSs have conducted reviews to estimate the transition probabilities between the mild and moderate, and moderate and compensated cirrhosis health states. The probability of transition from mild to moderate HCV was the same for both MSs; for moderate HCV to compensated cirrhosis, transition probabilities varied from 0.0038 (MSD) to 0.021 (Roche). The transition probabilities differ from those used in the SHTAC model (a summary of these values can be seen in Table 22). The choice of transition probabilities used in the SHTAC model is discussed more fully in subsequent sections.

The manufacturers differ in their choice of time horizon. MSD uses a lifetime horizon, whereas Roche uses a time horizon of 30 years and considers that this is long enough to capture important costs and effects arising from treatment. The appropriate approach for modelling treatments for chronic diseases that affect patients’ long-term prognoses, as recommended by NICE,69 is to use the lifetime horizon.

Roche assumes spontaneous SVR for children based upon its review of the literature, whereas MSD does not (as per the previous technology assessments20,41). However, it is noted that the probability of spontaneous SVR is small (< 2%) in the Roche submission and that it is unlikely to materially affect the cost-effectiveness results.

The health state utility values applied in the models of the two submissions can be seen in Table 23 and Appendix 7. It can be seen that the utility values applied for the mild and moderate HCV states, compensated and decompensated cirrhosis, HCC, liver transplant and post liver transplant, and the decrement applied during treatment, are the same in both submissions. These were all the utility values applied in the previous adult chronic HCV models with no adjustment made for the present population of children. 20,41 SVR from mild disease differs slightly between the two submissions (0.82 and 0.83 for MSD and Roche, respectively). Roche does not provide utilities for SVR from moderate disease or compensated cirrhosis, and it is unclear from the submission whether or not the utility weight used for the SVR from mild HCV was also used for these two health states.

Health state costs used in the MSD and Roche submissions are shown in Table 24 and it can be seen that the majority of costs were the same across the two submissions. Further details of the 95% CIs, distributions and parameters for the MSD submission are given in Appendix 7.

Overview

We developed a model to estimate the costs, benefits and cost-effectiveness of treatments for chronic HCV for children and young people. The scope for the appraisal, as issued by NICE, states that the interventions to be considered are:

• peginterferon alfa-2a with RBV

• peginterferon alfa-2b with RBV.

The comparator for these interventions is BSC, defined in the NICE scope as treatment without any form of interferon treatment, and one another. The perspective of the cost-effectiveness analysis is that of the NHS and PSS. The model estimates the lifelong costs and benefits from each treatment. The costs and benefits were discounted at 3.5% per year, as recommended by NICE. 69 The base price for the costs was taken from the most recently available data (2011–12). The intervention effect in terms of probability of SVR was derived from the systematic review of the clinical effectiveness reported in Chapter 4. The outcome of the economic evaluations is reported as the cost per QALY gained.

Model type and rationale for the model structure

The lifetime model of the natural history of chronic HCV aims to convert the principal outcome of interest in the clinical trials, that is, the probability of SVR, to long-term survival outcomes. To estimate the impact of this intermediate effect on final outcomes for patients, we required an appropriate model. No other models for chronic HCV in children were found and therefore the previous models for adults provide the best available peer-reviewed structure. We adapted our previously published models20,41 which were used in NICE guidance TA106 and TA200. 33,34 These models were developed for the progression of chronic HCV in adults. Where necessary, they have been modified to reflect the younger patient group in this analysis (discussed in full in subsequent sections). The model has a time horizon of 70 years, as this was considered long enough to capture all relevant costs and benefits, and a cycle length of 1 year.

The state transition diagram describing the health states within the model and the allowable transitions between these states is shown in Figure 8. For the current model, we have modified the structure to include health states for the fibrosis states (F0–F4), defined according to the METAVIR scoring system, instead of the previous health states of mild HCV, moderate HCV and compensated cirrhosis. This is based on more recent evidence on the progression of HCV by Thein and colleagues. 78 They conducted a systematic review of published prognostic rates to determine stage-specific fibrosis progression rates, based on a total of 111 studies of individuals with chronic HCV infection (n = 33,121). Although many of these studies had retrospective designs, the authors meta-analysed studies using both fixed and random effects, provided a number of sensitivity analyses and adjusted for covariates, and the results were reasonably robust to each of these. We therefore consider that these estimates are the most reliable data available. These data have allowed improvements to be made to the model, although at the present time not all chronic HCV data required for modelling fully complement these additional health states. Figure 8 shows nine health states. For clarity, mortality (an absorbing state) has not been included. In this diagram ellipses indicate health states and arrows indicate allowable transitions between health states.

FIGURE 8.

State transition diagram for SHTAC’s economic model. Shaded areas indicate those health states that lead to increased mortality risk compared with that of the general population.

The figure indicates that patients with chronic HCV (F0–F3) or compensated cirrhosis (F4) may have successful treatment (attain a SVR), remain in their current health state or progress to more severe stages of liver disease. The SVR state is assumed to be a permanent condition, with no spontaneous reactivation of HCV infection, although individuals are not immune to reinfection (this is outside the scope of the analysis). Individuals in the SVR health state are assumed to face the same mortality risks as the general population and face no greater risk of HCC than the general population.

For the utility values and health state costs, the previous health states of mild HCV and moderate HCV were used, where mild HCV relates to F0 and F1, and moderate HCV relates to F2 and F3. Targeted searches, undertaken as part of this assessment, did not identify any other new natural history evidence relating to progression or management of chronic HCV specific to children or young people. Utilities are associated with each health state and for the patient cohort the total number of QALYs is calculated. Patients on treatment with peginterferon alfa have a lower HRQoL than those not on treatment, owing to adverse events. This was assumed to be the same decrement for both treatments (see Health state values/utilities).

Patients with chronic HCV (F0–F4) face the same mortality risk as the general population. However, patients with decompensated liver disease or HCC and those who undergo liver transplantation face higher mortality rates, related to their stage of liver disease, than the general population.

Modelling assumptions

We included most of the assumptions from the previous HTA models. These were:

• that the patient’s stage of disease [mild HCV (F0, F1), moderate HCV (F2, F3) and compensated cirrhosis (F4)] prior to treatment influences their subsequent risk of progressive liver disease, post-treatment surveillance and also HRQoL

• that patients not exhibiting a SVR are expected to face the same risk of disease progression as untreated patients

• that the same SVR applies for patients with mild or moderate HCV, and for those patients with compensated cirrhosis

• that the model did not account for reinfection and onward transmission of HCV

• that the possibility of HCC patients receiving a liver transplant was not considered on account of its rarity

• that discontinuation due to adverse events was not accounted for as it is considered rare

• that costs associated with the management of adverse events were not accounted for as they were unlikely to be substantial

• that there is a reduction in utility while patients are being treated with peginterferon alfa and RBV.

In addition, we included the following assumptions after discussion with our expert advisors:

• The base-case analysis assumed that no patients would have spontaneous SVR.

• It was assumed that treatment would discontinue at 24 weeks if an EVR was not achieved at week 12 for genotype 1 or 4.

• Adult transition probabilities, utility weights and health state costs were applied for paediatric patients owing to lack of data.

• There is no change to parental utility values. Although there is some suggestion that parental QoL may be reduced, the evidence is not sufficient to be applied in the model.

Evaluation of uncertainty

The evaluation of the cost-effectiveness of treatment for chronic HCV in children is based on uncertain information about variables such as the clinical effectiveness, HRQoL and resource use. This uncertainty was evaluated using DSA, PSA and scenario analyses. One-way DSAs were conducted to evaluate the influence of individual parameters, model structure and assumptions on the robustness of the model (see Results of independent economic analysis).

Multiparameter uncertainty in the model was addressed using PSA (see Results of independent economic analysis). 79 In the PSA, probability distributions are assigned to the point estimates used in the base-case analysis. The model is run for 1000 iterations, with a different set of parameter values for each iteration, by sampling parameter values at random from their probability distributions. The uncertainty surrounding the cost-effectiveness of the treatment is represented on a CEAC according to the probability that the intervention will be cost-effective at a particular WTP threshold. Appendix 9 reports the parameters included in the PSA, the form of distribution used for sampling each parameter, and the upper and lower limits assumed for each variable.

Model validation

The SHTAC model was validated by checking the model structure, calculations and data inputs for technical correctness. The structure is similar to that used in previous HTAs20,41 but was redeveloped to include the health states F0–F4. The SHTAC model was checked for internal consistency against the MS economic models by running the SHTAC model with the inputs used in the MS models. The robustness of the model to changes in input values was tested using sensitivity analyses to ensure that any changes to the input values produced changes to the results of the expected direction and magnitude. Finally, the model results were compared with those from the MSs.

Southampton Health Technology Assessments Centre’s data sources

Results of independent economic analysis

This section reports the cost-effectiveness results for a cohort of 11-year-olds with chronic HCV, receiving either peginterferon alfa-2a or -2b (in combination with RBV), compared with BSC and with one another. The results for costs and QALYs are presented for each alternative.

The modelled, undiscounted duration in each health state for BSC, peginterferon alfa-2a and peginterferon alfa-2b are presented in Table 34. The results show increased survival for both treatments for chronic HCV when compared with BSC (47.5 years), and a slight survival advantage in patients receiving peginterferon alfa-2a (57.4 years, compared with 57.1 years in peginterferon alfa-2b).

The longer duration spent in more severe disease states by children in the BSC cohort is reflected in the health state costs for this group, presented in Table 35. Although this group does not incur treatment costs, the total undiscounted costs are substantially higher for this cohort than either of the treatment groups, particularly for the cirrhosis and decompensated cirrhosis health states. The additional long-term costs for the BSC cohort outweigh the treatment costs of peginterferon alfa-2a and -2b. The costs are slightly higher for peginterferon alfa-2b than peginterferon alfa-2a, for both the treatment and health state costs.

The base-case results, including discounted total and incremental costs, life-years, QALYs and the ICER for all genotypes, are reported in Table 36 for patient cohorts that contain all genotypes. Both peginterferon treatments have lower lifetime costs than BSC, with a reduction in costs of between £8874 (peginterferon alfa-2b) and £10,190 (peginterferon alfa-2a) over a patient lifetime. Furthermore, both treatments increase the lifetime life-years (by 1.82–1.88) and QALYS (by 1.66–1.72) compared with BSC. Treatment with peginterferon alfa-2a and -2b dominate BSC, that is, they have lower lifetime costs and are more effective.

Base-case results for patients with genotype 1 or 4 only are presented in Table 37 and the results for genotype 2 or 3 are shown in Table 38. The treatment effect (SVR) for genotype 2 or 3 is better than for genotype 1 or 4, and consequently the results from the model reflect this. There are more additional QALYs accrued in the genotype 2 or 3 subgroup (2.52–2.68 QALYs) than in the genotype 1 or 4 subgroup (1.44–1.47 QALYs), and more reduction in cost in the genotype 2 or 3 subgroup (£17,414–£18,043) than in the genotype 1 or 4 subgroup (£6929–£7967).

The results for genotypes 1 or 4 and 2 or 3 are similar to the base-case results; both peginterferon treatment options had lower lifetime costs and were more effective than BSC, that is, they dominate BSC.

The base-case results for the comparison of peginterferon alfa-2a with peginterferon alfa-2b can be found in Table 39. The results, including discounted total and incremental costs, life-years, QALYs and ICERs, are shown for all genotypes, genotype 1 or 4 and genotype 2 or 3. In the overall population base case, peginterferon alfa-2a has slightly lower lifetime costs (£19,055 compared with £20,371) and is slightly more effective than peginterferon alfa-2b (22.25 QALYs compared with 22.19). This leads to peginterferon alfa-2b being dominated by peginterferon alfa-2a. In the genotype 1 or 4 subpopulation, a similar outcome can be observed, with peginterferon alfa-2b being dominated. However, for those with genotype 2 or 3, peginterferon alfa-2b has lower lifetime costs and is more effective, and therefore peginterferon alfa-2b dominates peginterferon alfa-2a. This apparent difference illustrates how marginal the differences between treatments are. As stated previously, the estimates of effectiveness were very similar based on the included clinical effectiveness studies. These effectiveness estimates predominately drive the differences in costs and outcomes of the two treatments within the model. This is further explored below (see Scenario analyses).

Notwithstanding the limitations of the evidence base, overall the base-case analyses suggest that peginterferon alfa-2a and peginterferon alfa-2b are cost-effective compared with BSC. Peginterferon alfa-2a is cost-effective compared with peginterferon alfa-2b in the overall population and the subgroup with genotype 1 or 4, but is not cost-effective compared with peginterferon alfa-2b in the subgroup with genotype 2 or 3.

Probabilistic sensitivity analysis

In addition to the DSA and scenario analyses, a PSA was undertaken. Parameters were sampled probabilistically from appropriate distributions; these included the proportions of children distributed across genotypes, the transition probabilities, the health state utilities and monitoring, health state and treatment costs. Details of the PSA and the distributions applied are reported in Appendix 9. One thousand simulations were run. A summary of the results from the PSA are shown in Table 46.

TABLE 46 - Southampton Health Technology Assessments Centre’s base-case PSA results

PEG α-2a, peginterferon alfa-2a; PEG α-2b, peginterferon alfa-2b; IQR, interquartile range (shows the range between the 25% and 75% percentiles).

Treatment	Costs, £ (IQR)	QALYs (IQR)	ICER (£/QALY) vs. BSC	ICER (£/QALY) vs. PEG α-2a
BSC	29,689 (26,958–31,979)	20.50 (20.1–21.0)	–	–
PEG α-2a	19,226 (17,679–20,593)	22.22 (22.0–22.5)	Dominates	–
PEG α-2b	20,558 (19,202–21,777)	22.16 (21.9–22.4)	Dominates	Dominated

The PSA results closely reflect the deterministic base-case results; the total costs for both peginterferon alfa-2a and -2b (£19,226 and £20,558, respectively) are less than those incurred by patients receiving BSC (£29,689). Patients in the BSC group also accrue fewer QALYs than their counterparts in both treatment groups: 20.50 compared with 22.22 with peginterferon alfa-2a and 22.16 with peginterferon alfa-2b. Therefore, in the PSA peginterferon alfa-2a and -2b both dominate BSC. Peginterferon alfa-2a has lower lifetime costs and is more effective than peginterferon alfa-2b in this analysis, and dominates this treatment. As stated above (see Results of independent economic analysis), there is uncertainty around the reliability of the relative SVR effect size between peginterferon alfa-2a and -2b.

The scatterplots for cost and health outcomes of the treatment options in the PSA are shown in Figure 9. The CEAC is shown in Figure 10, and indicates that at all WTP thresholds, peginterferon alfa-2a has the highest probability of being cost-effective. At WTP thresholds of £20,000 and £30,000 per QALY, the probability of being cost-effective is 68% and 66% for peginterferon alfa-2a, and 32% and 34% for peginterferon alfa-2b, respectively.

FIGURE 9.

Scatterplot of the costs and health benefits for peginterferon alfa-2a, peginterferon alfa-2b and BSC.

FIGURE 10.

Cost-effectiveness acceptability curve for the PSA results.

Comparison of the cost-effectiveness results from the manufacturers’ submissions and Southampton Health Technology Assessments Centre’s model

The results of the manufacturers’ and SHTAC’s economic analyses are summarised in Table 47. Roche did not present its results for an all-patient cohort; rather, it presented its analyses based on genotype groups. The results seen from the three analyses are similar with respect to the overall cost-effectiveness conclusions; that is, that peginterferon alfa-2a and -2b (both in combination with RBV) are dominant compared with BSC. MSD and SHTAC also present analyses of the comparison of peginterferon alfa-2a with peginterferon alfa-2b as per the NICE scope. These differ, with MSD suggesting that peginterferon alfa-2b dominates peginterferon alfa-2a whereas SHTAC suggests the opposite; however, the difference between the results is marginal.

TABLE 47 - Southampton Health Technology Assessments Centre’s and the manufacturers’ baseline cost-effectiveness results

PEG α-2a, peginterferon alfa-2a; PEG α-2b, peginterferon alfa-2b.

Result	Analysis	BSC	PEG α-2a	PEG α-2b
Total cost (£)	SHTAC	29,245	19,055	20,371
	MSD, all patients	22,750	17,798	17,526
	Roche, genotypes 1 and 4	8199	12,170	–
	Roche, genotypes 2 and 3	8199	6336	–
Total QALYs	SHTAC	20.53	22.25	22.19
	MSD, all patients	16.77	19.16	19.24
	Roche, genotypes 1 and 4	14.20	15.21	–
	Roche, genotypes 2 and 3	14.20	15.77	–
Incremental cost vs. BSC (£)	SHTAC	–	–10,190	–8874
	MSD, all patients	–	–4952	–5224
	Roche, genotypes 1 and 4	–	3971	–
	Roche, genotypes 2 and 3	–	–1864	–
Incremental QALYs vs. BSC	SHTAC	–	1.72	1.66
	MSD, all patients	–	2.39	2.47
	Roche, genotypes 1 and 4	–	1.01	–
	Roche, genotypes 2 and 3	–	1.57	–
ICER vs. BSC (£/QALY)	SHTAC	–	Dominates	Dominates
	MSD, all patients	–	Dominates	Dominates
	Roche, genotypes 1 and 4	–	3915	–
	Roche, genotypes 2 and 3	–	Dominates	–
ICER PEG α-2b vs. PEG α-2a (£/QALY)	SHTAC	–	–	Dominated
	MSD, all patients	–	–	Dominates
	Roche, genotypes 1 and 4	–	–	–
	Roche, genotypes 2 and 3	–	–	–

The total costs vary widely between the analyses; for example, the cost of BSC varies between £8199 (Roche) and £29,245 (SHTAC). Roche uses a much shorter time horizon than the other two analyses, which largely explains these differences. The difference in costs between SHTAC and MSD (for BSC) is due to the length of time patients spend in the HCV health states; this is considerably shorter in the SHTAC analysis than in the MSD analysis, and therefore patients incur higher health-care costs. The incremental costs for peginterferon alfa-2a versus BSC vary between £3971 (Roche genotype 1) and −£10,190 (SHTAC).

There is also a wide range of total QALY estimates between the studies. For BSC, these vary between 14.2 (Roche) and 20.53 (SHTAC). The incremental QALYs vary between 1.01 (Roche genotype 1) and 2.39 (MSD). These differences are largely down to shorter time horizon (Roche) and the lower utility values used in the MSD than in the SHTAC analysis.

Clinical effectiveness

The results of seven studies evaluating peginterferon alfa treatment in children and young people were included in this systematic review. The evidence was limited to uncontrolled single-cohort studies (with one RCT treated as a single-cohort study) which had relatively small populations. Overall, the studies were of low methodological quality and, owing to the nature of the design, few statistical analyses were reported.

The studies varied in their population characteristics according to a number of factors, including HCV genotype mix, mean participant age, treatment history, mode of HCV transmission, baseline HCV RNA levels and different countries involved. Six studies excluded those co-infected with HIV or hepatitis B, three peginterferon alfa-2b studies46,47,51 specifically excluded younger children (< 8 years old), two48,51 (peginterferon alfa-2b) included a mix of treatment-naive and previously treated children (with a third study unclear47) and only one study57 (peginterferon alfa-2a) included UK participants. The design and quality of the studies, as well as other uncertainties, make assessment of the generalisability of the studies difficult and the variation in these factors may have implications for the interpretation of the findings.

The SVR rates in the included studies ranged from 53% to 66% in children treated with peginterferon alfa-2a and 29% to 75% in those treated with peginterferon alfa-2b, although the range was 49–65% if two studies with very small participant numbers46,47 are excluded. These rates are comparable with those seen in adults with chronic HCV (50–60%20). Observed patterns in the SVR subgroups (genotype 2 or 3, treatment naive, low viral load at baseline) also appear to be consistent with what is seen in adults, i.e. higher SVR rates are observed in these subgroups. However, the numbers of children in some of these subgroups were very small and none of the studies was powered for subgroup analysis; therefore, results should be interpreted with caution. Serious adverse events were defined differently in the studies and the numbers of data provided on adverse events varied considerably from one study to the next. However, overall the relatively mild adverse events typically associated with the use of peginterferon and RBV were frequent (e.g. flu-like symptoms were almost universal) and consistent with those in adults.

No studies were identified that directly compared peginterferon alfa-2a with peginterferon alfa-2b, and as any comparison would be by observation of data only, no conclusions can be drawn in the present review. Two recent (2010)89,90 meta-analyses of RCTs comparing peginterferon alfa-2a with peginterferon alfa-2b in adults both concluded that no recommendations could be made of one over the other. Whether or not the two peginterferon alfa forms are equally effective in children remains inconclusive.

Cost-effectiveness

Systematic searches did not identify any evidence that met the criteria for the systematic review of cost-effectiveness studies of peginterferon alfa treatments in children or HRQoL in children with chronic HCV. A small number of studies with some limited relevance were identified (although these did not meet the inclusion criteria) and were summarised for context to the cost-effectiveness analysis.

Two manufacturers submitted evidence for the cost-effectiveness of their respective treatments versus BSC, and one (MSD) also compared the cost-effectiveness of the two treatments. Both manufacturers based their model structure on one previously developed for NICE appraisals of these treatments in adults. There were differences, however, in the time horizon and the sources of the effectiveness data. MSD used a lifetime horizon and Roche a 30-year time horizon. Effectiveness data were sourced from a meta-analysis of single-cohort studies in the MSD model and a weighted average of single-cohort studies in the Roche model. Neither approach was assessed by the present review of clinical effectiveness to be robust, and these estimates should therefore be treated with caution. In the MSD model the base-case results found that both combinations of peginterferon alfa dominated BSC in all age and genotype subgroups. There were small differences in costs and health outcomes between peginterferon alfa-2a and peginterferon alfa-2b. Peginterferon alfa-2b dominated peginterferon alfa-2a for most age and genotype subgroups. In the Roche model the base-case results found that peginterferon alfa-2a is a cost-effective option for the treatment of paediatric HCV compared with BSC. Both submissions were assessed as meeting the majority of quality standards required of an economic evaluation.

An independent Markov model, based upon the previous SHTAC model, was undertaken by the assessment group to assess the cost-effectiveness of the two peginterferon alfa treatments in combination with RBV compared with BSC and one another, as per the NICE scope. A 70-year time horizon was used. Many of the assumptions used previously were employed, with the addition of assumptions that no children would have spontaneous viral clearance, that treatment would discontinue if an EVR was not achieved at week 12, and that adult transition probabilities, utility weights and health state costs could be applied for paediatric patients owing to lack of data. Effectiveness data were taken from the studies identified in the systematic review of clinical effectiveness, where a weighted average approach was used to establish the SVRs and EVRs. Utility data were taken from published sources although, as noted above, these were in adult populations. Results from this model, albeit based upon poor evidence, suggest that peginterferon alfa (-2a or -2b) in combination with RBV is more effective and has lower lifetime costs than BSC. Peginterferon alfa-2a has slightly lower lifetime costs and is more effective than peginterferon alfa-2b in the overall population and the subgroup with genotype 1 or 4, but not in the subgroup with genotype 2 or 3.

Strengths

The current technology assessment addresses a specific knowledge gap concerning the clinical effectiveness and cost-effectiveness of peginterferon alfa-2a and peginterferon alfa-2b in children and adolescents with chronic hepatitis C.

This technology assessment report has been undertaken by an independent evidence synthesis team free of any vested interest. The technology assessment addressed a clear question with a well-defined specific population, intervention, comparator and outcomes. An independent advisory group including clinical experts commented on the research protocol and the final report. The systematic review of clinical effectiveness followed standard principles of evidence synthesis recommended by the CRD to minimise bias. 43 The methods were set out a priori in a peer-reviewed research protocol that defined the research question, inclusion criteria, critical appraisal approach, data extraction process, and the methods of data synthesis to be used. The study selection step was conducted independently by two reviewers and all other steps were conducted by one reviewer and checked independently by a second reviewer, with all decisions recorded and transparently reported.

An economic model has been developed following recognised guidelines,44,45 and systematic searches have been conducted to identify data for the economic model. The searches, critical appraisal of the economic evidence, and development and reporting of the economic model were conducted by one reviewer and checked independently by a second reviewer.

Manufacturers’ submissions of evidence were systematically data-extracted and critically appraised by the review team using a standard health economic evaluation checklist (see Appendix 8) and, where appropriate, information from the submissions was used to inform the economic evaluation.

The model captures variability and uncertainty in effectiveness, costs and clinical practice using one-way DSA to evaluate the influence of individual parameters, model structure and assumptions on the robustness of the model. Multiparameter uncertainty was addressed using PSAs for a range of parameter distributions (see Appendix 9), while scenario analyses were used to explore differences between the treatments and the effect of varying the rate of progression to cirrhosis. Uncertainty around the cost-effectiveness of the interventions for different WTP thresholds is represented transparently in CEACs.

Limitations

Despite the extensive and systematic searches, relatively little clinical effectiveness evidence was found, with only two studies of peginterferon alfa-2a and five studies of peginterferon alfa-2b meeting the inclusion criteria for the systematic review. These were small studies, with only 7–107 participants. Although one study was a RCT, only one of its arms was relevant, meaning that no head-to-head comparisons of peginterferon alfa against BSC, or peginterferon alfa-2a against peginterferon alfa-2b were available. The available evidence thus relies solely on pre- to postintervention comparisons of outcomes within single cohorts of participants. On the whole, the studies were poorly reported, with high risk of bias, meaning that their results should be interpreted with caution. Although all studies reported the primary outcome (SVR), some other relevant outcomes were only reported in a few studies, meaning that no conclusions can be drawn regarding the effectiveness of peginterferon alfa-2a or -2b on biochemical response (normalisation of ALT levels), histological response (degree of inflammation, fibrosis or cirrhosis), QoL or growth (height or weight changes were reported in most studies but effects were not consistent).

Owing to heterogeneity among the studies in their participant characteristics and a lack of information on the variance of SVR estimates, quantitative meta-analysis to obtain pooled effect estimates was not appropriate. The smallest study (which had only seven participants and was not conducted in the UK) was an outlier in having a considerably lower SVR than the six other studies. With such small studies it can be difficult to determine whether they should be classed as a prospective cohort or case series (our inclusion criteria were conservative in order not to exclude potentially relevant evidence). The possible implications of including or excluding this study were considered as part of a structured narrative synthesis.

As noted previously, generalisability of the clinical effectiveness studies is difficult to determine. The generalisability to a UK paediatric population is uncertain and the results may also not be generalisable to participants with hepatic comorbidities, as the studies excluded participants with hepatitis B and HIV and, in some cases, other liver diseases. There were no studies identified in children and young people with HIV co-infection that met the inclusion criteria.

A systematic review of the literature for cost-effectiveness studies found none that met the inclusion criteria (two cost-effectiveness studies in children were identified, but one was reported superficially in an abstract only while the other, a full economic evaluation, included interferon rather than peginterferon treatment). A systematic review of studies of QoL in children with chronic HCV did not identify any relevant studies.

Parameters in the model (disease progression, utility and health state costs) have not been derived for the specific patient group in this assessment, i.e. children, as targeted searches undertaken for this review did not identify suitable data, and so parameter values have been taken for the adult population. It is uncertain how applicable these parameter values are. However, it should be noted that, as the model is a lifetime model, these parameter values are relevant during the adult years. There is also some degree of uncertainty over the intervention costs used, which include monitoring and outpatient costs. In the absence of clinical guidelines in children, these were based on the previous assessments in adults and uprated to current values. It is possible that the costs for the tertiary referral centres for children differ from the costs of secondary care in adults.

As there were no direct head-to-head comparisons of the interventions, a pragmatic approach was taken in the economic model to use the available SVR evidence through an unadjusted indirect comparison. Caution is recommended in the interpretation of the economic evaluation because, as no relevant RCTs were available, there is no means by which the similarity of the included studies can be assessed; however, it is generally recognised that the SVR of the control group is zero. The clinical effectiveness data for SVR that were included in the economic model are also limited by the small size and number of the primary research studies that met the inclusion criteria for the clinical effectiveness systematic review. Although the analysis approach has inherent risk of bias, the sensitivity and scenario analyses conducted provide an illustration of the likely estimate of the cost-effectiveness. Owing to the large number of parameters included in the sensitivity analyses, some parameters (e.g. for health state costs, transition probabilities and utility values) were combined and varied together rather than individually.

A potential limitation of the current evidence synthesis is that the searches for clinical effectiveness and cost-effectiveness data were limited to English-language publications. The approach can be justified as the context of the technology assessment is specifically the NHS in England and Wales. However, only one of the identified studies57 included any participants from the UK.

Uncertainties

Despite efforts to explore uncertainty in the evidence synthesis, some important uncertainties remain. There is uncertainty about the reliability of the relative SVR effect sizes between peginterferon alfa-2a and peginterferon alfa-2b, due to the poor quality of the studies and the lack of head-to-head trials. Variance around the estimates in the study samples is illustrated by SVR rates that range from 53% to 66% in children treated with peginterferon alfa-2a, and 49% to 65% in those treated with peginterferon alfa-2b when excluding two studies of peginterferon alfa-2b which had very small participant numbers (7 and 12 participants, respectively46,47). Some studies reported SVR rates in relation to prognostic factors (e.g. genotype or previous treatment history). However, these relationships were reported in few studies, and inevitably involved even smaller numbers of participants than were available for the primary outcome, so their reliability is uncertain.

The cost-effectiveness of peginterferon alfa-2a compared with peginterferon alfa-2b is proportionate to the relative SVR for these treatments. For strategies where SVR for peginterferon alfa-2a is greater than or the same as that for peginterferon alfa-2b, peginterferon alfa-2a is the optimal treatment; conversely, where SVR for peginterferon alfa-2b is greater than that for peginterferon alfa-2a, peginterferon alfa-2b is the optimal treatment. This demonstrates the extent of the uncertainty in the base-case results. The economic evaluation used a number of assumptions which had been discussed with our expert advisory group. We took a conservative approach to the assumption about stopping treatment in those with genotype 1 or 4 at 24 weeks if no EVR was obtained (at 12 weeks), because expert opinion was mixed. Finally, with regard to the parameters in the model, in the absence of data the model assumes that the treatment response is the same across all chronic HCV states (F0–F4); however, some uncertainty remains about this assumption.

The distribution of participants among the health states used in the model was based on a weighted average of the distribution of the participants in the studies included in the clinical effectiveness systematic review. It is uncertain how generalisable these populations are to the population of children with chronic HCV in the UK, and it is possible that in the economic model participants may be in more or less severe disease states than those seen in the UK.

Growth is an important outcome for assessing the effects of peginterferon alfa and RBV in children and adolescents with chronic HCV. The results appear to provide a weak indication that height and/or weight increases were reduced by peginterferon alfa-2b but not peginterferon alfa-2a, but this may not be reliable as only one peginterferon alfa-2a study reported this outcome and height/weight changes were not universal in the peginterferon alfa-2b studies.

Another outcome of particular interest in children and adolescents with chronic HCV is QoL. The experience of clinicians who informed the technology assessment suggests that children and adolescents may experience non-trivial QoL decrements associated with the stigma of the disease. However, there are very limited QoL data available for this population and only one of the included studies addressed this outcome, precluding any conclusions concerning the possible impact of peginterferon alfa on children and adolescents with chronic HCV. In the economic model, HRQoL was taken from studies in adults owing to a lack of data. It remains uncertain whether or not there is any impact on parent or carer’s QoL, and this has not been assessed in the cost-effectiveness model for the current assessment.

The scope of the current technology assessment is explicit about the age of the eligible population, namely children and adolescents aged 3–17 years. A number of primary research studies had populations that did not meet these criteria, as they included younger and/or older participants, and were therefore excluded (reasons for exclusion are reported in Appendix 4). It is unclear whether or not the inclusion of studies whose populations differed only marginally from the specified age range would have made any difference to the results. Although widening the age range might have allowed more evidence to be considered, it is difficult to determine at what point the age limits would become irrelevant to the decision problem. Also, deviation from the a priori-specified inclusion criteria may increase the risk of selection bias.

MEDLINE

1. Hepatitis C, Chronic/ (13,735)

2. (“hepatitis C” or HCV).tw. (48,770)

3. exp Hepatitis C/ (41,824)

4. Hepacivirus/ (20,518)

5. or/1-4 (56,080)

6. Ribavirin/ (7195)

7. (ribavirin* or copegus or rebetol or rebetron or rebretron or ribamide or ribamidil or ribamidyl or ribasphere or varazid or vilona or viramid or virazid or virazole or RibaPak).ti,ab,nm. (9111)

8. 6 or 7 (9111)

9. (peginterferon$ or peg-ifn or peg-interferon$ or (pegylat$ adj3 interferon$) or peg$ or (polyethylene glycol adj3 interferon$) or ViraferonPeg or pegintron or "peg-intron” or Pegasys).mp. (29,312)

10. Interferon-Alfa/ or Interferons/ (40,682)

11. (“IFN alfa” or “IFN alpha” or IFNalfa or IFNalpha or “interferon alfa” or “interferon alpha").tw. (24,503)

12. 10 or 11 (49,805)

13. Polyethylene Glycols/ (32,260)

14. 12 and 13 (3545)

15. 9 or 14 (29,368)

16. 5 and 8 and 15 (3824)

17. exp Child/ (1,432,200)

18. Child, Preschool/ (694,089)

19. Adolescent/ (1,468,478)

20. (child* or toddler* or adolesc* or teenage* or youth* or pediatric* or paediatric*).tw. (945,660)

21. or/17-20 (2,424,786)

22. 16 and 21 (325)

EMBASE

1. (hepatitis C or hcv).mp. (87,019)

2. exp Hepatitis C/ or exp Hepatitis C virus/ (75,723)

3. 1 or 2 (87,019)

4. (peginterferon$ or peg-ifn or peg-interferon$ or (peg$ adj3 interferon$) or (polyethylene glycol adj3 interferon$) or Pegasys or pegintron or viraferonpeg).mp. (13,639)

5. peginterferon/ or peginterferon alpha2a/ or peginterferon alpha2b/ (11,347)

6. 4 or 5 (13,639)

7. ribavirin/ (20,451)

8. (ribavirin* or copegus or rebetol or rebetron or rebretron or ribamide or ribamidil or ribamidyl or ribasphere or varazid or vilona or viramid or virazid or virazole or RibaPak).ti,ab,tn. (12,251)

9. 7 or 8 (21,598)

10. 3 and 6 and 9 (9778)

11. peginterferon alpha2a plus ribavirin/ or peginterferon alpha2b plus ribavirin/ (195)

12. 3 and 11 (186)

13. 10 or 12 (9809)

14. child/ (1,305,129)

15. preschool child/ (472,983)

16. adolescent/ (1,188,986)

17. (child$ or toddler$ or adolesc* or teenage* or youth* or pediatric* or paediatric*).tw. (1,347,475)

18. (preschool* or “pre-school*”).tw. (24,127)

19. or/14-18 (2,545,759)

20. 13 and 19 (421)

21. limit 13 to (preschool child <1 to 6 years> or school child <7 to 12 years> or adolescent <13 to 17 years>) (210)

22. 20 or 21 (422)

MEDLINE

1. Hepatitis C, Chronic/ (13,814)

2. (“hepatitis C” or HCV).tw. (48,925)

3. exp Hepatitis C/ (41,962)

4. Hepacivirus/ (20,593)

5. or/1-4 (56,255)

6. Ribavirin/ (7252)

7. (ribavirin* or copegus or rebetol or rebetron or rebretron or ribamide or ribamidil or ribamidyl or ribasphere or varazid or vilona or viramid or virazid or virazole or RibaPak).ti,ab,nm. (9176)

8. 6 or 7 (9176)

9. (peginterferon$ or peg-ifn or peg-interferon$ or (pegylat$ adj3 interferon$) or peg$ or (polyethylene glycol adj3 interferon$) or ViraferonPeg or pegintron or “peg-intron” or Pegasys).mp. (29,508)

10. Interferon-Alfa/ or Interferons/ (40,773)

11. (“IFN alfa” or “IFN alpha” or IFNalfa or IFNalpha or “interferon alfa” or “interferon alpha”).tw. (24,561)

12. 10 or 11 (49,919)

13. Polyethylene Glycols/ (32,457)

14. 12 and 13 (3591)

15. 9 or 14 (29,565)

16. 5 and 8 and 15 (3879)

17. exp economics/ (456,096)

18. exp economics hospital/ (17,917)

19. exp economics pharmaceutical/ (2332)

20. exp economics nursing/ (3862)

21. exp economics medical/ (13,273)

22. exp “Costs and Cost Analysis”/ (164,677)

23. exp Cost Benefit Analysis/ (53,972)

24. exp models economic/ (8599)

25. (cost* adj2 (effective* or benefit* or utilit* or minim*)).tw. (73,771)

26. markov chains/ or monte carlo method/ (23,357)

27. (decision adj1 (tree* or analys* or model*)).tw. (6891)

28. exp health care costs/ (40,596)

29. or/17-28 (520,897)

30. 16 and 29 (144)

31. (letter or editorial or comment or historical article).pt. (1,397,929)

32. 30 not 31 (141)

EMBASE

1. (hepatitis C or hcv).mp. (88,182)

2. exp Hepatitis C/ or exp Hepatitis C virus/ (76,553)

3. 1 or 2 (88,182)

4. (peginterferon$ or peg-ifn or peg-interferon$ or (peg$ adj3 interferon$) or (polyethylene glycol adj3 interferon$) or Pegasys or pegintron or viraferonpeg).mp. (13,980)

5. peginterferon/ or peginterferon alpha2a/ or peginterferon alpha2b/ (11,579)

6. 4 or 5 (13,980)

7. ribavirin/ (20,743)

8. (ribavirin* or copegus or rebetol or rebetron or rebretron or ribamide or ribamidil or ribamidyl or ribasphere or varazid or vilona or viramid or virazid or virazole or RibaPak).ti,ab,tn. (12,561)

9. 7 or 8 (21,934)

10. 3 and 6 and 9 (10,038)

11. peginterferon alpha2a plus ribavirin/ or peginterferon alpha2b plus ribavirin/ (203)

12. 3 and 11 (194)

13. 10 or 12 (10,071)

14. *Health Economics/ (16,281)

15. *Economics/ (11,217)

16. monte carlo method/ (17,113)

17. (cost* or economic*).ti. (126,413)

18. markov.tw. (11,647)

19. "monte carlo".tw. (22,996)

20. (cost* adj2 (effective* or utili* or benefit* or minimi* or consequence* or analys* or saving* or breakdown* or estimate* or variable* or allocation* or control* or illness)).tw. (130,225)

21. (econom* or pharmacoeconomic* or “pharmaco economic*” or budget*).tw. (214,740)

22. cost/ (49,867)

23. cost minimization analysis/ (2073)

24. cost of illness/ (12,986)

25. cost utility analysis/ (4167)

26. drug cost/ (51,793)

27. health care cost/ (110,482)

28. economic evaluation/ (7191)

29. pharmacoeconomics/ (5675)

30. budget/ (17,530)

31. “resource use”.tw. (5040)

32. “resource utili”.tw. (1)

33. (decision adj1 (tree* or analys* or model*)).tw. (9480)

34. (“unit cost*” or “hospital cost*” or “health care cost*” or “healthcare cost*” or “medical cost*”).tw. (27,557)

35. (managed adj2 (care or clinical or network)).tw. (19,055)

36. (resource* adj1 allocat*).tw. (5901)

37. (resource* adj1 utili*).tw. (6341)

38. or/14-37 (594,150)

39. 13 and 38 (533)

40. (cost and effective* and “hepatitis C”).ti. (175)

41. (cost and effective* and “hepatitis C”).ab. (608)

42. 40 or 41 (646)

43. 6 and 9 and 42 (208)

44. 11 and 42 (4)

45. 39 or 43 or 44 (543)

46. (comment or editorial or letter).pt. (1,197,206)

47. 45 not 46 (507)

MEDLINE

1. value of life/ (5222)

2. quality adjusted life year/ (5699)

3. quality adjusted life.ti,ab. (4572)

4. (qaly$ or qald$ or qale$ or qtime$).ti,ab. (3826)

5. disability adjusted life.ti,ab. (873)

6. daly$.ti,ab. (885)

7. health status indicators/ (17,959)

8. (sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirstysix or shortform thirty six or short form thirty six or short form thirtysix or short form thirty six).ti,ab. (12,492)

9. (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).ti,ab. (898)

10. (sf12 or sf 12 or short form 12 or shortform 12 or sf twelve of sftwelve or shortform twelve or short form twelve).ti,ab. (1941)

11. (sf16 or sf 16 or short form 16 or shortform 16 or sf sixteen or sfsixteen or shortform sixteen or short form sixteen).ti,ab. (18)

12. (sf20 or sf 20 or short form 20 or shortform 20 or sf twenty of sftwenty or shortform twenty of short form twenty).ti,ab. (303)

13. (euroqol or euro qol or eq5d or eq 5d).ti,ab. (2707)

14. (hql or hqol or h qol or hrqol or hr qol).ti,ab. (5603)

15. (hye or hyes).ti,ab. (52)

16. health$ year$ equivalent$.ti,ab. (36)

17. health utilit$.ab. (770)

18. (hui or hui1 or hui2 or hui3).ti,ab. (700)

19. disutil$.ti,ab. (169)

20. rosser.ti,ab. (72)

21. quality of well being.ti,ab. (297)

22. quality of wellbeing.ti,ab. (5)

23. qwb.ti,ab. (150)

24. willingness to pay.ti,ab. (1664)

25. standard gamble$.ti,ab. (580)

26. time trade off.ti,ab. (604)

27. time tradeoff.ti,ab. (191)

28. tto.ti,ab. (459)

29. (index adj2 well being).mp. (419)

30. (quality adj2 well being).mp. (736)

31. (health adj3 utilit$ ind$).mp. (531)

32. ((multiattribute$ or multi attribute$) adj3 (health ind$ or theor$ or health state$ or utilit$ or analys$)).mp. (206)

33. quality adjusted life year$.mp. (7493)

34. (15D or 15 dimension$).mp. (1018)

35. (12D or 12 dimension$).mp. (314)

36. rating scale$.mp. (74,498)

37. linear scal$.mp. (480)

38. linear analog$.mp. (784)

39. visual analog$.mp. (24,844)

40. (categor$ adj2 scal$).mp. (1040)

41. or/1-40 (150,176)

42. (letter or editorial or comment).pt. (1,134,806)

43. 41 not 42 (146,086)

44. Hepatitis C, Chronic/ (13,926)

45. (“hepatitis C” or HCV).tw. (49,230)

46. exp Hepatitis C/ (42,205)

47. Hepacivirus/ (20,742)

48. or/44-47 (56,582)

49. 43 and 48 (441)

EMBASE

1. quality adjusted life year/ (9303)

2. quality adjusted life.ti,ab. (6565)

3. (qaly$ or qald$ or qale$ or qtime$).ti,ab. (6358)

4. disability adjusted life.ti,ab. (1139)

5. daly*.ti,ab. (1265)

6. (sf36 or sf 36 or short form 36 or shortform 36 or sf thirtysix or sf thirty six or shortform thirtysix or shortform thirty six or short form thirty six or short form thirtysix or short form thirty six).ti,ab. (17,835)

7. (sf6 or sf 6 or short form 6 or shortform 6 or sf six or sfsix or shortform six or short form six).ti,ab. (1404)

8. (sf12 or sf 12 or short form 12 or shortform 12 or sf twelve or sftwelve or shortform twelve or short form twelve).ti,ab. (2980)

9. (sf16 or sf 16 or short form 16 or shortform 16 or sf sixteen or sfsixteen or shortform sixteen or short form sixteen).ti,ab. (31)

10. (sf20 or sf 20 or short form 20 or shortform 20 or sf twenty or sftwenty or shortform twenty or short form twenty).ti,ab. (376)

11. (euroqol or “euro qol” or “eq5d” or “eq 5d”).ti,ab. (4495)

12. (hql or hqol or “h qol” or hrqol or “hr qol”).ti,ab. (8357)

13. (“hye” or “hyes”).ti,ab. (63)

14. health* year* equivalent*.ti,ab. (41)

15. health utilit*.ti,ab. (1192)

16. (hui or hui1 or hui2 or hui3).ti,ab. (969)

17. disutil*.ti,ab. (270)

18. rosser.ti,ab. (83)

19. quality of well being.ti,ab. (332)

20. quality of wellbeing.ti,ab. (18)

21. qwb.ti,ab. (170)

22. willingness to pay.ti,ab. (2513)

23. standard gamble*.ti,ab. (695)

24. time trade off.ti,ab. (804)

25. time tradeoff.ti,ab. (208)

26. tto.ti,ab. (698)

27. (index adj2 well being).mp. (568)

28. (quality adj2 well being).mp. (994)

29. (health adj3 util* adj ind*).mp. (888)

30. ((multiattribute* or multi attribute*) adj3 (health ind* or theor* or health state* or util* or analys*)).mp. (299)

31. quality adjusted life year*.mp. (11,067)

32. health status indicator*.ti,ab. (327)

33. (15D or 15 dimension*).mp. (1511)

34. (12D or 12 dimension*).mp. (452)

35. "health related quality of living".ti,ab. (3)

36. "health related quality of life".ti,ab. (22,371)

37. rating scale*.mp. (115,178)

38. visual analog*.mp. (42,178)

39. (categor* adj scale*).mp. (702)

40. linear scal*.mp. (655)

41. linear analog*.mp. (962)

42. "quality of life".ti. (44,803)

43. or/1-42 (231,142)

44. (letter or editorial or comment).pt. (1,199,398)

45. 43 not 44 (224,733)

46. (“hepatitis C” or HCV).tw. (70,769)

47. Hepatitis C/ or Hepatitis C virus/ (76,817)

48. hepacivirus.tw. (58)

49. or/46-48 (87,733)

50. 45 and 49 (942)