Convincing or confusing?

Vaccine 25 (2007) 1355–1367

Review

Convincing or confusing? Economic evaluations of childhood pneumococcal conjugate vaccination—a review (2002–2006) Philippe Beutels a,b,∗ , Nancy Thiry a , Pierre Van Damme a a

b

Centre for the Evaluation of Vaccination, Epidemiology & Social Medicine, University of Antwerp (Campus Drie Eiken), Universiteitsplein 1, 2610 Antwerp, Belgium National Centre for Immunisation Research & Surveillance (NCIRS), School of Public Health, University of Sydney, Australia Received 24 June 2006; received in revised form 13 October 2006; accepted 18 October 2006 Available online 3 November 2006

Abstract We review 15 economic analyses of pneumococcal conjugate vaccines, published between 2002 and 2006, in terms of methodology, assumptions, results and conclusions. We found a great diversity in assumptions (eg, vaccine efficacy parameters, incidence rates for both invasive and non-invasive disease) mainly due to local variation in data and opinions. Accordingly, the results varied greatly, from total net savings to over D 100,000 per discounted QALY gained. The cost of the vaccination program (determined by price per dose and schedule (4 or 3 doses, or fewer)), and likely herd immunity impacts are highly influential though rarely explored in these published studies. If the net long-term impact (determined by a mixture of effects related to herd immunity, serotype replacement, antibiotic resistance and cross reactivity) remains beneficial and if a 3-dose schedule confers near-equivalent protection to a 4-dose schedule, the cost-effectiveness of PCV7 vaccination programs can be viewed as attractive in developed countries. © 2006 Elsevier Ltd. All rights reserved. Keywords: Cost-effectiveness; Vaccination; Economic; Pneumococcal; Herd immunity

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Differences in set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2. Differences in assumptions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.1. Vaccine efficacy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.2. Burden of disease . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2.3. Vaccination costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3. Differences in results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.1. Universal infants’ vaccination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.2. Catch-up vaccination . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3.3. Sensitivity analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

∗ Corresponding author at: Centre for the Evaluation of Vaccination, Epidemiology & Social Medicine, University of Antwerp (Campus Drie Eiken), Universiteitsplein 1, 2610 Antwerp, Belgium. Tel.: +32 3 8202523. E-mail address: [email protected] (P. Beutels).

0264-410X/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.vaccine.2006.10.034

1356 1356 1356 1356 1362 1362 1362 1363 1363 1363 1364 1364

1356

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

4.

Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1. Introduction Streptococcus pneumoniae is a bacterial pathogen that affects children and adults worldwide. It is a major cause of illness in children, especially those under the age of 24 months, in whom it can cause disseminated invasive infections (including meningitis and bacteraemia), lower respiratory tract infections (including pneumonia) and upper respiratory tract infections (including otitis media and sinusitis). In children, Streptococcus pneumoniae is currently one of the leading causes of meningitis and otitis media. Treatment of pneumococcal diseases is threatened by the emergence of pneumococcal strains resistant to penicillin and other antibiotics. The substantial disease burden and the availability of pneumococcal conjugate vaccines (PCVs) give the potential introduction of a universal childhood pneumococcal conjugate vaccination program a prominent place on the health policy agenda in many countries. Currently, eight, ten, eleven and thirteen valent vaccines (PCV8, PCV10, PCV11, PCV13) are or have been in phase II and III trials. The sevenvalent pneumococcal conjugate vaccine (PCV7) is currently licensed in Australia, North America, most parts of Europe and Central and South America. It is also part of the universal infant vaccination program in the US (since 2000, with a 4-dose schedule), Australia (since 2005, with a 3dose schedule), Canada (since 2005, with a 4-dose schedule in all provinces but Quebec, where 3 doses are given to low risk children), The UK and Norway (both since 2006, with a 3-dose schedule), as well as the Netherlands (since 2006, with a 4-dose schedule) [1]. Given the high investment costs associated with this program, countries considering its implementation would prefer to do so on the basis of sound assessments of its population effectiveness, budget-impact and cost-effectiveness. Given the country-specific nature of the prevalence of circulating pneumococcal serotypes, such assessments are likely to differ from one country to the next. This paper reviews the literature on economic evaluations of pediatric PCV use, focusing on the main differences in input data, methods and results. 2. Methods Published economic evaluations of options for use of PCVs were identified using Medline and EconLit with the search terms “pneumonia”, “pneumococcal”, “vaccine”, “cost” and “economic”. Abstracts of journal articles were reviewed to retrieve only full economic evaluations, as defined in Drummond et al. [2] (thus excluding pure cost analyses). Only the articles published between August 2002

1364 1366 1366

and April 2006 were selected, because a review article of economic evaluations on the same topic up to August 2002 had already been published by the current first author, and the rapidly changing insights make the more recent analyses much more relevant than older ones [3,4]. Our search identified 15 new studies [5–19], each of which was systematically reviewed in terms of methodology, assumptions, results and conclusions. As such, our review substantially differs from a recent overview, which interprets and presents results of studies, without such thorough review [20]. All cost data reported in our review were transformed to Euro 2002 values on the basis of local Consumer Price Indices and Purchasing Power Parities.

3. Results Our findings are mainly presented in comparative tables. Table 1 lists the selected studies and presents their general characteristics. Note that throughout this paper, we use the term “efficiency” in a restricted way, meaning the measures of relative efficiency used in health care delivery, expressed specifically as cost-effectiveness, cost-utility or cost–benefit ratios. The studies’ assumptions in terms of vaccine efficacy, epidemiological and economic burden are reported in Tables 2–4, respectively. Table 5 presents the studies’ results. 3.1. Differences in set-up Eleven published studies were performed in seven European countries (Finland [17], Germany [8], Italy [15], Spain [5,16], Switzerland [13,18], the Netherlands [6], UK [11,12,14]), three in Canada [9,10,19], and one in Australia [7]. All 15 studies analysed the efficiency of universal infant vaccination and three studies [5,8,9,18] additionally assessed the impact of catch-up programs for older children (Table 1). The universal infant program was usually defined in line with the pivotal clinical trials, as consisting of four doses of the PCV7 vaccine administered before the age of 18 months. In Melegaro and Edmunds [14] and Marchetti and Colombo [15], however, 3 doses of the vaccine were assumed to suffice to invoke an equally good protective vaccine efficacy as observed in the trials, and this approach was also explored in sensitivity analysis by Salo et al. [17]. In Ruedin et al. [13], the impact of universal vaccination programs using a hypothetical combined vaccine against nine serotypes of pneumococci and serogroup C meningococci (PCV9-MenC) was analysed. All these studies were model-based and inves-

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

1357

Table 1 Design and assumed vaccination costs in published economic evaluations of conjugate pneumococcal vaccinesa (08/2002–03/2006) Study

Country/region

Publication year

Study type

Perspective

Time span (years)

Discount rate (%)

Vaccination costs per dose (price + administration) in Euro 2002

Salo et al. [17]

Finland

2005

CEA, CUA, CBA

Payer Society

5

C:3 B:3

51.55 (49.97 + 1.58)

McIntosh et al. [12]

UK

2005

CEA

Payer

10

C:6 B:0; 6

72.23 (57.56 + 14.67)

Navas et al. [16]

Catalonia, Spain

2005

CEA, CUA

Payer

10

C:5

60.84 (53.38 + 7.46)

Society Marchetti and Colombo [15]

Italy

2005

CEA

Payer

B:5 14

Society

C:3

44.00 (44.00 + 0)b

B:NS

Butler et al. [7]

Australia

2004

CEA, CUA

Payer

5

C: 5 B:5

72.46 (68.65 + 3.81)

Asensi et al. [5]

Spain

2004

CEA

Payer Society

10

C:3 B:3

63.00b

Melegaro and Edmunds [14]

England & Wales

2004

CEA, CUA

Payer

Life long

C:3.5

58.66 (44.00 + 14.67)

B:1.5; 0 Mcintosh et al. [11]

England & Wales

2003

CEA

Payer

10

Ess et al. [18]

Switzerland

2003

CUA

Payer

5

C:3 B:0

53.47 (48.57 + 4.91)

Ruedin et al. [13]

Switzerland

2003

CUA

Payer

10

C:3 B:3; 0

81a

Claes and Graf von der Schulenburg [8]

Germany

2003

CEA

Payer

10

C:5

67.47 (64.60 + 2.87)

Society

72.23 (57.56 + 14.67)

B:0

Society Payer

C: 6

B:5

Bos et al. [6]

The Netherlands

2003

CUA

10

Lebel et al. [10]

Canada

2003

CEA

Payer Society

10

C:3 B:3

54.78 (54.78 + 0)

De Wals et al. [9]

Canada

2003

CEA, CUA

Payer Society

10

C:3 B:3

52.41 (47.07 + 5.34)

Moore et al. [19]

British Columbia, Canada

2003

CEA

Payer

5

NS

54.78 (54.78 + 0)

Society

C:4

45.49 (40.25 + 5.23)

B:4

CEA: cost-effectiveness analysis; CUA: cost-utility analysis; C: costs; B: benefits; NS: not stated. a All studies assess the seven-valent pneumococcal conjugate vaccine, with the exception of Ruedin et al. [13] who assess an hypothetical vaccine combining nine pneumococcus serotypes with meningococcus serotype C. b Not explicitly stated, deduced estimate.

tigated the impact of vaccination over a 5–10 year period after birth (over lifetime in Melegaro and Edmunds [14]). None of these studies modelled the positive impact of herd immunity and reduced antibiotic resistance, nor the negative impact of serotype replacement in their base-case analysis. In Melegaro and Edmunds [14] the potential impact of herd immunity and serotype replacement were estimated separately in the sensitivity analysis, and McIntosh et al. [12] extend on their earlier study [11] to explore the impact of herd immunity. Note that

the first major study to demonstrate herd effects in the US was only published in 2003 [21], when many of the studies under review were already submitted or published. Eight studies used quality-adjusted life years (QALY) [6,9,13,14,16–18] or disability-adjusted life years (DALYs) [7] as a measure of health outcome (i.e. they performed a cost-utility analysis (CUA)), while six studies focused on life-years (LY) [5,8,10–12,19] without quality adjustment (i.e. cost-effectiveness analysis (CEA)) [2]. Conceptually,

1358

Table 2 Baseline vaccine effectiveness assumptions of published economic evaluations of conjugate pneumococcal vaccines (08/2002–03/2006) Vaccine uptake (%)

Vaccine effectiveness (%) against

Adjusted effectiveness for serotypes circulating in home country

Duration of protection (years)

Waning of immunity per year

IPD

AOM

CAP

Other

Salo et al. [17] McIntosh et al. [12]

NR NR

89.1 97.4

6.0 NR

17.7 4.3

Tymp: 20.3

No IPD

5 10

No 1–3%a

Navas et al. [16]

95

89.1

6.4

22.7

Tymp: 23.2

No

IPD: 10 AOM, CAP: 2 Tymp: 3.5

No

Marchetti and Colombo [15] Butler et al. [7]

100b 100b

89.1 93.9

6.4 6.4

17.7 8.9

No (IPD in sens an) IPD

14 5

3% > 5y No

Asensi et al. [5]

100

97.4

5.8

11.4

IPD

5

3% > 5y

Melegaro and Edmunds [14] Mcintosh et al. [11]

100 95

63–87c 97.4

7.0 7.0

17.7 6.0

IPD IPD

10 1

No 1–3%a

Ess et al. [18]

70

97.0

7.0

11.0

IPD AOM–CAP

5

No

Ruedin et al. [13] Claes and Graf von der Schulenburg [8] Bos et al. [6]

80 100 100

89–87a 85.0 86–95b

6.0 6.0 5.8

11.0 9.1–32.2d 11.4

No IPD IPD

10 10 5

No No 3% > 5y

Lebel et al. [10]

100

89.1

5.8

11.4

RAOM: 10.6 SP: 33 MP: 24.9

No

5

3% > 5y

De Wals et al. [9] Moore et al. [19]

80 90

97.0 89.0

8.2 7.0

10.7 11.0

MP: 24.9

IPD No

10 5

1% > 3y No

RAOM: 10.6 SP: 33 MP: 24.9

IPD: invasive pneumococcal disease; AOM: acute otitis media; CAP: community acquired pneumonia; MP: myringotomy procedure; RAOM: recurrent acute otitis media; SP: severe pneumonia. a Varies according to age. b Dose distribution (% receiving only 1, 2, 3, or 4 doses assumed to be 9%, 9%, 24%, 58%, respectively, identical to the NCKP trial [22]). c Efficacy after adjustment for circulating serotype in England and Wales. d Varies according to time since vaccination.

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

Study

Table 3 Burden of disease assumptions of published economic evaluations of conjugate pneumococcal vaccines (08/2002–03/2006) Study

Incidence of IPD (per 100,000 population) All IPD

Meningitis

Case-fatality ratios (%) Bacteraemia

IPD

Non IPD

All IPD 0–5 y:1.4

Meningitis NA

Bacteraemia NA

Othera NA

Pneumonia 0

1–2 y:54.9

0.1 y:5.3 1–2 y:2.4

0–1 y:25.0 1–2 y:52.5

McIntosh et al. [12] Navas et al. [16] Marchetti and Colombo [15]

NR NR 0–1 y:27.1b 1–4 y:20.9b 5–14 y:5.0b

NR NR 0–1 y:7.6 1–4 y:2.1 5–14 y:0.5

NR NR 0–1 y:12.7 1–4 y:13.6 5–14 y:3.3

NR 1.4 NA

NR NA 0–1 y:14.0 2–4 y:7.0 5–10 y:1.0

NR NA 0.9

NA NA 0.9

NR 0.4 0

Butler et al. [7]

0–1 y:105.6 2–4 y:35.2

0–1 y:13.7 2–4 y:2.1

0–1 y:67.6 2–4 y:22.8

NA

0–1 y:11.5 2–4 y:7.1

0–1 y:0.4 2–4 y:1.4

0–1 y:1.5 2–4 y:0.5

0 y:0.65 1 y:0.15 2–4 y:0.03

Asensi et al. [5]

NA

0–10 yrs:3.4

0–10 yrs:27.5

NA

0–10 yrs:9.3

0–10 yrs:1.0

NA

0

Melegaro and Edmunds [14]

NA

0–1 y:14.6 1–4 y:1.6 5–9 y:0.2 10–14 y:0.2 15–19 y:0.1 20–24 y:0.2 25–44 y:0.3 45–64 y:0.5 65–74 y:0.9 75+ y:0.6

0–1 y:27.3 1–4 y:10.6 5–9 y:1.9 10–14 y:0.7 15–19 y:1.2 20–24 y:1.8 25–44 y:3.1 45–64 y:6.5 65–74 y:18.7 75+ y:42.5

NA

0–1y:4.0 1–4 y:4.0 5–9 y: 3.0 10–14 y:0 15–19 y:11.0 20–24 y:0 25–44 y:11.0 45–64 y:18.0 65–74y:29.0 75+ y:43.0

0–1 y:4.0 1–4 y:1.0 5–9 y:0 10–14 y:0 15–19 y:0 20–24 y:8.0 25–44 y:20.0 45–64 y:26.0 65–74 y:27.0 75+ y:40.0

NA

0–1 y:1.0 1–4 y:0 5–9 y:1.0 10–14 y:2.0 15–19 y:2.0 20–24 y:3.0 25–44 y:3.0 45–64 y:14.0 65–74 y:29.0 75+ y:46.0

Mcintosh et al. [11]

NR

NR

NR

NR

NR

NR

NR

Yes-NR

Ess et al. [18]

0–2 y:31c 0–5 y:11c

0–2 y:5.6 0–5 y:3.1

NA

0–5 y:9.0c

0–5y:9.0

NA

NA

0

Ruedin et al. [13]

0–9 y:7.4d 1–10 y:5.7d

NA

NA

0–10 y:5.0d

NA

NA

NA

0

Claes and Graf von der Schulenburg [8]

NA

1–2 y:8.0 3–4 y:1.6 5–10 y:0.04

1–2 y:12.2 3–4 y:3.4 5-10 y:0.8

NA

1–10 y:8.3

1–10 y:1.5

NA

1–10 y:0.08

Bos et al. [6]

NA

0–10 y:113e 0–10 y:114e

0-10 y:226e

NA

0–10 y:17.0

0–10y:6.0

NA

0

Lebel et al. [10] De Wals et al. [9]

NA NA

0–10 y:3.0 0–4 y:0.47–19.37f 5–9 y:0.46

0-10 y:25.0 0–4 y:12.8–94.8 5–9 y:4.6

NA NA

0–10 y:6.6 0–10 y:6.5g

0–10 y:1.26 0–10 y:2.0g

NA NA

0 0–10 y:0.1g

Moore et al. [19]

0–2 y:90–150h 2–4 y:10–50h

NA

NA

0–4 y:0.02

NA

NA

NA

0–4 y:0.0005

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

Salo et al. [17]

IPD: invasive pneumococcal disease; y(s): year(s); NA: not applicable; NR: not reported (or insufficiently clear to represent here). a Other invasive pneumococcal infections: Sepsis, peritonitis, bone and joint infection. b Not stated as such; derived from other estimates. c For all IPD minus meningitis. d For the nine serotypes included in the hypothetical PCV9-MenC vaccine. e Number of cases per year in Dutch children aged between 0 and 10 years.

1359

f Estimates vary according to age. g Not stated—estimates obtained from Petit et al. [42]. h High and low incidence estimates.

Study

1360

Table 4 Meningitis unit cost assumptions of published economic evaluations of conjugate pneumococcal vaccines (08/2002–03/2006) Pneumococcal meningitis costs (2002 Euro) Average cost of pneumococcal meningitis

Long-term sequelae of pneumococcal meningitis

Item

Neurological deficit

Cost

Salo et al. [17]

Meningitis case (no sequelae)

D 7504

NR

Navas et al. [16]

NR

NR

SE (per year)

D 10,063

Marchetti and Colombo [15]

Meningitis case (no sequelae?) Meningitis case (no sequelae) Meningitis case (including LT sequelae) NR

D 7,536a

NR

NR

D 6,356 D 6,356–D 3,498,915b

SE (per year) RC (per year)

D 4,983 D 47,197

Butler et al. [7]

Hearing deficit

Cost Deafness (assumed 13% of productivity losses from death) Cochlear implant (per procedure) NR

D 89,612

D 1,526 D 16,824

D 32,085 NR

NR

Yes

NR

Hearing aid (per unit) Cochlear implant (per procedure) Yes

Meningitis case (first year) Additional LT care (per subsequent year) Meningitis case (including LT sequelae)

D 6,897 D 210

Yes

NR

Yes

NR

D 6,992

Brain damage (per child)

D 1,189,806

Deafness (per child)

D 88,877

D 4,886 D 736

RC or SE (per year)

D 14,717

Yes

NR

Ruedin et al. [13]

Meningitis case (no sequelae) Additional LT health care for sequelae (per year) Additional RC or SE for sequelae (per year) IPD case (no sequelae)

SE (per year)

D 10,135

Cost auditive device (per year)

D 717

D 19,600

Claes and Graf von der Schulenburg [8]

Additional LT care for sequelae (per case) Meningitis case (no sequelae)

D 6,870

LT sequelae

D 47,855

D 23,319

Additional LT health care for sequelae (per episode) Additional RC or SE for sequelae (per episode) Meningitis case (uncomplicated) Meningitis case (complicated) NR Meningitis case (no sequelae) NR

D 23,319–D 84,019a

SE (per case)

D 33,084–D 82,790a

Hearing disorder (per episode) Cochlear implant (per procedure)

D 5,770

SE (per case)

D 166,215

D 19,450

RC (per case)

D 947,008

NR D 8,344 NR

Yes Yes No

NR NR

Asensi et al. [5]

Mcintosh et al. [11]

NR

Mcintosh et al. [12] Ess et al. [18]

Bos et al. [6]

Lebel et al. [10] De Wals et al. [9] Moore et al. [19]

NR: not reported; SE: special education; RC: residential care; LT: long-term. a Price level not adjusted, as original price level not clearly reported. b Varies according to the medical condition.

D 14,717 D 6,350

D 84,019

D 33,084–D 82,790a Hearing disorder (per episode)

D 4,629

Yes Yes No

NR NR

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

Melegaro and Edmunds [14]

Table 5 Results of published economic evaluations of conjugate pneumococcal vaccine (08/2002–03/2006) Country

Original currency (year)

Studies’ results (2002 Euro): incremental cost-effectiveness ratios (ICER)

Salo et al. [17]

Finland

Euro (2004)

Vaccination scenarios Infants: 4 doses (schedule not stated)

Payer’s perspectivea D 208,570 per disc LYG D 75,922 per undisc LYG D 44,563 per disc QALY gained

Societal perspectiveb D 133,563 per disc LYG D 23,725 per undisc LYG D 28,536 per disc QALY gained

McIntosh et al. [12]

UK

£ (2002)

Infants: 4 doses (2, 3, 4, 12–15 m)

D 6,932 per disc LYG D 6,394 per undisc LYG

NA

Navas et al. [16]

Catalonia, Spain

Euro (2000)

Infants: 4 doses (2, 3, 4, 12–15 m)

D 65,929 per disc LYG D 85,727 per disc DALY avertedc

D 15,908 per disc LYG D 47,307 per disc DALY averted BCR: 0.59

Marchetti and Colombo [15] Butler et al. [7]

Italy Australia

Euro (2002)d $AU (1997–1998)

Infants: 3 doses (2, 4, 6 m) Infants: 4 doses (2, 4, 6, 12 m)

D 38,286 per disc LYG D 175,540 per disc LYG D 92,374 per disc DALY averted

D 26,449 per disc LYG NA

Asensi et al. [5]

Spain

Euro (1999)d

Infants: 4 doses (2, 3, 4, 12–15 m) Infants catch-up: all <60 months

D 78,235 per disc LYG D 99,773 per disc LYG

Savings Savings

Melegaro and Edmunds [14]

England & Wales

£ (2002)

Infants: 3 doses (protected from 4 m)

D D D D

NA

Mcintosh et al. [11]

England & Wales

£ (2002)

Infants: 4 doses (2, 3, 4, 12–15 m)

D 46,214 per undisc LYG

D 41,292 per undisc LYG

Ess et al. [18]

Switzerland

CHF (2001)d

Infants: 4 doses (2, 4, 6, 12–15 m) Infants catch-up 1: all <24 months Infants catch-up 2: all <60 months

D 19,279 per undisc QALY gained D 16,483 per undisc QALY gained D 79,470 per undisc QALY gained

NA

Ruedin et al. [13]

Switzerland

Euro (2002)

Infants: 3 doses (2, 4, 6 m) PCV9-MenC

D D D D

NA

Toddler: 1 year—1 dose of PCV9-MenC Claes and Graf von der Schulenburg [8] Bos et al. [6]

166,060 per disc LYG 95,792 per undisc LYG 87,913 per disc QALY gained 57,087 per undisc QALY gained

39,000 per disc QALY gained 34,000 per undisc QALY gained 15,000 per disc QALY gained 13,000 per undisc QALY gained

Germany

Euro (2002)d

Infants: 4 doses (2, 3, 4, 12–15 m)

D 68,201 per disc LYG

Savingse

The Netherlands

Euro (2001)

Infants: 4 doses (2, 3, 4, 12–15 m)

D 80,006 per disc QALY gained

D 71,703 per disc QALY gained D 83,226 per disc LYG

Lebel et al. [10]

Canada

$CAN (2000)

Infants: 4 doses (2, 4, 6, 12–15 m)

D 125,469 per disc LYG

D 63,938 per discounted LYG

De Wals et al. [9]

Canada

$CAN (2000)

Infants: 4 doses (2,4,6,12–15 m)

D 152,584 per disc LYG D 142,033 per disc QALY gained

Infant catch-up: 3 doses (7–12 m)

–

Toddler catch-up: 2 doses (12–18 m)

–

Child catch-up: 1 dose (24–48 m)

–

D D D D D D D D

Infants: 4 doses <18 months

D 34,528 to D 73,457 per undisc LYG

Moore et al. [19]

Canada (British Columbia)

$CAN (2000)

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

Study

101,452 per disc LYG 94,148 per disc QALY gained 194,788 per disc LYG 193,165 per disc QALY gained 163,135 per disc LYG 163,947 per disc QALY gained 167,943 per disc LYG 163,947 per disc QALY gained

NA

1361

LYG: life-year gained; QALY: quality-adjusted life year; DALY: disability-adjusted life year; NA: not applicable. a Including only direct medical costs, this can arguably be interpreted as a restrictive societal perspective, if the denominator contains quality of life adjustment to account for non-health care related opportunity benefits of avoided morbidity (as in “QALYs”, as not only including mortality, as in “life-years gained”). b Including both direct medical costs and indirect productivity costs due to morbidity, unless specified under c. c In Navas et al., the authors erroneously refer to “costs per DALY gained”. In their Table 4, we interpreted the costs per life-year saved for society and “provider”, as being reversed (as this would be more consistent with the notion that society’s perspective produces more attractive ratios than the provider’s perspective, ceteris paribus.). d Assumed price level (based on publication year, or other information), as price level not explicitly or not clearly reported. e Societal perspective including also indirect productivity costs due to mortality.

1362

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

the latter studies therefore underestimated the effectiveness (limiting it to the consequences of mortality, rather than morbidity and mortality combined) and produced a simpler and more conservative measure of efficiency [3]. It is of course highly relevant to include quality of life aspects to evaluate clinical disease that has a large impact on quality of life, or occurs frequently or is rarely lethal (and virtually all disease expressions of pneumococcal infections fall under these descriptions). All studies adopted a health care payer’s perspective, i.e. including only direct medical costs and five of these also took on a societal perspective, in which indirect productivity costs (due to acute and chronic morbidity alone) were included in the numerator of the ratio [5,6,9–11]. In another three studies [8,16,17], not only were productivity costs due to morbidity monetised but also those due to averted mortality to express a societal viewpoint. By doing so as part of CEAs or CUAs, these analysts may have double counted mortality costs, as they appear as a gain in life-years in the denominator of the ratio as well. All these analyses were based on a static Markov cohort model approach, which is a standard well-established way to model ongoing risks and disease progression in an ageing population. However, without having access to these models, it is difficult to make an irrefutable judgement on the quality of execution of this approach. In general we are of the opinion that without access to the models, their quality can best be judged by the transparency in the input and structural choices within the Markov model framework (see also below). Good practice in this respect would include adapting IPD efficacy estimates from the US trial according to local serotype circulation, exploring the impact of herd immunity and serotype replacement, and performing multivariate sensitivity analysis. Note that none of the studies under review aimed to model the underlying transmission process. This would require a more complex dynamic population model and data on carriage of the different pneumoccoccus serotypes in all age groups. 3.2. Differences in assumptions 3.2.1. Vaccine efficacy Estimates for the efficacy of the vaccine against invasive and non-invasive disease varied greatly between studies (particularly for community acquired pneumonia, see Table 2), though they all relied on the same clinical trials (USA [22,23] and Finland [24]). In those trials, based on intention to treat (ITT) results, PCV7 was estimated to prevent about 97% of vaccine serotype invasive disease (89% of all invasive disease), 6% of all-cause otitis media (OM) in children and 6% (per protocol 4.3%, i.e. lower) of clinically diagnosed pneumonia (not necessarily with X-ray taken). Five studies did not adjust vaccine efficacy by estimates of serotypes circulating in their own country [10,13,16,17,19]. Their implicit assumption was thus that the prevalence of circulating serotypes in their country is identical to that in the US trial, and some authors indicated that they made a rough comparison to ver-

ify this. The duration of protection afforded by the vaccine remains uncertain and it was usually assumed that protection lasts for the model duration. Six studies have assumed that protection wanes after vaccination [5,6,9–11,15]. Another difference between the studies was the estimated vaccine uptake. Seven studies assumed unrealistically 100% vaccine uptake in the targeted population [5,6,8,10,14,15,17]. In the absence of herd effects, this should have no or a limited impact on the cost-effectiveness ratios, but it could produce misleading estimates of the impact of the program on the disease burden and health care budgets [25]. 3.2.2. Burden of disease The reliability of the input data clearly determines the accuracy of the projected cost-effectiveness results. Due to diagnostic divergence, however, and because clinically diagnosed OM and pneumonia can be caused by microorganisms other than Streptococcus pneumoniae, the incidence of non-invasive pneumococcal pneumonia is difficult to define, but could nonetheless be highly influential for the costeffectiveness of the program. Invasive pneumococcal disease (IPD), though not the most common manifestation of Streptococcus pneumoniae, is responsible for most severe pneumococcal disease. An accurate assessment of the disease burden of IPD (particularly in terms of incidence and lethality) is therefore also essential, and was noted for its influence in many of the economic analyses [5,6,10,11,14,15,18]. The assumed incidence and case-fatality ratios of IPD are presented in Table 3. All studies used local morbidity and mortality data (with the exception of the incidence of pneumonia and non-focal bacteremia, and case-fatality ratios, reported in Navas et al. [16] and case-fatality ratios reported in Bos et al. [6]), originating from databases (hospital, laboratory or surveillance records) [9], observational studies [5,8,10,13,18,19] or both [6,7,12,13,15,17]. Four studies reported estimates of the incidence and case-fatality ratios for all IPD in general (i.e. including pneumococcal meningitis, bacteraemia, sepsis, peritonitis) [7,13,17–19] while other studies limited the burden of disease to pneumococcal meningitis and/or bacteraemia [5,6,9,10,14]. Though the different age categories hamper easy comparisons, the incidence of IPD seems to vary greatly between countries. This could reflect differences in diagnostic practices between countries (i.e. whether blood or cerebrospinal fluid cultures are taken, especially for nonmeningitis IPD), as well as true divergence in incidence rate. In addition to showing similar differences in reported casefatality ratios, Table 3 also indicates that several studies lack transparency for (some of) these vital input estimates. The cost burden preventable by PCV7 is mainly influenced by the following factors: (1) The probability of acquiring a disease stage and the severity of that stage. For infections with Streptococcus pneumoniae, the costs of meningitis are high per case of meningitis, but the population-wide risk of pneumo-

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

1363

coccal meningitis is limited (see Table 3). Acute OM, on the other hand, is highly frequent, but generally mild and not costly to treat, whereas pneumococcal bacteremia and pneumococcal pneumonia are situated between these two extremes (the former much more like meningitis, the latter more like OM). (2) The PCV7’s protective efficacy, which is high for IPD, and relatively low for all-cause OM and all-cause pneumonia. Virtually all studies used the effectiveness of PCV7 versus pneumonia confirmed by X-ray and versus all-cause OM, as more specific incidence data on pneumonia and OM, caused specifically by pneumococcus (against which PCV7 is much more efficacious) was not available. Clearly, depending on the scope of the available incidence data on non-IPD, the appropriately corresponding measure of efficacy needs to be applied (and this is in this case usually not the most specific measure of protective efficacy from the clinical trials). Note that this may explain the divergence in assumed incidence of CAP (as noted above), though many papers lack transparency to verify this. (3) When considering herd effects, the costs of pneumonia, and the proportion of pneumonia caused by pneumococcus amongst adults >50 years of age is influential, because at those ages pneumonia is relatively frequent and severe.

Direct medical costs for an uncomplicated OM case were reported to be D 11 [6], D 48 [9], D 66 [8], D 76 [15] and D 103 [18]. In other studies no such estimates of OM costs per case (eg, instead “D 3 per OM consult” [14]), or only estimates with costs of complications included were reported (D 213 [10]), or it was not clearly stated what the estimates comprised (D 114 [11], D 216 [13]). Clearly, more transparency is needed for an easy interpretation and comparison of such reported estimates.

These three factors make it relevant to produce reliable estimates for each disease stage, in terms of costs per case and incidence. The estimated pneumocococcal meningitis incidence varies substantially between the various countries under review (see Table 3). This is also the case for the estimated treatment costs of pneumococcal meningitis, which is given in Table 4. Treating an uncomplicated case of acute bacterial meningitis (without sequelae) costs between D 4,886 in Switzerland [18] to D 8,344 in Canada [9]. In many studies the costs of long-term sequelae (neurological and hearing impairments) after pneumococcal meningitis were taken into account. The substantial costs of special education or residential care needs were explicitly considered in six studies [6–8,13,16,18]. The assumed probability of sequelae, given a case of pneumococcal meningitis, ranged from 4% [6,13], over 6–9.7% [11], to 16% [7,14,16] for neurological sequelae (“mental retardation”, “brain damage”, “seizures”, “focal neurological damage”) and from 4% [6], over 15.5% [11] 14–19% [14], 30% [7,8,16], to 32%[17] for hearing losses. In the other studies under review, these sequelae were not considered, or not reported in a specific (eg in Claes and Graf von der Schulenburg [8], 20% of pneumococcal meningitis would lead to “multiple sequelae” (excluding hearing losses)) or a transparent way, and hence cannot be reported here. For non-hospital costs, most studies resorted to expert opinion-based estimates. Since differing levels of severity were defined for OM, and not all studies produced an overall estimate, the costs per case of OM are difficult to compare.

3.3.1. Universal infants’ vaccination From a societal perspective, results for universal infant vaccination with PCV7 varied from total net savings [5,8] to over D 100,000 per discounted QALY gained [17] (cf. Table 5). It is noteworthy that both studies finding net savings seem flawed: Claes et al. [8] double count indirect costs of mortality in both the numerator and denominator of the incremental cost-effectiveness ratio (ICER), and Asensi et al. [5] made no adjustments for unemployment while assigning productivity costs to both morbidity and mortality. From the perspective of the payer (i.e. when only direct medical costs are considered), results ranged from D 19,279 [18] per undiscounted QALY gained to D 142,033 [9] per discounted QALY gained and D 6,394 [12] per undiscounted LYG to over D 200,000 [17] per discounted LY gained. Note that ratios with undiscounted health outcomes implicitly assume that a policy maker is indifferent between a life-year saved today, and a life-year saved in the future (be it 5, 30 or 200 years from now). In Finland [17], Australia [7], England & Wales [14], the Netherlands [6] and Canada [9], infant PCV7 vaccination was reported to be not as cost-effective as dialysis and breast cancer screening (Finland); breast or cervical cancer screening (Australia), meningococcal C or influenza vaccination (the Netherlands), adult pneumococcal (with the 23-valent polysaccharide vaccine) or varicella vaccination (Canada). By contrast, universal PCV7 infant vaccination programs were reported to have acceptable cost-effectiveness ratios in Spain [5], Canada [10,19] Germany [8] (more specifically

3.2.3. Vaccination costs Where it has been licensed, PCV7 is the most expensive pediatric vaccine to date, with the assumed price of a single dose ranging from D 40[6] to D 69[7] in this review (Table 1). In Ruedin et al, the cost of a vaccination course (vaccine price plus administration costs) with one dose of a hypothetical PCV9-MenC vaccine was set at D 80 [13]. Vaccine administration costs varied from D 1.6 per dose in Finland [17] (where the PCV7 can jointly be administered with other vaccines) to D 15 per dose in England and Wales [11,12,14] (where the full cost of a nurse consultation is charged). Three studies assumed the new PCV7 vaccination program would not require any additional administration costs [10,15,19]. 3.3. Differences in results

1364

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

when considering a societal viewpoint in these countries) and in Switzerland [18]. In Spain [5] and in Canada [10,19] universal PCV7 infant vaccination was reported to show cost-effectiveness ratios comparable with other local wellaccepted interventions (such as hepatitis A and hepatitis B vaccination of children in Spain). McIntosh et al. [11] stated that the ICER of universal PCV7 infant vaccination lies at the upper limit of acceptable ICERs in the UK. Finally, Ruedin et al. [13] concluded that, in Switzerland, universal infant vaccination with 3 doses of the combined PCV9-MenC vaccine is more cost-effective than vaccination with MenC alone. It is interesting to note that this divergence between the studies’ conclusions also occurs for studies pertaining to the same country. In Canada, Lebel et al. [10] and Moore et al. [19] were more favourable to universal infant PCV7 vaccination than De Wals et al. [9]. Without getting into much detail, it seems that this divergence between the studies’ conclusions mainly stems from different estimates of the costs of each disease stage and/or of the incidence rate of IPD (Lebel et al.’s [10] model resulted in a much higher number of pneumococcal meningitis cases per birth cohort than De Wals et al. [9], whereas Moore et al.’s [19] estimate of IPD incidence for British Columbia appeared to be higher than Canada in general). A more detailed discussion of these Canadian studies can be found in Beutels [3]. Also, for England and Wales, McIntosh et al. [11] reported more favourable results than Melegaro and Edmunds [14], presumably because McIntosh et al. [11] estimated a higher burden of disease in the absence of vaccination than Melegaro and Edmunds [14]. It is also of note that McIntosh et al. are employed by Wyeth, whereas Melegaro and Edmunds are employed by the Health Protection Agency, a British public body. Irrespective of such independency issues, these variations in assumptions and results highlight the need for consistent definitions in assessments of both clinical and economic input data [3].

3.3.2. Catch-up vaccination Three studies assessed the efficiency of supplementing the introduction of universal infant PCV vaccination with single dose catch-up PCV vaccination [5,9,18]. In these scenarios healthy children up to 24 [5,18] or 60 months [18] of age are all caught up with at the start of the program, or alternatively, more gradually, through the vaccination of infants (7–12 months), toddlers (12–18 months) and children (24–48 months), until the first cohort of vaccinated infants has reached their age (i.e. 7, 12 or 24 months, respectively). In two studies, the ICERs for catch-up vaccination were found to be less favourable than for universal infants’ vaccination alone (irrespective of the age of the catch-up group) [5,9]. In Ess et al. [18], additional catch-up vaccination of all infants <24 months when universal infant vaccination starts, was found to be more attractive than universal vaccination of younger infants alone (D 16,483 versus D 19,279 per undiscounted QALY gained).

3.3.3. Sensitivity analyses In view of the uncertainty inherent to many of the input parameters, thorough sensitivity analyses should be made. The current standard for such analyses is to assign (as much as possible, data driven) distributions to all parameters and sample from all these distributions simultaneously to obtain uncertainty intervals related to both incremental costs and incremental effects over a range of options. On the basis of these, cost-effectiveness acceptability curves (CEACs) could be constructed. Such CEACs were only presented in Melegaro and Edmunds [14]. The other studies usually confined sensitivity analyses to univariate, bivariate or threshold analysis. When ignoring herd immunity, the most influential parameters included often OM incidence and costs, as well as case-fatality ratios for IPD (see also discussion of input under Section 3.2). With inclusion of herd immunity effects induced by infant vaccination, the estimated pneumonia incidence in adults is also a very influential input. Irrespective of herd immunity, vaccination costs are highly influential, despite the fact that the plausible range for them is narrower than for other uncertain, but less influential, input parameters.

4. Discussion Given the lack of consensus between the studies’ results for both the perspectives of the payer and society, it is difficult to draw solid conclusions about the cost-effectiveness of universal infant vaccination with a PCV7. A key assumption determining the economic attractiveness of universal PCV7 vaccination in each country is the cost of vaccination. Indeed, given its large budget impact, the cost of the vaccine is one of the most influential variables determining its cost-effectiveness, and of course the easiest to adjust as a condition for reimbursement by governments. Some studies have not failed to report that a substantial reduction in the cost of the vaccine could bring the ICERs within an acceptable range. For example, without accounting for herd immunity, the price of the PCV7 would have to be reduced to a third of its current value for the ICER to be lower than the £30,000 (D 44,000) per QALY gained threshold for acceptable interventions in England and Wales [14]. Note that a previous review of PCV7 economic evaluations published before August 2002 concluded, in line with the current findings, that the attractiveness of PCV7 vaccination hinges on the potential for price reductions and the willingness for decision makers to adopt a societal perspective (i.e. to either include indirect costs or use QALYs as outcome measures in the denominator) [4]. Many studies have also pointed out the difficulty of assessing with much precision the efficiency of PCV7 vaccination due to uncertainties related to the current burden of pneumococcal disease, the duration of vaccine protection and the long-term effects of vaccination on the epidemiology of pneumococcal disease. Indeed, PCVs have the potential to decrease nasopharyngeal carriage, thereby reducing trans-

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

mission of pneumococci by herd immunity in the pediatric and adult population. Such indirect protection after infant pneumococcal conjugate vaccination has been observed in Northern California [26] and the USA in general [27,28] and this could potentially also have an impact on the costeffectiveness of the current pneumococcal polysaccharide vaccination strategies in (elderly) adults. However, after the introduction of PCV, non-vaccine serotypes may well replace vaccine serotypes, leading to a smaller reduction in disease burden over time. This raises various questions related to the potential long-term population effects. The possible effect of herd immunity and of complete substitution of vaccine serotype with non-vaccine ones was investigated by Melegaro and Edmunds [14]. As expected, their base-case cost-utility ratio (D 87,913 per discounted QALY gained) decreased dramatically when indirect protection to the unvaccinated was included (D 7,352 per discounted QALY gained). But the inclusion of complete serotype replacement increased substantially this ratio (D 39,132 per discounted QALY gained, still substantially lower than the base-case ratio). As was noted by these authors, there is as yet little quantified information on the magnitude of herd immunity and serotype replacement effects (i.e. the extent to which replacement occurs, and the severity of disease caused by non-vaccine types as opposed to vaccine types). Furthermore, serotype replacement may be tackled in future by appropriately timed conjugate (or polysaccharide) vaccines covering more than seven serotypes. Recently, at least three additional studies have explicitly considered the impact of herd immunity, and concluded that PCV7 vaccination in childhood would be cost-effective. McIntosh et al. [12] concluded that 4 doses of PCV7 at D 6,394 per life-year gained would be “highly cost-effective” in the UK. Beutels et al. [29] found (at A$14,645 per DALY averted in the baseline) three doses of PCV7 to be of comparable or better cost-effectiveness as routine meningococcal C conjugate vaccination in Australia (introduced in 2003), for varying time spans and assumptions regarding herd immunity and serotype replacement. Furthermore, at the time of finalising this review, a new economic analysis [30] was published for the US. This analysis included the herd effects that have been observed for 5 years into the US program, and compared it with another US analysis [31] that had been made before the program was introduced, and which did not account for herd effects (as these were too speculative to predict at the time, and indeed it was not known yet whether PCV7 would be able to induce herd immunity effects at all). It is therefore not surprising that the updated analysis reaches a much more favourable conclusion than the previous analysis (in the base-case $US7500 (∼D 6000) per life-year gained versus $US80,000–110,000 (∼D 64000–87000) per life-year gained) [30,31]. Additionally, we became aware of at least one other study, which appeared in a journal that is not indexed in the databases we searched (see methods section). Ford et al investigated the impact of herd immunity in Canada, and found this to be very beneficial [32].

1365

In many European countries and in the United States, the recommended vaccination schedule for PCV7 requires four doses per vaccinee. The infant immunization programme in Denmark, Finland, Italy, Norway, and Sweden is based on primary vaccinations at 3 and 5 months, and a third dose at 11 or 12 months of age, whereas in most other European countries the primary immunization schedule consists of 4 doses (three <1y, and one ≥1y). The use of a simplified schedule including 3 doses of PCV7 (a 2 + 1 schedule) administered concomitantly with the routine primary infant vaccinations at age 3, 5 and 11 or 12 months has been shown to confer an equivalent antibody level for any of the vaccine serotypes compared to the 4-dose PCV7 scheme [33,34]. These findings are valid for both pre-term and fullterm infants, and confirm earlier results demonstrating that the immune response induced by PCV7 using the reduced schedule is no different from that induced by the 4-dose schedule. Additionally, Goldblatt et al.’s trial [35] was specifically designed to make a direct comparison between a 2 + 1 and 3 + 1 schedule, using PCV9 of the same company that currently markets PCV7. There were no significant differences in immunogenicity levels for the serotypes contained in PCV7, both before and after boosting at age 12 months [35]. Kayhty et al. [34] showed that the administration of 2 doses of PCV7 induced a satisfactory antibody response, except for the serotypes 6B and 23F. However, at month 13, after the booster dose, the pneumococcal antibody concentrations were comparable with those observed with the 4-dose schedule [34]. Moreover, the important increase of antibody concentration after the administration of the third dose in the reduced schedule, suggests that 2 doses of PCV7 may induce a sufficient immunological memory. Additional data from the United States show that a remarkable decline in IPD among young children is seen despite vaccine shortages and with only a minority of children having received a fourth dose of PCV7 vaccine [26]. A post-licensure case-control study also showed that both the 2 + 1 and 3 + 1 schedule are highly effective (with no significant differences) in preventing IPD [36]. These findings provide important information for PCV7 vaccine introduction in countries routinely using 3 doses in the infant immunization schedule and could lead to substantial cost reductions (in terms of vaccine costs, vaccine supply and administration) at no apparent loss in effectiveness. An extensive recent overview of the immunogenicity, efficacy and safety of PCVs, is available in a separate paper [37]. Clearly, the incremental cost-effectiveness of a 3 + 1 schedule versus a 2 + 1 schedule (and not just both schedules versus doing nothing), should be considered if both schedules are possible in a particular setting. In Belgium a recent economic evaluation has elaborated this issue at length, and estimates a 2 + 1 PCV vaccination schedule to produce attractive ICERs from a payers’ and a societal perspective, particularly at vaccination costs of EUR 50 per dose or less. In comparison to a 2 + 1 schedule, a 3 + 1 schedule, however, was found to have a very unattractive ICER, under various scenarios of additional direct and indirect protection offered by this fourth inserted

1366

P. Beutels et al. / Vaccine 25 (2007) 1355–1367

dose [1]. Furthermore, a 3-dose program is more acceptable to both vaccinators and parents when faced with a crowded immunization schedule for infants (which is still expanding as more childhood vaccines are likely to be added in the near future). Another option is to replace the final dose of PCV7 by a dose of pneumococcal polysaccharide vaccine. Pneumococcal polysaccharide vaccines cover more serotypes (including the seven serotypes from PCV7) and are currently much cheaper than the PCV7 vaccine. Their ability to evoke long lasting immunity as a booster dose after priming with PCV remains an important topic of research. The divergences in studies’ conclusions are showing the difficulties with obtaining reliable burden of disease data, as well as the constantly changing insights about the effectiveness of this vaccine. However, the following is emerging in various countries: with a 3-dose schedule and taking observed herd immunity effects into account, childhood PCV7 vaccination is likely to be judged relatively cost-effective to the health care payer and potentially even cost-saving to society. Nonetheless, the long-term net effects on antimicrobial use and resistance, serotype replacement and cross reactivity need to be monitored, to verify that they remain beneficial to the overall cost-effectiveness of the program. It is encouraging that recent evidence from the US supports this [26,27,30,36,38–40]. The economic studies published hitherto, nearly always ignored the herd immunity impact and the possibility of administering 2 or 3 instead of 4 doses, and many studies ignored quality of life losses from pneumococcal disease. These aspects were all found to be very influential, and have now been quantified in some way. Therefore, it seems that the interest of many of the studies we reviewed lies predominantly in the input data that were used, rather than the results that were produced. Clearly future analyses of PCVs should explicitly take these three aspects into account. Furthermore, the upcoming higher valent PCVs (eg, PCV8, PCV10, PCV13, noted “PCV7+” hereafter) should be analysed in terms of the incremental direct (and total morbidity) costs per additional QALY gained of a range of options of use of PCV7+ versus PCV7 [1,41]. The aim is then to estimate the additional value for money offered by PCV7+ versus PCV7 at each dose, considering the locally evolving serotype distribution of pneumococci. Assuming that PCV7+ will be marketed at a higher price than PCV7, this could imply that the best use of these vaccines might be to supplement 1 or 2 doses of PCV7 at age 2–4 months with a dose of PCV7+ at a later age. At the same time, the potential role for PPV23 as a childhood vaccine should not be ignored in such analyses. While new safety and effectiveness information of these newer vaccines and schedules is collected from pre- and post-licensure studies, simulation models could further explore the incremental cost-effectiveness of their best use. That is after all what models are for to help understand by exploring. Some of these explorations may be convincing, but only when their

data and methods are transparent, can we fully confront the confusion over their differences.

Acknowledgements This study was commissioned by Belgium’s Federal Knowledge Centre for Health Care (KCE), contract number 2005-25-2, and benefited from discussions as part of POLYMOD, a European Commission project funded within the Sixth Framework Programme, contract number: SSP22-CT2004-502084. We are grateful to four anonymous referees for constructive comments. References [1] Beutels P, Van Damme P, Oosterhuis-Kafeja F. Effects and costs of universal infant pneumococcal conjugate vaccination in Belgium. Antwerp University, Federal Knowledge Centre Health Care, Brussels; 2006, 132 pp., http://kce.fgov.be/index nl.aspx?ID=0&SGREF= 5301&CREF=7195. [2] Drummond M, O’Brien B, Stoddart G, Torrance G. Methods for the economic evaluation of health care programmes. New York: Oxford University Press; 1997. [3] Beutels P. Potential conflicts of interest in vaccine economics research: a commentary with a case study of pneumococcal conjugate vaccination. Vaccine 2004;22(25–26):3312–22. [4] De Graeve D, Beutels P. Economic aspects of pneumococcal pneumonia: a review of the literature. Pharmacoeconomics 2004;22(11): 719–40. [5] Asensi F, De Jose M, Lorente M, Moraga F, Ciuryla V, Arikian S, et al. A pharmacoeconomic evaluation of seven-valent pneumococcal conjugate vaccine in Spain. Value Health 2004;7(1):36–51. [6] Bos JM, Rumke H, Welte R, Postma MJ. Epidemiologic impact and cost-effectiveness of universal infant vaccination with a 7-valent conjugated pneumococcal vaccine in the Netherlands. Clin Ther 2003;25(10):2614–30. [7] Butler JR, McIntyre P, MacIntyre CR, Gilmour R, Howarth AL, Sander B. The cost-effectiveness of pneumococcal conjugate vaccination in Australia. Vaccine 2004;22(9–10):1138–49. [8] Claes C, Graf von der Schulenburg JM. Cost effectiveness of pneumococcal vaccination for infants and children with the conjugate vaccine PnC-7 in Germany. Pharmacoeconomics 2003;21(8):587–600. [9] De Wals P, Petit G, Erickson LJ, Guay M, Tam T, Law B, et al. Benefits and costs of immunization of children with pneumococcal conjugate vaccine in Canada. Vaccine 2003;21(25–26):3757–64. [10] Lebel MH, Kellner JD, Ford-Jones EL, Hvidsten K, Wang EC, Ciuryla V, et al. A pharmacoeconomic evaluation of 7-valent pneumococcal conjugate vaccine in Canada. Clin Infect Dis 2003;36(3):259–68. [11] McIntosh ED, Conway P, Willingham J, Lloyd A. The cost-burden of paediatric pneumococcal disease in the UK and the potential costeffectiveness of prevention using 7-valent pneumococcal conjugate vaccine. Vaccine 2003;21(19–20):2564–72. [12] McIntosh ED, Conway P, Willingham J, Hollingsworth R, Lloyd A. Pneumococcal pneumonia in the UK—how herd immunity affects the cost-effectiveness of 7-valent pneumococcal conjugate vaccine (PCV). Vaccine 2005;23(14):1739–45. [13] Ruedin HJ, Ess S, Zimmermann HP, Szucs T. Invasive meningococcal and pneumococcal disease in Switzerland: cost-utility analysis of different vaccine strategies. Vaccine 2003;21(27–30):4145–52. [14] Melegaro A, Edmunds WJ. Cost-effectiveness analysis of pneumococcal conjugate vaccination in England and Wales. Vaccine 2004;22(31–32):4203–14.

P. Beutels et al. / Vaccine 25 (2007) 1355–1367 [15] Marchetti M, Colombo GL. Cost-effectiveness of universal pneumococcal vaccination for infants in Italy. Vaccine 2005;23(37):4565–76. [16] Navas E, Salleras L, Gisbert R, Dominguez A, Timoner E, Ibanez D, et al. Cost–benefit and cost-effectiveness of the incorporation of the pneumococcal 7-valent conjugated vaccine in the routine vaccination schedule of Catalonia (Spain). Vaccine 2005;23(17–18):2342–8. [17] Petit G, De Wals P, Law B, Tam T, Erickson L, Guay M, et al. Economic evaluation of pneumococcal conjugate vaccination in Finland. Scand J Infect Dis 2005;37(11):821–32. [18] Ess SM, Schaad UB, Gervaix A, Pinosch S, Szucs TD. Costeffectiveness of a pneumococcal conjugate immunisation program for infants in Switzerland. Vaccine 2003;21(23):3273–81. [19] Moore D, Bigham M, Patrick D. Modelling the costs and effects of a universal infant immunization program using conjugated pneumococcal vaccine in British Columbia. Can Commun Dis Rep 2003;29(11):97–104. [20] McIntosh ED. Cost-effectiveness studies of pneumococcal conjugate vaccines. Expert Rev Vaccines 2004;3(4):433–42. [21] Whitney CG, Farley MM, Hadler J, Harrison LH, Bennett NM, Lynfield R, et al. Decline in invasive pneumococcal disease after the introduction of protein-polysaccharide conjugate vaccine. New Engl J Med 2003;348(18):1737–46. [22] Black S, Shinefield H, Fireman B, Lewis E, Ray P, Hansen JR, et al. Efficacy, safety and immunogenicity of heptavalent pneumococcal conjugate vaccine in children. Northern California Kaiser Permanente Vaccine Study Center Group. Pediatr Infect Dis J 2000;19(3):187–95. [23] Black SB, Shinefield HR, Ling S, Hansen J, Fireman B, Spring D, et al. Effectiveness of heptavalent pneumococcal conjugate vaccine in children younger than 5 years of age for prevention of pneumonia. Pediatr Infect Dis J 2002;21(9):810–5. [24] Eskola J, Kipli T, Palmu A, Jokinen J, Haapakoski J, Herva E, et al. Efficacy of a pneumococcal conjugate vaccine against acute otitis media. N Engl J Med 2001;344(6):403–9. [25] Beutels P. Economic evaluations of hepatitis B immunization: a global review of recent studies (1994–2000). Health Econ 2001;10(8):751–74. [26] Black S, Shinefield H, Baxter R, Austrian R, Bracken L, Hansen J, et al. Postlicensure surveillance for pneumococcal invasive disease after use of heptavalent pneumococcal conjugate vaccine in Northern California Kaiser Permanente. Pediatr Infect Dis J 2004;23(6):485–9. [27] Poehling KA, Talbot TR, Griffin MR, Craig AS, Whitney CG, Zell E, et al. Invasive pneumococcal disease among infants before and after introduction of pneumococcal conjugate vaccine. JAMA 2006;295(14):1668–74. [28] Flannery B, Heffernan RT, Harrison LH, Ray SM, Reingold AL, Hadler J, et al. Changes in invasive pneumococcal disease among HIV-infected adults living in the era of childhood pneumococcal immunization. JAMA 2006;144(1):1–9. [29] Beutels P, McIntyre P, MacIntyre C, Trotter C. Proceeding of the 5th World Congress of the International Health Economics Association, Barcelona, Spain. 2005.

1367

[30] Ray GT, Whitney CG, Fireman BH, Ciuryla V, Black SB. Costeffectiveness of pneumococcal conjugate vaccine: evidence from the first 5 years of use in the united states incorporating herd effects. Pediatr Infect Dis J 2006;25(6):494–501. [31] Lieu TA, Ray GT, Black SB, Butler JC, Klein JO, Breiman RF, et al. Projected cost effectiveness of pneumococcal vaccination of healthy infants and young children. JAMA 2004;803(1):1460–8. [32] Ford M, Grace E, Wang E. The clinical and economic impact of pneumococcal conjugate vaccine associated herd immunity in Canada. J Med Econ 2004;7:85–92. [33] Esposito S, Pugni L, Bosis S, Proto A, Cesati L, Bianchi C, et al. Immunogenicity, safety and tolerability of heptavalent pneumococcal conjugate vaccine administered at 3, 5 and 11 months post-natally to pre- and full-term infants. Vaccine 2005;23(14):1703–8. [34] Kayhty H, Ahman H, Eriksson K, Sorberg M, Nilsson L. Immunogenicity and tolerability of a heptavalent pneumococcal conjugate vaccine administered at 3, 5 and 12 months of age. Pediatr Infect Dis J 2005;24(2):108–14. [35] Goldblatt D, Southern J, Ashton L, Richmond P, Burbidge P, Tasevska J, et al. Immunogenicity and boosting after a reduced number of doses of a pneumococcal conjugate vaccine in infants and toddlers. Pediatr Infect Dis J 2006;25(4):312–9. [36] Whitney CG, Pilishvili T, Farley MM, Schaffner W, Craig AS, Lynfield R, et al. Effectiveness of 7-valent pneumococcal conjugate vaccine against invasive pneumococcal disease. The Lancet 2006;368(9546):1495–502. [37] Oosterhuis-Kafeja F., Beutels P., Van Damme P. Immunogenicity, efficacy, safety and effectiveness of pneumococcal conjugate vaccines. Vaccine, submitted for publication. [38] Kyaw MH, Lynfield R, Schaffner W, Craig AS, Hadler J, Reingold A, et al. Effect of introduction of the pneumococcal conjugate vaccine on drug-resistant Streptococcus pneumoniae. N Engl J Med 2006;354(14):1455–63. [39] Stephens DS, Zughaier SM, Whitney CG, Baughman WS, Barker L, Gay K, et al. Incidence of macrolide resistance in Streptococcus pneumoniae after introduction of the pneumococcal conjugate vaccine: population-based assessment. The Lancet 2005;365(9462):855– 63. [40] Beall B, McEllistrem MC, Gertz Jr RE, Wedel S, Boxrud DJ, Gonzalez AL, et al. Pre- and postvaccination clonal compositions of invasive pneumococcal serotypes for isolates collected in the United States in 1999, 2001, and 2002. J Clin Microbiol 2006;44(3):999– + 1017. [41] Beutels P, Van Doorslaer E, Van Damme P, Hall J. Methodological issues and new developments in the economic evaluation of vaccines. Expert Rev Vaccines 2003;2(5):649–60. [42] Petit G, De Wals P, Law B, Tam T, Erickson L, Guay M, et al. Epidemiologic and economic burden of pneumococcal disease in Canadian children. Can J Infect Dis 2003;14(4):215–20.