Canada Communicable Disease Report

October 2008

Volume 34

Number 10

Monthly Report

Economic analysis of a public program for routine seven valent pneumococcal conjugate vaccine (PCV-7) in infancy, Alberta

A Chuck, PhD (1), P Jacobs, PhD (1,2), T Nguyen, MPH (3), A Hanrahan, MN (4), J Loewen, BN (4), L Mashinter, BScN, RN (4), A Ohinmaa, PhD (1,5), J Zhang, MPH, MSc (6), W Vaudry, MD, FRCPC (7), JD Kellner, MSc, MD, FRCPC (8)

1. Institute of Health Economics, Edmonton, Alberta, Canada
2. Department of Medicine, University of Alberta, Edmonton, Alberta, Canada
3. Alberta Health and Wellness, Edmonton, Alberta, Canada
4. Communicable Diseases Control, Capital Health, Edmonton, Alberta, Canada
5. School of Public Health, University of Alberta, Edmonton, Alberta, Canada
6. Alberta Cancer Board
7. Department of Pediatrics, Stollery Children's Hospital, University of Alberta, Edmonton, Alberta, Canada
8. Department of Pediatrics, University of Calgary and Alberta Children's Hospital, Calgary Alberta, Canada

Abstract

Two 6-month birth cohorts of infants; one born before and one after the introduction of a universal infant immunization program with PCV-7 in Alberta in 2002 were studied using electronic database analysis. Health care utilization hospital discharges, physician visits and ambulatory services) and costs for diagnoses potentially and certainly related to Streptococcus pneumoniae infection were compared. A reduction in utilization and costs was observed in the post vaccine birth cohort amounting to $75 per child.

Introduction

Streptococcus pneumoniae (pneumococcus) is an important human pathogen, particularly in infants, the elderly and those with particular risk factors. It causes a broad range of diseases from those which are minor but very frequent e.g., otitis media, to those which are less common but more severe e.g., pneumonia, sepsis and meningitis. A polyvalent (23- valent) polysaccharide vaccine (PPV-23) has been available for prevention of disease in older children and adults for decades. However, it has limited effectiveness in young children < 2 years of age, which is the group that has the highest incidence of pneumococcal disease^(1-3). A 7-valent conjugate vaccine (PCV-7) was approved in Canada in 2000 and recommended for routine use in children in Canada by the National Advisory Committee for Immunization (NACI) in 2002⁽³⁾. In 2002, PCV-7 was introduced in a publicly funded routine immunization program throughout the province of Alberta. All children in the Capital Health region (approximately 10,000 births annually) were offered PCV-7 as part of the universal childhood immunization program as of September, 2002.

The decision to publicly provide PCV-7 in Canada was based on a large scale efficacy study conducted in the United States;(4) this trial found that vaccine efficacy was 97% against vaccine-serotype invasive infections and that the overall reduction in invasive disease, regardless of serotype, was 89% in vaccine recipients. Using these results two Canadian groups conducted economic models of the net costs and cost effectiveness of the PCV-7 vaccine, if introduced in Canada. Although the two studies had very different results, they were based on comparable similar clinical studies: the differences were due to different model assumptions(5). For example, for the two studies used different epidemiological assumptions regarding three more serious diseases (pneumonia, bacteremia and meningitis). De Wals et al.(6) assumed a disease incidence of one-half that of Lebel et al.(7) Despite these differences, the net cost to the publicly funded health care system per child (including direct costs of disease and vaccines) were very close: $229 in De Wals et al.(6) and $222 in Lebel et al.(7)

Economic modeling is a hypothetical exercise based on assumptions, and the value of the model as a predictive device depends on how close the assumptions are borne out under actual circumstances. In this study, we have conducted an observational analysis of the introduction of PCV-7 in the Capital Health Region of Alberta. We interpolate our 2-year results to a 9-year basis using results from a modeling study⁽⁶⁾.

Methods

Setting and Administration

In Alberta's routine childhood immunization program, specified vaccines are provided at no charge to all infants, preschoolers and school-aged children following schedules and guidelines established by Alberta Health and Wellness (AHW)⁽⁸⁾. All routine universal immunizations are administered and data recorded by public health. The immunization records therefore reflect a comprehensive and accurate recording of immunizations conducted in the Capital Health Region. In the Capital Health region, vaccine delivery is provided by community health nurses (CHN) who have received additional training in immunization. Parents of infants and preschoolers can schedule immunizations during daytime hours, evenings and weekends at a public health clinic or on special request at their residence. A delayed monitoring system is in place to ensure follow-up of children found to have fallen behind in their immunizations. This includes phone calls and reminder letters to parents of children whose immunizations are delayed. Immunizations are also administered by CHNs in hospitals across the region to accommodate hospitalized children due for immunization. Only a small number of vaccinations (< 0.5%) are performed in hospitals. The regional immunization program also collaborates with physicians to reach at-risk populations, but the vaccinations are administered and recorded by the CHNs.

Vaccine schedule

In Canada, the recommended schedule for PCV-7 for children up to 2 years of age is as follows. The primary series is given at 2, 4 and 6 months, with a booster at 12 to 15 months⁽³⁾. However, in Alberta, the booster is given at 18 months and immunogenicity data supports this modified schedule⁽⁹⁾. The schedule is modified for children < 5 years of age whose vaccinations are started later, and who require fewer doses⁽¹⁰⁾ to complete the series.

Population

All mothers of newborns are contacted by a Capital Health CHN on day of discharge from hospital and 95% are visited within 24 to 36 hours. Records of children who move into the region are initiated at the time of first contact with the public health system and information regarding vaccine history is collected. Data are entered into Caseworks, the regional immunization database. The study was a comparative pre/post analysis of the change in health service utilization and costs before and after the introduction of PCV-7 program in the Capital Health region. The pre-PCV-7 population included all children, newborns and new residents within the CH region of Alberta (1997) boundaries with records in Caseworks, and dates of birth between 1 July, 1997 and 31 December, 1997 (n = 5,027). By 31 December, 1999 all children in the cohort were 2 years of age; for each child, data on utilization was obtained for the 2 years after birth (see below).

The immediate post-PCV-7 population included all children, newborns and new residents in the CH region of Alberta (2002 boundaries) with records in Caseworks and dates of birth between 1 July and 31 December, 2002 (n = 4,988). We collected data between 1 July, 2002 and 31 December, 2004 to obtain utilization data on each child in the cohort for the first 2 years of life.

The post-PCV-7 study population was categorized into two study groups, completely immunized (CI) and partially (PI) or not immunized (NI). A child was categorized as CI if he/ she received the recommended number of PCV-7 doses with adequate spacing between doses. If a child had started a series of vaccines but did not complete the recommended number of doses or had inadequate spacing between doses, that child was categorized as PI. A child was categorized as NI if he/she had not received any doses of the specific vaccine before his/ her second birthday. All children in the pre PCV-7 population were categorized as NI (note that some 'partially immunized' children can be fully protected).

Databases

Immunization data was obtained from Caseworks and includes information on date of birth, type of vaccine administered and date of immunization. Specifically, any infant born in Alberta who lives in a postal cod of the Capital Health Region will be captured in Caseworks including infants and children identified as having recently moved into the region. Utilization data was obtained from three Alberta provincial health ministry databases. The hospital inpatient (morbidity) database maintained by AHW provided information related to hospital utilization data. Information contained in the database included diagnoses (International Classification of Diseases 9 or 10 (ICD-9 or 10-CA) codes), procedures/interventions, length of stay and costing weights for specific hospital. Each hospital discharge is assigned to a Case Mix Group (CMG) using the acute-care patient grouping methodology developed by the Canadian Institute for Health Information (CIHI)⁽¹¹⁾. The CMG methodology aggregates acute-care patients with similar clinical and resource utilization characteristics. Each CMG has a corresponding resource intensity weight (RIW) which is a measure of hospital services and resources used compared to the average impatient and is further refined by age group and co-morbidities. The cost per weighted case for 2003 for the entire province was obtained from CIHI. This cost was applied to each case, according to its RIW.

The Alberta Health Care Insurance Plan database provided information related to billing services and ministry payments to physicians for medically insured services in Alberta. Information contained in the database included diagnosis (ICD-9 or ICD-10), service type/category/location, and amounts paid.

The Ambulatory Care Classification System database is AHW data base which provided information related to emergency visits and day services. Information contained in the database included diagnosis (ICD-9 or ICD-10), procedures, emergency room visits and community rehabilitation program services occurring in publicly funded facilities. Related procedures are grouped together and an average per visit cost is assigned to each procedural group, based on provincial cost averages for ambulatory services⁽¹²⁾.

Immunization records in Caseworks were merged with the Population Registry, the Hospital Inpatient Database, the Alberta Health Care Insurance Plan database and the Ambulatory Care Classification system database using anonymous identification codes.

Case Definition and Attribution

Utilization was reported for visits that were related to S. pneumoniae. Two infectious disease pediatric physicians independently reviewed all ICD-9 and ICD-10 codes and identified diagnoses that were either definitively or possibly linked to pneumococcal bacteria. Conditions that explicitly included the terms "pneumococcal" or "S. pneumoniae" e.g., "pneumococcal meningitis" were considered to be definitely linked. Syndromes known to be often or sometimes caused by S. pneumoniae, but with the etiologic agent not explicitly named e.g., "mastoiditis and related conditions" were considered to be possibly linked. Listings of definitively and possibly linked diagnoses are shown in Appendix 1 for ICD-9 and ICD-10 diagnosis systems.

In the hospital inpatient database there are up to 16 diagnoses, any of which can be identified as definitively or possibly linked to pneumococcal infection; three diagnoses identified in the Alberta Health Insurance Plan database; and in the ambulatory care database there are up to five diagnoses which can be identified a diagnosis and corresponding procedure were included in the utilization attribution if, firstly, the diagnosis was definitively linked with pneumococcal infections and secondly, if the diagnosis was possibly linked with pneumococcal infections. Therefore based on the diagnosis listed, each service date and procedure was categorized as not linked, definitively linked and possibly linked with pneumococcal infections. Diagnoses and services not linked to pneumococcal infections were assigned a cost of $0.

For both the pre-PCV-7 and post-PCV-7 populations, hospital, physician, and ambulatory care visits and costs were compared. Costs associated with the pre PCV-7 population were adjusted to reflect 2003 Canadian dollars using the Canadian consumer price index⁽¹³⁾. The number of hospital, physician and outpatient visits/services was also compared between the two populations.

Statistical Analysis

Independent samples t-tests were conducted to determine differences in direct health services costs between the pre- PCV-7 population and the post PCV-7 population. These groups were compared in terms of visits and costs. In addition, we conducted a sub-group analysis comparing the post-CI and post-PI+NI groups to determine whether there was "herd immunity" for non-immunized children in the group. Statistical significance was defined as p < 0.05.

Economic Analysis

The economic measure that we used was the cost of the vaccine minus the difference in costs of utilization of related services between the pre- and post-groups. Health services costs from years 3 through 9 were projected based on De Wals et al.⁽⁶⁾ estimates of utilization over time (63% of costs occurring in first 2 years) and then discounted at 5% (Discounted/Current costs = 58%). Vaccine costs were $67.50 per dose, for four doses, which was the public price at the time [Source: Wyeth Canada]. Vaccine administration costs were not included, as the vaccine was given in conjunction with other vaccines within the Alberta Immunization Schedule.

Results

There were 5,027 pre-PCV-7 and 4,988 post-PCV-7 children. Of children in the post PCV-7 population 84% were completely immunized against pneumococcal infection (Table 1). As previously mentioned, there are multiple service/procedures spanning 2 years listed for each child in the dataset. Table 1 shows the number of possible or definitive distinct diagnoses per child by immunization status. The pre-PCV-7 population had a higher number of possible diagnoses and definitive diagnoses than either the CI or NI children in the post PCV-7 population.

Visits were reported for those cases that were related to S. pneumoniae. The pre PCV-7 population had more hospital, physician and outpatient visits than the post PCV-7 group (Table 2). The cost of utilization per child is shown in Table 3. Definitive diagnoses are infrequent, and the total cost per child was $4.69 in the pre-stage and $1.26 in the post-stage (no significant difference). The possible related diagnoses were more frequent, and the total cost was $199 in the prestage and $124 in the post-stage. The difference between the children in the pre- and post-stage was $75.

Table 1. Services linked to pneumoccocal infections

Table 2. Comparison in health service utilization per child between pre-PCV-7 and post-PCV-7 populations

Table 3. Comparisons in health services costs per child between pre-PCV-7 and post-PCV-7 populations

When the 2 year data were extrapolated to 9 years, and discounted, the total cost for all related conditions was $183 in the pre-group and $114 in the post-group. When we add the vaccine costs to the post-group ($270 per child) the postvaccine group costs $201 more per child than the pre-vaccine group.

Within the post PCV-7 group, there were no statistically significant differences in health service utilization between children CI and children PI+NI. There were also no statistically significant differences in hospital, physician or outpatient costs between children CI and children PI+NI.

Discussion

Alberta introduced universal coverage for S. pneumoniae by providing the 7-valent pneumococcal conjugate vaccine in 2002. In this paper we estimated the difference in the hospital (inpatient and outpatient) and physician services' utilization and costs for pre- and immediately post-vaccination periods for those services that were related to the treatment of S. pneumoniae-related diagnoses. Our analysis was based on data which recorded services received by the child population of the Capital Health region of Alberta. The 2-year difference in cost per child, pre- and post-vaccination was $75 (the excess cost of the pre-immunization group over and above that of the post-group). There was no difference between the post-vaccination groups that were completely and incompletely (NI and PI) vaccinated.

Our (observational) results can be compared to those of two previous Canadian economic models. Lebel et al. (2003)⁽⁷⁾ estimated differences in utilization costs of $107 per person and De Wals et al. (2003)⁽⁶⁾ estimated cost differences to be $40, both over a 9-year period. Our results showed differences $201 over 9 years, including vaccine costs.

We should also note that recent economic models have focused on a herd immunity of PCV-7 in both children eligible to receive PCV-7 as well as older children and adults who are not eligible to receive it⁽¹⁴⁾. If we measure this effect by the net costs of the NI and PI compared to children CI, then we note that the difference in health service utilization and costs for this group was not statistically significant. This is consistent with the herd-immunity effect observed elsewhere. Other studies have also identified such an effect with S. pneumoniae-related vaccines⁽¹⁵⁾. Herd immunity extends beyond children in the current age group, to other children and adults. We did not consider these groups, and so our estimate of the net cost of vaccine is high.

We based our study on a regional immunization database and a linked provincial utilization data base. One of the key benefits of this approach is that it allows us to identify actual, rather than hypothetical, trends. All economic studies until now have been based on modeling efforts, which are conjectural, and thus a lower quality of analysis. In addition, our analysis covered the entire population of the region, which allows us to identify the net difference in costs for everyone, including the under- or non-vaccinated. We should note that because we used actual data, for the first 2 years, our data base did not include quality of life indicators; one of the benefits of modeling is that virtually any variable can be included, though perhaps artificially.

There are several drawbacks to our approach. First, health services utilization that was related to S. pneumoniae was often coded as "possibly related to S. pneumoniae" and therefore may not be directly related to the vaccine. However, this concern is mitigated by the fact that we have measured differences in utilization between groups, rather than overall levels of utilization. While we cannot attribute these differences to the vaccine, we have identified an association between the vaccine and utilization. Associated with this drawback is the fact that the economic benefits of PCV-7 are driven by reductions in less – invasive diseases and notby life – threatening conditions. A second drawback to our approach is that we cannot identify more refined outcomes such as life years saved. In addition, cohort data was only available up to 24 months of age. Of course, assumed values for these variables could be incorporated into our analysis as well (as we have done in our 9-year extrapolation); however, the level of supporting evidence for outcomes would be diminished by this action. A third drawback involves the use of the 2 years before the introduction of the vaccine as a control. No other control was available to us, and we could not control for other extraneous factors such as fluctuations in the yearly epidemiology of infectious diseases. Finally, in the prepopulation, ICD-9 data was available to us; in the postsample, ICD-10 data was available. In ICD-9, code 481 includes pneumonia related to streptococcal as well as lobar pneumonia (unspecified). In ICD-10, J13 includes only streptococcal pneumonia. However, because our main analysis focused on possible and definitive diagnoses, the overall results are unaffected.

In addition, we included two vaccine groups in the postimmunization population, fully immunized and partially or non-immunized. The sample size for the non – fully immunized group was small (about 20%) and results were largely not significant. However, there might be important insights to be gained from breaking down this group further into those who received none , and those who received one and two to four doses(16). Because of the sample sizes however, this would require a larger total sample in the postvaccine group. We suggest that this would be an important follow-on study.

In summary, using observational data, PCV7 vaccine does have a positive impact on utilization and costs. However, it would be fruitful to extend the analysis to groups in the nonimmunized age categories, to determine the full economic effects of the vaccine program.

Appendix 1. List of diagnoses which are classified as definitely and possibly linked to Streptococcus pneumoniae, using ICD-9 and ICD-10 codes

References

1. Dear K, Holden J, Andrews R et al. Vaccines for preventing pneumococcal infection in adults. Cochrane Database of Systematic Reviews [4], CD000422. 2003.
2. Centre for Communicable Diseases. Preventing pneumococcal disease among infants and younger children: Recommendations of the advisory committee on immmunization practices (ACIP). 2000;RR-9:1-35.
3. National Advisory Committee on Immunization (NACI). Statement on recommended use of pneumococcal conjugate vaccine. CCDR 2002;28(ACS-2):1-32.
4. Black S, Shinefield H, Fireman B et al. Efficacy, safety and immunogenicity of heptavalent pneumococcal conjugate vaccine in children. Pediat Infect Dis J 2000;19(3):187-95.
5. Beutels P. Potential conflicts of interest in vaccine economics research: A commentary with a case study of pneumococcal conjugate vaccination. Vaccine 2004;22:3312-22.
6. De Wals P, Petit G, Erickson LJ et al. Benefits and costs of immunization of children with pneumococcal conjugate vaccine in Canada. Vaccine 2003;21:3757-64.
7. Lebel MH, Kellner JD, Ford-Jones L et al. A pharmacoeconomic evaluation of 7-Valent pneumococcal conjugate vaccine in Canada. Clin Infect Dis 2003;36(1):259-68.
8. Honish A, Loewen J, Perrin M. Injecting quality into your immunization program. Alberta RN 2002;58(4):6-8.
9. Scheifele DW, Halperin SA, Ochnia JJ et al. Immunologic considerations for the timing of the booster dose of 7-valent pneumococcal conjugate vaccine in young children. Pediat Infect Dis J 2007;26(5):387-92.
10. Public Health Agency of Canada. Canadian Immunization Guide. 7th ed. Cat No. 0-660-19392-2. Ottawa: Public Health Agency of Canada, 2006.
11. Canadian Institute for Health Information. DAD resource intensity weights and expected length of stay (ELOS) 2007. 978-1-55465-057-6. Ottawa, CIHI, 2007.
12. Alberta Health and Wellness. Health costing in Alberta annual report. Edmonton: Alberta Health and Wellness, 2005.
13. Bank of Canada. Consumer price index: 1995 to present. 2007.
14. Lloyd A, Patel N, Scott DA et al. Cost-effectiveness of heptavalent conjugate pneumococcal vaccine (Prevenar) in Germany: Considering a high-risk population and herd immunity effects. Euro J Health Econo 2008;9(1):7-15.
15. Kellner JD, Church DL, MacDonald J et al. Progress in the prevention of pneumococcal infection. CMAJ 2005;173(10):1149-51.
16. Whitney CG, Pilishvili T, Farley MM et al. Effectiveness of seven-valent pneumococcal conjugate vaccine against invasive pneumococcal disease: A matched case-control study. Lancet 2007;368:1495-502.