Streptococcus pneumoniae (pneumococcus) causes both severe invasive infections and mild respiratory tract infections. In children, acute otitis media (AOM) is the most common disease caused by pneumococcus; for each culture-confirmed invasive pneumococcal disease, there are up to 1000 cases of AOM caused by pneumococcus.1–3 Although generally a mild disease, AOM disrupts the normal activities of children and families and causes large societal and health care expenditures.
In affluent countries, diagnosis and treatment of AOM requires lots of health care resources, and AOM is the most common indication for antimicrobial use in infants and young children.4 Furthermore, recurrent AOM, prolonged otitis media with effusion or otherwise complicated otitis is the most common indication for surgery requiring general anesthesia, that is, tympanostomy tube (or grommet or ventilation tube) placement (TTP), in infants and young children.5
In clinical trials, pneumococcal conjugate vaccines (PCV) have been shown to reduce AOM, especially complex AOM.3,6,7 The Finnish Invasive Pneumococcal (FinIP) vaccine trial showed that the 10-valent PCV (PCV10, GlaxoSmithKline, Rixensart, Belgium) reduced outpatient purchases of antimicrobials recommended for treatment of AOM by 8% (95% confidence interval [CI]: 1–14) and TTP by 13% (95% CI: −2 to 26).8,9
PCV10 was introduced into the Finnish National Vaccination Programme (NVP) in September 2010 using a 2 + 1 schedule. We evaluated the real-life public health impact of the PCV10 introduction with high vaccination coverage against antimicrobial use and TTP in vaccine-eligible children.
MATERIALS AND METHODS
We designed a retrospective observational register-based follow-up study in which we compared study cohorts before and after the introduction of the NVP.
National Vaccination Program
PCV10 was selected on the basis of a public request for tender and was introduced into the Finnish national vaccination program in September 2010 as a 3-dose schedule at 3, 5 and 12 months (2 + 1). No catch-up vaccinations were available for older children. The national vaccination program vaccinations are administered in local municipal well-baby clinics free of charge. There was no prior use of PCV7 in the NVP except for small groups at increased risk of pneumococcal disease. Private use of PCV7 before the PCV10 introduction was estimated below 2% based on national sales data.10 On the other hand, over 30,000 children received PCV10 in the nationwide cluster-randomized PCV10 trial (FinIP) during 2009–2010,11 which was approximately 20% of the corresponding birth cohorts. After the introduction to the NVP, the PCV10 coverage in the 2012 birth cohort was estimated at 92% for the full 3-dose vaccination series.12
The target cohort eligible for PCV in the NVP included children born from June 1, 2010 through September 30, 2014, irrespective of whether they had received the vaccine or not. No individual vaccination data were collected. We compared the target cohort to a season and age-matched reference cohort from calendar years 2004 to 2008 before national vaccination program (Fig., Supplemental Digital Content 1, http://links.lww.com/INF/C860). Calendar period January 2009 to August 2010 was excluded from the cohort analysis because of the FinIP trial.11
The study population for this nationwide study was defined using the Finnish Population Information System, which contains information of each permanent resident in Finland, including unique personal identity code. Permanent residents in Finland at the beginning of 2014 were used to construct the cohorts. Mortality, immigration and emigration were assumed similar in both cohorts. To verify the appropriateness of constructing the cohorts retrospectively, the ratio of exposure time in the 2 cohorts was compared with the aggregate-level birth cohort data from 2004 to 2014, available from Statistics Finland and found to be identical.
Case Definitions and Data Collection
All outcome information for this study was collected from national health registers and linked individually to the study cohorts using the personal identity code.
Outpatient antimicrobial purchases were collected through the administrative benefits register of the Social Insurance Institution of Finland (KELA). Use of antimicrobials in hospital settings was not available. All permanent residents in Finland are insured by the Finnish national health insurance. This includes reimbursements for costs of reimbursable prescribed medicines, including antimicrobials for treatment or prevention of infections. Antimicrobial drugs are available in Finland only through a physician’s prescription in registered pharmacies, who submit the reimbursement claims of purchased prescriptions for medicines automatically online to the Social Insurance Institution of Finland. The register data include the Anatomical Therapeutic Chemical (ATC) Classification System code and generic name of the drug, the quantity in absolute terms and defined daily doses, pharmaceutical form and concentration, date and costs of the drug purchased and the Finnish personal identity code of the patient. The actual indication for which the antimicrobial was prescribed is not systematically recorded in the register. However, some prescriptions include the indication when it can be extracted from the physician’s prescription text including dosing and other patient instructions.
Low-cost antimicrobial purchases were not reimbursed before 2006. Therefore, because of the absence of full data on antimicrobial purchases in the benefits register before 2006, both the antimicrobial prescriptions and the follow-up years 2004–2005 from the reference cohort and, respectively, 2010–2011 from the target cohort were excluded from analyses for antimicrobial prescriptions (Fig., Supplemental Digital Content 1, http://links.lww.com/INF/C860).
We assessed the relative reduction in outpatient purchases of antimicrobials recommended for treatment of AOM by the national current care guidelines13:
- Amoxicillin ATC J01CA04,
- Amoxicillin with enzyme inhibitor J01CR02,
- Cefuroxime J01DC02,
- Cefaclor J01DC04,
- Clarithromycin J01FA09, and
- Azithromycin J01FA10.
Although recommended in the guideline, penicillin (J01CE02) and sulfadiazine/trimethoprim (J01EE02) were not included in the analysis because data were unavailable because of discontinuation of reimbursement (and subsequent discontinuation of the administrative data collection) for the pediatric oral mixtures during 2013–2014. The prescription data on indication were not available for all prescriptions and thus, were not used in the case definition.
Tympanostomy Tube Placements
We collected data on TTPs through the Care register for health care at the National Institute for Health and Welfare (THL), which is a national hospital-based discharge register for inpatient and outpatient hospital visits and admissions. Private health care service costs are partly reimbursed for the patients by the national health insurance. To ensure complete collection of procedures performed in private practices, we also collected reimbursements for TTPs from the benefits register data from the Social Insurance Institution of Finland. The procedures were identified with the Nordic Medico-Statistical Committee (NOMESCO) code for TTP (DCA20). The Finnish personal identity code and date of the procedure (with the following day included) were used to identify and remove duplicate reports from the 2 registers. TTP was defined as a TTP in at least 1 ear.
Incidence rates were calculated by dividing the number of all outcome episodes during the follow-up in question with the corresponding person-time denominator data using annual birth cohort sizes available from Statistics Finland. Absolute rate reduction was calculated as the incidence rate difference between the 2 cohorts (target−reference).
To account for and investigate the recurrence of outcomes, event history analysis was applied. Individual event histories in the corresponding study cohorts were constructed with age as the underlying time scale. The children entered the follow-up at 3 months of age and were followed until end of year 2008 for the reference cohort and end of year 2014 for the target cohort (age at end 3–54 months of age). Because of potential dependence between repeated outcome occurrences on the same individual, generalized Cox regression was used for estimating the hazard ratio between the study cohorts (target versus reference). Andersen and Gill14 model was used to estimate the population-averaged effect. To investigate the vaccine impact on recurrent antimicrobial treatments, every individual’s prescriptions were ranked according to the sequence of prescription (as first, second, third, etc.), and the follow-up time was defined from 3 months of age until the nth outcome or end of follow-up. Marginal Cox regression15 was used to estimate the effect for every ranked endpoint—for example, comparing the hazard of having a seventh prescription in the reference and target cohorts.
Relative rate reduction (expressed as percent) was calculated as (1−hazard ratio)×100%. To account for dependence between repeated outcome occurrences, robust sandwich variance estimator16 was used to calculate the CIs in both versions of the generalized Cox regression models. Statistical software program R version 3.4.0 was used for the analysis.
As part of its statutory tasks, the National Institute for Health and Welfare (THL) is obliged to monitor the effectiveness and safety of vaccines used. The study plan was approved by the THL institutional review board (May 23, 2013). Permissions to use the register data for research were obtained from the relevant register controllers at THL (Dnro THL/1090/6.02.00/2013) and KELA (Kela 67/522/2013).
The incidence of outpatient purchases of antimicrobials with reimbursement showed yearly fluctuation, but decreased markedly after the PCV10 introduction (Fig. 1). Approximately half of the prescriptions since 2011 included a text field providing the indication for the antimicrobial treatment. In the target cohort during 2012–2014, otitis media was included as an indication in 84% of prescriptions, pneumonia 1.3%, other respiratory tract infection in 10%, skin infections in 3.5% and urinary tract infections in 2.5%. Some prescriptions had more than 1 indication. These data were not available for the reference cohort.
The rate of outpatient purchases of antimicrobials recommended for treatment of AOM was 1.09 per person-year in the reference cohort and 0.89 per person-year in the PCV10 target cohort. Thus, the relative rate reduction was 17.5% (95% CI: 17.0–18.1) and the absolute rate reduction 0.20 per person-year in the target cohort compared with the reference cohort. Table 1 shows the reduction by class of antimicrobial showing highest relative reduction for macrolides.
The incidence of purchases was highest during the second year of life (Fig., Supplemental Digital Content 2, http://links.lww.com/INF/C861). The reduction observed in the target cohort began early and persisted through to over 45 months of age.
The number of antimicrobial purchases per subject ranged from 0 to up to 79 (median 2) during the follow-up time of maximum of 36 months in the reference cohort. The reduction in the target cohort was especially prominent for recurrent purchases: for example, the relative reduction for the 10th purchase was 39% (95% CI: 36–41, Fig. 2).
Tympanostomy Tube Placements
The incidence of TTP done in the private health care (55% of all TTP) showed a slight increasing trend until 2007, whereas the incidence in public hospitals (45% of all TTP) was stable. From 2012 through 2014, there was a decrease (Fig. 3) in TTP procedures both in private and public health care.
The rate of any TTP was 5.41/100 person-years in the reference cohort and 4.56/100 person-years in the target cohort. The relative rate reduction was 14.8% (95% CI: 13.1–16.5) and the absolute rate reduction 0.86/100 person-years (Table 2). The procedures peaked just after 12 months of age, and the incidences were lower in the target cohort in all age groups except the oldest age group 45 months or above (Fig., Supplemental Digital Content 3, http://links.lww.com/INF/C862).
We demonstrated that the reductions in otitis media-related outcomes during NVP were similar to those previously shown in a clinical trial setting.8,9 Despite small relative reductions, the absolute reductions were large because of the high incidences of the outcomes, highlighting the public health and economic impact of these outcomes. Furthermore, the relative reductions were substantial in reducing the number of children with multiple antimicrobial purchases by more than one-third.
These results are concordant, yet slightly larger and more precise than our earlier estimates from the FinIP trial.8,9 During NVPs with high vaccination uptake, the overall impact may be higher by the development of the indirect effects, that is, less vaccine-type circulation in the population to cause disease, adding to the direct effect of vaccination. Reductions in antimicrobial use and TTP have been observed previously after PCV717–21 and PCV13,22,23 but we are not aware of published reports for PCV10, except for a reduction in hospital-based surveillance of otitis media from Iceland24 and an early report from the Netherlands with no demonstration of impact in young infants.25
The FinIP trial vaccinations may have affected the development of the indirect effect in the target cohort because we have demonstrated the effect of PCV10 vaccination on vaccine-type carriage in unvaccinated siblings of vaccinated children.26 However, we expect this indirect effect to be minor as only some 20% of the birth cohorts of January 2008 through May 2010 were vaccinated compared with the 92% for age cohorts since June 2010.
The reported estimates need to be interpreted with caution. First, the figures on the yearly incidences after the introduction suggest progressive declines compatible with the increase in the cumulative vaccination coverage in the pediatric population and the development of indirect effects. As we reported the reduction for the whole post-introduction period, our estimates should be conservative.
Second, the outcome of antimicrobial purchases is an unspecific surrogate for pneumococcal vaccine-type otitis media as otitis has multiple etiologies and antimicrobials are prescribed for many indications. If we assume that of the total antimicrobial consumption (1.09 per person-year in the reference cohort), 84% of the antimicrobials were used for treatment of AOM,8 up to 50% of AOM are caused by pneumococci and 77% of these caused by vaccine-type or vaccine-related serotypes 6A and 19A,3 and the vaccine effectiveness 54%,3 then the expected absolute reduction for vaccine-type AOM would be (1.09 × 0.84 × 0.5 × 0.77 × 0.54=0.19) which is nearly identical to the observed estimate (0.20 per person-year). However, as the serotype replacement dilutes the net impact, the reduction caused by the vaccine types should be higher. This would be achieved with the development of the indirect protection during NVP on top of the direct protection estimated in clinical trials, that is, vaccine effectiveness against vaccine-type AOM higher than 54%.
And finally, any before–after design, also ours, is prone to many sources of bias. Any changes in coding practices, access to care, diagnosis or treatment guidelines or in other interventions will affect the results, not to forget natural secular trends of various respiratory tract pathogens. There were major administrative coding practice changes regarding the antimicrobial purchase data complicating the analyses. The reimbursement since 2006 included all antimicrobials, also those with low cost. Additional changes occurred 2012–2014 as reimbursement ceased for some antimicrobials. These changes are easily seen as artifacts in Figure 1 for ATC codes J01CE02 (phenoxymethylpenicillin) and J01EE02 (sulfadiazine and trimethoprim). Restriction of the comparison to 2006 onwards and exclusion of selected antimicrobial types from the analyses at least partly corrected these problems.
Private health insurance coverage for young children has also increased from 36% in 2009 to 41% in 2014.27 Availability of private insurance coverage and subsequent easier access to care might be associated with lower threshold for private office visits, antimicrobial consumption and tympanostomy tube surgery with potential to bias our impact estimates downward. Indeed, the PCV10 impact was lower on private compared with public tympanostomy tube surgery.
The national guideline for treatment of AOM was updated in January 2010,13 that is, just before the PCV introduction. However, compared with the previous version published in 2004, the guideline did not change dramatically; as a rule, antimicrobial treatment is still recommended, if the diagnosis is adequately confirmed. Also, watchful waiting was mentioned already in 2004 version,28 and the guidance remained practically unchanged stating watchful waiting as an option in mild disease.
Antimicrobial stewardship activities may impact the antimicrobial consumption and bias our results. However, these activities have been ongoing for decades, and we did not observe any major trends before PCV introduction (Fig. 1) in the incidence of overall antimicrobial consumption. At the population level, the overall consumption of antimicrobials in Finland has remained quite stable between 1997 and 2011, after which progressive decline is seen.29 On the contrary, major trends were observed in different antimicrobial classes (Fig., Supplemental Digital Content 4, http://links.lww.com/INF/C863), especially the reduction in the use of macrolides as a result of resistance problems detected in early 2000s30 with continued reduction up to 2014. This reduction has been counterbalanced by increased use of β-lactams. Therefore, the analysis by class of antimicrobial may be biased because of these trends.
Influenza vaccination for children 6 to 35 months of age started in fall 2007. However, the coverage has remained low, estimated at 40% in 2007 through 2009, but considerably lower during the PCV10 era, estimated at below 20%. Regarding respiratory tract infections, the average number of notifications in the National Infectious Diseases Register was higher during the PCV10 era compared with pre-PCV era, with potential confounding effect toward conservative estimation.31
An additional limitation of our study was that we were not able to evaluate otitis media directly or otitis media diagnoses through the registers but we collected data on surrogate measures using antimicrobial purchases and TTPs. An outpatient care register for primary health care visits has been established in Finland, but it covers only the public sector and was implemented in 2011. Thus, it is not suitable for this kind of evaluation because of the lack of pre-vaccination baseline data. On the other hand, the resource-use outcomes are highly relevant from the public health perspective.
Despite the limitations, probably the best guarantee for validity of the current results is the concordance with earlier clinical trial results with similar outcomes.8,9 Another strength and novelty of the current work was the construction of individual event histories, which allowed us to investigate the risk on recurrent episodes using individually linked data. Indeed, we demonstrated higher reductions in children with multiple antimicrobial purchases. It is highly important that the children most prone to complicated diseases benefit from the vaccinations.
Although otitis media is not a life-threatening disease, its frequent occurrence makes it a public health problem. Our current results lend support to the public health value of PCV introduction; for each case of culture-confirmed invasive pneumococcal disease, more than 400, and, for each meningitis, more than 10,000 outcomes related to otitis media are prevented.32 Furthermore, the reduction in antimicrobial consumption helps combat the antimicrobial resistance.
1. Klemets P, Lyytikäinen O, Ruutu P, et alTrends and geographical variation in invasive pneumococcal infections in Finland. Scand J Infect Dis. 2008;40:621–628.
2. Kilpi T, Herva E, Kaijalainen T, et alBacteriology of acute otitis media in a cohort of Finnish children followed for the first two years of life. Pediatr Infect Dis J. 2001;20:654–662.
3. Eskola J, Kilpi T, Palmu A, et alFinnish Otitis Media Study Group. Efficacy of a pneumococcal conjugate vaccine against acute otitis media. N Engl J Med. 2001;344:403–409.
4. Grijalva CG, Nuorti JP, Griffin MRAntibiotic prescription rates for acute respiratory tract infections in US ambulatory settings. JAMA. 2009;302:758–766.
5. Cullen KA, Hall MJ, Golosinskiy AAmbulatory surgery in the United States, 2006. Natl Health Stat Report. 2009;11:1–25.
6. Fireman B, Black SB, Shinefield HR, et alImpact of the pneumococcal conjugate vaccine on otitis media. Pediatr Infect Dis J. 2003;22:10–16.
7. Prymula R, Peeters P, Chrobok V, et alPneumococcal capsular polysaccharides conjugated to protein D provide protection against otitis media caused by both Streptococcus pneumoniae
and nontypable Haemophilus influenzae: a randomized double blind efficacy study. Lancet. 2006;367:740–748.
8. Palmu AA, Jokinen J, Nieminen H, et alEffect of pneumococcal Haemophilus influenzae protein D conjugate vaccine (PHiD-CV10) on outpatient antimicrobial purchases: a double-blind, cluster randomised phase 3-4 trial. Lancet Infect Dis. 2014;14:205–212.
9. Palmu AA, Jokinen J, Nieminen H, et alEffectiveness of the ten-valent pneumococcal conjugate vaccine against tympanostomy tube placements in a cluster-randomized trial. Pediatr Infect Dis J. 2015;34:1230–1235.
11. Palmu AA, Jokinen J, Borys D, et alEffectiveness of the ten-valent pneumococcal Haemophilus influenzae protein D conjugate vaccine (PHiD-CV10) against invasive pneumococcal disease: a cluster randomised trial. Lancet. 2013;381:214–222.
13. Working Group Appointed by the Finnish Medical Society Duodecim, the Finnish Paediatric Society, the Finnish Otolaryngological Society and the Finnish Association for General Practice. Otitis media (acute). Current Care guideline. Finnish Medical Society Duodecim, Helsinki 2010. Available at: www.kaypahoito.fi
. Accessed February 20, 2017.
14. Andersen PK, Gill RDCox’s regression model for counting processes: a large sample study. Ann Stat. 1982;10:1100–1120.
15. Wei LJ, Lin DY, Weissfeld LRegression analysis of multivariate incomplete failure time data by modeling marginal distributions. J Am Stat Assoc. 1989;84:1065–1073.
16. Lin DY, Wei LJThe robust inference for the Cox proportional hazards model. J Am Stat Assoc. 1989;82:1075–1078.
17. Grijalva CG, Poehling KA, Nuorti JP, et alNational impact of universal childhood immunization with pneumococcal conjugate vaccine on outpatient medical care visits in the United States. Pediatrics. 2006;118:865–873.
18. Palmu AA, Verho J, Jokinen J, et alThe seven-valent pneumococcal conjugate vaccine reduces tympanostomy tube placement in children. Pediatr Infect Dis J. 2004;23:732–738.
19. Sarasoja I, Jokinen J, Lahdenkari M, et alLong-term effect of pneumococcal conjugate vaccines on tympanostomy tube placements. Pediatr Infect Dis J. 2013;32:517–520.
20. Poehling KA, Szilagyi PG, Grijalva CG, et alReduction of frequent otitis media and pressure-equalizing tube insertions in children after introduction of pneumococcal conjugate vaccine. Pediatrics. 2007;119:707–715.
21. Jardine A, Menzies RI, Deeks SL, et alThe impact of pneumococcal conjugate vaccine on rates of myringotomy with ventilation tube insertion in Australia. Pediatr Infect Dis J. 2009;28:761–765.
22. Marom T, Tan A, Wilkinson GS, et alTrends in otitis media-related health care use in the United States, 2001-2011. JAMA Pediatr. 2014;168:68–75.
23. Lau WC, Murray M, El-Turki A, et alImpact of pneumococcal conjugate vaccines on childhood otitis media in the United Kingdom. Vaccine. 2015;33:5072–5079.
24. Sigurdsson S, Kristinsson KG, Erlendsdóttir H, et alDecreased incidence of respiratory infections in children after vaccination with ten-valent pneumococcal vaccine. Pediatr Infect Dis J. 2015;34:1385–1390.
25. Gefenaite G, Bijlsma MJ, Bos HJ, Hak EDid introduction of pneumococcal vaccines in the Netherlands decrease the need for respiratory antibiotics in children? Analysis of 2002 to 2013 data. Euro Surveill. 2014;19;pii:20948.
26. Palmu AA, Toropainen M, Kaijalainen T, et alDirect and indirect effectiveness of the 10-valent pneumococcal conjugate vaccine against carriage in a cluster randomized trial. Pediatr Infect Dis J. 2017;36:1193–1200.
28. Käypä hoito- toimitus. Äkillinen välikorvatulehdus. Duodecim. 2004;120:2129–2130.
30. Bergman M, Huikko S, Pihlajamäki M, et alFinnish Study Group for Antimicrobial Resistance (FiRe Network). Effect of macrolide consumption on erythromycin resistance in Streptococcus pyogenes
in Finland in 1997-2001. Clin Infect Dis. 2004;38:1251–1256.
32. Jokinen J, Rinta-Kokko H, Siira L, et alImpact of ten-valent pneumococcal conjugate vaccination on invasive pneumococcal disease in Finnish children – a population-based cohort study. Plos One. 2015;10:e0120290.