Vaccination, which is the second most effective way (after clean water) to save lives and promote good health , saves around 3 million lives each year worldwide ; reduces suffering; has vastly reduced the incidence of various diseases; and results in substantial cost savings. In an analysis by the Centers for Disease Control and Prevention in the US, every US$1 spent on immunization saves US$6 in direct medical costs and US$12 in indirect costs . It has also been estimated that each birth cohort vaccinated with the seven main vaccines can save $10 billion in direct medical costs and $33 billion in indirect costs .
The benefit of introducing a widespread vaccination program can only be judged on the basis of an accurate estimation of country-specific disease burden. The study reported here adapted a previously described model for the UK , but used data from 12 Istanbul hospitals  and Turkish Ministry of Health statistics  on burden of disease, resource use and costs. These were combined with vaccine efficacy estimates, adjusted to the specific pneumococcal serotype distribution in Turkey .
The projections reported herein suggest that introducing either PCV-13 or PHiD-CV would greatly reduce pneumococcal infection disease burden compared with PCV-7. Not surprisingly, both the PCV-13 and PHiD-CV vaccines were also predicted to have a greater beneficial impact on cost savings due to pneumococcal disease and AOM. These observations are consistent with those made by other authors [13, 40, 41].
The modeled projections indicate that PCV-13 is expected to offer a greater impact on IPD and pneumonia (by virtue of its greater complement of S. pneumoniae serotypes) than PHiD-CV. This translates into a saving of one additional death every year after the launch of a vaccination program (65 LYs gained over a 1-year period at a vaccine steady state). However, in terms of cost and QALYs, this relative reduction in IPD and pneumonia was grossly outweighed by the substantial reduction in AOM cases estimated with PHiD-CV. Moreover, when comparing the cost profiles of the two vaccines, the additional costs saved by PCV-13 in IPD and pneumonia (around US$37,139) were approximately 200-fold less than the additional savings in AOM estimated with PHiD-CV (around US$8.2 million). The relatively greater impact of PHiD-CV on AOM also contributed to the prediction that the vaccine would save 933 more QALYs than PCV-13 over the 1-year period. It was these substantial differences in direct cost savings and quality of life that led to PHiD-CV being identified as the most cost-effective vaccine under the modeled conditions.
The observation that AOM treatment costs drive the cost-effectiveness of PCVs in children <10 years of age is as expected, given the substantial prevalence of the disease within this age group. Previous estimates suggest that >80% of children will experience ≥1 episode of AOM by the age of 3 years . AOM is a major burden for healthcare systems, resulting in around 1.7 times as many hospitalizations as pneumonia in children aged 0–9 years in the UK; and 73 times as many GP consultations . Further reducing the incidence of AOM compared with PCV-7 was estimated in this study to decrease the number of AOM hospitalizations by 7,956 and 2,234 for PHiD-CV and PCV-13, respectively, and the number of GP visits by 267,212 and 75,019, respectively. AOM burden of disease is reflected by the sizeable annual cost of AOM treatment predicted under all three of the vaccination schedules considered in the model, ranging from US$87 million to US$99 million. One limitation of this study is that AOM incidence was not available in Turkey, therefore data from other countries were used to estimate the prevalence of AOM in Turkey [8, 18, 19]. A separate scenario was analyzed to test the base case incidence estimate of 33,000 per 100,000 person years obtained from Garcés-Sánchez et al. study . The base case incidence rate of AOM was reduced by - 50% in 0–5 and 5–9 age group down to 16,500 and 10,095 per 100,000 person respectively. In this scenario PHiD-CV was projected to dominate PCV-13 as it generates 491 more QALY and US$5.5 million cost savings.
Historically, researchers have been intrigued by the differences in efficacy of pneumococcal vaccines against clinical AOM between the POET  and FinOM  trials. However, a direct comparison between the 34% overall impact against AOM in the POET trial (11-valent pneumococcal polysaccharide conjugate vaccine conjugated to H. influenzae-derived protein D [11-Pn-PD]) versus 7% in FinOM (PCV-7) cannot be made due to the different settings and designs of these trials. Palmu et al. , former investigators of FinOM, assessed the impact of variations in case definition, design and local epidemiology between trials, and concluded that these factors only account for part of the difference. An extension of this re-analysis with the POET data set, using a standardized case definition, confirmed that the original difference in case definition was not the cause of the observed AOM efficacy difference. De Wals et al.  assessed various scenarios, adjusting for different factors. Replacement due to non-vaccine serotypes, not observed in POET, was identified as a main factor to explain the divergence. The observed reduction in AOM episodes due to NTHi in POET with the 11-Pn-PD vaccine was identified as a secondary factor to explain the divergence. Tangible differences in calculated maximal efficacy against AOM are thus consistent with those presented in independent evaluations .
The vast amount of antibiotics prescribed for childhood AOM has the potential to increase antibiotic resistance , which may increase treatment costs and decrease quality of life. The prevention of AOM by vaccination therefore has an important role in the reduction of antibiotic prescriptions . However, while reducing the number of AOM cases is likely to reduce the rate of antibiotic prescription, modeling the impact of vaccination in this context is complex and was beyond the scope of the analysis reported here.
The predictions of all simulation models are dependent upon the approximations and assumptions used to configure them. Where possible, we endeavored to use region-specific surveillance data and information from controlled clinical trials. However, assumptions were required for certain parameters in the model. In the absence of any specific efficacy data, the efficacy of the additional serotypes in PHiD-CV and PCV-13 had to be estimated using the average efficacy of the serotypes in PCV-7, but their true efficacy may lie at either extreme of the efficacy range (87% for 19F to 99.9% for 9V). However, this assumption is in line with WHO guidelines and does not appear to overtly bias the model [13, 48].
A second approximation in the model is the absence of indirect vaccine effects (e.g. herd protection and serotype replacement), which is an important limitation. PCV-7 has been shown to reduce nasopharyngeal carriage of vaccine-type S. pneumoniae serotypes in vaccinated and unvaccinated individuals [49–51]. PCVs are therefore capable of having an additional impact on the overall transmission of S. pneumoniae within the population, which would potentially translate into protection for the unvaccinated population [52–54]. Lessons from the large-scale PCV-7 vaccination program in the US indicate that vaccination can result in substantial indirect herd effects, e.g. IPD has been reported to have decreased by 15% in unvaccinated children aged <5 years , and by 29% in unvaccinated children >5 years and adults . Most deaths from pneumococcal disease occur in elderly adults , as they are more likely to have compromised immune systems. Excluding a herd effect on the elderly is therefore likely to have resulted in an underestimation of the true health gains that a PHiD-CV, PCV-7 or PCV-13 vaccination program could achieve. Indeed, indirect effects that increase the number of IPD cases averted may have a crucial impact on economic savings per QALY gained [19, 57, 58]. Opposing this effect is serotype replacement, which results in increased disease from non-vaccine serotypes . At present in the US, beneficial herd effects appear to outweigh negative serotype replacement [14, 59]. Indirect effects of the vaccines (herd effect and serotypes replacement) were not included. Since PCV-7, PHiD-CV and PCV-13 are directly compared and the potential differences in indirect effect induced by each vaccine are not known, the inclusion or exclusion of equal (not differential) indirect effect for all vaccines won’t impact the results, because the incremental differences between three vaccines would be the same with inclusion or without inclusion of indirect effect. On the other hand herd protection plays an important role in impact on cost-effectiveness ratio when the vaccines are compared to a no vaccination strategy.
Another weakness of the current analysis is that long-term complications after meningitis (neurological sequelae and severe hearing loss) were not included, due to the absence of Turkey-specific incidence and cost data. However, if available incidence data are used (neurological sequela 7% ; hearing loss 13% ), this would result in six fewer cases of long-term complications for PHiD-CV vs PCV-7 and eight fewer for PCV-13 vs PCV-7. This small number of cases is not expected to influence the outcome or conclusions of this study.
Additional assumption is related to the extrapolated incidence rates of IPD, CAP and AOM in 5–9 age group. This assumption may have a minor impact on the results and conclusions, since in 0–10 age group approximately 90% of IPD, CAP and AOM cases are occurring in children below 5 years of age.
Another important approximation is related to the modeling of the impact of the vaccines based on the IPD serotypes distribution before PCV-7 implementation. The pre PCV-7 IPD serotypes distribution was used in a base case in order to have a consistent data set of both incidence rates and serotypes distribution from the same time period of pre PCV-7 introduction before 2008. The impact of PCV-7 is projected using the model. An alternative scenario was analyzed using the latest serotypes distribution from Ceyhan et al. and estimating the impact of vaccination on costs and QALYs gained . In the scenario analysis PHiD-CV was projected to yield 759 more QALYs and cost savings of US$11.6 million when compared with PCV-13. The analysis described here assumes a 3+1 dosing schedule in order to gain maximum impact from the vaccination program. Although some countries have adopted 2+1 schedules of PCVs , the 3+1 schedule is used in Turkey. Our decision to focus on 3+1 schedules of PCVs was also based on a previous cost simulation using UK statistics, which showed that implementing a 3+1 schedule had incremental health benefits over the 2+1 schedule for PCV-7 currently in place in the UK . This four-dose program is in line with the original recommendations for vaccination with PCV-7 and is supported by data demonstrating that efficacy increases with increasing numbers of doses .
With over 70 million people, Turkey has the 17th largest population in the world and the 3rd largest in Europe . Turkey’s huge population is diverse and heterogeneous, with discrepancies in income, poverty, infrastructure, and services between those living in the east and the west, and between those in urban and rural areas . This analysis has shown that the use of PHiD-CV and PCV-13 are more cost-effective than PCV-7 (in terms of more QALYs gained and costs saved) in this upper middle income country. These results are in line with other cost-effectiveness studies of PCVs of various valencies in high (Canada ) and low (The Gambia ) income countries, suggesting that these results may be applicable to the other, economically heterogeneous, countries in the Middle East and North Africa (MENA) region, assuming similar disease incidence rates and treatment costs.