Vaccine probe studies emerged in the past 15 years and are particularly useful for pathogens for which the true burden may be hidden due to the absence of accurate laboratory testing or limited sensitivity of available diagnostics Vaccine probe studies can estimate VPDI, as well as the proportion of a syndrome caused by the pathogen, ideally, via randomized clinical trials [13] (note that while less precise, VPDI can also be estimated post licensure by evaluating changes in outcome incidence during the pre- and post-vaccine period and using time-series analysis can also be used). Probe analysis has been used for several vaccines with known efficacy to assess the total burden of disease preventable by vaccine whether disease burden is etiologically confirmed or documented clinically [14–16].

Another measure of some interest to public health discussions of vaccine utility is the calculation of syndromic etiologic fraction, defined as the vaccine effectiveness against syndromic disease divided by effectiveness against etiologically confirmed disease. For example, in a randomized controlled trial of pregnant women in Bangladesh, inactivated influenza vaccine given to mothers reported an effectiveness among young infants of 62.8% (95%CI:5.0–85.4) against etiologically confirmed disease and an effectiveness of 28.9% (95%CI: 6.9–45.7) for all clinically documented febrile respiratory infection [17]. The calculated etiologic fraction would be 28.9%/62.8% = 46%, and provides an estimate of the fraction of febrile respiratory illness due to influenza.

Similar calculations can be done for the efficacy of pneumococcal conjugated vaccine (PCV) in reducing radiologically confirmed pneumonia (defined as pleural effusion or alveolar infiltrate) in children. Bacteriologic studies from children with pneumonia [18–22] reported PCV-vaccine effectiveness against radiologically confirmed pneumonia (i.e., pneumonia with an alveolar consolidation or pleural effusion) ranging from −2% in the USA to 37% in the Gambia (Table 1). Taking the example of the Gambia, if PCV prevents 50% of cases of invasive pneumococcal disease and 37% of radiologically confirmed pneumonia, then the etiologic fraction of radiologically confirmed pneumonia due to vaccine serotype pneumococcal disease was 74%, assuming that the efficacy against invasive pneumococcal disease and radiologically confirmed pneumococcal pneumonia is similar. If, however, efficacy against severe pneumococcal pneumonia is actually lower than what it is for invasive disease, then the proportion of pneumonias due to pneumococcus would be estimated to be even higher. As for flu in Bangladesh, the vaccine probe approach has provided information beyond that available from measuring vaccine effectiveness alone.

Vaccine probe analysis may also add evidence to potential health impact of a vaccine with low efficacy (e.g. malaria) by demonstrating impact against non-specific syndromes (like fever) or outcomes (like hospitalizations, antimicrobial use or clinic visits). Another potential area for the use of vaccine probe analysis is to estimate the overall VPDI in areas where this information is lacking. This would provide the data needed for Ministries of Health and Finance to evaluate the health impact value of introducing a vaccine into an immunization program. Assumptions for a typhoid conjugate vaccine probe study with primary outcome of hospitalization with 5 days of fever showed that such trials are feasible and could overcome many current limitations in estimating overall typhoid burden [23].

Many decisions on the introduction of new vaccines into immunization programs are guided by economic analyses. The two main models used to evaluate cost-effectiveness of vaccination programs are static (decision analysis) models and dynamic models (e.g. the SIR Model). The two models are otherwise identical except that the former account only for direct protection of vaccines while the latter take account of changes in infection risk to unvaccinated persons resulting from a decreased population transmission [25]. As reported by a review of modelling approaches, the vast majority of economic evaluations of vaccines use static models, meaning that they do not take into account the indirect impact of vaccination programs [24].

The importance of incorporating the indirect effect of vaccination in the assessment of vaccine cost-effectiveness has been investigated in several studies that compared dynamic and static approaches. Comparison of the two models with regard to mass immunization of infants with varicella vaccine showed that as compared to static approach, dynamic models predicted a higher number of cases prevented, a concomitant increase in the average age at infection, and inter-epidemic periods [25]. These effects are common to many vaccine preventable diseases, particularly many childhood diseases that stimulate relatively long-term immunity. Other vaccine (disease) specific effects such as serotype replacement following PCV vaccination [26] are also important factors that could impact vaccination programs and are not fully captured by static models. Timing of vaccination is another element with major impact on cost-effectiveness of vaccination in the context of outbreaks and again static models are not appropriate for this purpose because optimizing schedules depends on total and not just direct disease prevention. The use of dynamic models also provided evidence that a cholera vaccination coverage of 50% in Bangladesh would reduce the number of cholera cases in unvaccinated subjects by 89% [27]. Thus, investment in use of dynamic models to capture indirect effects of vaccines, particularly those with high coverage rate, is crucial because they affect distribution of disease in the population and influence optimal vaccination strategy. This will lead to more appropriate decision making worldwide.

Streptococcus pneumoniae (S. pneumoniae) causes a variety of clinical diseases ranging from non-severe otitis media and sinusitis to invasive pneumococcal diseases (IPD), including bacteremia, septicemia and meningitis with high mortality risk. Nasopharangeal carriage plays an important role in transmission. Prevention of vaccine serotype pneumococcal diseases is promoted by pneumococcal conjugated vaccines (PCV) containing 7 (PCV-7; Prevnar^™), 10 (PCV10 Synflorix^™) or 13 (PCV-13; Prevnar13^™) serotypes. Estimation of the overall burden of disease preventable by vaccine faces some difficulties related to poor sensitivity of blood culture for identifying pneumococcal pneumonia (since most pneumococcal pneumonia is non-bacteremic), etiological assignment (pneumonia) and overlapping clinical presentations (otitis media). The Finnish Invasive Pneumococcal disease vaccine effectiveness trial (FinIP) is a phase III/IV probe vaccine design, cluster-randomized, double-blind trial in children <19 months that aimed at investigating vaccine effectiveness of the pneumococcal Haemophilus influenzae protein D conjugate vaccine (PHID-CV10) and absolute rate reduction against (1) laboratory-confirmed IPD and (2) clinically suspected IPD, hospital-diagnosed pneumonia and otitis media by use of diagnoses coded in hospital discharge and antimicribial purchases registries [28,29]. PCV effectiveness was the highest for laboratory-confirmed IPD and decreased substantially for less specific outcomes such as clinical pneumonia and otitis media, while the VPDI value increased in the opposite way. Indeed, the VPDI per 100,000 person-years was 75 for laboratory-confirmed IPD, 205 for non-laboratory-confirmed but clinically suspected IPD, 340 for hospital-diagnosed pneumonia, and 12,000 for any antimicrobial outpatient prescription mainly due to otitis media [28,29]. The higher rate of preventable disease among clinical rather than laboratory-confirmed diseases indicate that vaccine has greater public health value and cost-effectiveness than would be estimated from assessment of laboratory-confirmed illness alone. Where laboratory diagnostics are poor, laboratory confirmed disease rates may more substantially underestimate pneumococcal disease burden. In Finland, PCV-10 was introduced to the national immunization program in September 2010 for children born from June 2010 (2 + 1 schedule at 3, 5 and 12 months of age) without catch-up. The impact of this vaccine in Finnish children has been evaluated by a population-based, observational follow-up study of 2 cohorts (eligible and older ineligible children) [30]. The results showed significant reduction (80%; 95%CI: 72–85) in the overall rate of vaccine-serotype IPD among eligible children and a 48% (95%CI: 18–69) reduction among ineligible unvaccinated children due to herd immunity. The VPDI value for laboratory-confirmed IPD, non-laboratory-confirmed IPD and pneumonia were 50, 122 and 90 per 100,000 children-years (age 3–42 months), respectively. These estimates are lower than what has been reported by the FinIP trial and illustrate the added value of vaccine probe studies and assessment of VPDI for the evaluation of overall vaccine benefits.

Rotavirus is the leading cause of severe diarrhoea and accounts for about one-third of all diarrhoea-related hospitalizations among children <5 years of age. Two licensed oral rotavirus vaccines (Rotarix and RotaTeq) are currently used in more than 75 countries around the world. Rotavirus vaccine effectiveness against severe rotavirus gastroenteritis in high and middle income countries ranged from 84% to 98% [31–37]. By contrast, vaccine effectiveness was only moderate (51–64%) in Africa and in less developed areas of Asia [38–40]. Many potential factors such as malnutrition, presence of concomitant infections (in particular enteric infections), and environmental enteropathy could interfere with the performance of these vaccines. Yet even with lowered vaccine effectiveness, the higher disease burden in these settings might mean that vaccine has a greater public health impact in low-income countries than in high-income settings. In this case, VPDI might be a more appropriate marker of the real impact of vaccination program as it is related to the etiologically specific disease incidence. This has been demonstrated by a randomized-placebo controlled trial that showed lower vaccine effectiveness in Malawi than in South Africa (49.4% vs. 72%); while the number of prevented episodes of severe rotavirus gastroenteritis was greater in Malawi than in South Africa (67 vs. 42 per 1000 infants vaccinated per year) [38]. Moreover, VPDI value in these African countries was greater than those estimated in high and middle income countries where vaccine effectiveness was higher.

Plasmodium falciparum and other malaria species caused 198 million cases (124–284 million) and 584,000 (367,000–755,000) deaths in 2013 [41]. It is a global problem in the tropics but due to the high density of the parasite in Africa, the highest transmission rates and about 90% of mortality are on the African continent [42]. Currently, the only candidate malaria vaccine close to licensure is RTS,S/AS01 that is directed against the pre-erythrocytic stage of the parasite (i.e. before its entry into the red blood cells). Based on one of the severe malaria definitions used in the trial, vaccine efficacy during the 12 months after vaccination was 38% among infants vaccinated at 6–12 weeks of age and 47% among children vaccinated at 5–17 months of age. While efficacy was relatively low compared to other vaccines used in African national immunization programs, the high background malaria incidence meant that these efficacies equated to a reduction of 900 and 2300 severe malaria cases per 100,000 study subjects, respectively [43,44], a substantially higher severe disease burden reduction than that calculated for some routine infant immunizations [45]. No definite, protective effect against death was demonstrated in the clinical trials, possibly because almost no malaria deaths occurred in either arm due to improvements in clinical management in the trial setting.

There are other arguments backed-up by epidemiological data that suggest RTS,S may have additional benefits. Several studies have reported that malaria infection predisposes the infected individual to bacteraemia. This hypothesis has been tested in a matched case-control study of Kenyan children in which cases were those with invasive bacterial disease [46]. The authors reported a significant inverse relationship between the malaria-protective phenotype of sickle-cell trait (HbAS) and bacteraemia (odds ratio0.36; 95%CI: 0.20–0.65). Beside, longitudinal data [46] showed that a decline in bacteraemia paralleled a decrease in malaria admission [47], reinforcing this hypothesis. Malaria reduction may also impact all-cause mortality. Following extensive malaria control interventions in Bioko Island, Equatorial Guinea, under-5 mortality has been reduced by 2/3, from 152 per 1000 births (95%CI = 122–186) to 55 per 1000 (95%CI = 38–77; hazard ratio = 0.34 [95%CI = 0.23–0.49]) [48]. This study assessed non-vaccine interventions and observed impacts may not necessarily apply to vaccines. Nevertheless, it provides suggestive evidence that an effective malaria vaccine could potentially prevent cases and deaths for which malaria is a sufficient cause and many more deaths for which malaria is a necessary but not sufficient part of the causal chain. In sum, the example of RTS,S provides an important message that even a vaccine with relatively low efficacy could have huge impact on overall population morbidity and mortality, a feature that which has to be taken into consideration in addition to vaccine effectiveness.

Dengue is one of the most important emerging infectious diseases of the 21st century. Dengue epidemiology has changed in the past 40 years as the result of expanding geographic distribution of both the viruses and mosquito vectors, increased epidemic activity, the development of hyperendemicity and the emergence of severe disease [49,50]. The observed dramatic increase in the frequency and magnitude of epidemic dengue is driven by population growth, urbanization, environmental changes and modern transportation (i.e. globalization). Primary prevention through vector control has failed due to only partially effective prevention and control tools and inappropriate implementation and assessment [51]. New dengue disease burden estimates put the global at-risk population at 3.6 billion, with 390 million infections and 21,000 deaths annually, the latter being most probably underestimated [52]. Vaccine development for dengue has been hampered by several factors including the perceived need for a balanced protection against all 4 serotypes, lack of funding and limited understanding of dengue virology and pathogenesis. Currently, three dengue vaccines are in phase II and III clinical trials. One vaccine has completed phase III clinical trials, and showed efficacy against laboratory-confirmed clinical dengue (65%), hospitalized dengue (80%) and severe dengue (93%) [53]. The vaccine demonstrated variable efficacy against the four dengue serotypes (47–83%) but was highly effective (52–81%) in persons with previous exposure to dengue [53]. The majority of children in hyperendemic areas have already been infected with dengue at least once, and since infections with the third and fourth serotypes are generally mild or asymptomatic [54,55], a dengue vaccine with this efficacy profile should have a significant public health impact by preventing severe disease, decreasing health care utilization, particularly the hospitalization rate, and by having a priming effect in populations with previous dengue infection. Further study is required to confirm (i) whether the vaccine virus will induce the same level of immunity as natural infection, (ii) the role of virus strains with variable virulence and epidemic potential, (iii) the role of temporal distribution of infection with different serotypes and (iv) the paucity of knowledge about the disease severity in 3rd and 4th dengue infections and the immunity induced by the vaccine in these individuals. While this efficacy profile can help reduce the burden of disease and achieve public health objectives, successful dengue prevention and control is likely to be obtained only by using the vaccine in combination with other new innovative tools in the pipeline as well as already existing tools and strategies [56] implemented by various public health agencies. Programmatic demonstration projects may be useful to show how the full impact of these coordinated interventions can be achieved. To facilitate this new paradigm for dengue prevention and control, a new initiative, the Partnership for Dengue Control (PDC), was created in July 2013 at an international consensus conference attended by a cross-sectional representation of the global dengue and public health communities, including representatives from academic institutions, international health agencies, NGOs, and the private sector [56].