Community Research and Development Information Service - CORDIS

FP7

OPTIMUNISE Report Summary

Project ID: 261375
Funded under: FP7-HEALTH
Country: Denmark

Final Report Summary - OPTIMUNISE (Optimising the impact and cost-effectiveness of child health intervention programmes of vaccines and micronutrients in low-income countries)

Executive Summary:
Project Final Report (261375 optimunise)
Title: Optimising the impact and cost-effectiveness of child health intervention programmes of vaccines and micronutrients in low-income countries
Acronym: OPTIMUNISE
Contract/Grant agreement number: FP7-HEALTH-F3-2011-261375
Executive summary (max 1 page)
With OPTIMUNISE we have demonstrated that in low- and middle income countries, health and demographic surveillance systems (HDSS) sites can play a major role in the continuous monitoring and evaluation of health interventions. We have developed a publicly available free tool for data collection and data analysis. We have trained a number of young researchers in Guinea-Bissau, Ghana and Burkina Faso in using these tools to evaluate the overall real-life health impact of health interventions. We are also in the process of establishing other markers of a good health than mortality, which can be used as outcomes in the continued evaluation of health interventions in the future.
Building on this platform, we have corroborated many of our previous observations of non-specific effects of BCG, measles vaccine (MV), diphtheria-tetanus-pertussis (DTP) and vitamin A supplementation (VAS), helped also by the fact that WHO carried out a review, which supported many of the conclusions. We have discovered that several other interventions have NSEs, confirming the patterns we had already established: the live oral polio vaccine (OPV) has beneficial NSEs, whereas the non-live penta and the new RTS,S malaria vaccine have negative NSEs for females. Sex-differential effects were far more common than usually assumed and we confirmed the predefined hypotheses about DTP having a negative effect for female survival. Vaccines and micronutrients interacted and combination and sequence of vaccinations were very important for the effect on survival.
The potential effects of campaigns on child survival has not been assessed before but we have shown that the beneficial NSEs of OPV and the many campaigns with OPV during the last 15 years go a long way in explaining why mortality has decreased so incredibly in Africa. Similarly, we have shown that general MV campaigns also have major beneficial NSEs. There have been far more OPV campaigns than MV campaigns, so OPV campaigns may have been a major driver in explaining why two of the three HDSS sites have reached the Millennium Development Goal 4 (MDG4).
We have furthermore discovered that vaccination in the presence of pre-existing immunity, from the mother or from previous vaccinations, confers additional beneficial NSEs, thus adding to the evidence for the importance of the OPV and MV campaigns for the decline in mortality.
We did not, in a randomised controlled trial (RCT), confirm the finding from previous studies of a particularly beneficial effect of providing an early MV at 4-6 months of age. This could be due to the low power of the studies because the mortality rate was much lower than expected but we find it more likely that it could be due to the many OPV campaigns carried out during the trial, which tended to blur any effect of early MV. The RCT revealed that children in both Burkina Faso and Guinea-Bissau are susceptible to measles infection much earlier than hitherto anticipated, and an early MV could establish protective immunity and be an effective tool in the eradication of measles infection.
We are in the process of documenting how taking into account both the NSEs and the overall real-life effect of health interventions, rather than their assumed effects, can refine the cost-effectiveness analyses and lead to completely different conclusions about which interventions to prioritise. The first analyses have shown that MV campaigns are highly cost-effective and that the restrictive vial policy to reduce wastage of vaccines in multi-dose vial (BCG, MV) is not cost-effective if the effect of these vaccines on overall child survival is taken into consideration.

Project Context and Objectives:
Project Final Report (261375 optimunise)

Title: Optimising the impact and cost-effectiveness of child health intervention programmes of vaccines and micronutrients in low-income countries
Acronym: OPTIMUNISE
Contract/Grant agreement number: FP7-HEALTH-F3-2011-261375

EC contribution: 2.999.970 €
Duration: 66 months
Starting date: 01/03/2011
Final report: 31-08-2016

Project context and the main objectives (max 4 pages)
Context: the non-specific effects of interventions
Child health programmes in low-income countries are usually justified in term of their impact on child survival and how they may contribute to reaching the Millennium Development Goals. The international news and the scientific policy debate are therefore also filled with assessments indicating that particular programmes saved the lives of a certain number of children. For example, the vaccination programme is said to have saved over 20 million lives in the last two decades and the vitamin A programme over a quarter of a million lives each year (http://www.child-survival.org/childsurvival/whatiscs.php). Such assessments are mostly based on measurements of performance indicators for a particular programme, e.g. vaccination coverage, assumptions about the burden of the particular health problem (disease or deficiency), and assumptions about efficacy of the intervention, which are often based on results from small-scale pilot studies. There is no consideration of unintended positive or negative health side-effects of the common health interventions. Possible sex-differences in response to the interventions are not assessed. The possibility that health interventions may interact because they all affect the immune system is also not considered.

The approach to assess performance indicators and use these to calculate the effect on mortality based on assumed burden of disease and assumed effect of the intervention on the target disease is deemed sufficient. So far, it has been the general understanding - the paradigm - that the intervention programme has only a specific effect; for example, measles vaccine prevents measles infection and vitamin A supplementation (VAS) prevents vitamin A deficiency. Once the effect is established, for example, that VAS reduces mortality with around 25% as found in the 1980s, it is assumed that this effect is constant.

It is now becoming increasingly clear that this approach is insufficient. Observational studies and randomised trials conducted in several African and Asian countries have consistently shown that when it comes to assessment of real life effects of an intervention on mortality, the assumptions about the effect on mortality are not reliable. First, vaccines and micronutrients have beneficial or negative non-specific effects; i.e. effects which are not explained by prevention of the targeted infections or deficiencies. Second, these non-specific effects are frequently sex-differential. Third, interventions may interact producing stronger beneficial or negative non-specific effects; for example, vitamin A given together with DTP may increase mortality for girls. To the extent there are interactions between interventions, it also means that some intervention, which once reduced mortality by 25%, may now have a totally different effect because it also interacts negatively with some other intervention. The implication is that we need to continue to monitor the effect of our main interventions to continue to be sure that they have the intended overall (beneficial) effects.

Hence, there is a need to assess the real life impact of the existing child health intervention programmes in the places where they are being used, taking into account the variability in programme implementation, that vaccines and micronutrients can have non-specific sex-differential effects on mortality, and that interventions may interact. Only real-life estimates of effects can lead to valid assessment of cost-effectiveness and thereby to sound evidence-based policy making.

A prerequisite for testing the real life impact and cost-effectiveness of our child health interventions is to have individual-level data on health intervention uptake as well as on health status. Most low-income countries do not collect individual data on health interventions and morbidity and mortality. However, many countries have Health and Demographic Surveillance Systems (HDSS) sites. The INDEPTH Network is a network of such HDSS sites in Africa and Asia. These sites have typically been initiated as demographic surveillance sites, collecting data on births, deaths, and migration, but in recent years, many sites have integrated collection of health data into the routine data collection. The research from the Bandim Health Project HDSS has proved that HDSS sites can be extremely valuable tools for monitoring and testing the real life effects of health interventions.

The present project involved three HDSS sites in West Africa, the Bandim Health Project in urban and rural Guinea-Bissau, Centre Recherche en Sante deNouna, Burkina Faso, and Navrongo Health Research Centre, Ghana. Based on a few modifications of the current data collection at these sites, including the routine collection of information on all interventions in childhood, such as vaccines, micronutrient supplementation, de-worming, bed-net distribution and bed-net impregnation, given at the health centres or in campaigns, we show that these HDSS sites can provide the platform for 1) assessing the real life effect and cost-effectiveness of child health intervention programs in observational studies; 2) testing modifications of the current health intervention programs in randomised controlled trials (RCTs), and 3) testing new interventions and their potential interactions with existing interventions in RCTs. The long-term hope is that the method will be adapted by other HDSS sites in Africa and Asia.

Already implemented health interventions can generally only be evaluated in observational studies, because once a WHO-recommended intervention has been established in collaboration with international donors, it would be considered unethical to conduct a randomised controlled trial (RCT) of its “real life effect”, since it would imply withholding an intervention considered beneficial from some children. Hence, when it comes to testing the real life effect of the current child health program we must rely on observational studies for assessing impact and cost-effectiveness of health programmes. Such studies are prone to bias and the hypotheses need to be formulated in ways that minimise the risk of bias. A paper in Tropical Medicine and International Health (TMIH) formulated four testable hypotheses regarding the potential non-specific effects of the currently used diphtheria-tetanus-pertussis (DTP) and measles vaccines and their potential sex-differential effects (Fine and Smith, TMIH 2007). Using the HDSSs as a platform we aimed to test these hypotheses: first, that DTP vaccination after BCG vaccination is associated with an increase of at least 25-50% in female mortality (TMIH-hypothesis-I); second, among BCG-vaccinated children there is an increase in the female to male (F/M) mortality ratio when DTP vaccinations are given (TMIH-hypothesis-II); third, among DTP-vaccinated children there is a decrease in the F/M mortality ratio following measles vaccination (TMIH-hypothesis-III); fourth, in both boys and girls, measles vaccine (MV) given with DTP is associated with at least 50% higher mortality than that associated with MV given alone (TMIH-hypothesis-IV).

The comparison of children who get an intervention with children who did not get the intervention is likely to be affected by different forms of bias. To account for this in our analyses, we emphasised the assessment of the determinants of programme adherence - from socio-economic factors to health related determinants like nutritional status or previous history of infections - and controlled for these factors in the analyses.

In some situations it is ethically justified to test modifications of the current programmes in RCTs and obtain unbiased estimates of the effect of certain aspects of a health programme. This may be useful because it evaluates the impact of possible modifications in an unbiased way. The HDSS platform is ideal for conducting such RCTs and we aimed to utilize this possibility in the present project. Specifically, we wanted to test a recent finding from an RCT in Guinea-Bissau: providing early measles vaccine at 4.5 and 9 months of age compared with the recommended measles vaccine at 9 months of age reduced overall mortality between 4.5 and 36 months of age by 30% in the per-protocol analysis and 22% in the intention-to-treat analysis, the effects being even stronger if the children had not received neonatal VAS (NVAS).

The HDSS sites already collect data on mortality, and mortality was the main outcome in the present evaluation. Fortunately, mortality has declined in Africa in the last decade and it may become increasingly difficult to measure the overall impact on mortality of existing and new interventions. We therefore also aimed to identify the best relevant comparable outcome parameters, e.g. hospitalisation or growth, which correlate with child mortality/survival, and which can be used to assess the overall health impact of interventions in future assessments.

Overall objective
The OPTIMUNISE project had as it overall objective to develop the capacity of existing Health and Demographic Surveillance Systems (HDSS) in Africa to monitor the real life effects, including mortality, morbidity and growth, of our routine child health intervention programs in order to promote evidencebased policy which leads to the much needed reduction in the still unacceptable high child mortality. Hence, the context was that the work should contribute to reducing child mortality, to facilitate reaching the Millennium Development Goal 4 (MDG4) of reducing under-five mortality with 2/3 between 1990 and 2015.

The specific objectives were defined as
1) To develop HDSS sites into a platform for monitoring real life effects of current child health programmes with the aim to assess the specific and non-specific, sex-differential effects of child health intervention programmes, as well as intervention interactions. This implies collection of routine data on all health interventions in childhood (WP1-3,WP7). A manual should be developed during the first year and will be updated based on the experience with implementing routine surveillance. The manual should be available through the INDEPTH Network.

Based on this platform:

2) To measure the health impact and cost-effectiveness of the existing major child health programmes with vaccines and vitamin A controlling for known determinants of programme compliance (WP4, WP7). At least five papers were planned to be submitted during the 3rd to the 6th year of the project.

3) To evaluate in an RCT a specific modification of the current child health programme: To provide an additional measles vaccine at age 4.5 months, in addition to the recommended measles vaccine at age 9 months (WP5-6, WP8). Separate analyses from the two sites and a pooled analysis were to be available in the last year of the study.

4) To assess the relevance of different health outcomes parameters for the evaluation of the real life effect of child health programmes (WP9). This analysis was to be available in the last year of the study.

The proposal was related to a call for assessment of the “Impact and cost-effectiveness of existing major health programmes”. First, it emphasised collaborative health research within Africa to ensure better health for Africans. Second, it emphasised capacity building in Africa. Third, it directly addressed the quest for reliable evidence on the overall impact and cost-effectiveness of major health programmes and more impact research, thereby contributing directly to evidence-based policy making. Fourth, it would result in a new methodology for measuring the real life effect on health of our child health programmes. Fifth, along with an emphasis on the most important outcome mortality, it would identify other reliable comparable outcome parameters. Sixth, as a key feature it would help to identify possible unintended positive and negative health side-effects beyond the targeted condition, with potentially very large implications for child health. Seventh, the testing of a specific modification in programme could potentially contribute significantly to improving child survival in low-income countries. Eight, the developed methodology and the identified outcome parameters would be easy to adapt to a wider range of health programs and to other geographical settings.

Consortium
The partners in the consortium were:
Participant no. Participant organisation name Country
1 Statens Serum Institut Denmark
2 Bandim Health Project Guinea-Bissau
3 Centre de Recherche en Sante de Nouna Burkina Faso
4 Navrongo Health Research Centre Ghana
5 Heidelberg University Germany
6 Medical Research Council, The Gambia United Kingdom
7 INDEPTH Network Ghana
8 Rijksinstituut voor Volksgezondheid en Milieu (RIVM) The Netherlands
Note: Partner 6 withdrew from the consortium because they no longer had the measles antibody assay available.
For references see Final Report Foreground

Project Results:
Description of the main S & T results/foregrounds (max 25 pages)
This report summarises what has happened within each of the objectives defined in the application.

I. The platform for monitoring real life effects on morbidity and mortality of health interventions (WP1-3)
Objective: To develop HDSS sites into a platform for monitoring real life effects and cost-effectiveness of our current child health programmes with the aim to assess the overall effect of the interventions on mortality, including the specific and non-specific and sex-differential effects of the interventions and their potential interactions. This implies collection of routine data on all health interventions and campaigns in infancy and childhood, as well as collection of data on health care use and health care cost.
At the three HDSS sites - Nouna, Burkina Faso, Navrongo, Ghana, and Bandim, Guinea-Bissau – we developed a method for routine monitoring of health interventions, growth, morbidity and survival (WP1-3,WP7, Deliverable 1). The monitoring system is built on regular home visits to all families in the study areas at least every 4 months. At these home visits we collect information on pregnancies, deliveries, information on interventions from vaccination cards (immunisations, vitamin A supplementation (VAS) and campaigns), on growth (arm-circumference), on morbidity (hospital admissions, consultations), and on survival.
The method is described on the web-page of the INDEPTH Network (Deliverable 1):
http://indepth-network.org/projects/optimunise/optimunise-key-activities
The work was done by Partners 2-4 and with help from Partners 1,5, and 7.
Work packages (WP) 1-3: Building the platform
Over the 66 months the OPTIMUNISE project has lasted we have established and consolidated this platform. For WP1 the work was done by Partners 2 and 1, for WP2 by Partners 3 and 5, and for WP3 by Partner 4.
Since the beginning of the OPTIMUNISE project, 90,609 children below the age of 3 year have been under surveillance in the three sites: The average annual birth cohort for the three sites was 11,800. A total of 34,791 children below 3 years of age in Ghana, 21,866 in Burkina Faso and 33,952 in Guinea-Bissau. We saw vaccination card for 84% (75,888) of these children: 30,105 (87%) in Ghana; 15,314 (70%) in Burkina Faso; and 30,469 (90%) in Guinea-Bissau.

While the core of the vaccination programme has been common in the three sites, many new vaccines have been introduced during the OPTIMUNISE project. The core of the vaccination programme has been BCG+Oral Polio Vaccine (OPV) at birth; Pentavalent + OPV at 6, 10, 14 weeks of age in Ghana and Guinea-Bissau but 2, 3, 4 months in Burkina Faso; Measles and Yellow fever at 9 months of age in all three countries.

During the course of the project Pneumococcal vaccine was introduced at 6, 10, 14 w since May
2012 in Ghana, at 2, 3, 4 months: since November 2013 in Burkina Faso, and at 6, 10, 14 w: since June 2015 in Guinea-Bissau.
Rubella vaccine was introduced from 9 months of age in Ghana since September 2013 and since May 2015 in Burkina Faso.

Inactivated polio vaccine was introduced at 14 w since July 2016 in Guinea-Bissau but not yet in the other countries.

A second dose of measles vaccine was introduced at 18 months since May 2012 in Ghana and at 15 months since October 2014 in Burkina Faso. It is not yet introduced in Guinea-Bissau.

In addition to the routine vaccinations, the OPTIMUNISE sites have also been collecting data on all the child health campaigns which have been implemented in the three sites. A very large number of campaigns have been conducted during the project:

OPV campaigns: 9 in Ghana, 28 in Burkina Faso, and 9 in Guinea-Bissau.
Vitamin A supplementation (VAS) campaigns: 1 in Ghana, 10 in Burkina Faso, and 11 in Guinea-Bissau.
Measles vaccination campaigns: 1 in Burkina Faso and 2 in Guinea-Bissau.
Measles and rubella vaccination campaigns: 1 in Ghana and 1 in Burkina Faso.

MDG4 context: Monitoring child survival
The ultimate goal for EU FP7 sponsored research projects in low-income countries was that they should support the process of reaching the MDG4. Over the project period from 2011-2016 under-five child mortality has declined remarkably. Bandim and Navrongo have data going back more than 20 years; followup started a bit later in Nouna.
MDG4 was reached in both the urban and the rural HDSS of the Bandim Health Project (Bandim) and in Navrongo. There has been a 68% reduction in mortality between 1995 (241/1000) and 2014 (76/1000) in the rural areas followed by Bandim. In the urban areas data goes back to 1979; between 1990 (227/1000) and 2014 (65/1000) mortality decline was 71%. In Navrongo the decline was 78% between 1996 (235/1000) to 2015 (52/1000). Based on extrapolation of the data from 1990 to 2015, the under-5 mortality rate has decline with 54% in Nouna.

Given the experiences and expectations when we wrote the OPTIMUNISE proposal this marked decline is surprising; the process may have been particularly surprising in Guinea-Bissau which has had a decade which started with a large part of the national health staff fleeing to Europe in the aftermath of the civil war (1998-1999), and continued with numerous military coups, political instability, repeated changes in government, donor fatigue and no major economic progress.
It has been of utmost interest for all Partners (1-8) in the OPTIMUNISE project to understand the background for this enormous decline in mortality. As discussed below, the beneficial non-specific effects (NSEs) of vaccines and vaccine campaigns with measles vaccine (MV) and oral polio vaccine (OPV) may have been far more important than usually assumed (see Section II). The decline in the mortality has obviously also affected our statistical power to assess effects of interventions, very specifically in relation to the randomised control trial (RCT) of early measles vaccination (see section III).

II. Measure the impact and cost-effectiveness of the existing major child health programmes with vaccines and vitamin A controlling for known determinants of programme compliance (WP4, WP7).

Objectives: To describe the real life implementation of child health programs in West Africa. To assess the overall impact on health of these programs. To test the four hypotheses regarding non-specific and sex-differential effects on mortality of vaccines published in 2007 (WP4). Estimate the effect on overall morbidity and mortality of changes in programme implementation over time and of the introduction of new vaccines and their potential interactions with other interventions (WP7).
The basis for OPTIMUNISE was the observation that vaccines and micronutrient programmes have nonspecific effects (NSEs) because they affect the immune system and may induce immune training, which affect susceptibility to unrelated infections and may therefore affect overall morbidity and mortality. Since the process may reduce or enhance susceptibility, NSEs can be both beneficial and deleterious. The effects are often sex-differential and different health interventions may interact with each other. The data processing and communication of results for WP4 was done by Partners 1-5 and 7 and for WP7 by Partner 4 with some help from Partner 1.
This area of research has grown much faster than initially anticipated. In this respect, it was very important that WHO’s Strategic Advisory Group of Experts on Immunisation (SAGE) in 2012 decided to conduct a thorough review of the NSEs of BCG, diphtheria-tetanus-pertussis (DTP) and measles vaccine (MV) on mortality for children less than five years of age. The review was conducted during 2013 and presented to SAGE in April 2014. SAGE recommended further research into the NSEs of vaccines and delegated the IVIRAC (Immunization and Vaccines related Implementation Research) committee to develop protocols and conduct such research.
During the OPTIMUNISE project a paradigm conflict has developed between the current “specific solution” paradigm and the “systemic effects” paradigm that we have emphasized. In the “specific solution” paradigm vaccines have only specific effects and their overall benefit on survival is determined by the effect on the specific disease; vaccines have same effect for boys and girls; the specific effect of a vaccine is rarely modified by other health interventions; maternal immunity reduces the specific response to vaccination; the first dose of live vaccines is usually protective and little additional effect is anticipated from repeated doses; vaccination may be terminated once the target disease pathogen is eradicated; vaccines may induce long-term adaptive immune memory against the specific disease pathogen; the specific effects are mediated through development of specific memory B- and T-cells.
When the OPTIMUNISE proposal was developed we emphasized the following principles of a “systemic effects” paradigm: first, live vaccines have beneficial NSEs, which have marked beneficial effects on general health (Principle 1); second, non-live vaccines have harmful NSEs, which have marked negative effect on general health in spite of specific disease protection (Principle 2); third, NSEs are sex-differential and girls are most affected by harmful NSEs (Principle 3), and fourth, NSEs are strongest for the most recent
vaccination, but vaccines interact with other vaccines and vitamins, so sequence and combination are important (Principle 4).
During the course of OPTIMUNISE we have developed at least 5 new principles: fifth, maternal immunity enhances the beneficial NSEs of live vaccines (Principle 5); sixth, repeated doses of live vaccines enhance their beneficial NSEs (Principle 6); seventh, termination of live vaccines may lead to increases in morbidity and mortality (Principle 7); eighth, vaccines may have long-term imprinting effect on the host, including both the innate and adaptive immune system (Principle 8); and ninth, live vaccines and non-live vaccines have different effects on the innate immune system as live vaccines induce innate immune training whereas non-live vaccines induce innate tolerance (Principle 9).
Below we summarized the results which have been obtained for these principles during the implementation of WP4 and WP7 in the OPTIMUNISE project.

Principle 1(P1): Live attenuated vaccines have beneficial NSEs
Partners 1 and 2 had previously shown that BCG, MV, and smallpox vaccine (Vaccinia) had beneficial NSEs. Several studies during OPTIMUNISE have strengthened these observations and furthermore we have added that oral polio vaccine (OPV) apparently also has beneficial NSEs, something which has not been emphasized previously.
Routine measles vaccine (MV). Due to GAVI’s interest in knowing more about the determinants of the fully immunized child (FIC) and the consequences of being FIC for subsequent child survival, Partners 1-4 and 7 obtained funding from GAVI for additional data workshops and analyses. We used data from the three West African OPTIMUNISE sites as well as from three other INDEPTH sites (Kintampo, Ghana; Nairobi, Kenya; Chakaria, Bangladesh) being part of a parallel research training project sponsored by DANIDA.
The observed patterns were consistent over the 6 sites and the results can be summarized as follows: Vaccination data collected between 2001-14 from 109,473 12-23-month-old children at five African and one Asian INDEPTH HDSS sites was used to analyse the trend over time and determinants of being “fully immunized children” (FIC) and the consequence for subsequent child mortality of being FIC compared to not being FIC.
There was an upward trend over time in the proportion being FIC at all centres except one, the coverage in 2013 ranging between 71% and 88%. As expected, cultural and socio-economic factors indicating better conditions were positively associated with FIC. However, encouragingly with increasing coverage the differences in FIC associated with education and wealth tended to disappear. None of the centres found differences in the proportion of being FIC among females and males.
While the age of DTP-containing vaccines and OPV went down over time at all centres, the patterns were more variable for BCG and measles vaccine (MV). For MV, several centres showed slight increases in the age of vaccination. This is unfortunate since there is a limited time-window from 274-365 days of age to get MV and become a FIC.
The predominant cause of not being FIC was lack of MV, explaining from 75% to 100% of not being FIC at the six centres. Controlling for back-ground factors, being FIC was associated with 22% lower mortality (95% confidence interval: 12-31%) than not being FIC. The main reason for not being FIC was lack of MV, and our analyses indicated that lack of MV is associated with 28% (14-45%) higher mortality (25). None of the centres with cause-specific mortality data reported measles epidemics, suggesting that the effect of MV is likely to be non-specific. This has been analysed specifically by Partner 4 in Navrongo that has the largest data set; the higher mortality of children who have not had MV is not explained by more measles deaths (42,51)
In conclusion, to improve FIC coverage and child survival, a stronger emphasis should be given to ensure that all children are measles vaccinated on time.

Hence, the key messages were:
▪ FIC coverage has increased over time and ranged between 71% and 88% in year 2013
▪ No difference in FIC coverage between boys and girls
▪ Place of residence and delivery, and maternal education are important factors for FIC
▪ Increasing FIC coverage diminishes importance of education and wealth for being FIC
▪ BCG age at vaccination decreased very significantly in Navrongo but is still a challenge in other sites
▪ Lacking measles vaccination is the main cause for not being FIC
▪ Being FIC is associated with 22% lower mortality

Measles vaccine campaigns. Partners 1 and 2 also conducted two epidemiological studies of the effect of MV campaigns in Guinea-Bissau. First, in an after-before comparison of the general mortality rate for children aged 6 months to 5 years in rural Guinea-Bissau, mortality was reduced by 20% after the general campaign in 2006 (22). Second, in urban Bissau, participation-versus-non-participation in the MV campaign in 2012 was associated with 72% lower mortality in the following year (54). The regular 3-yearly MV campaigns are conducted to eradicate/control measles infection to assure that everybody is reached. One should therefore expect that the strongest effects for children who had not previously received routine MV. However, in both studies the beneficial effect of campaign MV was much stronger for those who had previously received routine MV (22,54). Navrongo data also suggests that the MV campaigns have had a major effect on child survival reducing the mortality rate after these campaigns (67).
BCG. Between 2002 and now, Bandim (Partners 1 and 2) has conducted 3 RCTs of BCG among low-birthweight (LBW) children because they are not normally given BCG at birth. Two trials have previously been published (Aaby JID 2011; Beiring-Sørensen PIDJ 2012) and a third has now been submitted (46,60). In the first two trials randomization to BCG reduced neonatal mortality with 48% (95% CI 18-67%). Over the period neonatal has declined dramatically and the reduction in the third trial was 30%. In a combined analysis of the three trials, early BCG-vaccination was associated with a 38% (95% CI 17-54%) reduction in neonatal mortality, most of the reduction being due to less neonatal sepsis and respiratory infections (46).
Though BCG vaccine is recommended at birth the vaccination is often delayed. Given the beneficial effects of the vaccines it is very unfortunate that there are such delays in administration of this vaccine. An important reason is that the 20-dose vials of BCG (which have to be discarded 4-6 hours after opening) are not opened unless at least ten children are present for vaccination; therefore many contacts with the health system are not used to vaccinate. BCG coverage is usually reported as the coverage at 12 months of age, not disclosing the delay in vaccination. Within OPTIMUNISE we showed that among children born in 2010 in rural Guinea-Bissau BCG coverage by 1 week of age was 11%, 38% by 1 month, and 92% by 12 months. If BCG had been given at first contact with the health system, 1-week coverage would have been 35% and 1-month coverage 54%. After the introduction of the monthly visits (July 2012-June 2013), 1 month coverage increased to 88%. A risk factor analysis identified many factors associated with delay of BCG vaccination, including a number of socioeconomic factors, but these factors were not strongly associated with BCG vaccination when BCG-vaccination became available during the monthly visits (16).
As a case study within WP7, Navrongo (Partner 4) has examined how the median age of BCG vaccination has changed over time and how neonatal mortality has developed. BCG vaccination age declined from 46 days in 1996 to 4 days in 2012. Similarly, neonatal mortality rates also declined from 46/1000 livebirths in 1996 to 12/1000 livebirths in 2012. In the 1990s and early 2000s, the reduction in the age of BCG vaccination was linked to the experiment with community nurses in the Navrongo area, both the coverage and the age of vaccination being much lower in the communities, which had community nurses than in control community. The further reduction in the late 2000s is linked to decisions by local public health authorities to always open a 20-dose vial of BCG even though there might only be one child present for vaccination (65).

Routine OPV: In an RCT in Guinea-Bissau (conducted by Partners 2 and 1), OPV-at-birth was associated with 32% (0-57%) reduction in infant mortality when the effect of the many additional OPV campaigns was controlled by censoring at the time of subsequent OPV campaigns (105). Partner 1 has verified this striking observation in Denmark, which used OPV until 2001. In a nation-wide study, being OPV vaccinated was associated with a 27% (13-39%) lower risk of admissions for lower respiratory infections (106).

OPV campaigns. As mentioned in the beginning, there have been a very large number of campaigns in low-income countries in the last two decades. Due to the plans to eradicate polio infection globally, the largest number of campaigns has been with OPV. The recurrent outbreaks of polio infection in Northern Nigeria have meant that new OPV campaigns have been organized.
Surprisingly there has been virtually no analysis of whether these campaigns with OPV had an effect on survival. From a disease-specific perspective they should have no effect because polio is rarely a killing disease and the infection has virtually been eliminated from West Africa for the last many years. However, when the first general OPV campaign was implemented in Guinea-Bissau in 1998, Bandim (Partners 1 and 2) conducted a study which suggested that participants in the campaigns had significantly lower mortality than the non-participants (117).
Within OPTIMUNISE, using the HDSS data we have now tried to estimate how OPV campaigns affect the general mortality rate by comparing the mortality rate after-campaign with the mortality rate before campaign. Participation has been very high in these campaigns (>90%) and we have therefore assumed that all eligible children received the campaign OPV. We first conducted this kind of analysis within all the randomised trials we have conducted in Bandim between 2002 and 2014 (Partners 1 and 2). The analyses have controlled for age, season, time trend and other campaigns (52).
In the first analysis (52) the mortality rate for children aged less than 3 years was 19% (5-32%) lower after the campaigns than before. With each additional dose of campaign-OPV the mortality rate declined 13% (421%) per dose (test for trend, p=0.005). We have subsequently extended the analysis to include all children in the HDSS. In Bandim Partners 1 and 2 analysed the period 2002-2014 and the reduction in mortality rate was 25% (15-33%) and again there was significant beneficial effect with each additional dose of OPV (58). In Navrongo (Partner 4) mortality for the period 1996-2014 was analysed and the effect was a 12% (4-19%) reduction in mortality rate and beneficial effects for each additional dose. When the effects were estimated in a simulation model with random dates for hypothetical campaigns no effect was seen (52) so the effect cannot be explained as a result of a time trend in mortality. Interestingly, VAS campaigns were not associated with a positive effect on the mortality rate. Hence, the benefits of OPV campaigns did not seem to represent merely a beneficial effect of the additional attention during campaigns.
Measuring the effects of interventions: The role of campaigns. As a consequence of these observations, we discovered that OPV campaigns can change the effect of other interventions or the time-trend being studied. Intervention RCTs usually test the standard treatment versus a proposed modification assumed to be better. However, if OPV campaigns are implemented during the conduct of the intervention trial and given to all participants, OPV is likely to reduce the difference between the groups being compared in the RCTs. We have shown this several times during conducts of trials of MV (45), BCG (46), OPV-at-birth (105), and VAS (49):
• In the conduct of a RCT of an additional MV at 4½ months, measuring mortality from 4 to 36 months, the overall mortality reduction was 30% (6-48%); but if we censored at the time of the OPV campaigns, the effect of early MV was 47% (13-68%).
• In an RCT of OPV at birth (OPV0) vs no OPV0, infant mortality was reduced with 17% (-13-39%); but if we censored at the time of the OPV campaigns, the effect of OPV0 was 32% (0-57%) (105)
• Likewise in two trials of BCG-at birth to LBW children, infant mortality and neonatal mortality was reduced with 17% (-8-37%) and 30% (-4-53%), respectively; but if we censored at the time of the OPV campaigns, the effect of BCG-at-birth 20% (-6-40%) and 34% (0-56%), respectively (46)
It has therefore become more difficult to show that an intervention has a significant beneficial effect as long as OPV campaigns are being conducted.
OPV campaigns may also reduce differences between other groups; for example we have examined as one of the objectives of the OPTIMUNISE project the effect of out-of-sequence vaccinations with MV and DTP (TMIH-hypothesis-IV). As part of WP7 Partners 4 tested in Navrongo the potentially negative effects of getting MV and DTP out-of-sequence (OOS), i.e. DTP with or after MV. In the adjusted analysis, the overall result was 30% (6-47%) lower mortality for those who had the recommended sequence of MV-after-DTP3 compared with OOS vaccinations (42). However, the beneficial effect of MV-after-DTP3 was much stronger before campaigns, the reduction in mortality being 63% (12-83%); after the campaigns were implemented the reduction in mortality associated with MV-after-DTP3 was only 27% (2-46%) (42,51). Similarly in another study from Navrongo, the reduction in mortality linked to being MV versus not being measles vaccinated was 28% (13-39%), but if we censored follow-up at the time of OPV campaigns the benefit associated with MV was a 39% (19-54%) reduction in mortality.
In both RCTs and observational studies, the mortality rate declines after the OPV campaigns and this reduces the differences between the groups being compared. Hence, these examples strongly indicate that OPV has beneficial NSEs. If the effects of campaigns are not taken into consideration we may fundamentally misinterpret the time trends in mortality (49,57).
Specific and non-specific effects of live vaccines on mortality. In the RCTs with live vaccines, which had a beneficial effect on overall survival, we examined how much of the reduction in mortality was due to the specific effect and much to the non-specific effect:
• In a meta-analysis of three RCTs of BCG at birth and with censoring for OPV-campaigns, the reduction in neonatal mortality was 41% (19-57%). None of this reduction was due to specific disease protections and hence the non-specific effect was 41% (19-57%)(46)

• In an RCT, censoring for OPV-campaigns, OPV0 was associated a reduction in Infant mortality of 32% (0-57%). None of this reduction was due to specific disease protections and hence the non-specific effect was 32% (0-57%) (105)

• In an RCT of early MV, the intervention was associated a 30% (6-48%) in mortality between 4-36 months of age. 4% of this reduction was due to the specific prevention of measles infection and 26% (0-45%) was due to beneficial non-specific effect (107).

Hence, the reduction in mortality due to NSEs was much greater than the reduction due to specific-disease prevention.

P2: Non-live vaccines - though protective against target diseases - have harmful NSEs
Partners 1-5 and 7 have strengthened the evidence considerably for deleterious effects of non-live vaccines during the OPTIMUNISE project (1,6,10,23,24,32,34,35,39,40,42,43,48,53,57,59,63,73). We have only been able to test the DTP danger-signal in natural experiments (43,59) since it is difficult to get ethical permission to test WHO-recommended vaccines in RCTs. However, new RCTs have tested the non-live RTS,S malaria vaccine (24,32), and we have actively pursued the outcome.
The SAGE review and the impact of DTP on child survival. The SAGE review found that BCG and MV were associated with nearly a halving of the mortality rate compared with not being vaccinated with these vaccines (see below). However, the review found a 38% (-8-108%) increase in mortality associated with DTP. Though not significant in its own right it was clearly significantly different from the effects of BCG and MV in the same studies and from standard expectations as normally the healthiest children who are vaccinated first. Nonetheless, the reviewers concluded that the literature was inconsistent since two studies had found a beneficial effect of DTP. We have subsequently shown that the WHO-SAGE analysis of DTP was flawed due to inclusion of several studies with survival bias and frailty bias. If only studies with prospective follow-up are included, the available studies suggest 2-fold higher mortality for DTP vaccinated children, the MRR being 2.00 (1.50-2.67) (34).

The WHO-SAGE review concluded that though there were sex-differential effects of MV (better for girls) (109-110), there was no sex-differential effect of DTP. Since sex-differential effects were one of the focus areas for OPTIMUNISE, we reviewed the literature. Again this was a flawed conclusion because the SAGE reviewers included studies with major survival bias (34). If these studies were excluded, DTP was associated with a strongly negative effect for girls (48); among DTP vaccinated children girls had 53% higher mortality than boys (15 studies) (48). A new study from Nouna (Partner 3 and 5) has also found that DTP-containing vaccine is associated with higher mortality for girls than for boys (63). Likewise, Partners 1 and 2 found that Pentavalent vaccine (DTP+HBV+Hib) was associated with increased female mortality (39).
Introduction of DTP-vaccination in 1981 associated with increased mortality: The SAGE review emphasised that it is difficult to assess the “true” effect of DTP because it is always given with OPV. Furthermore, it was emphasised that there were herd immunity to pertussis while the studies were conducted, and one would therefore not see the benefit of preventing pertussis. Given the continuing discussion of the true effects of DTP, Partners 1 and 2 went back to examine what happened when DTP was introduced in Bissau in the early 1980s. In studies of both infants and older children, we found that DTP (+/- OPV) was associated with 4-5 times higher mortality than not being DTP-vaccinated (43, 59). Since it is the healthiest children who are vaccinated first (34), the effects of DTP may be difficult to see in many studies because most studies compare healthier DTP-vaccinated children with frail DTP-unvaccinated children. However, when the vaccines were introduced the situation resembled a natural experiment and it was clear that there was a very marked negative effect of DTP (43,59). Vaccines were provided in connection with nutritional examinations from 3 months of age in Guinea-Bissau. Children were called for examinations every 3 months. Hence, due to variation in date at birth some children were vaccinated shortly after age 3 months, whereas others were nearly 6 months old. In this natural experiment, the children who received DTP early had an age-adjusted hazard ratio (HR) for mortality of 5.00 (1.53-16.3) between 3-6 months of age compared with not yet DTP-vaccinated children. In children aged 12-35 months, the HR was 4.13 (0.93-18.40) for DTP-vaccination compared with no DTP vaccination even though the DTP-vaccinated children had better nutritional status than the not DTP-vaccinated children (59). The effect was significantly negative for girls (59).
Malaria vaccine associated with increased mortality: The GSK malaria vaccine RTS,S was tested in two RCTs. When the first results were published, the RTS,S group had slightly higher mortality. This made little sense if RTS,S prevented malaria infection. Hence, we asked for results to be reported by sex, from the hypothesis that if malaria vaccine, apart from protecting against malaria, also had negative NSEs, then it might be reflected in higher female mortality as seen for other non-live vaccines. In the final report (2015), the mortality data indicated 24% (-3-58%) higher mortality and a significant long-term negative effect; but data was not reported by sex. We wrote to Lancet (24) and WHO, requesting that data be presented by sex, as the increased mortality could represent negative NSEs in females. After request from SAGE members, GSK released the data by sex. As we had hypothesised, RTS,S was associated with 2-fold higher mortality for girls in both RCTs, but made no difference for boys (32). Hence, the pattern is similar to what we have seen for other non-live vaccines (see P3).

P3: NSEs are sex-differential; girls are most affected by harmful NSEs
Potential sex-differential effect of interventions were emphasised as part of the original OPTIMUNISE protocol and we specifically planned to test the TMIH-hypotheses about NSEs. Using the SAGE review as a basis, we have reviewed the existing literature on DTP. In the 7 available studies, DTP given after BCG was associated with 2.54-fold higher mortality for girls but not for boys (TMIH-hypothesis I). In 15 available studies, among DTP vaccinated children girls had 53% (21-93%) higher mortality than the boys (TMIH-hypothesis-II) (48). As described above, the malaria vaccine was associated with two-fold higher female mortality in RCTs. Hence, so far we have now shown increased female mortality for all the non-live vaccines we have examined:
• DTP vaccinated versus DTP unvaccinated was associated with a MRR of 2.00 (1.50-2.67) in 8 studies; in 7 studies comparing DTP-vaccinated female with DTP-unvaccinated females the MRR 2.54 (1.68-3.86) and in 15 studies comparing the mortality of female and male DTP-vaccinated children the MRR was 1.53 (1.21-1.93) (34, 48).

• In 3 trials of IPV comparing the mortality of female and male IPV-vaccinated children the MRR was 1.52 (1.02-2.28).

• In a natural experiment with HBV, HBV-vaccinated versus not HBV-vaccinated was associated with a MRR 1.81 (1.19-2.75); comparing female HBV vaccinated versus HBV-unvaccinated females the MRR was 2.27 (1.31-3.94) and the female/male MRR among HBV children was 2.20 (1.07-4.54).

• In a natural experiment with H1N1 vaccine (51), H1N1-vaccinated versus not H1N1-vaccinated children was associated with a MRR 1.86 (1.02-3.42); comparing female H1N1-vaccinated versus H1N1-unvaccinated females the MRR was 2.32 (1.19-4.52) and the female/male MRR among H1N1 children was 2.68 (0.44-16.4).

• In two RCTs of RTS,S malaria vaccine, RTS,S-vaccinated versus RTS,S-unvaccinated had a MRR of 1.24 (0.97-1.58); in the two trials the MRR for female RTS,S vaccinated versus controls were 1.81 (1.04-3.14) and 2.00 (1.18-3.39), respectively. The female/mala MRR among RTS,S-vaccinated children was 1.33 (1.02-1.74) (24,32).

• In the only study of Penta, the female/male MRR was 1.73 (1.11-2.70) (39)

The SAGE review concluded that MV was associated with a stronger beneficial effect for girls than for boys and thereby confirmed TMIH-hypothesis III (109,110). A stronger beneficial effect of MV for girls have also been confirmed in several consortium studies (7,39,57)

P4: NSEs are strongest while the vaccine is the most recent vaccination; vaccines interact with other vaccines and vitamins, so sequence and combination are important
DTP with or after MV have negative effects on child survival. The observation that a non-live vaccine given after a live vaccine changes the overall effect from beneficial to negative, at least for girls, has been strengthened during the OPTIMUNISE period. The SAGE review (109) summarized the current evidence: MV+DTP compared with MV-only was associated with a HR of 2.30 (1.56-3.39); for DTP after MV compared with MV-only, the HR was 2.16 (1.25-3.74). Partner 4 subsequently corroborated these trends in a large study in northern Ghana; receiving MV and DTP out-of-sequence (i.e. DTP with or after MV) was associated with a HR of 2.58 (1.14-5.84) as long as the children had not yet received OPV campaigns (42). In a smaller study from Guinea-Bissau (Partners 1 and 2), the HR of death was higher among children with DTP>=MV (HR=1.44 (0.97-2.15)) (57). These studies do not separate the effects of DTP given after MV and DTP given with MV but they support the TMIH-hypothesis IV that DTP given with MV is associated with higher mortality than MV given alone. In a study from Guinea-Bissau which specifically studied the effect of coadministration of DTP/pentavalent vaccine with MV, the adjusted HR was 3.24 (1.20-8.73) compared with MV given alone or MV given with Yellow Fever vaccine (6).
DTP given after MV has also been found to have a negative effect, for girls, in 5 RCTs from Guinea-Bissau, Senegal, The Gambia and Sudan (35).
Combined BCG and DTP reduce the negative effects of DTP: WHO recommends BCG at birth and DTP at weeks 6, 10 and 14. However, many children receive BCG later, often with DTP. In a large dataset from Bangladesh, Partner 1 showed that children who received BCG+DTP1 simultaneously had an adjusted-HR from 6 weeks to 9 months of age of 0.52 (0.38-0.70) compared with children who followed the WHO-schedule of BCG-first-then-DTP (53). We have found similar trends in smaller studies from India, Senegal (23, 40) and Guinea-Bissau.
Combined DTP and OPV vaccinations reduce the negative effects of DTP. Interestingly the studies of the introduction of DTP and OPV in urban (43, 59) and rural (113) Guinea-Bissau showed that giving OPV together with DTP reduced the negative effects of DTP compared with DTP-unvaccinated children (43):
In urban Bissau where DTP was introduced in 1981 (43), the HR DTP-only vs unvaccinated was 10.0 (2.6-39) but “only” 3.52 (1.0-12.9) if DTP had been administered with OPV. In another study from rural Bissau where DTP was introduced in 1984 (113), the HR DTP-only vs unvaccinated was 5.00 (0.6-40) but “only” 1.90 (0.9-4.0) if DTP had been administered with OPV. In a meta-analysis of these two studies, the HR for DTP-only vs unvaccinated was 8.1 (2.6-25) but “only” 2.2 (1.2-4.2) if DTP had been administered with OPV, suggesting a significant modifying effect of co-administering DTP and OPV.
To the extent this is true it might suggest that the negative effect of DTP may increase once OPV has been removed from the vaccination programme in 2020.
Vitamin A supplementation (VAS) is a two-edged sword: Bandim (Partner 1 and 2) completed the first RCT testing the WHO-policy to administer VAS with vaccines to children after 6 months of age. Contrary to current beliefs, VAS did not reduce mortality by 25%. We found no overall effect, but a strong sex-differential effect; VAS halved mortality for girls yet doubled it for boys. We corroborated previous observations of a beneficial effect of repeated VAS in females. Currently there is a heated debate regarding whether VAS should continue; we are arguing that large-scale RCTs should be done to ensure that VAS is provided to those who benefit, but not those harmed (29).
There has been much work to extend the assumed benefit of VAS to younger ages. In the 2000s, our RCTs showed 10-15% increased mortality after neonatal VAS (NVAS). WHO-Gates sponsored three NVAS trials with >100,000 children. The two African trials found slightly increased mortality, supporting our RCTs. We have posited that the negative effect of NVAS may be linked to NVAS enhancing female mortality after DTP vaccination (29). WHO has not made the final decision, but it is unlikely that NVAS will become policy, as it may be harmful under some circumstances.

P5: Maternal immunity enhances the beneficial NSEs of receiving live vaccines
One new insight in the OPTIMUNISE period has been the importance of maternal immunity in priming for positive NSEs of live vaccines in her offspring. Consistent with an enhancing effect of maternal immunity, vaccination earlier in life, when maternal immunity is more likely to be present, is associated with a much stronger beneficial effect for both OPV(45,105) and BCG(17).

Maternal antibodies enhance the beneficial NSEs of MV: We examined whether maternal antibodies enhance the NSEs of MV. In two RCTs of early MV, Partners 1 and 2 measured measles antibody levels at vaccination. In contrast to current thinking that it is best to use MV after maternal antibody has disappeared, being vaccinated with MV in presence of maternal measles antibody enhanced the beneficial NSEs (12). This was not explained by confounding. Combining both RCTs, MV in the presence of maternal measles antibodies compared with no maternal measles antibody was associated with a 78% (36-93%) reduction in mortality between 4 months and 5 years of age (12).

Maternal BCG-vaccination may enhance the NSEs of BCG: In a study from Guinea-Bissau, the reduction in mortality between 4 months and 3 years of age associated with having a BCG-scar was 46% (13-67%) when the mother also had a BCG-scar but only 24% (-17-51%) when the mother did not have a BCG-scar (81). [In a RCT in Denmark, BCG at birth had no overall effect on hospital admissions for infectious disease. However, among children of BCG-vaccinated mothers, randomisation to BCG-at-birth was associated with 35% (6-55%) lower risk of admission with an infectious disease, 23% (-1-42%) lower risk of atopic dermatitis, less parental-reported morbidity and fewer GP-visits up to 3 months of age]

P6: Repeated doses of live vaccines enhance their beneficial NSEs
Since the effect of primary MV and BCG seemed to depend on the existence of maternal immunity, we hypothesised that repeated doses of live vaccines (i.e. second and subsequent doses given in the presence of existing immunity) would enhance the NSEs. This turned out to be the case. Partner 1 reviewed available evidence for MV, BCG, OPV and Vaccinia. Repeated doses enhanced the beneficial effect on survival for all four vaccines (37). Since the first dose is protective or because there is no threat from the specific infection, these effects are clearly non-specific. MV: Two RCTs of early MV allowed comparison of two versus one dose of MV; two doses versus one dose reduced all-cause mortality by 63% (23-83%). CVIVA conducted the first studies of MV-campaigns and child survival. In both studies campaign-MV had a beneficial effect (HR=0.80 (0.64-0.96); HR=0.28 (0.10-0.77)); this effect was stronger among those who had previously received MV (HR=0.59 (0.36-0.99); HR=0.15 (0.04-0.63)). BCG: Two RCTs showed that BCG revaccination enhanced BCG’s protective effect against child mortality significantly. OPV: In campaigns in Guinea-Bissau, each additional dose of OPV reduced mortality by 13% (4-21%). In Denmark, the protective effect of OPV on admissions for infections increased with additional doses. Vaccinia: The protective effect of Vaccinia on survival and HIV-1 increased with the number of Vaccinia scars.
The beneficial non-specific effects of live vaccines and the beneficial effect of boosting during campaigns could be a major reason for the development towards MDG4. The boosting mechanism (which needs to be studied mechanistically) may provide a large part of the explanation for why mortality has dropped so dramatically in the last 15-20 years in low-income countries. Since there has been far more OPV campaigns than MV campaigns, the largest part of the reduction is probably due to the OPV campaigns. This clearly raises the issue of what will happen when the OPV campaigns are stopped because polio has been eradicated.

P7: Termination of live vaccines may lead to increases in morbidity and mortality
One of the implications of the NSEs is that if a vaccine with beneficial NSEs is stopped after eradication, then one may do more harm by depriving all people of beneficial immune training than the good one did by protecting a few individuals from dying of the disease. Since polio, measles and rubella are likely to be eradicated in the coming decades and vaccinations may be stopped we examined what happened when smallpox was eradicated (1977) and vaccinia vaccine stopped (1980). From this perspective Partners 1 and 2 have been able to show both in Guinea-Bissau (115) and Denmark (116) that smallpox vaccination apparently had long-term beneficial effects on survival. Individuals in Denmark who were Vaccinia and BCG vaccinated in the 1970s had 46% (19-64%) lower mortality between school entry and 45 years of age from natural causes of death, whereas there was no association with accidents and suicides (116). We have also shown in Guinea-Bissau and Denmark that smallpox vaccination was associated with 35% lower risk of being HIV-1 infected (56).
This experience may be critical now that we are about to stop OPV in 2020, and measles and rubella vaccines may be the next to follow. Partners 1 and 2 are about to start a cluster RCT of OPV and MV to document whether these campaigns have an impact on child survival and health in general.

P8: Vaccines may have long-term imprinting effect on the immune system
NSEs from a specific vaccine have been most pronounced while that vaccine is the most recent vaccination, and may change with the receipt of a new vaccine of a different type (see P4). However, some vaccines may also have long-term NSEs on health. Indications of this are the observations on beneficial NSEs of Vaccinia, which were seen even decades after smallpox vaccination was stopped (56). Another indication relates to BCG: Among BCG-vaccinated children a BCG-scar is a marker of beneficial long-term NSEs reducing mortality also after other vaccines have been given. There is no real indication that in the majority of cases BCG-scarring depends on the prior immunological status of the child. If BCG is administered by an experienced administrator 95-98% get a BCG scar (16, 38, 72). In rural Guinea-Bissau, only half of BCG-vaccinated children had a scar, most likely due to poor vaccination technique (17). Mortality at 0-4 years of age was 26% (4-44%) lower and hospital admissions were 26% (8-40%) lower for children with a BCG-scar compared with BCG-vaccinated children without a scar (17). In a meta-analysis of the effect of BCG-scar versus no BCG-scar among BCG vaccinated children, having a BCG-scar was associated with 46% (32-57%) lower mortality in the first year of life (17; 103; Roth 2005; Roth 2006; Garly 2003).
The direct implication of this finding is that BCG-scarring could be used to monitor the quality of BCG vaccinations; increasing the proportion having a BCG-scar might have a strong impact on child survival.

P9: Live and non-live vaccines have different effects on the innate immune system
While OPTIMUNISE did not directly fund immunological studies, there has been a parallel research agenda where we have conducted immunological studies to find possible explanation for the non-specific effects. The suggestion so far is that live vaccines, BCG, vaccinia and Yellow Fever, induce innate immune training. In contrast, the non-live trivalent influenza vaccine (TIV) had opposite effects of BCG and induced innate tolerance; the same was the case for the non-live typhoid fever vaccine (TFV).

Cost-effectiveness of health interventions
Taking the NSEs into consideration will have major consequences for the estimated cost-effectiveness of intervention programmes. We have evaluated the cost effectiveness of the “restrictive vial opening policy” in Guinea-Bissau (14). In brief, the “restrictive vial opening policy” refers to the principles implemented in many low-income countries of only opening a vial of live BCG or MV, if there are a certain number of children present to be vaccinated. The reason is that these vaccines, once opened and dissolved, need to be used within 4-6 hours. Hence, opening a bottle for just one child will lead to high wastage rates. Countries are under pressure from donors to reduce wastage. Thus, in Guinea-Bissau and Burkina Faso, a 20-dose vial of BCG is only opened if there are 10-12 children present to be vaccinated with BCG, and a 10-dose vial of MV is only opened if there are 6-7 children present. In addition, for MV, since children only count in the statistics if vaccinated before 12 months of age, it is often required that 6-7 children between 9-11 months of age be present before a vial is opened. One OPTIMUNSE paper addressed the incremental cost-effectiveness of abandoning the restrictive MV policy, and providing MV to all children regardless of age and number of children present. Since 2011, the BHP has conducted a cluster-randomized trial, assigning children to either receive MV according to “the restrictive MV policy” or to receive MV regardless of age and number of children present (MV-for-all policy), in rural Guinea-Bissau. We used MV coverage estimates from this study to calculate the effect of the MV-for-all policy. We estimated the costs of delivering MV taking into account its positive non-specific effects and using different MV wastage scenarios; 40% wastage under the restrictive MV policy (at least 6 children vaccinated per vial) and 90% wastage under the MV for all policy. In the villages followed by Partner 2, MV coverage was 84% under the restrictive MV policy and 97% under the MV-for-all policy. Among 54,573 children born in Guinea-Bissau in 2011 and alive at 9 months of age, at 90% wastage, the MV-for-all policy was cost-effective at USD 10.7 per life year gained (LYG) and USD 282 per death averted. At 87% wastage, the MV-for-all policy was cost-saving (55).
Information on household costs of seeking measles vaccination in Guinea-Bissau has been collected. Based on interviews with 1308 mothers of children aged 9-21 months living in rural Guinea-Bissau, on their experiences with seeking MV, we calculated the costs of- and time spent on seeking MV. We assessed the MV status of the children through the HDSS: 34% of children who were measles unvaccinated at time of interview had sought MV at least once and 19% of children already measles vaccinated had been taken for MV more than once. In total, 80% had gone for MV but the coverage was only 70%. Mothers on average took their child for MV 1.4 times, the cost of one time being USD 1.33 (41). Hence, with a restrictive vial policy, we may save some doses of vaccines, but we are also transferring the cost of going in vain to the parents. Furthermore, the result of the restrictive vial policy is lower coverage.
The two first papers are focusing on different cost-implication of the restrictive vial policy (41,55). Further papers regarding the cost-effectiveness of measles vaccination campaigns, rotavirus vaccination and pneumococcal vaccination are in preparation by Partners 4, 1 and 2 (68-70,76).

III. To evaluate in a randomised trial a specific modification of the current child health programme: To provide an additional measles vaccine at age 4.5 months, in addition to the recommended measles vaccine at age 9 months (WP5,WP6,WP8)

Objective: To evaluate in a randomised controlled trial the effect on child survival and other health indicators of providing an additional dose of EZ measles vaccine as soon as possible after 4 months of age and before the standard measles vaccine at 9 months of age for children who have received DTP3 at least 4 weeks before (WP5,WP6). Within the early measles vaccination RCT (WP5-6) we planned to measure measles-specific antibody levels and test the effect of age and maternal measles-specific antibody levels at the time of the primary measles vaccination for the long-term measles-specific antibody level.
The RCT of early two-dose MV strategies was conducted in Burkina Faso and Guinea-Bissau by Partners 1-3 and 5 (WP5,WP6). Partner 8 analysed the measles antibody samples in connection with these RCTs (WP8).
WHO recommends the first dose of measles vaccination at 9 months of age in areas with measles transmission. The vaccination age is based on a compromise between the age at which the child is expected to lose its maternal measles antibodies (and therefore becomes susceptible) and avoiding vaccination in the presence of high levels of maternal antibodies (which blunt the antibody response to measles vaccine). However, the rationale behind this policy is based on numerous assumptions and extrapolations, rather than real data.
When the current measles vaccination strategy was defined in the 1970s, most mothers had experienced measles infection. Mothers naturally infected with measles transmit higher levels of maternal antibodies to their children. However, most children are now born to mothers who have not had natural measles infection but received MV in childhood. They therefore transmit less measles antibodies to their children and the children may become susceptible to measles infection already by 3-4 months of age. For example, during an epidemic in Bissau, 4-5% of the children had had measles infection before they were enrolled in a trial of early MV at 4 to 5 months of age. Hence, the children are susceptible to measles infection much earlier now than when the policy was defined.
In addition to protecting against measles infection, accumulating evidence both from observational studies and randomised trials indicates that MV has beneficial NSEs, lowering all-cause mortality due to other infectious diseases (see P1 above). In contrast, the non-live DTP vaccine is associated with increased mortality. The most recent vaccine profiles the immune system (35), hence by providing a live MV the negative effect of DTP may be abrogated.
An early additional dose of MV may also have beneficial effects later, after the control group received MV. Accumulating evidence from observational studies suggest that additional doses of MV may lower all-cause mortality. In a trial of early MV from Guinea-Bissau, early MV was associated with 29% (-1-50%) lower mortality between 9 months and 3 years of age (119).
To test these prior findings that an additional early MV dose could reduce all-cause mortality up to 3 years of age by 30-50% (119), OPTIMUNISE conducted an RCT. The RCT took place between July 2012 and May 2016 and tested the effect of an additional early dose of MV given 4 weeks after the third dose of pentavalent vaccine. It was conducted in the Nouna HDSS in Burkina Faso (Partners 3 and 5) and in three regions of the rural Bandim HDSS in Guinea-Bissau (Partners 2 and 1).
In both countries BCG is recommended at birth together with oral polio vaccine (OPV), and 3 doses of pentavalent (DTP, hepatitis B and H. influenza type B) with OPV are recommended with 4 weeks intervals starting from age 6 weeks in Guinea-Bissau and age 2 months in Burkina Faso. At age 9 months children are scheduled to received MV and yellow fever vaccine (YF). A 13-valent pneumococcal vaccine was added to be given with pentavalent vaccine 1, 2 and 3 in Burkina Faso in November 2013 and in June 2015 in Guinea-Bissau. Rota virus vaccines were added to the vaccination programme in Burkina Faso in November 2013 (3 doses with pentavalent vaccine 1, 2 and 3) and in November 2015 (2 doses with pentavalent vaccines 1 and 2) in Guinea-Bissau. In October 2014 Burkina Faso introduced a second dose of MV at 15 months of age.
In the RCT the children were enrolled and randomised at 4-6 months of age, 4 weeks after Penta3 scheduled at 14 weeks (Guinea-Bissau) or 4 months (Burkina Faso). To ensure timely vaccination and that children would be eligible to enter the study before 215 days, all children were visited at home every month to remind the mother to seek age appropriate vaccinations. Children aged 121-215 days were eligible for enrolment 4 weeks after Penta3 provided that they were registered as residents of the HDSSs. Potentially eligible children were visited at home. Mothers/guardians were explained that we were investigating whether an additional early dose of MV had beneficial effects on child mortality. Following the explanation of the study, mothers interested in participating were asked to seek the health
centre/vaccination post on the same or subsequent day depending on site. At the vaccination post/health centre the mother/guardian was given information about the study orally and in writing. After the opportunity to ask and receive answers to questions, the mother/guardian was asked if she wanted her child to participate in the study and provided consent, she would sign the consent sheet.
After consent the mother was interviewed on past hospital admissions and the health status of the child documented. Children were randomised 1:1 to early MV or no early MV in blocks stratified by sex. Following randomisation the child received either a standard dose of Edmonston-Zagreb measles vaccine (Serum Institute of India) or no vaccine. Since we were interested in testing the non-specific effects of the vaccine, we did not use a placebo vaccine which could also have had NSEs. The early MV dose was recorded on study forms, but the information was not transferred to the child’s vaccination card to ensure that it would not be mistaken as a 9 months MV or affect the health care of the child. All children were followed through the HDSS routines and at the first visit after 9 months of age the child was invited back to the vaccination post to receive the routine measles vaccine.
In a subgroup of trial participants, we collected blood samples at enrolment, just prior to the 9 months vaccination and at 15 months (Burkina Faso) /24 months (Guinea-Bissau). Serum was analysed for measles antibody concentration using a multiplex assay by Partner 8.
Prior studies have indicated that early measles vaccination is safe (47). Every third month the data safety and monitoring board received and reviewed extracted data containing the randomisation table, the identification numbers and dates of all registered deaths, hospital admissions and consultations within the first month after enrolment.
A total of 8309 children were enrolled, 4559 in Burkina Faso and 3750 in Guinea-Bissau. 104 children were excluded due to protocol violations and all the remaining 8205 children were followed in the trial. The mortality rate in the trial was much lower than expected in Guinea-Bissau and slightly lower in Burkina Faso. We registered 145 deaths in the per-protocol analysis whereas the sample size calculation was based on 300 deaths. Mortality was similar in the intervention and control groups, the Hazard Ratio (HR) comparing the mortality rate in the intervention group versus the control group being 1.05 (0.75-1.46). The intention-to-treat analysis included 243 deaths and the HR was 1.12 (0.87-1.44). The results did not differ by sex.
Blood samples were collected and analysed for 869 children and their mothers at enrolment. Antibody concentrations were described by geometric mean concentration and the log-transformed concentrations compared by group using Students t-test. Furthermore, we compared the proportion of samples with concentrations indicating protective levels using a cut-off of >=125 mIU/ml.
Pre-vaccination antibody levels in children were low, only 21% (90/422) of children in Guinea-Bissau and 4% (16/447) of children in Burkina Faso had protective antibody levels at enrolment. At 9 months of age, there were only 14 children in the two control groups who had a protective measles antibody level. All of these children, except one, had an increase in antibody level between 4 and 9 months; presumably they had had an unregistered measles vaccination elsewhere or a measles vaccination in a campaign. Hence, with one exception, all children in both Burkina Faso and Guinea-Bissau had lost protective maternal antibody levels before 9 months of age. In both sites, early vaccinated children responded well to the vaccine and >=90% had protective antibody levels at 9 months of age. At final follow-up at 15/24 months, 97-100% was protected in both groups in both Bandim and Nouna. In Bandim 97% (96/99) had protective antibody levels after 2 doses of MV and 97% (102/105) after one dose at 9 months of age. In Nouna 100% (90/90) and 97% (94/97) had protective antibody levels after 2 and 1 dose of MV.

The findings did not support that early MV reduces all-cause mortality. This could potentially be explained by the low mortality in the trial and by the frequent OPV campaigns with beneficial NSEs targeting all study children, potentially preventing the deaths, which the early MV would otherwise prevent. An interference and interaction between OPV campaigns and early MV was seen in the previous trial of early two-dose MV
(45).
Most children were susceptible to measles infection at 4-6 months and almost all children in the control group were susceptible at age 9 months. Children responded well to the early MV with high antibody levels. Since early MV did not seem to hamper the induction of lasting protective levels, a two-dose MV strategy with an early MV followed by MV at 9 months would improve measles immunity for the population.

IV. To assess the relevance of different health outcomes parameters for the evaluation of the real life effect of child health programmes (WP9)

Objective: To assess consistency and cost-effectiveness of different health outcomes for the evaluation of child health programmes in order to produce relevant comparable outcome parameters.
Mortality is fortunately decreasing in Africa and it is becoming increasingly difficult and costly to measure the impact of health interventions on mortality. Hence, there is a need to identify reliable markers which are indicators of a good health and which are cheap and easy to measure, and can be used in future evaluations to assess whether current and new interventions are likely to have a positive effect (specific and non-specific) on health.
The HDSSs collects different health outcomes as a part of the routine data collection:
• Mortality. Reliable data on child mortality are available from all the HDSSs.
• Morbidity. Mothers/caretakers are interviewed at home visits about medical consultations and hospitalisations. Routine data on consultations and hospitalisations are collected at the health care institutions where possible.
• BCG-scarring. BCG-scars are registered and measured during the home visits.
• Growth. Arm-circumference data are collected during the home visits.

Partners 1 and 2 had the main responsibility for WP9 but other partners have and will contribute to WP9.
This data has been used to assess the association between various morbidity outcomes (hospitalisation/consultations) and growth in relation to mortality. In addition, some immunological markers have been assessed as part of specific studies.

Using this data, we have identified that hospitalisations to a limited extent overlap with mortality; hence, in a situation with decreasing mortality, a composite outcome of hospitalisations and mortality will be a powerful indicator of “bad health”. Furthermore, having a BCG scar is very closely linked with lower mortality, and in line with this observation, having a high IFN-gamma response to in vitro stimulation with PPD is associated with lower mortality.

Hospital admissions. Previously, using data from an RCT of early MV (at 4.5 months of age) in Guinea-Bissau, Partners 1 and 2 have published that the effect of MV on mortality and hospital admissions are similar (7, 61, 107). OPTIMUNISE investigated these unique data on MV, hospitalization, and mortality more closely using event history analysis, competing risk methods, and mediation analysis techniques. It was found that effect of early MV on hospitalisation is not only another way of measuring mortality as many hospitalised children do not die. By dividing event of hospitalisation into severe and mild (with and without fatality), very equivalent effects of MV were found; MV reduced both type of events by around 30% in the interval 4.5 to 9-10 months of age. This clearly indicates that a composite outcome measure of hospitalisation and mortality is useful in future trials of NSEs of MV and possibly also other vaccines.
BCG-scarring. We have confirmed previous observations that not having a BCG scar is a very strong marker of mortality (17,103). Since the same effect is found irrespective of whether 50% or 5% do not have a scar, it is unlikely that this effect mostly reflects that some children are immunologically “weaker” and therefore do not make a scar (as this proportion would not vary between 50% or 5%). In line with this, our studies confirmed that non-scarring is predominantly an issue of vaccination technique and strain of BCG vaccine (17,38). Hence, factors which affect BCG-scarring (strain, vaccination technique, amount of BCG injected intra-dermally (wheal size)) are per se determinants of mortality. The interesting implication, which has grown out of this project and was highly emphasised at the final stakeholder meeting is that it would be worthwhile to monitor BCG-scarring as an indicator of the quality of BCG vaccinations, which is a major determinant of mortality in its own right. As a consequence the OPTIMUNISE consortium has submitted a letter of intent (LOI) to EDCTP to further study how to optimize the use and monitoring of BCG vaccine to reduce infant mortality.
Growth monitoring (50). Partners 1-3 and 5 intend to use the RCT and the careful monitoring of the growth of the children within the RCT to assess whether anthropometry is still a strong marker of severe morbidity/mortality. Most of the studies, which have been used to justify the focus on anthropometry, are from the period where mortality was still very high in low-income countries. Hence, we will use data from the current situation with low mortality to assess whether anthropometry is still a strong marker. Given the delay in finishing the RCT, these analyses are delayed. WE expect to be able to conduct the analysis in the beginning of 2017.
Cytokine levels. Partners 1 and 2 have used immunological data obtained from subgroup studies to assess the correlation between in vitro cytokine production after stimulation with specific and non-specific antigens and mitogens and mortality, with the hope to identify immunological markers of a good survival probability. Unfortunately, we were not able to identify simple immunological markers. Maybe not surprisingly, the analyses showed that for most cytokines there was a U-shaped association with mortality – the risk was increased for both very high and very low cytokine responses (9). The only clear linear association was seen for the IFN-gamma response to PPD – the higher the response, the lower the mortality. The finding corroborates the observation that among BCG vaccinated children, developing a BCG scar and a large PPD response in vivo is associated with reduced all-cause mortality. It remains to be tested whether a strong response reflects an underlying stronger immune system or whether the association reflects that the children who were vaccinated correctly had better survival, i.e. a beneficial nonspecific effect of BCG.

V. Conclusions: The NSEs of vaccines: Evidence and implications (WP11)
Objective: To communicate more broadly the premises of the study in order to encourage more such studies, to disseminate the results of the study, to ensure all the major stakeholders are reached and engaged in the discussion of the results, and to the extent possible to assure that results are utilized through changes in policy.
It was intended that the results of the studies conducted by OPTIMUNISE should be disseminated as widely as possible through conference presentations and publications. The final meeting of the consortium was therefore also a stakeholder meeting in which all Partners participated as well as several international agencies, national health authorities and industry.
The OPTIMUNISE project has strengthened the evidence for the importance of NSEs:
• We have corroborated that the live vaccines (BCG, MV and vaccinia) have strong beneficial effects for survival, the NSEs likely being more important than the specific effects. We have added information that the same is the case for OPV.
• We have corroborated that the non-live DTP vaccine is associated with higher mortality, particularly for girls. We have shown that the same is the case for Penta and the new malaria vaccine RTS,S.
• We have corroborated that vaccines and micronutrients very often have sex-differential effects; studies should not report only for “children” but for girls, boys and overall.
• We have corroborated that vaccines and micronutrients very often interact; combinations and sequence have major consequence for survival. It cannot be assumed that specific interventions continue to have the same overall effect. Hence, continuing monitoring of efficacy is needed when new interventions are being introduced, as they are continuously.
• We have discovered that priming from the mother is critical essential for stimulating the beneficial NSEs in the child.
• We have discovered that boosting enhances the beneficial effects of live vaccines and may go a long way to explain the strong beneficial effects of the campaigns to eradicate polio and measles infections. Removing a live vaccine with beneficial NSEs after eradication could have major negative effects for child health in low-income countries. This problem may be important in relation to the approaching eradication of polio, measles and rubella infections.
• There are also long-term programming effects of vaccination, which are only starting to be studied.

Implications:
The OPTIMUNISE project has shown that NSEs have numerous implications for evaluation of vaccination practices, cost-effectiveness assessments, research practices, and policy planning.
Many practices for vaccinations, micronutrients and other interventions acquire a new meaning when they are seen from the perspective of NSEs.
First, wastage and restrictive vial policy. To limit wastage of vaccine doses it has become policy in many national vaccination programmes to only open a multi-dose vial if a sufficient number children are present, usually 10-12 for BCG (20-dose vial) and 6-7 for MV (10-dose vial) (14). This will delay the age of vaccination and will ultimately lead to lower coverage as we have shown in several studies. Since BCG and MV have the strongest beneficial effect on mortality it is obviously unfortunate that these vaccines are delayed (or not given). The study by Partner 4 of the age of BCG vaccination in Navrongo clearly suggested benefits from always opening a vial of BCG (65).
Second, combination and sequence of vaccinations interact. In the current programme missing vaccines are recommended to be given whenever there is an opportunity and there is no consideration of whether altered combinations or sequences have an effect. We have shown consistently that sequence matters. In general it would be advisable to have a live-vaccine-last policy. Until now MV has often been the last vaccine the children received. It is therefore of concern that WHO is planning a 2nd year of life platform for further vaccinations with several non-live vaccines, including booster DTP, Meningococcal vaccine, and possibly RTS,S malaria vaccine.
Third, micronutrients and vaccine interact. The possible interactions between micronutrients and vaccines have been ignored but OPTIMUNISE has shown that vitamin A supplementation may have beneficial effects in some situations but also major negative effects in other situation (29).
Fourth, sex-differential effects. We have repeatedly found sex-differential effects so all programmes should assess the intervention effect for both sexes.
Fifth, quality of vaccines and vaccinations. Normally a vaccine is just considered given when it is administered according to the vaccination card. There may be a few children who have not responded with a protective antibody level but this has no general consequences for immunization practice. However, the data for BCG-scarring suggest that this is not correct for BCG. There is a major difference in survival depending on whether the child developed a scar or not. Hence, it would be logical to consider revaccination and to use BCG-scarring to monitor the quality of BCG-vaccination. Both vaccination technique and to lesser extent strain of BCG appears to have an effect (17,37).
Sixth, monitoring indicators. Currently DTP3/penta3 is the most commonly used indicator for programme performance. Programme markers will naturally affect how the programme is performing and where it focuses (14). Given that DTP3 is associated with increased female mortality (48) this may have unfortunate consequences. The focus on DTP3 has deemphasized the importance of MV. The study of the fully immunized child (FIC) showed that delaying/not giving MV has strong negative effects for child survival (25). It would make much more sense to emphasize the vaccine indicators which are positively associated with survival, e.g. the age of and scarification after BCG vacation and the age and coverage for MV.
The NSEs Endgame: WHO and the SAGE review
The NSEs are not new products or new methods which can be assesses independently for their contribution to survival. If they were they would have been “sold” a long time ago and have improved child survival. NSEs are about how immune training or immune misdirection may affect child survival. The NSEs therefore also pose a major critique of assumptions underlying the current programmes for childhood interventions; because interventions interact, and may have negative effects, it becomes far more complex to find the optimal strategies.
The NSEs question many WHO policies but since it is WHO which decide global health policies it also WHO which will decide how and to which extent it will use the NSEs. This is the basis for an ambivalent position.
The WHO-SAGE review of the potential NSEs of BCG, DTP and MV on the survival of children under-five years of age has been a first major step by acknowledging that there are probably important NSEs. In this review there were strong beneficial effects of BCG and DTP the reduction in mortality being 47% (28-60%) and 46% (35-55%), respectively. However, the estimate for DTP went in the other direction being a 38% (-8-108%) increase in mortality. Hence, the effects of different vaccines were highly significantly different (P for the same effect of all vaccines<0.001; BCG versus MV p=0.92; BCG versus DTP, p<0.001; DTP versus MV, p<0.001. )

SAGE recommended further studies of the potential NSEs of vaccines and delegated the responsibility of prioritising further studies and protocol development to the IVIR-AC committee. Though two years have passed it has not been revealed which studies will be prioritised. So though the SAGE review does indicate an acknowledgement that NSEs may be important, it is still unclear how this perspective will be used by WHO.

Perspectives
In a world where NSEs are now increasingly acknowledged there are at least two different approaches. The WHO has requested more studies of NSEs of vaccines, but also emphasised that these studies should be randomised studies, as observational studies are prone to bias. RCTs are seen as the gold standard for assessing effects of interventions. This might be true in a “single solution” paradigm. However, if the NSEs perspective is pursued to its logical implications, the global health community needs to get beyond the “single solution” paradigm. From the perspective of OPTIMUNISE we need to assess the contextual conditions of interventions (45). We also need to recognise that effects are not necessarily constant, but may change with implementation or removal of other health interventions. Hence, we need not only other methods for determining the contextual conditions, but we need some form of continuous monitoring to assess whether intervention programmes continue to have the same effects. With OPTIMUNISE we have taken the first steps in demonstrating how this can be done through HDSSs within the INDEPTH Network.
These sites provide a good platform for combining monitoring of real-life effects with research training of the next generations.

Potential Impact:
OPTIMUNISE: Potential impact
Please provide a description of the potential impact (including the socio-economic impact and the wider societal implications of the project so far) and the main dissemination activities and the exploitation of results. The length of this part cannot exceed 10 pages.
Potential impact: socio-economic impact – and wider societal implications
The subject of OPTIMUNISE was how to reduce child mortality in the best and most cost-efficient way in resource-poor countries. An impact in this area could come from working with specific populations partially managing the health care system, from working with health policy makers at different levels to introduce new policies or change existing ones, and from working with other researchers and changing their perceptions of how evaluations of childhood interventions should be done.
Study populations
As described in the previous section, there has been a surprisingly sharp decline in under-five child mortality during the last 15-20 years. Though the three HDSS institutions in Burkina Faso, Ghana and Guinea-Bissau are not the main responsible for health care delivery to the affected populations, they have undoubtedly contributed to the process by maintaining focus and monitoring key indicators. OPTIMUNISE has also contributed to direct health care as part of the interventions studies going on in the study areas.
The fact that OPTIMUNISE documented that we had actually reached MDG4 in two sites is important. It came as a total surprise in Guinea-Bissau and has given an empowering “feeling” that it is possible to do much more than is usually assumed. It has been discussed with INASA, the national institute of health in Bissau, to speed up the process of suggesting evidence-based policy changes to the ministry of health (MOH). [The process has slowed due to repeated changes of government]. It was also quite clear during the Stakeholder meeting in Ghana that it was the marked decline in mortality in Navrongo which caught the Ghanaian media attention.
Showing this marked decline will also entail a responsibility to follow future changes when new interventions are introduced (e.g. IPV, RTS,S malaria vaccine, meningococcal vaccine) or existing ones are removed (e.g. OPV). Having documented the marked decline in specific populations provides an opportunity to study the process in greater details. The speed of the change fits poorly with current expectations about effects of different interventions. Hence, there is the possibility of obtaining a better understanding of the driving forces behind reaching MDG4 and we hope to use that opportunity in future research.

Policy implications
As described in the result section OPTIMUNISE has defined a whole series of areas where health intervention policies to reduce child mortality could be improved and which should therefore be taken into consideration in health policy planning and assessment. Some of these relate to changes in policy which could be implemented now and some relate to how assessment of interventions with vaccines and micronutrients should be carried out generally.
The most important changes which could be done now with existing knowledge or a little bit of additional research are:
• Live vaccines should be given earlier in life and with higher coverage since these vaccines tend to have beneficial non-specific effects (NSEs).
• As far as possible it should be a live-vaccine-last policy, or in other words non-live vaccines should not be given with measles vaccine (MV) or after MV.
• Live vaccines with beneficial NSEs should not be stopped because the targeted disease has been eradicated.
• BCG-scarring could be used to monitor the quality of BCG vaccinations.
• The restrictive vial policy for BCG and MV should be abandoned as it increases the age of vaccination and reduces the vaccination coverage.
• DTP, the most commonly used non-live vaccine, may have an overall negative effect on child survival. This has not been contradicted by any study with assessment of vaccination status and prospective follow-up. Hence, ways should be found to minimize those negative effects, e.g. by co-administering a live vaccine or using a live vaccine shortly after the non-live vaccine.
• The fact that we have documented declining maternal measles antibodies level and therefore a widening gap of measles susceptibility may help to test earlier measles vaccination.

Principles for assessing the impact of health interventions
• Non-live vaccine may have an overall negative effect on survival. Hence, every time a new non-live vaccine is introduced it should be examined that it is not associated with an overall negative effect. For example, this did not happen when RTS,S malaria vaccine was tested and we could document a long-term negative effect for girls (24,32).
• Girls and boys react very differently to immuno-modulatory interventions. Possible sexdifferential effects of new (and old) interventions should therefore always be assessed
• Interventions which act through the immune system may easily interact. Hence, it is necessary to always assess possible changes in effects due to interaction between existing and new interventions.
• Interaction between interventions may also mean that intervention effects which were measured at one point in time may no longer be valid if important other interventions are implemented or stopped. Vitamin A supplementation (VAS) may no longer have the strong beneficial effect it was believed to have (29) and we were not able to confirm a major reduction in mortality with early measles vaccination. There is therefore also a need for trying to understand how different interactions may change the overall effects, e.g. DTP coverage for the effect of VAS and neonatal VAS (29) or the OPV campaigns for the effects of early measles vaccination (45). As a consequence we also need to continue to monitor the main interventions to assure that they still have the assumed effects.
• The intervention programmes should use program indicators which are positively linked to the desired outcome, i.e. enhances child survival. The current situation in which the international programmes use DTP3 as main performance indicator is indefensible since DTP3 is linked to increased mortality, in particular female mortality (34,43,48,59).

These policy implications are far-reaching given existing modes of implementing health interventions. As discussed briefly below, there are even wider implications of the OPTIMUNISE work. We are summarizing how far the OPTIMUNISE consortium has come in relation to health policy makers and the research community.
Health policy makers
Since NSEs of vaccines were first documented in 1990 in connection with the trials of high-titre measles vaccine (HTMV), it has been the headquarters of WHO and various WHO committees which have dealt with the possible NSEs of vaccines. The HTMV trials showed that a protective measles vaccine (HTMV) was associated with two-fold higher mortality for girls but not for boys, thus contradicting key assumptions in the immunization programme (35). WHO-experts initially maintained that it was impossible; the studies from Guinea-Bissau and Senegal had not been planned and there was no plausible biological mechanism. However, when the same trend was observed by American researchers on Haiti WHO withdrew HTMV. This experience generated the research which has led to more and more NSEs being detected for different vaccines, their sequence, their combinations and their interactions with micronutrients. The OPTIMUNISE consortium has moved this research agenda considerably forward as presented in the previous section on results.
It has been important that WHO’s Strategic Advisory Group of Experts on Immunization (SAGE) established a working group on the NSEs in 2012-2013 and commissioned a review of the NSEs of vaccines which was presented to SAGE in Geneva in April 2014. The review suggested that BCG and MV may have major beneficial effects halving mortality which is not related to tuberculosis and measles infections whereas the majority of studies suggested that DTP is associated with increased mortality. For MV the review recognized that there were clear indications of sexdifferential effects. Furthermore, out-of-sequence vaccinations had NSEs for survival. SAGE recommended further studies of the NSEs. Professor Benn from Partner 1 was a member of the SAGE’s working group; contributing to the review took a considerable part of partner 1’s time in 2013-2014.
Though the SAGE review recognized several of our key observations, the review was still fraught with methodological problems particularly in relation to the analysis of the NSEs of DTP. The SAGE review said that though the majority of studies showed a detrimental effect of DTP (7/10) the studies were inconsistent because two studies showed a positive effect. SAGE also claimed that there was no sex-differential effect of DTP. The SAGE review included several studies with major survival or frailty bias. Due to lack of information some children get classified as “unvaccinated” because there is no information on their vaccination status. However, one of the key reasons for “lacking information” is that the child has already died. Due to such misclassification the mortality rate in the “unvaccinated” group becomes unnaturally high and as a result the estimated effect of DTP becomes very favorable. However, if we use only studies with registration of vaccination status as vaccinated or unvaccinated and prospective follow-up, then all studies show a negative effect of DTP, around two-fold higher mortality (34). These studies show also that DTP is associated with around 50% higher mortality for girls than for boys (48). Hence, OPTIMUNISE played a major role in pointing out these shortcomings in the SAGE review.
It is obvious that the international organisations which are involved with general delivery of vaccines (WHO, GAVI, UNICEF) have major problems with the NSEs because the NSEs question many basic assumptions on which the current system is based. We have had contact with GAVI and UNICEF but both organisations refer all decisions regarding general policy to WHO-SAGE. This means that potential policy implications of OPTIMUNISE’s work have to be directed to WHO-SAGE. We have tried to maintain this discussion by writing brief reports about selected subjects, e.g. the beneficial NSEs of OPV, the potential increase in mortality associated with RTS,S malaria vaccine, and the faulty analysis of the impact of DTP (WHO-Letters-I to V).
SAGE recommended further studies of the NSEs and transferred the matter to WHO’s implementation research committee (IVIR-AC) in April 2014. There have apparently been ongoing discussions in IVIR-AC about prioritization of the NSEs research question and the development of research protocols. However, OPTIMUNISE has not been made part of that process and the results of the discussions have not been made public so it is difficult to say at the moment to which extent OPTIMUNISE has influenced the WHO prioritization.
As documented in WHO-Letters-II to IV, the new RTS,S malaria vaccine which is moderately effective against malaria may in fact be associated with higher overall long-term mortality (24) and in particular higher female mortality (32). The Global Advisory Committee on Vaccine Safety (GACVS) has defined female mortality as a danger-signal for the RTS,S malaria vaccine. Until now this may be our most specific immediate impact within the WHO system.
The research we conducted before the OPTIMUNISE project showed that neonatal vitamin A supplementation was likely to have negative effects for female survival in areas with high vaccination coverage, possibly due to a negative interaction between NVAS and DTP (29). The subsequent trials conducted in Ghana and Tanzania with Gates and WHO funding supported no overall beneficial effect. In the meta-analysis of the African studies (where vaccination coverage is high) NVAS was associated with significant excess mortality (29). Though a final decision has not been made yet, it seems likely that NVAS will not become global policy. The work on the NSEs of vaccines will have contributed importantly to NVAS not becoming policy.
We will keep SAGE, GACVS and WHO/UNICEF Steering Committee for Guidelines on vitamin A supplementation informed about new developments in the area of NSEs of vaccines and micronutrients. WHO did send a representative to the OPTIMUNISE stakeholder meeting in Accra August 2016. However, given that WHO has been ambivalent and reluctant to discuss priorities, we have to base future work on the assumption, that policy implications will have to be enacted through research networks showing consistent patterns which could entail major health benefits.

Other researchers
There has been considerably more interest among other researchers in terms of utilizing the potential implications of OPTIMUNISE’s work on the non-specific effects.
Global health authorities have tended to dismiss the NSEs as biologically implausible but an increasing number of immunological studies are showing that both innate and adaptive immune mechanisms may enhance/diminish protection against unrelated infections. Several groups of immunologists in USA, Canada, Australia, Uganda, and Holland are working actively on possible mechanisms which would explain why vaccines have other effects than merely inducing specific-disease immunity (15). Research done in collaboration with the Dutch group has shown that live vaccines, like BCG and vaccinia, induce innate immune training. In contrast the non-live trivalent influenza vaccine (TIV) may have opposite effects and induced innate tolerance; the same was the case for the non-live typhoid fever vaccine (TFV). A network has been created for immunologists and epidemiologists who are interested in exploring the mechanisms of NSEs: the Optimmunize network (http://cviva.dk/Optimmunize.aspx).
A group of researchers working with vaccines in the national public health institute in the Nordic countries (Finland, Sweden, Norway and Denmark) are seeking funding to study the possible NSEs of vaccines in high-income countries which have good vaccines registers and good registers for hospital admissions. Researchers from Holland and UK are also examining the possible NSEs of vaccines in high-income settings. Given the huge costs of health care in high-income settings it might have wide implications if it could be shown that changes in age of vaccination or sequence of vaccinations had implications for hospital admissions. It would be ironic if it was the health care costs in high-income countries – rather than the child mortality implications in low-income countries - that meant that the NSEs were taken into consideration because they were cost-effective.
The OPTIMUNISE consortium is seeking funding to continue its work and a LOI has been submitted to EDCTP. The focus will be on how to optimize the beneficial effect of BCG studying such issues as the age of vaccination, the possibility of monitoring the quality of BCG vaccinations as measured by BCG scaring, and revaccination with BCG.
A small network of PhD researchers and students within the INDEPTH has also been made and has shown excellent work documenting NSEs of vaccination in Ghana, Kenya, Bangladesh and Burkina Faso. It is expected that this group will continue and explore the potential NSEs of vaccines.
Other activities are also planned within the INDEPTH Network. Since the HDSS sites have timely data from study populations sometimes 10 or 30 years back it is possible to trace the decline in mortality so we will seek collaboration to examine the road to MDG4, the effects of campaigns and what happen when the campaigns stops.
Industry may also become interested. Sanofi and Merieux organized a meeting on the off-targeted effects of vaccination in June 2015 which Partner 1 helped to plan (33). GSK has organized a task force on the non-specific of vaccines which took part in the final stakeholder meeting in Accra August 2016.

Wider implications
The OPTIMUNISE work questions many basic assumptions for current childhood intervention programmes, for example, that there are only specific effects, that any vaccine with a protective effect is beneficial, that effects are the same for girls and boys, and that interactions between the interventions do not affect their main outcomes. We have maintained for a long-time that these assumptions are flawed and we have not been disproven. This may sound surprising but is actually fairly simple: The current public health paradigm for childhood interventions only examined the specific effects, e.g. that DTP protects against whooping cough, but not the wider health implication, e.g. that DTP enhance susceptibility to unrelated infections. Hence, it is actually possible – as supported by all data (34,35,43,48,59) – that DTP protects against whooping cough but also increases mortality from unrelated infections. Hence, the OPTIMUNISE work is part of a paradigm shift.
Public health agendas should be defined based on solid data and health policies should be defined so that they optimize health. But the agendas are also part of bureaucracies which will defend the current system. OPTIMUNISE has strengthened the claim that the current policies are not the best, and with what we know at present, we could improve health. However, we also question basic assumptions and this is likely to be met by resistance. How one measures ‘potential impact’ in the middle of a paradigm shift is not self-evident.
As described in the result section, the OPTIMUNISE emphasis on the NSEs of vaccines and micronutrients have generated many consistent patterns which will have far-reaching implications, for example, that non-live vaccines have negative effects for girls, that live vaccines usually have beneficial effects, that eradication campaigns with MV and OPV have had major beneficial effect on child survival and that stopping vaccination after eradication may be dangerous. Assuming these ideas are right they will presumably eventually win but it may take a long time.

Main dissemination activities
Conferences: Results from the OPTIMUNISE project have first been presented at scientific meetings at the local institutions and then at the annual meeting of INDEPTH Network to reach a wider audience of researchers working at HDSS sites and who would have the possibility of implementing studies in similar settings.
Dr. A. Rodrigues from Partner 1 represented the consortium at the European Tropical Medicine Congress meeting in Basel September 2015. The European Commission had organized the meeting to discuss how research is utilised in policy. Christine Benn, Ane Fisker and Peter Aaby were invited as keynote speakers at several international meetings. Several members of the consortium have made presentations on the non-specific effects of vaccines and the work of the consortium in Geneva (September 2015), Paris (September 2015), Copenhagen (September 2015), Basel (September 2015), Bissau for EDCTP visit (October 2015), Bissau (October-November 2015), Addis Ababa (November 2015), Stockholm (November 2015), Copenhagen (March 2016), Leiden (May 2016), Johannesburg (August 2016), Accra (August 2016), Cape Town (August 2016), Cambridge (September 2016). Peter Aaby received an honorary doctorate from Universidade Nova de Lisboa in June 2015 for the work on the non-specific effects of vaccines.

Publications: So far OPTIMUNISE has submitted 64 papers for publication in leading international journals:
• 50 are published or accepted for publication,
• 6 are in revision,
• 9 are under review,
• 17 papers are at various stages of being drafted

Key results were presented the final stakeholder meeting held in the end of August 2016 in Accra. The power point presentations are available: Deliverable 11 The consortium partners have within the last project period had invitations to write editorials or review articles for PNAS, Lancet, Trends in Immunology, Expert Reviews of Vaccines, Nature Immunology, and Tropical Medicine and International Health. Transactions of the Royal Society of Tropical Medicine published a special issue on the NSEs of vaccines.

Other dissemination activities. We have used the press, radio or film to communicate the message that non-specific effects are important and that policy needs to take these effects into consideration.
A Danish crime novel by Sissel Jo Gazan is based on our research about the nonspecific effects of vaccines.
(http://salomonssonagency.se/php/book.ph p?lang=en&bookid=261). The book won the readers’ prize in Denmark in 2014 and has also been published in English (The Arc of the Swallow), American, German, and French.

An Austrian and a German film team are currently making films about vaccines and have included the theme of non-specific effects of vaccines.

Exploitation of results
Intellectual property rights:
There should no intellectual property rights in the form of patents in relation to the ideas presented here. The research results should become public domain. The data collected by the consortium will gradually be made available as data sharing once the results have been analysed by the consortium scientists.

List of Websites:
http://www.indepth-network.org/projects/optimunise

Related information

Contact

Torben Theilmann, (Financial director)
Tel.: +45 32683614
Fax: +45 32683165
E-mail
Record Number: 197115 / Last updated on: 2017-04-10
Follow us on: RSS Facebook Twitter YouTube Managed by the EU Publications Office Top