The eradication paradox: can stopping oral polio vaccine have negative consequences?

Introduction
“One drop can save many lives” has been the slogan for oral polio vaccine (OPV) campaigns to achieve the goal of global polio virus eradication, where the aim has been to reach all children under 5 years of age. In the last three decades, children under 5 years of age in low-income countries have therefore experienced numerous OPV campaigns. Simultaneously, there has been a remarkable decline in under-5-year child mortality.

According to the current paradigm for vaccines, the two factors, OPV campaigns and the decline inchild mortality, should have nothing to do with each other. Wild poliovirus is almost eradicated andthere has been very little disease from poliovirus to protect against, and in addition, polio is rarely adeadly disease. However, epidemiological studies have shown that live vaccines, such as OPV, mayhave beneficial non-specific effects (NSEs) providing protection against other infections than theinitially targeted one.

We therefore set out to investigate whether OPV campaigns could in fact be part of the reason forthe marked decline in child mortality. Did “one drop” save more children’s lives than anyone hadever imagined?

Methods
We evaluated the quality of the OPV campaign information by comparing campaign informationfrom various data providers. We compared the World Health Organisation (WHO) and The RotaryFoundation (Rotary) campaign information, collected for administrative purposes, with the campaigninformation collected by local health and demographic surveillance systems (HDSS) sites (Article I). 

 With prospective population cohort data from 2004 to 2019 from the rural Chakaria HDSS insoutheastern Bangladesh, we analysed the effect of OPV campaigns on child mortality from day 1 to3 years of age by comparing mortality after versus before OPV campaigns (given withoutintervention) in Cox proportional hazards models, adjusted for age, time trends and other healthinterventions (Article II). In the Chakaria data from 2012 to 2019, we analysed the effect of OPVcampaigns on different causes of death (Article III).

In three randomized controlled trials (RCTs) of an early 2-dose measles vaccine (MV) strategy fromGuinea-Bissau conducted from 2003 to 2019, we investigated whether OPV campaigns modified the7mortality effect of the RCT intervention, and furthermore, we investigated the overall effect of OPVcampaigns on child mortality (Article IV).

In prospective population cohort data from 2002 to 2014 from the urban Bandim Health ProjectHDSS in Guinea-Bissau, we modelled the impact of OPV campaigns on under-3-year mortality in twodifferent counterfactual scenarios. We assumed that children had never or had always been exposedto OPV campaigns, as alternatives to the original OPV campaign exposure (Article V).

Finally I summarised the evidence of the NSEs of OPV and OPV campaigns. Key observations wereincluded both from this PhD dissertation, but also from the existing scientific literature. Theobservations came from different datasets with different underlying bias structures. I generatedseveral alternative hypotheses and triangulated which hypothesis could provide the best explanationfor the evidence.

Results
Regarding the quality of various sources of campaign data, we found that the campaign datacollected by the WHO was incomplete and sometimes wrong. This was in particular the case forvitamin A campaigns, which was often co-administered with OPV. This could be misleading if onlyWHO campaign data was used to assess the impact of OPV campaigns without other interventions.Using WHO campaign information could cause misclassification of OPV campaigns co-administeredwith or without VAS. Rotary had better information, but had only collected information on OPVcampaigns and not on other interventions.

In the Chakaria HDSS data we corroborated the previous findings from West Africa, as we found OPVcampaigns to be associated with an overall reduction of child mortality between day 1 and 3 years ofage of 31% (95% confidence interval (CI): 10-48%) comparing time after vs. before OPV campaigns.No other campaign type had similar beneficial effects. The beneficial effect of OPV campaigns wasprimarily explained by a significant reduction in deaths caused by respiratory infections, the riskreduction being 62% (95% CI: 20-82%).

In three RCTs from Guinea-Bissau, OPV campaigns interacted with the early 2-dose MV strategy.Among the children who had received no campaign-OPV-before-enrolment the 2-dose/1-dose MVhazard ratio (HR) was 0.79 (95% CI: 0.62-1.00) while the corresponding HR for those who hadreceived campaign-OPV-before-enrolment was 1.39 (95% CI: 0.97-1.99) (test of heterogeneity,p=0.01). These effects were further amplified by campaign-OPV-after-enrolment and having received8OPV0 within two weeks of birth. A beneficial effect of receiving OPV campaigns was found amongchildren who received one dose of MV, here each additional dose of campaign-OPV was associatedwith a HR of 0.85 (95% CI: 0.76-0.95). The opposite tendency was seen in the 2-dose group with theHR being 1.09 (95% CI: 0.96-1.23) per additional dose of campaign-OPV (test of heterogeneity,p<0.001).

From 2002 to 2014, the Bandim Health Project in Guinea-Bissau followed prospectively a cohort of55,814 children between day 1 and 3 years of age. During this time 2,834 children died (5.1%).During this period all-cause mortality declined by 69%. Comparing a scenario where children hadnever been exposed to OPV campaigns with the original exposure, our model predicted the average(in 1,000 simulations) all-cause child mortality rate was 9.9% (95% CI: 5.1-14.5%) lower because ofthe conducted OPV campaigns. This corresponded in absolute numbers to an average of 292 deaths(2.5-97.5% sample quantiles: 148-435) averted by OPV campaigns conducted within the HDSS studyarea. The number of deaths averted was further increased to 939 deaths (2.5-97.5% samplequantiles: 803-1,082) if children had always been exposed to OPV campaigns compared to thescenario where no children had been exposed to OPV campaigns.

Conclusion
The current results from Guinea-Bissau and Bangladesh corroborate previous findings on the NSEs ofOPV campaigns. Triangulation of the evidence indicates that the beneficial NSEs of OPV and OPVcampaigns is the only plausible hypothesis to explain not only significant reduced all-cause childmortality, in particularly for deaths due to respiratory infections, but also the altered effect of anearly 2-dose MV strategy observed consistently in three RCTs. The modelling supported that OPVcampaigns have had a role in the decline in child mortality in the last decades, and may have loweredall-cause mortality by around 10% in Guinea-Bissau given the number of OPV campaigns conductedin the country.

 When wild poliovirus is eradicated, OPV will be stopped. This is now planned to happen in 2026. Thisrepresents a paradox since stopping OPV will also remove the beneficial NSEs. It is therefore crucialto further investigate the NSEs of OPV and look for ways to mitigate the potential negative effects ofstopping OPV.

When there are “no more drops left” to save children’s lives, do we then know what will happen?