We assessed the effect of vaccination campaigns that used IPV alongside OPV on the transmission of wild-type and VDPV. In Nigeria, we observed a substantial reduction in the incidence of poliomyelitis associated with cVDPV2 and in the detection of cVDPV2 in environmental samples after mass campaigns with IPV+tOPV. This reduction was greater than that observed after campaigns that used tOPV alone, although the difference was not statistically significant. This finding suggests that the substantial boost to intestinal immunity after vaccination with IPV, as observed in recent OPV challenge studies (12,13), translates into a measurable effect of IPV campaigns on poliovirus transmission in the community. These encouraging findings support the use of IPV in campaigns during the polio endgame to eradicate remaining wild-type polioviruses, increase population immunity in advance of withdrawal of OPV serotypes, and respond to any outbreaks of vaccine-derived poliovirus. These results are particularly relevant in the context of OPV withdrawal, which began with the globally synchronized withdrawal of OPV serotype 2 (OPV2) in April 2016, because IPV has the advantage of providing a boost in mucosal immunity while avoiding release of new Sabin viruses that could evolve to VDPVs.

In Pakistan, an effect of mass campaigns with IPV+tOPV on persistent WPV1 circulation was not apparent. This finding may partly reflect a lack of statistical power. In the sensitivity analysis, using a longer time window of 150 days resulted in a significant reduction in the incidence of poliomyelitis: IRR 0.56 (95% CI 0.33–0.95) for IPV+tOPV campaigns and IRR 0.67 (0.53–0.86) for tOPV-only campaigns. However, this reduction was not apparent for WPV1 isolated in the environment (PR 0.85 [95% CI 0.59–1.23] for IPV+tOPV campaigns, and PR 1.00 [95% CI 0.76–1.30] for tOPV-only campaigns). In Pakistan, the lack of evidence for an effect from IPV campaigns may also reflect low coverage during the campaigns and restrictions on the age groups that were targeted for vaccination. Campaign coverage was suboptimal in Pakistan during much of 2014–2015; in September 2015, in areas at high risk for polio, just 39% of union councils were estimated to have achieved >80% campaign coverage, although this improved to 62% in November 2015 (18). Estimates of campaign coverage during January 2014–June 2015, which were based on the vaccination histories of children reported with nonpolio acute flaccid paralysis, also suggest that coverage among undervaccinated communities in Pakistan was poorer than that in Nigeria (19). Suboptimal vaccination coverage means that fewer children will benefit from a boost in intestinal immunity by IPV, not only as a result of smaller numbers receiving the vaccine, but also because fewer children will have been mucosally primed by OPV given in earlier campaigns or though routine immunization systems. According to WHO/UNICEF, routine immunization coverage with OPV in 2015 was suboptimal (≈75% on average) in both Pakistan and Nigeria.

During vaccination campaigns, it is more challenging to deliver IPV than OPV because IPV must be administered by trained healthcare workers and is thus offered at specific locations only (e.g., health centers) rather than being delivered directly to households, as with OPV. Nonetheless, in Nigeria and Pakistan, IPV campaign coverage has been reported to be comparable to that for OPV-only campaigns (20,21). However, in Pakistan, IPV was administered only to children <2 years of age, and in Nigeria (with the exception of 1 campaign), children <5 years of age were included. The wider age group in Nigeria may have contributed to the greater effect of IPV on poliovirus transmission because children 2–4 years of age may shed poliovirus and contribute to transmission despite being protected against poliomyelitis.

The difference in the effect of campaigns that used IPV in Pakistan and Nigeria may also reflect differences in the circulating poliovirus types (cVDPV2 in Nigeria and WPV1 in Pakistan). However, this seems unlikely because the transmissibility and pathogenicity of cVDPV2 in Nigeria appears equivalent to that for wild-type poliovirus (22).

Our analysis has several limitations, including its observational nature and lack of randomization of vaccines used during campaigns, which could have resulted in systematic differences in the areas that used IPV+tOPV compared with tOPV alone. We attempted to account for these differences in our statistical analysis by allowing for random variation in the incidence of poliomyelitis and prevalence of virus isolation by state or province and by vaccine(s) used in the campaign. The statistical power of our analysis was also limited by the low incidence of poliomyelitis in areas with IPV+tOPV campaigns. In Nigeria, the estimate of the effect of IPV+tOPV campaigns against poliomyelitis was driven by the results from Borno State, where an emerging cVDPV2 outbreak was apparently stopped by these campaigns. Environmental surveillance data from other states (Sokoto, Kaduna) support the effectiveness of IPV in campaigns, but the number of sampling sites informing the estimates was limited. As further experience with the use of IPV in campaigns is acquired, and with the planned expansion of environmental surveillance, it will be possible to refine our estimates. It may also become possible to increase the strength of our analyses by using additional statistical techniques, such as interrupted time-series methods, which are currently not appropriate, given the limited number of observations. Although a cluster-randomized trial would deal with any biases introduced through a lack of randomization in our study, such a trial would not be ethical, and to achieve sufficient statistical power would require such a large study as to be impractical.

Another limitation of our study is that we used poliovirus isolation in environmental (wastewater/sewage) samples as a proxy for poliovirus transmission in the community. Although environmental surveillance is known to be highly sensitive for poliovirus circulation in the catchment population, even in the absence of poliomyelitis cases, it may not capture more subtle effects of vaccination campaigns on the extent of poliovirus transmission within an area (23). Last, our before-and-after comparisons may be confounded with seasonal and longer term trends in poliovirus transmission. However, in Nigeria and Pakistan, the median months for IPV+tOPV campaigns were May and June, respectively, compared with June and July, respectively, for tOPV campaigns; the similarity of this timing suggests that the comparison of these campaigns was not confounded by seasonal trends. In addition, the absence of any effect of bOPV against cVDPV2 in Nigeria, as was expected, suggests that any confounding, if present, was minimal.

Our analysis offers support for the use of IPV in mass vaccination campaigns to stop poliovirus transmission, provided good coverage can be achieved. This use of IPV in campaigns should be pursued while maintaining sufficient supply of IPV for routine immunization in high-risk OPV-using countries (24). Any vaccine shortage could be compensated for by the use of fractional dose IPV, provided there is sufficient evidence of its efficacy in boosting immunity. The greater effect in Nigeria of IPV+tOPV campaigns compared with tOPV-only campaigns suggest that IPV-only campaigns may also offer substantial benefit. However, campaigns with IPV alone have not been implemented. Evaluation of the potential benefits of IPV-only campaigns in the context of waning intestinal immunity, a growing cohort of children who have not received OPV2, and global containment of OPV2 after withdrawal of the vaccine serotype should be a priority. Results of continuing programmatic experience will also inform efforts to improve the levels of coverage and the benefits of including children 3–4 years of age in IPV campaigns. Regardless, it is now clear that IPV will play a major role in securing the global eradication of poliomyelitis and in maintaining a polio-free world.