Evaluating the effects of polio vaccines on general child health in Guinea-Bissau

Background
While OPV has been key in preventing and controlling polio, evidence also suggest that OPV has contributed to lowering child mortality by up to 15-30% inlow-income countries through a non-specific effect (NSE) protecting against non-polio infections. However, the use of OPV carries a risk of vaccine-associated paralytic polio and vaccine-derived polioviruses spreading between children. Therefore, to eradicate poliovirus, the global plan is to replace OPV with the inactivated polio vaccine (IPV). IPV will ensure protection against polio. However, IPV induces a weaker mucosal immune response and does therefore not prevent transmission of poliovirus. Furthermore, limited evidence suggests that unlike OPV, IPV does not have beneficial NSEs. We therefore need real-life data on what may happen when OPV is replaced by IPV.

Aim
The aim of this thesis is to evaluate the potential implications of replacing OPV with IPV, specifically whether this may augment child morbidity and mortality in Guinea-Bissau due to depriving children of the benefit of OPV and exposing them to a potential negative effect of IPV.

Methods
Study 1. In a cluster-randomised trial, we used the Bandim Health Project’s (BHP) health and demographic surveillance system (HDSS) to test the effect of campaign-OPV (C-OPV) vs no C-OPV on non-accidental mortality/hospital admission, consultation, illness, and growth among children aged 0-8 months. In Cox proportional hazards models with age as the underlying timescale, we compared rates of non-accidental mortality/hospital admission (composite outcome) during 12 months of follow-up by trial arm. Hazard ratios (HR) were estimated with 95% confidence intervals (CI) using cluster-robust standard errors to account for intra-cluster correlation. The effects of C-OPV on growth, illness, and consultation were estimated using linear and log-binomial regression models to obtain mean growth differences and relative risks (RR) with 95% CIs.
Study 2. During an OPV-shortage in 2004, we investigated the effect of not having received OPV with diphtheria-tetanus-pertussis (DTP) vs DTP+OPV on the risk of hospital admission/at-hospital mortality (severe morbidity) among children aged 6 weeks to 14 months attending outpatient consultations. Using BHP’s urban HDSS, including registry of all outpatient consultations and admissions at the National Hospital Simão Mendes (HNSM), we assessed crude and adjusted estimates of risk of severe morbidity in log-binomial regression models using propensity score matching and cluster-robust standard errors to account for children who had more than one consultation.
Study 3. In August 2016, IPV was introduced in Guinea-Bissau, to be given alongside the 3rd pentavalent vaccine (penta), and due to an IPV-shortage in May 2017, IPV was re-introduced in January 2019. Using the BHP’s urban HDSS, including data from health centres and the HNSM, we compared rates of outpatient consultations and hospital admissions between children who received penta3+OPV3+IPV vs penta3+OPV3. Outcome rates between receipt of penta3 and age 9 months were compared overall and by sex in Cox proportional hazards models with age as the underlying timescale. HRs were estimated with 95% CIs adjusted for weight-for-age z-score. Furthermore, HR for consultation was adjusted for the underlying consultation rates, and HR for hospitalisation and mortality was adjusted for season of penta3 vaccination.

Results
In study 1, we found no marked overall effect of C-OPV vs no C-OPV on nonaccidental mortality/hospital admission (HR: 0.87, 95% CI 0.68-1.12), reported illness (RR: 0.95, 95% CI 0.84-1.09), risk of consultation (RR: 0.86, 95% CI 0.70-1.06) nor growth among children aged 0-8 months in rural Guinea-Bissau. However, interactions were observed with timing of OPV at birth (OPV0) for the composite outcome (p=0.04), driven mainly by a differential effect on mortality.
In study 2, we did not find an increased risk of severe morbidity among children attending outpatient consultations who received DTP-only vs DTP+OPV as the most recent vaccine during the OPV-shortage in 2004 (RR: 0.85, 95% CI 0.60- 1.21). However, children attending outpatient consultations who received DTPonly appeared to have a lower risk of admission due to measles infection.
In study 3, contrary to our hypothesis, children who received IPV with penta3+OPV3 had no increased risk of non-accidental consultation, hospital admission and mortality compared with children who only received penta3+OPV3 (consultation: HR 0.94, 95% CI 0.86-1.03; hospital admission: HR 0.88, 95% CI 0.56-1.37; mortality: HR 0.47, 95% CI 0.15-1.48). However, IPV administration may be associated with an increase in the female/male mortality ratio.

Conclusion
The beneficial effect of OPV was less than anticipated, and no difference in risk of consultation, hospital admission and mortality was observed after co-administration of IPV with penta3+OPV3. However, potential interactions such as with sex, timing of OPV0, and time since vaccination need to be explored further in studies assessing the effect of OPV vs IPV. We recommend to further clarify the overall and sex-differential effect of OPV vs IPV using real-life data before replacing OPV with IPV to ensure that a polio free world does not deprive children the possibility of better health and survival.