Background: Routine Immunisation (RI) refers to the administration of vaccines at standardised touchpoints largely to infants aged < 24-months-old. Vaccination saves lives: the World Health Organisation (WHO) estimate that 3.5 to 5 million deaths are averted annually through immunisation. RI performance is monitored using coverage – the percent of target populations reached – and diphtheria-tetanus-pertussis containing vaccine third-dose (DTP3) coverage is often used as a key indicator. DTP3 coverage globally increased from 20% in 1985 to 85% in 2000 and plateaued for the next 20 years. In 2020, the COVID-19 pandemic placed unprecedented challenges on health systems and routine health services delivery worldwide.
Objectives: Firstly, quantify country- and global-level disruption to RI coverage compared to levels expected based on prior trends (i.e., in the absence of the pandemic – an important distinction compared to standard approaches comparing to absolute 2019 values). Secondly, explore the drivers of disruption globally by developing a conceptual framework and underlying dataset of potential determinants of RI resilience. Thirdly, use this framework to quantitatively test hypotheses to identify potential drivers of disruption. The overarching aim is to inform priorities areas for recovery and resilience building efforts, in the context of limited resources.
Methods: We use Auto Regressive Integrated Moving Average (ARIMA) forecasting to predict country DTP3 coverage levels in the absence of the pandemic and calculate annual ‘coverage deltas’ – reported coverage minus modelled (i.e., expected) coverage. A conceptual framework for vaccine disruption determinants was developed, and dataset consolidated and cleaned (in some cases using Principal Components Analysis to synthesise multi-dimensional correlated clusters of variables into synthetic variables in reduced dimensions). Multiple methods are employed to explore potential associations between coverage deltas and explanatory indicators: (i) Discriminant Analysis of Principal Components (DAPC) to predict classification of country performance (below, within, or above expectations); (ii) linear regression and random forest regression to predict the continuous outcome of modelled coverage deltas.
Results: RI coverage globally declined throughout the pandemic, with the greatest disruption in 2021 when DTP3 coverage was 3.4 percentage points [95% Confidence Interval: 2.5-4.4%] below expectations (i.e., ARIMA-modelled). This decline reflects coverage reverting to 2005-levels, i.e., an almost 20-year backsliding in progress. In 2023, after the end of the pandemic period, RI remained statistically lower than expected in almost 20% of countries, with many others exhibiting stagnation. No clear factors were identified to classify country performance using DAPC. Linear regression indicated that stronger pre-pandemic immunisation programmes (p < 0.0001) and more health workers (p = 0.0065) once above minimum thresholds (78% DTP3 coverage and 58 health workers per 10,000 people), are associated with improved RI resilience. These findings were supported by the random forest, which also suggested the importance of health financing and health system strength – potential underpinning factors for immunisation programmes and health workforce.
