Pandemic Preparedness and COVID-19: lessons learned from national and subnational response, what we can learn from existing preparedness metrics, and how to prepare for novel threats

Abstract

The COVID-19 pandemic has been one of the most catastrophic health emergencies of all time, leading to millions of deaths and hundreds of millions of infections worldwide. Yet, a global pandemic impacting all of humanity was not unforeseen, with countless studies articulating the spillover potential of various pathogens into humans and the ability for such viruses to replicate. Similarly, pandemic preparedness frameworks and metrics existed pre-COVID to quantify pandemic risks and vulnerabilities for a given country to emphasize both strong existing capacities as well as areas for improvement for emerging outbreaks and pandemics of international concern. The unprecedented scale of COVID-19 has renewed focus on pandemic preparedness and response. This research aims to understand the drivers of differential COVID-19 outcomes across various countries and within key countries and to quantify how current preparedness indices of immunization measure up against retrospective COVID-19 vaccination uptake. In the first aim, we consider whether existing pandemic preparedness indicators nationally were informative of better COVID-19 outcomes, and investigate what other political, social, health or demographic covariates influenced heterogeneities in COVID-19 across the globe. In the second aim, we focus our analyses subnationally to investigate drivers of within-country heterogeneities, and again investigate whether pre-pandemic preparedness was informative of COVID-19 outcomes, and whether national patterns were persistent subnationally. In the third aim, we build on our findings from the first and second aim that pandemic preparedness composite measures were not predictive of COVID-19 successes, and decompose such metrics to investigate one specific indicator of routine immunization in order to understand whether our measurement of vaccine readiness was truly informative of pandemic vaccine delivery. In the first chapter Pandemic preparedness and COVID-19: an exploratory analysis of infection and fatality rates, and contextual factors associated with preparedness in 177 countries, from Jan 1, 2020, to Sept 30, 2021, we used multi-stage log-log regression models to understand drivers of cross-country differences in COVID-19 infections and mortality from January 2020 through September 2021, expanding on previous research looking at COVID-19 outcomes in relation to pandemic preparedness scores. We first controlled for immutable factors including daily seasonality and age profile, and secondly controlled for baseline risk factors like age profile, seasonality, density, and BMI. Following these adjustments, we modeled our standardized COVID-19 infections and infection fatality rates against policy-amenable factors, including pandemic preparedness indicators, health care readiness, and social and political characteristics. We found that the largest drivers of reduced COVID-19 infections were not associated with existing pandemic preparedness metrics, but instead to higher trust in the government and in other people. As trust is an essential driver of effective risk communication and behavioral modification such as vaccine uptake or social distancing, improving trust prior to the next pandemic is essential. In the second chapter An exploratory analysis of improved COVID-19 outcomes in subnational locations across two countries: the United States and Brazil, January 2020 through May 2022, we again used multi-stage log-log regression models to understand within-country drivers of COVID-19 outcomes in Brazil and the United States. These two countries were chosen for their high overall COVID-19 burdens, but also for heterogeneous COVID-19 burdens, responses, and high political polarization. This time, our analysis ran from January 2020 through May 2022, and we controlled for daily seasonality and variant prevalence in the first stage, followed by a standardization for baseline risk factors. We again modeled these standardized estimates versus policy-amenable factors, including pandemic preparedness indicators, health care readiness, and social and political characteristics, though many of the covariates from the first chapter were not available at the state-level and either had to be modeled or omitted. Although there were observable differences within these countries, we identified no significant policy-amenable drivers of COVID-19 differences within countries following baseline standardization, though hospital beds per capita were found to be significantly related to higher infections. Trust was not a key driver of COVID-19 outcomes in Brazil and the United States, though the sample sizes of our trust variables were small and had wide confidence intervals. Similarly, modeled pandemic preparedness indicators were not predictive of improved COVID-19 outcomes subnationally. Our research additionally suggests that access to high quality health care is a potential avenue to explore to reduce the burden of disease in future pandemics. Within-country efforts to prepare for the next pandemic may be best focused on improving access to care and reducing existing burdens of comorbidities such as obesity and cancer which drive not only high mortality in general but are exacerbated in pandemics like COVID-19 where undue morbidity and mortality were observed among the chronically ill and elderly, and to improve estimates of trust and pandemic preparedness at a local level to better understand true disparities nationally. In the final chapter Considering measles containing vaccine as a proxy for pandemic preparedness in the context of COVID-19: are we truly measuring what matters?, we look at a country’s routine measles containing vaccine (MCV) coverage – a proxy for immunization readiness and vaccine delivery in preparedness metrics like the Joint External Evaluation (JEE) – and model it against at least one dose COVID-19 vaccination between December 1, 2020 and December 1, 2022 for national and subnational locations separately. Vaccination has been an incredible tool throughout the COVID-19 pandemic in reducing morbidity and mortality, but has been highly inequitable in its distribution, so we wanted to understand how closely heterogeneities in MCV as a routine measure mapped to those seen in COVID-19 vaccine uptake. Moreover, composite scores of pandemic preparedness proved non-informative of COVID-19 outcomes, and so we sought to understand whether specific indices were beneficial in understanding specific aspects of the COVID-19 pandemic, such as immunization. In each location, we consider the time to various thresholds of coverage (1%, 5%, and 10%) to understand the relationship between pre-pandemic immunization and speed of novel vaccine roll out, controlling for pre-pandemic vaccine hesitancy and percentage of the population over 65 years of age. We consider the maximum number of persons vaccinated in a single day (smoothed and averaged over a one-month period to adjust for noisy data) as a measure of speed of scale-up and separately estimate the maximum level of coverage achieved for at least one dose COVID-19 vaccination coverage. We model these additional COVID-19 vaccine outcomes against MCV as well, again controlling for vaccine hesitancy and percentage of the population over 65 years of age. Our research suggests that the level of pre-pandemic one-dose measles vaccine coverage across 134 countries was successful in predicting the time to vaccine roll out at varying thresholds and the overall vaccination level achieved. In the subnational model, we found no significant relationships between routine immunization coverage and one-dose COVID-19 vaccine delivery, a relationship that persisted across all data subsets and additional indicators of routine immunization. While composite metrics of pandemic preparedness are not effective at predicting pandemic COVID-19 outcomes, specific, targeted indicators have stronger predictive validity than the composites. Specifically, this analysis demonstrates that measles vaccine coverage is an effective metric for quantifying immunization readiness at the national level, and can provide utility for considering equitable delivery of vaccines and therapies for future threats. This dissertation takes a comprehensive look at pandemic preparedness measures and assess their validity with a COVID-19 lens. We find that pre-pandemic composite scores held little validity for predicting better COVID-19 outcomes across the globe, nor within two key countries. However, when drilling down to individual components of these composite scores, we found a correlation between pre-pandemic immunization readiness and pandemic vaccine delivery at a national level, suggesting the possibility of wider validity of individual indicators for various facets of pandemic preparedness and response. Trust in other people and in the government was identified as a key driver of lower COVID-19 infections in the national analysis, though lack of focused, high-quality subnational data limited the extension of these findings in the second, subnational aim. Considering trust as an essential tool to build prior to subsequent pandemics and monitor on an ongoing basis, as well as focusing dedicated resources and efforts to better quantify and understand trust nationally and subnationally will have extensive payoffs in better understanding current patterns of trust, as well as improving messaging and adherence for novel threats. COVID-19 was the first of what is likely to be many outbreaks and pandemics in our lifetimes given a growing number of interactions between humans and wildlife due to climate change and urbanization, allowing for spillover of novel and re-emerging pathogens into human populations. Lessons learned from this dissertation would be useful for guiding pandemic preparedness plans to consider how to reformat existing metrics to be most suited for planning for a multitude of different diseases in the future, and to ensure that subnational capacities and vulnerabilities are thoroughly addressed.