Analysis of Prevention Policies and Health Improvements Applied to the COVID-19 Pandemic

Abstract

In November 2019, a respiratory virus known as SARS-CoV-2, which causes the COVID-19 disease was first detected in China, rapidly leading to a global lockdown with significant impacts on economies, health systems, and societal structures. Traditionally, research has shown seasonal drivers in respiratory viruses due to factors like UV radiation, temperature, humidity, and human immune response, adding complexity to the classical SEIR epidemic models. As well, behavioural patterns, age, and social contact intensity have been seen to impact on overall transmission rates. The pandemic highlighted the need for robust systems to manage health crises and beyond the immediate health-related response, it caused economic and social disruptions, requiring companies to balance operations and employee safety. A major response involved shifting to remote work, affecting mental and physical health, productivity, and workforce engagement. Additionally, the pandemic triggered an economic crisis due to demand shocks from confinement. Moreover, cultural factors might have an influence on virus transmission and vaccination acceptance. Studies on societal behaviour might have opposite findings, while some authors highlight that individualistic societies may accelerate virus spread due to lower adherence to collective interests compared to collectivistic ones. Others discuss that individualistic societies might be expected to present less transmission rates due to lower intensity of physical contacts. Also, besides cultural aspects have been seen to significantly impact on vaccination intentions, healthcare infrastructure, government policies, and economic factors also play crucial roles in actual vaccination rates. Throughout this doctoral thesis, we present the data-science based decision-making process that led to define key measures taken by an important international financial institution headquartered in Spain to manage the pandemic within their corporate headquarters to ensure a secure workplace for personnel on premises, offering a detailed analysis of the protocols and tools used, and how they can serve as a model for future health crises. In this regard, the present work focuses on Madrid, Spain, using Machine Learning algorithms to propose both a reactive and preventive measurements. Among the reactive measurements we include the detection of focal infection points, statistical analyses to test new diagnostic kits, the design of a "return to office" protocol and a continuous follow-up of the evolution of the pandemic. Within the preventive measurements. We include behavioural analyses considering worker nature, seasonal effects, and psychosocial demands. Professionals conducted regular COVID-19 diagnostic tests utilising a data-driven decisionmaking process to optimise resource deployment for early detection and isolation of COVIDpositive cases, resulting in a total of 55,789 tests. The sanitary team conducted individual follow-ups for all personnel and recorded the information in databases. Individualised control panels enabled real-time monitoring to adjust restrictive measures as necessary. This process ensured that actions were appropriate for the evolving situation, such as identifying hotspots for quarantine assignment when needed. A positive correlation was observed between the cumulative incidence reported by Madrid’s Ministry of Health and the headcount. Moreover, 1.7% of individuals continued to test positive for COVID-19 after completing a 14-day quarantine period, which justified the decision to maintain this quarantine duration. Six occupational outbreaks were identified between the second and sixth waves, exhibiting varying degrees of severity and impact. Despite the high community incidence rate, occupational infections within the bank remained notably low, representing only 1.9% of the total positive cases. The study revealed significant variations in COVID-19 transmission based on demographic, organisational, and seasonal factors. Higher infection rates were found among higher seniority levels and certain departments, suggesting the need for tailored interventions. Seasonal variations also played a role, with higher transmission rates noted during the winter months compared to the summer along with a higher subjective identification of associated symptoms in summer. A combined approach using medical and computational tools was implemented to ensure workplace safety during periods of high transmission. The study demonstrated that targeted action strategies, such as identifying high-risk individuals and implementing specific preventive measures, were more effective and cost-efficient compared to extensive screenings across the entire workforce. Algorithm-based medical screenings yielded higher detection rates, allowing for more accurate identification of potential cases. Additionally, the study highlighted the importance of adapting preventive measures to various demographic, organisational, and seasonal variables to maximise efficiency. In summary, the COVID-19 pandemic has highlighted the critical importance of preparedness and adaptability amidst health emergencies by integrating both biomedical and computational tools. The insights obtained from this study can inform future strategies for managing similar situations, emphasising the necessity for a comprehensive and data-centric approach for dynamic and responsive strategies, ensuring that interventions remain relevant and effective in different contexts. Finally, these approaches can serve as a model for other organisations facing future health crises.

En abierto se puede consultar la parte no embargada de la Tesis Doctoral