General discussion 9 289 Moreover, untangling the key pre-infection risk factors of long COVID is also prone to the introduction of bias, given that several pre-COVID characteristics also increase susceptibility of severe disease. For instance, cohort studies focussing on hospitalised COVID-19 patients are prone to collider bias[121], given that certain risk factors (such as obesity) are associated with the chance of inclusion into the study. In addition, severity of acute COVID-19 acts a mediator between pre-COVID risk factors and long COVID. Adjusting for acute COVID-19 severity can therefore undermine the ability to estimate the total effect of the risk factor of interest. Adopting a causal inference approach may therefore help researchers explore the relationship between predisposing factors and the outcome of interest in a methodologically sound way. Long COVID researchers should be vigilant of these methodological complexities and share best practices as we develop an evidence base for the condition. 9.4 CONCLUDING REMARKS In summary, in this thesis I have outlined the epidemiology and long-term clinical features of COVID-19 in Amsterdam, the Netherlands. Our findings illustrate that existing health inequities are accentuated during a time of crisis. Working collaboratively to further understand and combat the drivers of these inequities should be a priority. We are also only beginning to understand the cumulative burden, clinical features, determinants, and pathological processes of long COVID and its various sub-types. There is an urgent need to invest in diagnostic and therapeutic tools for this often-debilitating condition; ideally in studies that involve patients from all backgrounds and regions, and encourage participants to take part in setting the research agenda. In response to the COVID-19 pandemic, several developments in Dutch disease surveillance have taken place to enhance our capability to detect disease outbreaks. On a global scale, however, highincome regions can do more to support capacity-building in low-resource settings to ensure timely, high-quality surveillance of infectious diseases. We have also learned that utilising existing population-based studies can help to rapidly generate crucial information during a crisis. This has been particularly beneficial in research around long COVID, where pre-infection data has been invaluable to reducing some of the key biases in long COVID research. Moving forward, I suggest actively incorporating the ‘lessons learned’ presented in this thesis to help mitigate the impact of a future pandemic.
RkJQdWJsaXNoZXIy MTk4NDMw