Learnings from a pandemic: Evaluation of Outcomes, Risk Factors and Digital Applications in COVID-19

During the COVID-19 pandemic, three studies were conducted to understand the implications of the
disease in hospitalized patients and how our definition of endpoints can impact results in this
population group, and to evaluate risk stratification by either level or respiratory support or chest Xrays as a severity disease marker. Lastly, we supported hospital operations during the pandemic
with a digital health application.
Study I: The first paper challenges the use of the outcome recovery in COVID-19 trials and
compares outcomes before and after the introduction of dexamethasone and remdesivir in COVID19 treatment. In a retrospective observational analysis of more than 3,000 hospitalized COVID-19
patients, we identified a high incidence of readmissions and post-discharge mortality among
patients hospitalized with COVID-19. Time-to-recovery outcomes do not account for such events,
and they were, therefore, not representative of the “true recovery” without disease relapse. Thus, in
the same cohort, we estimated both a recovery (defined as discharge from hospital) and a sustained
recovery outcome, which considered post-discharge events up to 14 days after hospital discharge.
Doing this ultimately lowered the cumulative incidence of sustained recovery compared to
recovery. In addition, we found patients with critical disease to have a prolonged time to sustained
recovery, suggesting an ideal follow-up time between 60 and 90 days. Also, we found a higher
cumulative incidence of sustained recovery after the introduction of treatments in the second half of
2020.
Study II: In the second study, we modified and implemented a chest X-ray score for hospitalized
COVID-19 patients. In a small pilot study retrospectively assessing 37 hospitalized patients, we
analyzed the effect of a new modified Brixia (MBrixia) chest X-ray score for COVID-19 patients.
Compared to the Brixia score, the MBrixia score doubled the number of zones to be scored by their
lesion severity and required consolidations in the CXR to reach the maximum score. The CXR
MBrixia score displayed an incremental association with COVID-19 severity, as measured by the
level of respiratory support at the time of CXR. Several predefined biomarkers were also correlated
with the MBrixia score value.
Study III: In the third study, we aimed to validate the results from Study II and assessed the
MBrixia scores' association with mortality. In a more extensive analysis of 279 hospitalized
patients, we validated the results from Study II, with small disparities in biomarker correlations. We
11
found that a MBrixia score in the upper half of the total score range at the time of admission was
associated with a 3-fold increased hazard ratio for mortality.
Digital Health Application: We built a digital health application on an existing data warehouse
infrastructure that supported and streamlined SARS-CoV-2 PCR testing in hospitals. It came to play
an essential role in the hospital's PCR testing efforts and screening of travelers to Greenland. It
provided an updated overview of hospitalized patients and their COVID-19 severity, including
levels of respiratory support and MBrixia scores, for stakeholders to assess. Further, the application
supported less trained clinical staff in new COVID-19 departments.
Conclusion: The studies underscored the complexity in assessing COVID-19 outcomes and
underlined that post-discharge events occur and results from recovery and sustained recovery
endpoints differ and, therefore, may impact interpretations. Risk stratification by radiographic
imaging can be a valuable diagnostic tool in assessing severity and prognostics, and especially in
integration with digital health applications. Digital health applications have been shown to play a
central role in enhancing local hospital operations, ensuring efficient and safe patient management,
and supporting broader public health initiatives during the pandemic.