204 Chapter 9 surveillance. Finding a way to benefit from the new possibilities of AI and complement traditional surveillance methods, is an important future objective. Part II: Causal inference and real-world data Randomised Controlled Trials (RCTs) have many advantages when it comes to causal inference, most importantly by the elimination of confounding26. However, it is not always feasible or ethical to conduct a randomised trial. Moreover, RCTs are expensive, time-consuming and in certain settings almost impossible, as with rare exposures or outcomes, for example27. Observational studies can, in specific circumstances, also answer causal questions and are generally easier to execute28. Especially during outbreaks, when there is a great need for rapid results that can timely inform clinical practice and policies, it may be preferable to use routinely collected observational data. There has been an increasing interest in the use of realworld data (reflecting clinical practice rather than a controlled clinical trial setting) to increase the speed of clinical research29. In the second part of this thesis, the aim was to answer specific clinical questions in the field of viral infectious diseases using large and observational data sets. In Chapter 4, the effect of SARS-CoV-2 co-infection with influenza viruses, respiratory syncytial virus (RSV) or adenovirus on in-hospital mortality and the need for invasive mechanical ventilation in hospitalised patients was analysed. Understanding the effect of a viral respiratory co-infection with SARS-CoV-2 and endemic respiratory viruses on clinical outcomes is essential to anticipate the clinical requirements throughout the traditional respiratory virus season. The International Severe Acute Respiratory and emerging Infections Consortium Coronavirus Clinical Characterisation Consortium (ISARIC4C) Clinical Characterisation Protocol UK (CCP-UK) database was used, a national cohort study including SARS-CoV-2 patients in over 300 hospitals in England, Wales and Scotland30. Over 212.000 adult COVID-19 patients were included, and over 17.000 (8%) underwent testing for additional respiratory viruses using Reverse Transcription Polymerase Chain Reaction (RT-PCR) (influenza viruses, adenovirus and/or RSV). 6965 of these patients had virological test results available, including 583 who had a confirmed respiratory viral co-infection. 227 (39%) patients had a co-infection with one of the influenza viruses, 220 (38%) had an RSV coinfection and 136 (23%) had adenovirus as the co-infecting pathogen. In-hospital mortality and the need for invasive ventilation was compared between SARS-CoV-2 co-infected patients and SARS-CoV-2 mono-infected patients. After correcting for confounders, the odds of receiving invasive mechanical ventilation were higher for SARS-CoV-2 patients that had an influenza virus co-infection compared to SARS-CoV-2 mono-infected patients (OR 1.80, 95% CI 1.22-2.63). Co-infection with influenza virus and adenovirus both were significantly associated with increased odds of death compared to mono-infection (OR 1.61, 95% CI 1.12-2.30 and OR
RkJQdWJsaXNoZXIy MTk4NDMw