Metadvice Medical Mysteries: Covid 19 - part III

By Prof. Andrew J Krentz, MD FRCP FFPM, Head of Cardiometabolic Division - Metadvice Ltd
September 03, 2020
Reading time: 9 minutes

This Metadvice blog series explores some of the mysteries surrounding the complexities of SARS-CoV-2 infection on people and healthcare systems, and how advanced analytics and technologies like artificial intelligence (AI) can help navigate these complexities.


Central issues here are the highly heterogeneous clinical manifestations of COVID-19 allied to the short time that has been available in which to unravel a highly contagious multi-organ illness with many novel features. While time and tested epidemiological and experimental methodologies have brought some early successes in containing the pandemic and improving outcomes for severely affected individuals much uncertainty persists about many facets of SARS-CoV-2.

In a previous article we considered the evidence for increased vulnerability of certain groups with compromised immune function or pre-existing cardiometabolic comorbidities. The flip side also merits consideration, i.e. whether some individuals may have pre-existing immune defences that confer protection against SARS-CoV-2. The quest for effective vaccines is proceeding at a rapid pace with several candidates already in phase 3 trials. Whether SARS-CoV-2 will take its place along the four strains of seasonal coronaviruses that already circulate every winter and become a fifth seasonal coronavirus causing epidemics in coming years remains to be determined.

As for patients requiring hospital admission, treatment options may be considered in two categories - general supportive measures and antiviral therapies. The latter category may be subdivided into repurposing of existing drugs originally licensed for indications other than COVID-19 and novel therapies directed specifically at SARS-CoV-2. Many different drug classes are currently in clinical trials. The results of these trials should help seal the knowledge vacuum that has allowed misinformation and extrapolation to detract from the mission of providing evidence-based care. A prominent example is the high-profile support that has been voiced for the antimalarial drug hydroxychloroquine. Data generated from randomized controlled trials (RCTs) and quasi-experimental studies suggest that hydroxychloroquine offers little or no benefit over standard care for the treatment of COVID-19. That said, the quality of the current evidence is considered low. While the possibility remains that the drug may have as yet unproven protective effects if given early after exposure to COVID-19, concerns persist about the potential for serious cardiovascular side-effects. As ever in pharmacotherapeutics, risks and benefits must be carefully weighed.


The good news is that progress has already been made in preventing the progression of COVID-19 in hospitalized patients with reductions in mortality rates reported for patients requiring intensive care. However, new threats to COVID-19-infected patients, e.g. invasive aspergillosis, have been reported while, as discussed below, the longer-term consequences of infection are gaining more attention. Let us now consider therapeutic advances in the management of some of the issues we introduced in earlier articles, alongside the continuing knowledge gaps that require additional attention.

Managing hypoxia - COVID-19 is viewed primarily as a respiratory infection that in severe cases can lead to hypoxaemic respiratory failure. Treatment of hypoxic COVID-19 patients has evolved in real time - the “science of experience” that we alluded to in our introductory article in this series. Simple measures, such as nursing in the prone position, have proved helpful, particularly in obese patients. Other aspects of care, such as decisions about which patients to intubate and when this is best timed, remain more problematic.

Anticoagulation - The place of anticoagulant therapy in the management of COVID-19 has not been settled. While standard low-dose prophylactic anticoagulation is recommended, current data based on retrospective studies of anticoagulation are considered insufficient to support wholesale moves to more intensive anticoagulant therapy with its attendant risk of bleeding.

Corticosteroids - The value of the high-dose synthetic corticosteroid dexamethasone was confirmed in a multi-arm randomized controlled trial (RECOVERY) led by University of Oxford investigators. RECOVERY is a notable example of an efficient multi-arm adaptive clinical trial that has performed well in the context of the pandemic. It is well recognized that corticosteroids may aggravate another negative prognostic indicator for COVID-19, i.e. hyperglycaemia. However, while nearly a quarter of the trial participants had a history of diabetes no relevant biochemical data were collected in RECOVERY.

Metabolic control - Hyperglycaemia has been identified as a risk factor for severe COVID-19 illness and carries a higher mortality risk. Whether hyperglycaemia and ketosis are proportionate markers of severe systemic illness or are directly linked to the pathophysiology of COVID-19 is presently a topic of debate. Clinical trials of insulin therapy are in progress to determine the benefits and/or risks of tight metabolic control in patients hospitalized with COVID-19. On the strength of its cardio- and reno-protective actions, the sodium-glucose contransporter-2 (SGLT-2) inhibitor dapagliflozin is also under evaluation in patients with COVID-19-associated respiratory failure. Of note in this clinical scenario is the hazard presented by diabetic ketoacidosis in vulnerable patients treated with SGLT-2 inhibitors. Non-randomized observational data suggesting that the commonly used glucose-lowering agent metformin might confer protection against severe COVID-19 infection, at least among patients with type 2 diabetes, are intriguing and hypothesis-generating.

Antiviral therapies - The COVID-19 pandemic has stimulated frenetic research activity to identify antiviral treatments. The crucial importance of well-designed studies that have adequate statistical power has recently been re-stated by expert groups. In parentheses, the tendency for the results of important COVID-19 clinical trials to be announced as press releases and pre-prints awaiting full peer review has generated concerns among some clinicians.

  • Remdesevir - This broad-spectrum antiviral agent inhibits SARS-CoV-2 RNA polymerase to prevents viral replication. In placebo-controlled clinical trials in adults, remdesivir therapy led to clinical improvements in a range of clinical outcomes and fewer serious adverse events. Specifically, remdesevir reduced the need for supportive measures including mechanical ventilation and shortened time to recovery in patients with mild or moderate, or severe COVID-19 disease who were on supplemental oxygen therapy. However, no statistically significant differences have been reported for mortality.

  • Convalescent plasma and monoclonal antibodies - A recent Cochrane living review has concluded that it is presently very uncertain whether convalescent plasma is beneficial for patients hospitalized with COVID‐19. At the time of writing there were nearly 100 ongoing studies evaluating convalescent plasma and hyperimmune immunoglobulin, of which 50 were RCTs.


It is becoming increasingly apparent that mysteries about the recovery phase of COVID-19 are expanding to include the emergence of complications that may threaten the medium to long-term health of survivors.

Post-COVID-19 syndrome - In addition to persistent respiratory issues and low-grade fever in some patients a polysymptomatic post-covid-19 syndrome has been described, full details of which remain to be delineated. Persisting symptoms of so-called ‘long covid’ reported by UK doctors who had experienced COVID-19 included fatigue, reduced exercise capacity and difficulties with concentration.

Paediatric Inflammatory Multisystem Syndrome - In children, a rare delayed multi-organ Kawasaki-like syndrome (Paediatric Inflammatory Multisystem Syndrome Temporally associated with SARS-CoV-2 - PIMS-TS), has led to a small number of deaths.

Cardiovascular complications - In adults, concerns about long-term COVID-19-associated cardiovascular complications centred around a risk of developing heart failure were highlighted in a previous article in this series.

Neurological sequelae - Temporary loss of taste and smell in some patients with COVID-19 is evidence of invasion of the central nervous system by the virus. More serious concerns are reports of an increased risk of stroke and a dementia-like syndrome. A global study to track and evaluate the short- and long-term impact of COVID-19 on cognition and behaviour has been announced.

Renal injury - Acute kidney injury necessitating renal support in intensive care units has been a prominent complication of severe COVID-19; microthrombosis has been identified in post-mortem studies. A proportion of affected patients who recover may have lasting degrees of impaired kidney function.

How can Metadvice analytics help improve treatment?

The COVID-19 pandemic has placed unprecedented burdens on healthcare systems across the world. While the advances in care discussed above have been enthusiastically welcomed it should be acknowledged that the avalanche of research data on COVID-19 exceeds the capacity of healthcare professionals to keep up to date with all aspects of best practice. The synthesis of data from electronic medical records with the latest clinical trial results has potential to usefully empower clinicians in this challenging endeavour. Metadvice can assist by extracting the latest research from reliable sources to inform optimal clinical decisions for individual patients in real time.

A plethora of recent publications attest to the potential of AI may have in unravelling the mysteries of COVID-19. Areas of interest include improving the accuracy of diagnosis (notably in interpreting medical imaging data), predicting demand for critical care beds, aiding drug repurposing and developing new therapies. However, the intrinsic strengths and limitations of AI need to be borne in mind. While AI has its place in data analysis, it cannot compensate for fundamental methodological flaws such as an inadequate sample size or other deficiencies in study design or data quality. Metadvice’s expertise in AI is complemented by clinical expertise and knowledge of drug discovery and therapeutic trials. Identifying hitherto unrecognized patient characteristics that modulate responses to treatment provides an example of how Metadvice can complement the development and implementation of new therapies for COVID-19.

How can resilience against future infection be strengthened?

There is a need for better stratification of individual susceptibility to COVID-19 if preventive measures are to be effectively implemented ahead of future waves or localized resurgences. However, concerns have been expressed about the performance and reliability of some COVID-19 risk prediction models published to date, indicating a need for further studies.

The vulnerability to COVID-19 associated with pre-existing conditions such as obesity and associated cardiometabolic disorders has prompted consideration of measures to improve resilience against future waves of infection. The UK’s Prime Minister recognized the need to reduce his body weight followed his own hospitalization with COVID-19. This personal effort has been accompanied by a public health initiative to help counter the UK’s high rates of obesity. Taking a broader perspective, however, it is noteworthy how little advice has been forthcoming to guide optimization of general health and immune status in the face of the pandemic. The paucity of actionable information reflects uncertainties about the effectiveness of non-pharmacological measures to bolster immunity. Vitamin D status has received attention in recognition of the role of the hormone in innate and adaptive immunity, albeit with no COVID-19-specific recommendations to date.

As mentioned, additional challenges are now becoming apparent with respect to the longer-term assessment and management of patients who have continuing symptoms following recovery from acute COVID-19 infection. AI may be able to provide useful insights into this partially characterised symptom complex.


Much has been learned about the transmission and pathophysiology of COVID-19 in a short space of time. Effective supportive measures and treatments for critically ill patients have been identified from clinical observations and well-designed and executed clinical trials, respectively. These measures have been incorporated into clinical practice with good effect. However, emerging longer-term health implications of COVID-19 have yet to be fully characterised. AI-informed evidence-based insights coupled to clinical data offer the prospect of more effective preventive measures and optimal application of novel therapeutics according to the principles of precision medicine.