key: cord-0869768-pu1wmrlh
authors: Banerjee, A.; Katsoulis, M.; Lai, A. G.; Pasea, L.; Treibel, T. A.; Manisty, C.; Denaxas, S.; Quarta, G.; Hemingway, H.; Cavalcante, J.; Nousardeghi, M.; Moon, J. C.
title: Clinical academic research in the time of Corona: a simulation study in England and a call for action
date: 2020-04-17
journal: nan
DOI: 10.1101/2020.04.14.20065417
sha: d931cfd93de25e385046491e62f4b2ac62cf8ef0
doc_id: 869768
cord_uid: pu1wmrlh

Background: Coronavirus (COVID-19) poses health system challenges in every country. As with any public health emergency, a major component of the global response is timely, effective science. However, particular factors specific to COVID-19 must be overcome to ensure that research efforts are optimised. We aimed to model the impact of COVID-19 on the clinical academic response in the UK, and to provide recommendations for COVID-related research. Methods: We constructed a simple stochastic model to determine clinical academic capacity in the UK in four policy approaches to COVID-19 with differing population infection rates: Italy model (6%), mitigation (10%), relaxed mitigation (40%) and do-nothing (80%) scenarios. The ability to conduct research in the COVID-19 climate is affected by the following key factors: (i) infection growth rate and population infection rate (from UK COVID-19 statistics and WHO); (ii) strain on the healthcare system (from published model); and (iii) availability of clinical academic staff with appropriate skillsets affected by frontline clinical activity and sickness (from UK statistics). Findings: In Italy model, mitigation, relaxed mitigation and do-nothing scenarios, from 5 March 2020 the duration (days) and peak infection rates (%) are 95(2.4%), 115(2.5%), 240(5.3%) and 240(16.7%) respectively. Near complete attrition of academia (87% reduction, less than 400 clinical academics) occurs 35 days after pandemic start for 11, 34, 62, 76 days respectively, with no clinical academics at all for 37 days in the do-nothing scenario. Restoration of normal academic workforce (80% of normal capacity) takes 11,12, 30 and 26 weeks respectively. Interpretation: Pandemic COVID-19 crushes the science needed at system level. National policies mitigate, but the academic community needs to adapt. We highlight six key strategies: radical prioritisation (eg 3-4 research ideas per institution), deep resourcing, non-standard leadership (repurposing of key non-frontline teams), rationalisation (profoundly simple approaches), careful site selection (eg protected sites with large academic backup) and complete suspension of academic competition with collaborative approaches.

The pandemic SARS-CoV-2 virus (causing the disease, is unprecedented in its impact on individuals, populations and health systems (1) . Since the first cases in Wuhan, China in November 2019, every country has been affected(2-3) but with wide variations in the ability and capacity to respond, with only half estimated to have operational readiness (4) .

Countries hit later can benefit and learn from acquired knowledge and experience of preceding countries. As part of this response, the research effort is crucial for development, testing and adoption of effective preventative and treatment (5, 6) .

A Pubmed search (November 2019 to April 2020), using terms "coronavirus" and "COVID-19" showed 2206 and 1604 articles respectively, suggesting swift global research mobilisation. However, the publication mix shows the vast majority are reviews, opinions and commentary rather than formal research. Many publications on COVID-19 are not clinically led, and many are not directly clinically informed. "Learning is difficult in the midst of an emergency" (7) , but our ability to deliver timely, high-impact clinical research, relevant to patients and populations, is critical across the academic spectrum (8) , from "bench to bedside to big data", whether basic biology, repurposed and novel therapeutic approaches, vaccines or modelling. Obstacles to and strategies for delivering research during a pandemic are poorly characterised.

Anecdotally, many countries have a baseline shortage of clinical academics in translational science (9) and many leading pathfinder health institutions are within major international transport hubs (London, Madrid, New York), which are affected early in the pandemic.

Lockdowns close university departments and funding bodies, with alternative funding sources (charities, philanthropy) hit by stockmarket falls and competing demand. Frontline remoteness impedes communication of urgency to decision makers, themselves usually selected for process delivery rather than dynamic adaptability. Critical researchers with relevant virology/immunological/intensive care knowledge are drawn in to local or national clinical responses. Other academic staff most likely to redeploy to COVID-19 research selfselect for immediate response roles (10) with universities prioritising repurposing to frontline care (11) . High disease rates, required self-isolation periods (12, 13) and distractions of remote working degrade the focus needed to create new or repurposed research delivery structures.

We therefore wanted to understand the pandemic research process and describe early lessons. Our aims were to: (i) model potential impact of the pandemic on clinical academic . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint capacity in England relating to COVID-19; and (ii) develop evidence-based recommendations to inform the optimal scientific response to COVID-19.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint

Based on our previous analysis of COVID-19 cases and excess deaths in England (14), we considered four scenarios of government interventions associated with different levels of population infection rates: 80% ("do-nothing"), 40% ("relaxed mitigation"), 10% ("mitigation") and 6% ("Italy model"), since "partial suppression"(1%) and "full suppression"(0.001%) were no longer feasible. The analyses of excess deaths used data in a cohort design with Helsinki. Key variables were population infection rate, background mortality risk based on underlying conditions, and relative risk (RR) of mortality associated with COVID-19. We used real-time data until 7 April 2020 for the number of confirmed cases and deaths (17) .

We designed and implemented a simple stochastic model to predict number of new cases in the population. Since the number of new cases are proportional to the active cases of the previous date (see Web appendix), we used official data from 10 April and calculated the ratio We assumed that an individual remains infected for 2 weeks, followed by death or immunity, and that actual cases were ~20 times more than confirmed cases, as people with mild or no symptoms are not routinely tested(based on prior estimates of 5-to 100-fold) (18) . Further details are specified in the Web appendix.

We used NHS Digital data (December 2019) to quantify number of doctors in England (n=125,119) (19) . Baseline number of clinical academics was estimated as 5% of . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 17, 2020. (19) . We assumed doctors of any specialty could contribute to the COVID-19 academic response, and necessary research skills and training were homogeneously available throughout the medical workforce.

Clinical academics are not available for research if: (i) they are delivering frontline care due to health system strain, or (ii) they are off sick. We modelled two scenarios with no medical academic capacity at 10% (low strain on the health system) and 5% (high strain on the health system) infection rates respectively. Our outcome was the available number of medical academics during the pandemic in England. We assumed that the number of potentially available 100% FTE clinical academics in research is 3200, but it is obvious that this number has decreased from the early days of the COVID-19 pandemic. We present in detail the assumptions of modelling of available clinical academics in the Web appendix. 3. Nightingale Hospital: A new NHS field hospital has been established, providing extra medical and intensive care capacity for provision of care to COVID-19 patients with a maximum theoretical capacity of 4000 beds, mainly intensive care (24) . We describe the scenario, staff involved and clinical and research priorities, and constraints.

Based on our model and our case studies, we have developed pragmatic recommendations for clinical research priorities relating to COVID-19.

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . Figure 1 illustrates different scenarios of infection growth rate and population infection rate (point estimate from x-axis; and cumulative estimate from area under the curve) and the "flatten the curve" phenomenon. The higher the growth rate, the higher the peak infection rate and the quicker the system is overwhelmed by cases of COVID-19. In addition, the course of the pandemic is longer. Conversely, if the infection growth rate is reduced, the curve is "flattened" and the pandemic course is shorter. The peak infection rate will be 2.4% 

Assuming the "low strain on the health system" model (where there is no academic capacity at population infection rate of 10%), Figure 2 shows that less than 400 100%FTE clinical academics (~13%) will available after April 10 for 11, 34, 62, 76 days for the scenarios of "Italy model", "mitigation", "relaxed mitigation" and "do-nothing" respectively. In the "do nothing scenario", no clinical academics are available to do research for 37 days (3/5/2020 to 8/6/2020). The predicted dates to reach 2560 clinical academics (80% normal capacity) are 23/6/2020, 1/7/2020, 3/11/2020 and 10/10/2020 for the scenarios of "Italy model", "mitigation", "relaxed mitigation", and "do-nothing", respectively.

In the "high strain on the health system" model (where there is no academic capacity at population infection rate of 5%), in the "relaxed mitigation" scenario, no clinical academics can do research for 18 days (13/5/2020 to 30/5/2020) and in the "do-nothing" scenario, from 23/4/2020 to 28/6/2020. The predicted dates to reach 2560 clinical academics (80% normal capacity) is 23/6/2020, 2/7/2020, 7/11/2020 and 11/10/2020 for the scenarios of "Italy model", "mitigation", "relaxed mitigation", and "do-nothing" respectively.

Northern Italy: As the first Western region to be affected (Lombardy, Bergamo) there was effectively no warning. Almost overnight a huge surge of severely ill patients hit us. It was the beginning of a nightmare. With no treatments, we had to re-organize the hospital wards, ITU beds, transform simple general into sub-intensive units, commit all doctors from all . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint specialities and research to COVID-19 in a matter of hours-days. It was "catastrophe medicine", research was impossible and approaches were empiric based on analogy to other diseases. Autopsy was our only science. "Mors ubi gaudet succurrere vitae -Where the dead are happy to help the living" and we started to appreciate the high rate of thrombotic complications and pulmonary pathology, initiating empiric anticoagulation and corticosteroids. Only months later with external partnerships were formal randomized trials initiated.

The COVID-HCW Study: Italy informed our strategy. The team nucleus was 5 senior lecturers and 1 professor, all of whom had all their clinical work stopped as irrelevant in the pandemic (cardiac MRI) and who had a track record of monthly large grant writing and detailed systems knowledge. The hospital had no emergency department, so it was protected with a large institution behind it. "Exponential teams" were created to deliver national and local components of research permissions -permissions took 100 documents and ~40 staff working at least part-time to deliver in 7 days (covid-consortium.com).

Following scoping, we rejected all but the most basic of aspirations: to capture is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint patient therapeutic trial patient will be recruited on day 8 after first patient admitted. Other studies (data, staff surveys) can be conducted at other sites (Table 1) or after the initial exponential wave peaks. In addition, there are opportunities for non-consented research, such as epidemiologic and advanced data analytics; e.g. initiatives such as DeCOVID (25) to mobilise data, computer scientists, analysts and analytic infrastructure, including and clinical expertise. There is potential to link effective learning directly to and from clinical questions.

After discussion among co-authors, and consensus among a stakeholder group at the Nightingale Hospital, we produced recommendations for a COVID-19 clinical research strategy ( Table 2) .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint

In the first study of clinical academic capacity in the COVID-19 era, we show the existential threat to research responses facing the UK and other countries. Urgent recognition and mobilisation are required to ensure prioritisation of the most appropriate and clinically imperative science. We have developed recommendations relevant to all health systems.

The healthcare and public health emergency caused by COVID-19 is not in question, fuelling global discussion, modelling and multidisciplinary research at a pace rarely seen (26) (27) . Emergencies as far-reaching as the current scenario require total rethinking of research delivery, and aspects that work better when some of the processes are accelerated and the permissions expedited, may well yield long-term benefits outside of COVID-19 research.

Here we have modelled clinical academic time in terms of numbers of staff and time in the pandemic. However, a far deeper examination of the role of clinical academics beyond "hours at the desk" is warranted in times of public health emergency to include the "what"

and "how" of their work. For example, certain tasks such as research permissions and data analysis may be diverted away from clinical academics, who may be better placed to act as conduits between the clinical and public health spheres and teams of non-clinical . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint researchers. The needs of the hour are patient-centred, data-driven and time-responsive, and it may be time to usefully change the role and function of the clinical academic. It is worth noting that this is occurring against a backdrop of declining clinical academic numbers (21) .

Our simulations suggest the pandemic will create health system strain for many critical months. Depending on a range of COVID-related factors, we show that the clinical academic workforce may be depleted when it is needed most to lead and conduct clinical research, even in a relatively well-resourced context such as the UK, whether by funding, number of universities and staff, infrastructure or policy. Therefore, other countries are likely to be worse affected. COVID-19 research is least likely to occur where it is most needed, magnifying the well-documented "10-90 research gap", where only 10% of resources for global healthcare research are devoted to low-income settings where 90% of preventable deaths occur (34) . Although COVID-19 is a unique threat, there are lessons to be learned from prior health research strategies to address structural inequities, such as the Global Fund for Malaria, TB and HIV/AIDS (29) . Without coordinated international responses, including urgent funding and infrastructure, research will be retrospective, patchy and unlikely to have an effect.

We provide six clear recommendations for science in the UK and globally in relation to COVID-19 (Radical prioritisation, Leadership, Rationalisation, Resourcing, Careful site selection and Dissemination). Radical prioritisation is important where field hospitals are being established in rapid timescales in different countries with delivery constraints. Highquality evidence can be obtained, but studies need to be lean with minimal complexity for key operational steps: consenting, randomisation, drug delivery, monitoring, outcomes and follow-up. The number of patients recruited to deliver definitive answers needs to be large, with fast recruitment across multiple sites. Furthermore, adaptive trial designs are preferred as new arms (e.g. multi-drug) can be generated swiftly and other arms dropped (e.g. supportive care if one arm has a signal of efficacy) without restarting permissions, via substantial amendments (38, 39) .

Leadership and rationalisation are the next key steps. Balance needs to be struck between clinical researchers in contact with the "frontline" so that research questions are clinically relevant and timely, and having research active leaders who will not be protected from frontline work. Rationalisation involves a study selection strategy that is deeply resourced for a limited number (1 or 2) studies per COVID cohort. In selecting these studies, a single study . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . of one investigational medical product versus standard of care (supportive care) with 50:50 randomisation is inefficient compared to studies with multiple therapeutic arms. Most single agent approaches to COVID-19 are likely to have, at most, a modest effect.

We used a stochastic model accounting for infection rate, infection growth rate and clinical academic capacity using up-to-date official statistics. There are limitations to our model and its assumptions. Our model was simple and was based only on observational patterns of the number of new cases and actual cases from publicly available data (17) . We conducted analyses on 10 April, and on 11 April, some extra ~3000 cases were added retrospectively and distributed over the past 10 days-we did not include these data. It did not take into account infectious disease epidemiology parameters, such as the basic reproductive number (R0), and we did not consider differing levels of risk of infection (35) (36) (37) . Our model on the availability of clinical academics makes several assumptions (Web Appendix), including the total number of 100%FTE academics as ~3200, with a uniform skillset across the workforce.

In the first study to model and estimate the impact of COVID-19 on the coordinated clinical academic response at system level, we show that all countries face depletion of their clinical academic workforce for several months, which will greatly hamper research in prevention and treatment. The number of studies needs to be rationalised urgently and background problems in clinical academia need to be overcome quickly. To quote Sir Jeremy Farrar, "The only exit from this pandemic is through science" (40) and that requires staffing.

Research question: JCM and AB. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. .

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. . . CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

The copyright holder for this preprint this version posted April 17, 2020. We inserted values for the ratio

from official statistics(17) from day 1 (March 5) to day 36 (April 10). Then we assumed different scenarios that are illustrated in web table 1.

We made the following assumptions:

1. The number of academics dropped linearly from 3200 to 2400 between day 0 and day 36, because 800 (25% were self-isolated on the 10 April). We further assumed that 800 academics will be quarantined from day 36 to day 90 and will return to work linearly until day 150.

2. The number of clinical academics available on 10 April was 400.

3. We modelled two scenarios for medical academic capacity at 10% (low strain on health system) and 5% (high strain on health system) infection rates respectively, assuming that 1000 clinical researchers would have to work in hospital per 1% of active cases in the population, if the active cases are <2.07%, i.e. infection rate of the 10 th of April). In the 10% and 5% cut-off models respectively, we assumed ~50 (i.e. 400/7.93) 2) and ~136

(i.e. 400/2.93) extra researchers can be available per 1% of active cases (≥2.07% i.e.

infection rate on 10 April).

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 17, 2020. 

. CC-BY-NC-ND 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)

The copyright holder for this preprint this version posted April 17, 2020. . https://doi.org/10.1101/2020.04.14.20065417 doi: medRxiv preprint

Novel Coronavirus and Old Lessons -Preparing the Health System for the Pandemic

Early Transmission Dynamics in Wuhan, China, of Novel Coronavirus-Infected Pneumonia

Clinical course and risk factors for mortality of adult inpatients with COVID-19 in Wuhan, China: a retrospective cohort study

Health security capacities in the context of COVID-19 outbreak: an analysis of International Health Regulations annual report data from 182 countries

Wellcome statements on novel coronavirus (COVID-19)

COVID-19: Real-time dissemination of scientific information to fight a public health emergency of international concern

Thinking Globally, Acting Locally -The U.S. Response to Covid-19

A call for action to establish a research agenda for building a future health workforce in Europe

Academic factors in medical recruitment: evidence to support improvements in medical recruitment and retention by improving the academic content in medical posts

Royal College of Physicians. COVID-19 and its impact on NHS workforce

Clinical academics freed up to support NHS during Covid-19 pandemic

Protecting healthcare personnel from 2019-nCoV infection risks: lessons and suggestions. Front Med

COVID-19: protecting health-care workers

Estimating excess 1-year mortality from COVID-19 according to underlying conditions and age in England: a population based cohort using NHS

UK phenomics platform for developing and validating electronic health record phenotypes: CALIBER

A chronological map of 308 physical and mental health conditions from 4 million individuals in the English National Health Service. Lancet Digit Health

COVID-19 coronavirus pandemic country estimates: UK. 5

A fiasco in the making? As the coronavirus pandemic takes hold, we are making decisions without reliable data. StatNews

NHS Digital Workforce statistics

One in four NHS doctors 'sick or in isolation

COVID-19: Healthcare Worker Bioresource: Immune Protection and Pathogenesis in SARS-CoV-2 (COVID19-HCW

Coronavirus: Nightingale Hospital opens at London's ExCel centre

Call for COVID-19 rapid response data science taskforce

Locally Informed Simulation to Predict Hospital Capacity Needs During the COVID-19 Pandemic

Projecting demand for critical care beds during COVID-19 outbreaks in Canada

International Severe Acute Respiratory and emerging Infection Consortium

Genomic diversity of SARS-CoV-2 in Coronavirus Disease 2019 patients

Modified SEIR and AI prediction of the epidemics trend of COVID-19 in China under public health interventions

Operational and implementation research within Global Fund to Fight AIDS, Tuberculosis and Malaria grants: a situation analysis in six countries. Global Health

National Institute of Health. Coronavirus news, funding and resources for global health researchers

The developing world in The

The effect of control strategies to reduce social mixing on outcomes of the COVID-19 epidemic in Wuhan, China: a modelling study

Centre for Mathematical Modelling of Infectious Diseases COVID-19 working group. Early dynamics of transmission and control of COVID-19: a mathematical modelling study

Estimates of the severity of coronavirus disease 2019: a model-based analysis

Randomised Evaluation of COVID-19 thERapY

Adaptive Platform Trial for Community-Acquired Pneumonia

Opinion: The only exit from this pandemic is through science. We must fund it. Globe and Mail

Our simulation with 640 000 hypothetical individuals (1/100 of the population in the UK) started on the 5 of March, when 23 people had COVID-19 and the rest were healthy (17) . We selected 23, because we selected the first date that the confirmed cases were >100 in worldometer (116 cases on the 5/3). So, we assume that on that date, we had 116*20~= 2300 cases, so we used in our simulation 2300/100=23.