Guidance for clinicians about the COVID-19 Clinical Risk Assessment Tool

You can use the COVID-19 Clinical Risk Assessment Tool to:  

• inform clinical discussion and care planning around COVID-19 risk   
• have a patient-centred conversation to reach a shared understanding of estimated risk  
• reinforce public health messaging   

QCovid® is a multivariable risk prediction model similar to QRisk®. It gives a weighted, cumulative risk score of known risk factors and has performed well in predicting severe outcomes from COVID-19 when formally evaluated. The model assigns evidence-based weighted values to risk factors and combines them.

QCovid® is a ‘living’ risk prediction model. This means that the model will be updated periodically when new research becomes available. The original QCovid® model was incorporated by NHS Digital into the COVID-19 Clinical Risk Assessment Tool in February 2021 and the same model was used nationally for the population risk assessment.

The QCovid® model calculates absolute and relative risk values for several outcomes. The COVID-19 Clinical Risk Assessment generates results for the risk of dying from coronavirus following a positive PCR test only. 

The input factors that QCovid® uses are:  

• age  
• sex registered at birth  
• ethnicity  
• housing category   
• postcode for Townsend deprivation score
• height and weight for BMI   
• vaccination status
• medical history   

You can read the publication about QCovid® in the British Medical Journal (BMJ). These papers include detailed information about risk factors and how the model calculates risk assessment results: 

• Living risk prediction algorithm (QCOVID) for risk of hospital admission and mortality from coronavirus 19 in adults: national derivation and validation cohort study
• Risk prediction of covid-19 related death and hospital admission in adults after covid-19 vaccination: national prospective cohort study

Medical history questions in the tool are grouped by themes to allow broad questions such as ‘any heart problems’ to be covered quickly.   

Wherever possible, make sure you enter all the information about the patient that you can. This will generate the most accurate result to inform the clinical discussion or decision.  

Use the following guidance when entering a patient’s medical history in the tool.

We’ve put the chemotherapy drugs into Group A, B and C. These groups are based on the PHE Systemic Anti-Cancer Therapy dataset. It covers hospital-administered cancer chemotherapy treatments. You can find the list in the Supplementary Box A in the BMJ QCovid® paper. Group C are the most immunosuppressive followed by Group B and then Group A.  

If the patient has been on multiple chemotherapy drugs, select the most immunosuppressive one.  

The immunosuppressants described are medicines which fall within chapter 8.2 of the British National Formulary.  

These drugs should not be selected if they have been used for treatment of COVID-19 (for example in clinical trials).

Full list of immunosuppressants  

 If recording the result in the patient’s medical record, enter:

• the date it was generated
• the release version number of the tool. You’ll find the release number on the results page, near the absolute and relative risk results
• the factors that contributed to the results. You’ll find these on the results page, in the 'How your patient's risk is calculated' section

Making a note of what version you used for the patient’s assessment will help you refer back to your notes and know where our understanding of the virus was at that point.

If you do choose to record the results in the patient’s medical record, the relevant data controller of those records will assume responsibility for the risk assessment results. 

The defaults for the tool were determined as part of this development of the QCovid® model.

Postcode

The patient's postcode is used to calculate a Townsend deprivation score, which is a marker of deprivation. A greater Townsend score implies a greater level of deprivation. The range is -12 to 12. 

If no postcode is entered, the average UK score of 0 for the Townsend deprivation score is used.  

A default BMI value of 31 is used where no information is entered for height and weight. If BMI is calculated below 15, a default BMI of 15 is used. If it's calculated above 47, a default BMI of 47 is used.

A default score in the midpoint of risk is used if you select ‘prefer not to say’ for ethnicity.

Vaccination status

The value for unvaccinated is used when ‘Don’t know’ is selected in response to vaccination status.

Certain factors which influence risk such as occupation, individual behaviour (such as hand washing, wearing face coverings and visiting friends or family), new strains of virus and the local infection rates are not covered by the tool.  

The result does not give a guarantee of any outcomes and should not be used to communicate a patient’s actual individual predicted risk. This is because it does not use factors such as their exposure to the virus. It gives information about the chances of an outcome occurring and can facilitate patient-centred discussions around level of risk.  

The research data was collected between January 2020 and June 2021. Because there was not enough research data available for some groups of people, the tool can not accurately assess risk for:

The research was done on adults aged from 19 to 100, so the tool can only be used for patients within this age range.  

The research was done using sex registered at birth.  

The tool can not give an accurate result for people who are pregnant due to the small number of pregnant women in the underlying research data. 

You can use the tool to estimate risk for people who were advised to shield during the pandemic, or who have a condition that qualified a person for shielding during that time, as it may inform useful discussions with patients. The government has updated its guidance for people previously considered to be clinically extremely vulnerable  

However, risk may be underestimated in this group, because when the research was being carried out, these people may have followed shielding advice and may have been protected from catching COVID-19. 

Some rare conditions may also increase COVID-19 risk. However, there is currently insufficient data to quantify this. Therefore, for patients with rare diseases, we recommend using clinical judgement and comparison to similar clinical conditions when assessing individual risk.

At the start of the coronavirus epidemic in the UK, modelling from the Scientific Advisory Group for Emergencies (SAGE) suggested that if people who were clinically vulnerable (CV) to the virus could be identified and shielded from the coming wave, it could have a significant impact on preventing mortality.

In response, the CMO set clinical criteria for the development of the shielded patients list (SPL), a subset of those likely to be at the highest risk, the clinically extremely vulnerable (CEV). The criteria included clinical conditions which were thought likely to present the greatest risk of poor outcomes.

NHS Digital identified the initial patients for the SPL using national health datasets. Clinicians were also able to add and remove individual patients from the SPL using similar criteria and clinical judgement. Patients on the SPL were advised to follow shielding guidance. The list continues to be kept up to date, and new conditions are added to the list by the UK CMOs as the evidence base develops. For example, dialysis, CKD stage 5 and Down’s syndrome have recently been added.

As the epidemic progressed, we obtained more detailed data about the clinical risk of COVID-19. England’s CMO commissioned the New and Emerging Respiratory Virus Threats Advisory Group (NERVTAG) to create an evidence-based risk prediction model.

To do this, Oxford University led a group of researchers from across the UK to produce QCovid®. They studied the patient records of over 8 million people using GP and hospital record data from late January to April 2020. They looked at which individuals caught COVID-19 and were subsequently hospitalised with or died from the virus during the 97-day period of the first wave.   

Further validation took place between May and June 2020. The model has also been validated in an independent dataset by the Office for National Statistics (ONS), with concordance scores in the ‘excellent’ range. View a pre-print of the ONS validation. QCovid® showed that factors such as age, sex registered at birth, ethnicity, BMI and some medical conditions influence risk.  

Further research was done using more up-to-date data from the second wave (September 2020 to June 2021), and the model has been refined and updated based on the latest findings. Factors such as vaccination status and current infection levels are now reflected in the model, and some conditions have been removed as risk factors where sufficient data was not available.

The COVID-19 Clinical Risk Assessment Tool was updated on 25 November 2021 to implement this latest version of QCovid®. 

This information for your patient should help give them an understanding of:  

• how NHS staff can use the tool  
• how the tool works  
• understanding the results  
• limitations of the tool   

You must ensure your patient understands where they can find the privacy notice covering your organisation’s use of the tool. NHS Digital has created a template privacy notice that you can use for this purpose if needed.  

As with all shared decision-making conversations, you should support the patient to inform direct care decisions where COVID-19 risk is a factor, such as in-person attendance of hospital appointments.  
You should advise your patient that they:  

• could potentially reduce their risk by controlling diabetes and losing weight through a healthy balanced diet and doing regular physical activity  
• can protect themselves and others by adhering to the latest public health guidance
   

The national shielding programme has now ended, and people are no longer being identified as high risk (clinically extremely vulnerable) or being added to the Shielded Patient List (SPL

The Chief Medical Officer (CMO) for England, in consultation with senior clinicians, set a threshold for risk as assessed by the QCovid® model above which people should be considered Clinically Extremely Vulnerable (CEV).

These thresholds were:

• an absolute risk of catching and dying of COVID-19 of 0.5% and above
• a relative risk of catching and dying of COVID-19 of 10 and above

Patients who met either of these thresholds were advised to be added to the SPL unless the patient chose against it or there was a clinical reason not to. 

These thresholds were applied in the population risk assessment completed by NHS Digital inn February 2021 to identify people who should be added to the SPL made as part of the population risk assessment.

Because the population risk assessment uses data from multiple national data sets, including some different default values, there will have been some variation between its results and the online tool.

Where any variation impacts clinical decision making (for example, addition to the SPL), clinicians were advised to use their clinical judgement alongside the patient’s wishes to make a shared decision.

Clinicians were previously able to use the tool to inform useful discussions around removal from the SPL. 

For those patients who were advised to shield during the first wave of the pandemic, or who have a condition that qualified a person for shielding during that time, QCovid® could underestimate their risk of severe outcomes from COVID-19.

This is because when the research was being conducted patients may have been advised to shield, meaning they may have been protected from catching COVID-19 and therefore severe outcomes such as hospitalisation and death.

For this cohort of patients,  clinicians were advised to use clinical judgement alongside the results of the tool to inform your decision on removal from the SPL.

Research from the Winton Centre at the University of Cambridge suggests that clinicians’ and the public’s understanding of COVID-19 risk is often not proportionate to the actual risks as we’re now coming to understand them. The results from this tool will support the standardisation of risk assessment within consultations.

As society learns to live with COVID-19, we hope that the tool will facilitate discussions between clinicians and patients around a shared understanding of risk, tailored to the individual and taking into account their own personal circumstances and risk appetite. It will also provide an opportunity to signpost to relevant public health and health promotion advice and resources.

The Winton Centre has presented  a useful research article, published by the Royal Society of Open Science on this topic: Communicating personalised risks from COVID-19: guidance from an empirical study.