Citizens' Jury on QCovid: Report on the jury's conclusions and key findings

Executive Summary

Introduction

The QCovid® risk model was developed at the University of Oxford in 2020 to provide a sophisticated way of identifying people who were at the highest risk of death or a poor outcome should they contract Covid-19.

There are four main ways of using QCovid® that have been explored by the Scottish Government to date. These are referred to as 'tools' or 'use cases' and include:

• a clinical tool (for use by clinicians such as GPs to determine patient risk)
• a public-facing tool (for use by general public to determine their own risk)
• a population tool using non-anonymised data at the population level (to identify and notify people at high risk)
• a population tool using anonymised data at the population level (to inform planning and resourcing)

The Scottish Government recognised that there could be ethical issues associated with any use of a risk prediction model like QCovid®, but particularly in running one through large datasets of personal health information.

As part of the Scottish Government's commitment to ensuring open, honest and transparent government, Ipsos Scotland was commissioned to conduct a Citizens' Jury to better understand the public's views on the use of a risk prediction model like QCovid®.

This work was overseen by an independent Ethics Panel which included external advisors with expertise in the use of ethics and data. The findings will feed into any wider assessment should the Scottish Government use a model like QCovid in the future to ensure that ethical considerations are at the heart of how any risk prediction models are deployed.

Objectives and methodology

The objectives of the jury were to explore attitudes towards QCovid® and similar risk prediction models and the different ways in which they could be deployed. The public engagement aimed to understand specifically:

• any ethical concerns around deploying a model like QCovid® in different ways, particularly running the risk model through population level health records
• any ethical concerns around deploying a model like QCovid® in different scenarios
• any circumstances or scenarios when the public benefits of using QCovid® would outweigh private concerns over the use of personal data

The jury of 25 people representing a cross-section of the population from across Scotland met online across six three-hour workshops throughout February and March 2022. The jury was convened to answer the following key question:

"What are the risks and benefits of using public health data to predict people's risk of dying from Covid-19?"

More specifically, they explored the risks, benefits and ethical concerns related to each potential use of a model like QCovid®, and the principles that would make its use acceptable.

Main findings

Having learned about QCovid® as an example of a risk prediction model, and deliberated its relative risks and benefits, some clear themes emerged which cut across each of the tools. These were:

Efficacy and accuracy

The jury generally felt reassured by the fact that the QCovid® model had been extensively validated by experts. There remained some concerns over the completeness of the data underlying the QCovid® model and the impact of gaps in medical records - such as information on ethnicity or through some health conditions not being known to the GP - potentially leading to inaccurate scores. The jury therefore emphasised the need for QCovid® or a similar model to be kept up-to-date and capable of adapting to changing circumstances such as new variants or booster vaccines.

Data security

The jury recognised that there were clear protocols in place for accessing public health data and felt that data security should be in place for any use of the model, but particularly when using non-anonymised data at the population level.

Transparency and communication

The reasons for applying a risk prediction model were understood and generally accepted by the jury. However, it was strongly agreed that clear communication would be necessary for informing the general public about the rationale for using QCovid® or similar risk prediction models in Scotland and explaining how this is done. This level of transparency was considered important for building public trust in the model.

Targeted support

There was clear consensus among the jury that sufficient and targeted support mechanisms must be in place. These mechanisms would need to include emotional support (particularly for those receiving a high score), support to interpret what the score means, and practical support to help people take appropriate action.

Justification

Attitudes towards risk prediction models could vary depending on the status of a virus like Covid-19. For example, in a low prevalence situation, there was a view that a clear rationale from the Scottish Government would be needed to justify its use.

As well as these overarching themes, the jury agreed principles and "red lines" for each of the four tools (summarised below). The principles act as guidance for the Scottish Government to consider if implementing the particular tool, while "red lines" are points that in their opinion, if crossed, would in the jury's opinion make the use of this tool unacceptable. These principles can be applied to future, similar, tools.

Clinical tool

Principles: Use of the tool is acceptable if…

• Information is provided to explain to patients what the tool is and how it will be used
• There is clear communication on the use of the tool
• Practical and emotional support is provided to help patients according to their risk score.
• GPs or other healthcare professionals can help patients understand their risk score.
• GP resources are not placed under too much burden.
• GPs or other healthcare professionals are trained to use the tool effectively.
• Patients have the option to ask for their score and to refuse the option to discover their score.
• Results are confidential.
• The tool is kept up to date in case of people moving from low risk to high risk and vice versa.

Red lines: Use of the tool is unacceptable if…

• It collects personal information which is not needed for the tool to work.
• The data is kept after you've received your score.
• If it detracts from GPs' ability to address other, more critical, health needs.
• If your score is shared with other parties (i.e., anyone other than your GP) without your consent.

Public-facing tool

Principles: Use of the tool is acceptable if…

• There is sufficient support in place to help people understand their risk score.
• It is accessible (e.g., alternative formats – language translations, large print, braille or text to speech, and simple language).
• There are alternative ways of accessing the score for those who are not online or who require additional support.
• There is clear, simple guidance for using the tool and there are consequences for misuse.
• The tool is kept up to date in case of people moving from low risk to high risk (and vice versa).

Red lines: Use of the tool is unacceptable if…

• There is not adequate support in place to help people understand their score.
• It is introduced on its own without the clinical tool being available
• There is no alternative for people excluded from using an online tool.
• The information an individual inputs can be accessed and/or used by anyone else.
• Identifiable information is requested and/or stored.
• It doesn't reach everyone who needs it.
• There is any obligation for people to pass on information about their risk score.
• It cannot be guaranteed that the data put in is accurate.

Population-level tool using non-anonymised data

Principles: Use of the tool is acceptable if…

• There is sufficient targeted support in place to help people at high risk.
• There is clear information about the sources of support available and the support is easy to access
• Information about the use of the tool is available and clearly communicated to the general public.
• The score is confidential to the individual, with no legal requirement to share.
• The tool is kept up to date in case of people moving from low risk to high risk (and vice versa).
• There is a mechanism to challenge or change the outcome.
• There are data security protocols in place.

Red lines: Use of the tool is unacceptable if…

• Data about individuals is shared with third parties for purposes that do not align with healthcare-related public benefits relating to the pandemic.
• There is not adequate ongoing support in place to help people who are identified as being at high risk.
• The data is not held securely.
• The risk to public health from Covid-19, or another virus, at the time is minimal.
• It is used to discriminate against individuals (e.g., in the workplace or in accessing services such as insurance).

Population-level tool using anonymised data

Principles: Use of the tool is acceptable if…

• Information about the tool is available and clearly communicated to the general public.
• The tool is kept up to date in case of people moving from low risk to high risk and vice versa
• The data is agile and able to adapt should new situations arise.
• The data is to be used by Scottish Government and NHS Scotland only.
• There are data security protocols in place.

Red lines: Use of the tool is unacceptable if…

• Data is shared with third parties for purposes that do not align with healthcare-related public benefits relating to the pandemic.
• Data is not protected from commercial companies accessing it.
• There is any collaboration with data farm companies.^[1]
• It is used to discriminate against certain groups (e.g., being denied access to certain services based on age or ethnicity).