Many Different Flows of Data: an Interview with Geraint Lewis

Detail of image by r2hox, via Flickr. Click for more information.
Detail of image by r2hox, via Flickr. Click for more information.

Finding effective and efficient means of sharing information is a key challenge for anyone involved in science communication – and that’s particularly true for large-scale health systems, in which data inform a wide range of decisions related to both policy and practice.

To learn more about some of the challenges and opportunities in this arena, I talked to Dr. Geraint Lewis, chief data officer of the National Health Service in England.

Communication Breakdown: Before we get into your work as chief data officer of the NHS, I’d like to learn a little bit more about you. What’s your educational background?

Geraint Lewis: After finishing my A-levels at my local comprehensive school in Cardiff, I spent a year studying music in France before going to Cambridge to study medicine. As a junior doctor, I studied for my membership exams for the Royal College of Physicians, followed by a Masters in Public Health from the London School of Hygiene and Tropical Medicine, and then finally my membership exams for the Faculty of Public Health. I finished my training and qualified as a consultant in public health medicine in 2008. I’m currently studying for a qualification in Major Project Leadership at Oxford University’s Saïd Business School.

CB: You’ve worked in acute medicine and emergency medicine, dealing with patients that need urgent medical care. What led you to move from the emergency room to dealing with public health and policy issues?

Lewis: I still remember the moment I decided to switch from clinical medicine to public health medicine. I was working in an emergency department outside Sydney and it just dawned on me how interested I was in the similarities and differences between the Australian health care system and the NHS in England. It was at that point that I decided I would come back to the UK and specialise in public health; however, I continued to work weekend shifts in Accident & Emergency for many years afterwards.

CB: In 2004, you began developing something called the “Virtual Wards” project. Can you explain what that is?

Lewis: Two nurses and I conceived the idea of a “virtual ward” as a way to improve the coordination of health and social care services for high-risk patients in the community. A virtual ward uses some of the systems and routines of a hospital ward for patients in their own homes, such as a single set of notes, regular multidisciplinary team meetings, and a ward clerk to coordinate the team. The word “virtual” is used because the “virtual ward” does not have any physical premises. Instead, the patients are cared for at home.

I had recently moved from being a hospital doctor to working in the community and one of the things that struck me was how many great services were being offered to patients in the community; however, it was disappointing that these services were not better joined up. For example, I heard stories about patients receiving the tests and investigations multiple times in the course of a single week simply because different community teams did not know what each other were doing. Virtual wards aim to improve communication to avoid this type of inefficiency.

After returning from Australia, my first job as a public health doctor was at Croydon Primary Care Trust (PCTs) in South London. At the time, the Department of Health had funded a project in two parts of England – Croydon and South Warwickshire – to build a new predictive model that would advise GPs which of their patients were at highest risk of an unplanned hospital admission in the next 12 months. In return for participating in this project, the two local PCTs were given early access to the tool.

The question I was asked by the Director of Public Health in Croydon was what services we should offer to patients who were identified as being at high risk by the new model. I worked with two nurses to develop a new preventive intervention to be offered to high-risk patients. Admission to a virtual ward is offered to those patients who have the highest risk of an unplanned hospital admission based on the predictions of the predictive model.

CB: Has the Virtual Wards project been successful?

Geraint Lewis. Photo courtesy of Geraint Lewis.
Geraint Lewis. Photo courtesy of Geraint Lewis.

Lewis: An evaluation of virtual wards in three parts of England found that they were associated with a significant reduction in both outpatient attendances and elective admissions but not in unplanned admissions.

More recently, a study of a virtual ward in Hong Kong found a significant reduction in unplanned emergency readmissions as well as a significant improvement in patients’ quality of life. This suggests that the ‘Virtual Ward’ concept could be refined and adapted to focus on different outcomes.

CB: During your time as a Harkness Fellow, you did work on predictive risk modeling and “upstream” care. What is upstream care? And what did you learn through that work?

Lewis: “Upstream care” is the name given to preventive care that is offered to a patient who has been identified as being at high risk of a future unplanned hospital admission. The aim of upstream care is to reduce the risk that the patient will experience a future adverse outcome (e.g. a “downstream” unplanned hospital admission).

I spent my Harkness Fellowship studying how predictive modelling was being offered in various parts of the U.S. health care system. I conducted a series of interviews with academics, analysts, researchers and policymakers. I found that there was widespread and growing interest in a concept called “impactibility modeling.”

Impactibility models are a form of filter that are applied to the group of high-risk patients identified by a predictive risk model. The aim of an impactibility model is to filter out those patients who are unlikely to benefit from “upstream care,” leaving behind the subgroup of high-risk patients who are most likely to benefit (i.e., those patients who are most likely to have their risk of hospitalisation mitigated by the preventive care being offered).

My research suggested that impactibility models may be a way to improve the effectiveness and efficiency of “upstream” care because they can prevent resources being wasted on patients who will not benefit from the preventive care.

However, while I found that some types of impactibility model may reduce health care inequalities (e.g., impactibility models that prioritise patients who are currently receiving suboptimal care), other types of impactibility model may worsen inequalities (e.g., impactibility models that de-prioritise patients whose first language is not English).

CB: You also spent time working for Walgreens, a drug retailer based in the U.S. What led you to move from the public sector to the private sector?

Lewis: After finishing my training in public health, I was appointed Senior Fellow of the Nuffield Trust, a health policy think tank based in London.

A few years later, in 2011, I was approached by Walgreens who encouraged me to apply for a post in their Outcomes and Analytics division. I spent 2012 working at Walgreens as Senior Director for Outcomes and Analytics, based near Chicago. During my time with the company, I helped evaluate several worksite-based employer health programs and I helped establish a series of Accountable Care Organisations (ACOs) as part of the Medicare Shared Savings Program. These ACOs were partnerships between Walgreens and physician groups in Florida, Texas, and New Jersey.

CB: Did your time in the private sector inform the way you view your work now that you’re back in public health?

Lewis: One of the things that impressed me most about working for Walgreens was the investment in innovation. In the NHS we have recently made strides towards harnessing innovation more effectively in healthcare.  For example, the NHS Five Year Forward View included a commitment to finding innovative solutions to the mounting pressures on the system. I think there are lessons that we can learn from the private sector on diffusing successful innovations more rapidly.

CB: I’m very curious about your current job. What is the job description of the chief data officer of the NHS?

Lewis: My team and I have a broad range of responsibilities relating to data, mostly relating to what we call “secondary uses” of data (i.e., uses of data other than for direct clinical care). These uses include healthcare planning, public health, clinical audit, benchmarking, performance improvement, medical research and policy development. My responsibilities include aspects of information governance, data standards, data extractions and collections, as well as the publication of open data and the promotion of greater transparency. We work closely with different parts of NHS England as well as with partner organisations such as the Department of Health and the Health and Social Care Information Centre.

CB: Did you have specific goals for what you wanted to accomplish when you took the job in January 2013? Have those goals changed over the past two years?

Lewis: My principal goal was to expand the existing Hospital Episodes Statistics (HES) service to cover all settings of care. Since the 1980s, the NHS in England has collected data for the whole population, spanning all hospitals. In 2003, HES was expanded to include outpatient data and then in 2008 it was expanded again to include Accident & Emergency data. I am keen to see HES expanded to other settings of care, such as ambulances, primary care, and community health services.

This principal goal of mine remains unchanged; however, the timescales have!

I have come to realise that achieving change at the national level takes much longer than I had anticipated, principally because of the need to engage with such a wide range of stakeholders in order to secure their input and to build and maintain their trust.

CB: How does NHS organize all of the data it collects, and then communicate it to all of the audiences that can use it to inform various decisions? For example, I’m guessing that NHS data is used internally to help guide policy decisions; shared with doctors to inform clinical practice; shared with researchers to help steer future research; shared with the public to keep them abreast of what NHS is doing and (maybe) to help them make informed decisions about their own health. How do you organize and disseminate information to so many disparate audiences?

Lewis: At a local level, there are many different flows of data, which vary from area to area and from region to region.

At a national level, a new organisation was established in April 2013, the Health and Social Care Information Centre (HSCIC). Its role is to serve as the statutory ‘safe haven’ for NHS data: collecting, analysing and presenting national health and social care data, creating indicators that can be used to measure the quality of health and care services, and helping health and care organisations improve the quality of the data they collect.

NHS England has the legal power to direct the HSCIC to collect information from all providers of NHS funded care.

CB: I know that it’s important to protect the privacy of patients. How does NHS protect patient information, while still making sure that key audiences are getting the information they need to make informed decisions? Are there trade-offs involved, between protecting patient privacy and sharing data?

Lewis: It is important to stress that protecting the confidentiality of patient information will always remain of paramount importance to the NHS.

We and our partner organisations comply with a wide range of laws and regulations as well as the NHS Constitution. For example, the Care Act (2014) places a legal obligation on the HSCIC to respect and promote the privacy of patients, and the NHS Constitution states that patients have the right to privacy and confidentiality and to expect the NHS to keep your confidential information safe and secure.

CB: You’ve worked in the U.S. and Australia, as well as the U.K. How, if at all, have your international work experiences informed your approach to your current job?

Lewis: My former boss described the U.S. health care system as a bit of a jungle: overall it’s a bit of a mess but within this mess there are some amazing examples of clinical practice and the use of technology, which I have learnt from.

For example, Intermountain, Kaiser, Geisinger, and the Mayo Clinic all hold useful lessons for the NHS, as does the Veterans Health Administration.

CB: The high-profile cases of Ebola over the past year have highlighted the international nature of public health. How do you facilitate data-sharing with other countries and international organizations to inform public health decisions?

Lewis: There are very strict rules about sharing data outside the UK, and especially outside the European Economic Area (EEA). For example, Principle 8 of the Data Protection Act 1998 states that “Personal data shall not be transferred to a country or territory outside the EEA unless that country or territory ensures an adequate level of protection for the rights and freedoms of data subjects in relation to the processing of personal data.”

When it comes to aggregate data that have been rendered fit for publication, it greatly helps analysts with their international comparisons if the data adhere to international standards such as SNOMED-CT.

More recently, I have been impressed by the work of the International Collaboration for Health Outcomes Measurement, whose aim is to define a global set of standards by which health care providers across the world measure the outcomes for different treatments that matter most to patients.

CB: Technology, particularly social media and information technologies, is constantly evolving. Have you seen any meaningful changes in data-sharing technology in your first two years on the job?

Lewis: Not really.

CB: Given the sheer size of the NHS, what do you do to ensure that the overall system is up to speed on current data management and data-sharing tools and techniques?

Lewis: In my opinion, it’s not the size of the NHS that is the challenge but rather its complexity. The NHS is comprised of a large number of organisations, including Clinical Commissioning Groups, Commissioning Support Units, NHS Trusts, NHS Foundation Trusts, and a range of Arms-Length Bodies of the Department of Health. Many of these organisations have overlapping responsibilities and it’s important we work together to ensure that our work is coordinated.

The National Information Board was established last year to help coordinate informatics policy across the whole of the health and care sector and to ensure that the NHS adopts state-of-the-art tools and techniques. For further details see

CB: If you could get science communication and health communication researchers to tackle one specific issue – an outstanding challenge that you think needs to be addressed – what would it be? Why?

Lewis: I think it would be to explain the concept of linked data (i.e., data that have been joined up at the individual level over time and across different health and social care providers). Linked data are an extremely precious resource because they allow analysts to track individuals as they interact with different parts of the health service. Without linked data we cannot know how well different parts of the NHS are working together to provide high quality, joined-up care for patients.


One thought on “Many Different Flows of Data: an Interview with Geraint Lewis

  1. Pingback: [BLOCKED BY STBV] Morsels For The Mind – 27/02/2015 › Six Incredible Things Before Breakfast

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s