Healthcare has traditionally been conservative in its adoption of new technologies. Many clinicians still routinely use pen and paper to record medical information during a face-to-face encounter with a patient on the ward or in the clinic.

Information technology was first adopted in order to manage the administrative running of hospitals such as appointment scheduling and tracking the location of paper records. Traditionally, there has been a separation of these patient administrative systems (PAS) from clinical systems used by health professionals, even if such systems were available.a

However, there are a number of transformative and disruptive influences in health that, together with new technology, are driving and will continue to drive new ways of working. I’ve grouped these influences into three broad related categories:

  • Changes in the models of care
  • Patients as active participants
  • ‘Hospitals without walls’

Changes in the models of care

The first group of transformative influences relate to the changing way in which healthcare is provided, primarily as a result of changing working practices, a need for safe and secure channels of communication between the patient and multi-disciplinary teams from different organisations and different sectors and a focus on measures of outcome and not process. Such changes are particularly relevant to an ageing population with individuals living longer with multiple long-term health conditions.

Changes in how patients experience their care

The traditional approach to healthcare is based on face-to-face consultations with a primary care physician in a community-based facility, an outpatient clinic with a specialist in a hospital or an unscheduled or scheduled stay within that hospital for a medical emergency or an elective procedure.

As a consequence of this model, patients with diabetes mellitus used to be called for specialist outpatient review every 3-months and care would be provided by a fragmented patchwork with little or no overall coordination between services.

A modern approach to healthcare requires support for:

  • Management of multiple long-term health conditions requiring multi-disciplinary teamwork, cross-organisation care coordination as well as involvement of non-health partners such as social care and the third-sector when appropriate.
  • Shared care between teams with remote monitoring using modern technology.
  • Avoidance of ‘information silos’ that would usually result in information not being available at the right time or place for clinical decision making. Information should follow the patient and not be fixed in the organisation.

As a result, a patient with diabetes mellitus should now be able to monitor their disease themselves with support from community teams including the general practitioner who can call upon a range of specialist services for specific complications or surveillance. Rather than a patient being screened for retinopathy (a complication of diabetes mellitus) in a clinic by an inexperienced junior doctor, they are called to a dedicated diabetic retinopathy screening programme with retinal photography and the results made available to all others in the team.

Changes in how we measure care

An important parallel change has been a focus on health outcomes rather than assessing services according to simply how many patients are seen or how many have a procedure. A focus on outcomes permits services to judge whether an intervention such as a knee replacement, actually improves outcomes in groups of patients with similar characteristics to yours. By analysing aggregated outcomes for groups of patients, we can modify our services to achieve the best outcome for patients and benchmark local services with those from elsewhere.

Changing our practice to consider outcomes requires an infrastructure that allows the routine and systematic capture of structured clinical data before and after by both clinicians and patients. Outcomes may be based on questionnaires but can just as easily be other outcomes such as return to work, number of steps walked per day, ability to exercise or managing to live independently in one’s home.

“Big data”

As a result, we face a future in which we have ever increasing amounts of data on patients. Information sourced from multiple organisations, increasingly structured outcome data and information from technology worn by the patient or used in the home to monitor progress and health status.

Genomics is an important additional potential source of “big data”. Genomics provides genotype data which, when linked with phenotypic data (what the patient looks like), allows us to assess genotype-phenotype correlations. Such analysis allows us to understand the causes of disease and the modifiers that affect progression of disease or predict responsiveness to treatment.

All such data is sensitive and it is in the linkage of these datasets that is most important. While all medical information is confidential, it is linkage that those of us who are interested in protected privacy should focus.b

Patients as active participants

The second group of transformative influences relate to a repositioning of a patient from a person unto which care is performed to an active and involved participant in their care, controlling their interactions and deciding how and with whom their information should be shared. Such a repositioning is necessary given the changes in models care care outlined above.

As a result of this paradigm shift, it becomes necessary to design our information systems around the patient and not our organisations. Instead of fragmented services being run using fragmented software, information technology is needed to support an approach in which the patient is at the centre.

Accordingly, patients must be able to:

  • View and contribute to their own records.
  • Record their aims and objectives as well as their own outcomes for use in monitoring their problems and assessing the efficacy of any intervention.
  • Communicate with members of their healthcare teams as well as other professionals and volunteers who are involved in their care with those communications forming part of a longitudinal record.
  • Control how and what information is shared about them, including choosing to share their information for clinical research.

Communication between patients and their services

The Government Procurement Service (GPS) has estimated that the NHS in England spends £79 million each year on postal costs. While many patients do try to communicate with their healthcare teams via email, there are several problems that restrict its wider use:

  • Current email is insecure with both content and metadata unencrypted with no way of proving identity. While cryptographic technologies are available, tools for signing and encrypting communications, such as ‘PGP’, are too difficult for most users to use routinely.
  • Email communications are not usually filed into medical records and so the record is not available to the wider team, unless printed and placed into paper records.
  • Even encrypted email communications leak confidential information by virtue of the metadata required to appropriately route those communications. For example, one can infer a relationship between a patient and the HIV clinic if they regularly communicate with [email protected] even if the content of that communication is encrypted.

Communication may be appointment letters, cancellations, clinical communications but also data from devices such as a video recorded of an epileptic seizure, a step-count from an ambulatory step-monitor or an implantable device monitoring heart rhythm.

Patient consent for information sharing may be implicit, in which, consent is assumed until a patient opts out, or explicit, in which a specific opt-in is required before information retrieval or processing can begin.

Typically, an opt-in model is required for patients to participate in clinical research. It is recognised that patients are keen to be involved in research but wish to be asked for their consent before their identifiable information is shared. However, that report also notes that “more could be done to increase awareness of the benefits of research”. As a result, intermediary brokering services are needed to allow patients to signify their interest in clinical research and permit them to subsequently be informed of relevant research projects without divulging their private information. ‘Research brokers’ should match patients to projects and aim to maximise participation. To enable such functionality, private information such as diagnosis and treatments must be processed but via mechanisms that prevent the inappropriate use of that information for other purposes. A patient who opts-in to receive information about research suitable for them must (i) expect to have their information not used for any other purpose except brokering, and (ii) not be shown projects that are obviously unsuitable for them.

Much like Google connects users to advertisers, there is value in brokering connections between patients and researchers. The use of SNOMED-CT as an ontology (algorithmically I can understand that ‘multiple sclerosis’ IS-A type of ‘demyelinating disease’ for example) can help to support such connections based on diagnosis, procedure or other facets.

While simple brokering requires little more than matching on SNOMED-CT, how can more complex requirements be assessed such as, “patients with JME who have not yet had lamotrigine” - how can allow a broker permit a third-party to check suitability without divulging identity?

Hospitals without walls

The third group of transformative influences relate to the decentralisation and democratisation of healthcare enabled by new technologies. These technologies comprise:

  • Increasing availability of personal mobile devices
  • Increasing capability for smart home monitoring with an ‘internet of things’ in which multiple devices are connected together and combine to provide a unified view of a patient
  • New commercial models of care, including advanced analytics, supported clinical decision making with applied machine learning for ‘artificial intelligence’.

Mobile devices, home monitoring and the ‘internet of things’

New technology means that patients can now increasingly receive their care and monitoring in their home using a range of medical grade devices. For example, implantable loop recorders can be inserted under the skin and, in the event of a cardiac event, the results sent wirelessly to the healthcare team. In addition, there is increasing acceptance of commodity consumer-grade devices, frequently marketed as fitness devices, in order to gain objective data on health outcomes in order to monitor a disease or to assess the effect of an intervention. There are parallels with the technology used for video-conferencing in which professional, enterprise-grade devices installed in business are increasingly replaced with consumer-grade devices.

As a result, there is now consumer-grade technology available to allow patients to obtain ongoing advice and care from a range of professionals from a range of organisations via electronic communication channels such as electronic mail or voice or video calls rather than attending for routine outpatient clinic appointments. Such communications need to be safe and secure, with guarantees about identity at both sides of the communication, with guarantees of non-repudiation and verifiability with an appropriate entry recorded contemporaneously in the medical record.

However, just as outlined above, data transfers from a patient and their devices to clinical services must be encrypted and be kept private with avoidance of information leakage. As a result, not only must the content of those transfers be encrypted but the metadata needed to describe and route those transfers must be kept private. For example, network traffic analysis may be able to identify all elderly frail patients living alone with dementia in a region simply by looking for specific types of electronic monitoring data streams in a similar way to being able to identify when a person leaves their house by eavesdropping on smart meter data.

Innovation, data sharing and analytics

There is significant potential benefit to the use of software within healthcare in order to improve outcomes, safety and efficiency.

Such software might provide functionality to improve the care of a single patient, or might work across groups of patients to provide analysis of services and outcomes providing a view for clinicians and managers that combines data from multiple sources. In parallel, clinical users are now expecting to be able to use their own or the organisation’s mobile devices in order to work effectively when on the ward. Such applications support team-working, communication and alerts regarding patients under their care.

The traditional approach has been to bring innovation ‘within the walls’ of health organisations which can work when procuring specific products from vendors in order to perform a specific task. However, this model is less suitable for products which provide services via ‘cloud computing’ in which large-scale computing power may be provided across the network. For example, if a commercial entity provided automated or computer-assisted interpretation of histological slides or radiology imaging to multiple organisations using the results of deep learning, that functionality may be better provided across a network rather than being installed within the server rooms of a health organisation.

As a result, the “hospitals-without-walls” trend requires a focus on the network with a mixed economy of core functionality upon which innovative applications can operate in secure sandboxes, in which data access and interactions can be monitored and controlled.

My next post discusses the current solutions already created in response to these disruptive influences. The third post discusses how an open platform might better deal with these disrupting influences in the future.


a: For more information on the separation of administrative and clinical systems, see my blog post on patient safety.

b: I’m not sure that everyone realises this.