A report on “Delivering the benefits of digital health care” from Nuffield Trust in the UK recently came across my desk. It covers a bigger scope of technology than the project I’m working on (which is a project about transforming clinical care (and implementing an electronic health record across three large health organizations to support this clinical transformation), but does not include telehealth and some of the other IT “solutions” talked about in this report), but some of the “lessons learned” that they share resonate with what we are doing.
Some highlights:
“Clinically led improvement, enabled by new technology, is transforming the delivery of health care and our management of population health. Yet strategic decisions about clinical transformation and the associated investment in information and digital technology can all too often be a footnote to NHS board discussions. This needs to change. This report sets out the possibilities to transform health care offered by digital technologies, with important insight about how to grasp those possibilities and benefits from those furthest on in their digital journey” (p. 5, emphasis mine)
- this report suggests that rather than focusing on the technology with an eye to productivity gains, “the most significant gains are to be found in more radical thinking and changes in clinical working practices” (p. 5).
- it’s “not about replacing analogue or paper processes with digital ones. It’s about rethinking what work is done, re-engineering how it is done and capitalising on opportunities afforded by data to learn and adapt.” (p. 6)
- This reminds me of what my IT management professor in my MBA program liked to say: “If you automate a mess, all you get is an automated mess”. It’s much better to focus on getting your processes right, and then automating them, rather than just automating what you have.
- “”It’s fundamentally not a technology project; it’s fundamentally a culture change and a business transformation project” (Robert Wachter, UCSF)” (p. 22)
- in a notable failure, the NHS in the UK spent 9 years and nearly £10 billion and failed to digitise the hospital and community health sectors with reasons for the failure being “multiple, complex, and overlapping” including “an attempt to force top-down change, with lack of consideration to clinical leadership, local requirements, concerns, or skills” (p. 14)
- it is noted that implementing an electronic health record (EHR) [which is what the project I’m working on is doing) is particularly challenging
- they also note that things take longer than you expect:
- “The history of technology as it enters industries is that people say ‘this is going to transform everything in two years’. And then you put it in and nothing happens and people say ‘why didn’t it work the way we expected it to?… And then lo and behold after a period of 10 years, it begins working.” (Robert Wachter, University of California San Francisco (UCSF)” (p. 20)
- and they note that “the technologies that have released the greatest immediate benefits have been carefully designed to make people’s jobs or the patient’s interaction easier, with considerable investment in the design process.” (p. 20)
- poorly designed systems, however, can really decrease productivity
- getting full benefit of the system “requires a sophisticated and complex interplay between the technology, the ‘thoughtflow’ (clinical decision-making) and the ‘workflow’ (the clinical pathway)” (p. 21)
- systems with automated data entry (e.g., bedside medical device integration, where devices that monitor vital signs at the bedside automatically enter their data into the EHR, without requiring a clinician to do it manually) really help maximize the benefits
Seven Lessons Learned
- [Clinical] Transformation First
- it’s a “transformation programme supported by new technology” (p. 22)
- Culture change is crucial
- “many of the issues face […] are people problems, not technology problems” (p. 23)
- you need:
- “a culture that is receptive to change
- a strong change management process
- clinical champions/supporting active staff engagement” (p. 23)
- User-centred design
- you need to really understand the work so that you design the system to meet the needs of the clinician
- “the combination of a core package solution with a small number of specialist clinical systems is emerging as the norm in top-performing digital hospitals” (p. 8)
- Invest in analytics
- data analytics allows you to make use of all the data you collect as a whole (in addition to using it for direct clinical care)
- requires “analytical tools available to clinicians in real time” (p. 8)
- Multiple iterations and continuous learning
- you aren’t going to get it right the first time, no matter how carefully you plan [this is something that our new Chief Clinical Information Officer is always reminding us of] and so you will need “several cycles – some quite painful – before the system reaches a tipping point where all of this investment starts to pay off” (p. 26)
- Support interoperability
- to provide coordinated care, you need to be able to share data across multiple settings
- “high-performing digital hospitals are integrating all their systems, to as low a number as possible, across their organisation” (p. 9)
- Strong information governance
- when you start to digitize patient information, the size and scope of privacy issues change (i.e., while there is risk that an authorized person could look at a patient’s paper record or paper records could be lost when being transported between places, with digitized record there is a risk that all of your patients’ record could be accessed by an unauthorized person and that it is much easier to search electronic records for a specific person, condition, etc.)
- you need “strong data governance and security” (p. 9)
Seven Opportunities to Drive Improvement
- More systematic, high-quality care
- health care “often falls short of evidence-based good practice” (p. 31)
- “technologies that aid clinical decision-making and help clinicians to manage the exponential growth in medical knowledge and evidence offer substantial opportunities to reduce variation and improve the quality of care” (p. 31)
- integrated clinical decision support systems and computerized provider order entry systems:
- reduce the likelihood of med errors (they cite a review paper (Radley et al, 2013) [which I have now obtained to check out what methods the papers they reviewed used to measure med errors]
- reduced provider resource use
- reduced lab, pharmacy & radiology turnaround times
- reduced need for ancillary staff (p. 32)
- at Intermountain Healthcare, “staff are encouraged to deviated from the standardised protocol, subject to clear justification for doing so, with a view to it being refined over time” (p. 34) – “hold on to variation across patients and limit variation across clinicians” (p. 35) as “no protocol perfectly fits each patient” (p. 35)
- need to avoid alert fatigue – by only using them sparingly (or else they will get ignored and the really important ones will be missed) and targeting them to the right time (e.g., having prescribing alerts fire while the provider is prescribing)
- be on the lookout for over-compliance – “Intermountain Health experience problems where clinicians were too ready to adopt the default prescribing choice, leading to inappropriate care in some cases” (p. 37)
- More proactive and targeted care
- “patient data can be used to predict clinical risk, enabling providers to target resources where they are needed most and spot problems that would benefit from early intervention” (p. 38)
- drawing on patient data, computer-based algorithms “can generate risk scores, highlighting those at high risk of readmission and allowing preventative measures to be put in place” (p. 39)
- “it may also have a role in predicting those in the community who are likely to use health care services in the near future” (p. 39)
- “monitoring of vital signs, [which are then] electronically recorded, [can be used to] calculate early warning scores [and] automatically escalate to appropriate clinicians [and] “combine these data with laboratory tests to alert staff to risks of sepsis, acute kidney injury or diarrhoeal illness” (p. 39)
- Cerner estimates using early warning system for sepsis “could reduce in-hospital patient mortality by 24% and reduce length of stay by 21%, saving US$5,882 per treated patient” (p. 41)
- there’s also opportunity to “check a patient’s status from remote location within the hospital, as well as facilitating handover between staff and task prioritisation using electronic lists” (p. 39)
- monitoring of vital signs throughout the whole hospital is best to maximize benefits
- predictive analytics is only as good as the quality of the data you put into the system
- lots of data is unstructured – need to find ways to use these data (e.g., natural language processing)
- Better coordinated care
- coordinated care leads to a better care experience, reduces risk of duplication or neglect
- “if all health care professionals have access to all patient information in real time, there is significant potential to reduce waste (e.g., duplication of tests). It can help make sure things are done at the right time, at the right place and not overdone” (p. 45)
- “chasing report or a result […is…] an inefficient use of time, effort and energy and doesn’t really give confidence to the patient and carers” (p. 47)
- but note that “systems to share results/opinions digitally can remove the opportunity for informal exchange of views and advice across teams, which often enrich and improve clinical decision-making” (p. 48), so alternative ways of doing this may need to be provided.
- Improved access to specialist expertise
- telehealth (not part of the project I’m working on)
- Greater patient engagement
- this section referred to tools, like wearable tech (e.g., Fitbit) or patient portals that empower patients to take more control of their own health (not part of the project I’m working on)
- “patient co-production of data into a hospital EHR will redefine the interaction with care services” (e..g, questionnaires that patients fill out before they even come to the healthcare facility, tracking of long-term data (e.g., blood pressure, weight))
- Improved resource management
- e-rostering (i.e., of staff), patient flow management, business process support (e.g, HR, facilities, billing) all discussed (not relevant to the project I’m working on)
- ability of staff to remotely access health records “can transform the way hat staff in the community deliver care” (p. 66)
- System improvement and learning
- “feeding learning from clinical and non-clinical data back into existing processes is essential to fully realising the benefits of digital technology” (p. 70)
- Intermountain Healthcare:
- captures 3 type of data:
- intermediate & final clinical outcomes
- cost data
- patient satisfaction and experience
- “clinical registries are derived directly from clinical workflows” – currently has “58 condition-specific registries – tracking a complete set of intermediate and final clinical and cost outcomes by patient” (e.g., 71)
- remember that data collection is costly, so only collect data routinely if you are using it for some purpose that adds value (“Intermountain Healthcare does this through small individual projects, before building data collection into existings processes”) (p. 76)
- captures 3 type of data:
What could the future look like?
- operational improvement from:
- combining impact of a bunch of small changes [this assumes that (a) the different elements of the system are additive, as opposed to complex, and (b) the “benefits” outweigh the unintended negative consequences]
- getting the “full benefit” out of all the technologies (i.e., it will take time for people to implement the available technologies and to optimize their use) [this doesn’t even include technologies that are not yet available)
- “benefits” they expect are most likely to see:
- “reduced duplication and rework
- removing unjustified variation in standard clinical processes
- identifying deteriorating patients and those at risk
- predicting the probability of an extended stay or readmission
- cutting out unnecessary steps
- improving communication and handoffs
- removing administrative tasks from clinical staff
- scheduling and improving flow
- inventory & procurement management
- rostering, mobile working, and staff deployment
- patient self-service for administrative tasks such as booking
- other automation, e.g., robotics in back office” (p. 80-1)
- redesigning the whole pathway:
- “reduced variation
- ability to ensure the most appropriate level of care
- fitting staffing skill mix to demand more effectively” (p. 81)
- population health management
- “early intervention & targeting
- enabling patient self-management
- shared decision-making
- measuring outcomes and value rather than counting activities” (p. 82)
- all this requires better data and analytics, learning & improvement processes, and supporting patients with self-management and supporting shared decision-making (p. 82)
“Early strategic priorities should be the areas where technology is able to facilitate some relatively easy and significant wins. Most notable are the systematic and comprehensive use of vital signs monitoring and support for mobile working. In the short to medium term, the use of EHRs, telehealth, patient portals and staff rostering apps can also generate savings and improve quality. However, these require sophisticated leadership with support for organisational development and change management to ensure that the full benefits are realised. In the longer term, the really big benefits will come from the transition to a system and ways of working premised on continual learning and self-improvement.” (p. 88, emphasis mine)
Potential intended consequences mentioned in the report:
- decreased productivity if the system is poorly designed (e.g., time spent on data entry, time spent responding to unhelpful alerts)
- “over-compliance” – “Intermountain Health experience problems where clinicians were too ready to adopt the default prescribing choice, leading to inappropriate care in some cases” (p. 37)
- “systems to share results/opinions digitally can remove the opportunity for informal exchange of views and advice across teams, which often enrich and improve clinical decision-making” (p. 48),
Limitations:
- they noted there was little evidence on this type of work in the literature, particularly in terms of return on investment
