At the core of Dr. Robert Wachter’s brilliant The Digital Doctor (McGraw Hill, 2015) is the near death of a boy who receives a 38 1/2-fold overdose as a result of an EMR alert that is “a model of bad design.” The book describes a world-class acute care facility infested with alerts, alarms, and notifications designed by multiple device and software manufacturers, each seeming to imagine that their device would be the only one in the hospital and generally willing to “over-alarm.”
In this frightening world, senior physicians train young residents to “ignore the alerts,” and experienced nurses pile pillows on the most annoying alarms.
If you are a primary care physician—in family practice, an internist, a geriatrician—you should be very afraid. For a couple of decades, the “establishment” hasn’t seemed to care about you very much…your income has declined steadily, your ability to control referrals and specialist authorizations has eroded, and your regulatory paperwork has exploded.
Now, suddenly, everyone wants to talk to you. You will be the gatekeeper and the bearer of risk, the coordinator of care, and the creator of the care plan. And everyone seems to know how you should do your job. Massive databases will be mined, patient records will be aggregated longitudinally, predictive analytics will generate actionable insights, and your reimbursement will be based on how you perform in response to all the wisdom coming your way.
There is a great deal of goodness in all this. We need to put primary care in the control position, to reward it as the “top license,” and to feed it the very best that science and statistical analysis can deliver.
But if we repeat the design mistakes of the EMR and device manufacturers in acute care, and apply them to population health, we will stop accountable care in its tracks. In Dr. Wachter’s words, “The issue of alarm and alert fatigue is a clear and present danger.” Vendors of many kinds (EMR, population health, care management, HIE, et al.) are busily planning to apply analytics and big data techniques to generate population health alerts, alarms, warnings, care gaps, practice improvement reports, personalized patient encounter guidance, and individualized care plans.
Yet we know that the primary care community will ignore this flood if it drowns them as alerts flood the ICU today. Already, adoption of even simple alerts like care gaps remains frustratingly low.
How can we fix this?
Dr. Wachter identifies the key remedies for the acute setting, summarized below. The purpose of this blog post is to add a couple of ideas that address the differences between the acute and ambulatory environments.
Dr. Wachter’s keys to effective alerts and alarms:
- Parsimony. “Be sparing with alerts, because every alert makes it less likely that people will pay attention to the next one.” The five ICUs in Dr. Wachter’s book experienced 2.5 million alerts per month.
- Tiering. Dr. Wachter describes how rigorously Boeing’s cockpit designers tier warnings, cautions, and advisories to make sure that top-tier alarms grab the pilot’s attention.
- Task Orientation. Every alert or alarm should be presented with a “guide to a solution,” ideally in the form of a clear checklist.
- User-Driven Design. Dr. Wachter offers a withering assessment of health IT designs that lack a critical “underlying philosophy” to involve actual users to see if “alerts (and everything else) work the way the designers intended.”
- Common Sense. Nonsensical alerts, which undermine confidence and attention, can often be avoided by having someone with “knowledge of the domain” review and correct such results during the quality assurance process. For example, blood pressure alarms are often calibrated to alarm when detecting values that are commonly expected and managed during an acute care episode.
These themes will all apply to population health, but may take on different meaning. For example, the tasks of primary care-based accountable care are more focused on assessment and prevention than acute care. Thus, as opposed to direct medical intervention or procedure, the major population health tasks are often scheduling of encounters and labs, the implementation of referrals and authorizations, and the preparation or gathering of documentation.
Here are some design principles for population health communications that are complementary to Dr. Wachter’s acute care concepts:
- Target by User Type. Many tasks recommended or implied by population health alerts will NOT be performed by the doctor, but by other members of the care team. Communications should be targeted to the appropriate staffer. Since doctors practice at different locations for different contract entities, supported by distinct care teams and support staff, HIPAA-safe delivery of clinical information and associated alerts will depend on mapping patients not only to their individual physician, but also to the appropriate team member(s) who support the specific patient for the specific physician at the specific location.
- Workflow is Integral. In contrast to acute care alarms, actions suggested by population health management systems are often effected and/or tracked on the same computer system that is carrying the alerting communication (“schedule health status assessments for the following patient roster”). Therefore, it is critical to embed population health communications within familiar and well-adopted workflows. For example, if Medical Management needs information from the EMR or HIE to support analysis of a recommended Care Plan, this “chart pull” might be embedded in the related Pended Authorization workflow, which already has established work processes and allocated staff resources in most provider locations. Thus, the user who receives the chart pull communication request should be able to “pull the chart” within the same computer session and track progress of this work among his or her other tasks.
- Standards-Based Integration Will Gate Progress. Most users continue to prefer a “home system” for the bulk of their work. Many will work in an EMR, Practice Management System, or Payer-Provider Portal—all of which will be producers and consumers of population health communications. Successful population health management will require a reliable flow of artifacts between these systems.
- Simplify the UI to Minimize Variation. Providers are dealing with many payers who each have multiple insurance products. These products are typically based on distinct contract terms and may be supported by unique payer informatics and population health vendor systems, each of which generates multiple population health artifacts at the population and individual level. Successful population health communication systems must rationalize and standardize the appearance and workflow “verbs” of the common alerts, alarms, warnings, care gaps, practice improvement reports, personalized patient encounter guidance, and individualized care plans.
Successful population health communication systems will ensure that the right team member receives comprehensible information at the right time in well-adopted workflows. To avoid “drowning the doctor,” particular care must be taken to target the individual physician only when she is the right person to act on a given specific communication.