Way back in 2009 I penned a research paper with a long and complicated title that could also have been, simply, Population Health Management and Business Process Management. In 2010 I presented it at MedInfo10 in Cape Town, Africa. Check out my travelogue!


Since then, some of what I wrote has become reality, and much of the rest is on the way. Before I dive into the weeds, let me set the stage. The Affordable Care Act added tens of millions of new patients to an already creaky and dysfunctional healthcare and health IT system. Accountable Care Organizations were conceived as virtual enterprises to be paid to manage the clinical outcome and costs of care of specific populations of individuals. Population Health Management has become the dominant conceptual framework for proceeding.
 
I looked at a bunch of definitions of population health management and created the following as a synthesis: “Proactive management of clinical and financial risks of a defined patient group to improve clinical outcomes and reduce cost via targeted, coordinated engagement of providers and patients across all care settings.”
 
You can see obvious places in this definition to apply trendy SMAC tech — social, mobile, analytics, and cloud — social, patient settings; mobile, provider and patient settings; analytics, cost and outcomes; cloud, across settings. But here I want to focus on the “targeted, coordinated.” Increasingly, it is self-developed and vendor-supplied care coordination platforms that target and coordinate, filling a gap between EHRs and day-to-day provider and patient workflows.

The best technology on which, from which, to create care coordination platforms is workflow technology, AKA business process management and adaptive/dynamic case management software. In fact, when I drill down on most sophisticated, scalable population health management and care coordination solutions, I usually find a combination of a couple things. Either the health IT organization or vendor is, in essence, reinventing the workflow tech wheel, or they embed or build on third-party BPM technology.
 
Let me direct you to my section Patient Class Event Hierarchy Intermediates Patient Event Stream and Automated Workflow in that MedInfo10 paper. First of all you have to target the right patients for intervention. Increasingly, ideas from Complex Event Processing are used to quickly and appropriately react to patient events. A Patient Class Event Hierarchy is a decision tree mediating between low-level events (patient state changes) and higher-level concepts clinical concepts such as “on-protocol,” “compliant”, “measured”, and “controlled.”
 
Examples include patients who aren’t on protocol but should be, aren’t being measured but should be, or whose clinical values are not controlled. Execution of appropriate automatic policy-based workflows (in effect, intervention plans) moves patients from off-protocol to on-protocol, non-compliance to compliance, unmeasured to measured, and from uncontrolled to controlled state categories.
 
Population health management and care coordination products and services may use different categories, terminology, etc. But they all tend to focus on sensing and reacting to untoward changes in patient state. But simply detecting these changes is insufficient. These systems need to cause actions. 
And these actions need to be monitored, managed, and improved, all of which are classic sterling qualities of business process management software systems and suites.
 
I’m reminded of several tweets about Accountable Care Organization IT systems I display during presentations. One summarizes an article about ACOs. The other paraphrases an ACO expert speaking at a conference. The former says ACOs must tie together many disparate IT systems. The later says ACOs boil down to lists: actionable lists of items delivered to the right person at the right time. If you put these requirements together with system-wide care pathways delivered safely and conveniently to the point of care, you get my three previous blog posts on interoperability, usability, and safety.
 
I’ll close here with my seven advantages of BPM-based care coordination technology. It…

• More granularly distinguishes workflow steps
• Captures more meaningful time-stamped task data
• More actively influences point-of-care workflow
• Helps model and understand workflow
• Better coordinates patient care task handoffs
• Monitors patient care task execution in real-time
• Systematically improves workflow effectiveness & efficiency

Distinguishing among workflow steps is important to collecting data about which steps provide value to providers and patients, as well as time-stamps necessary to estimate true costs. Further, since these steps are executed, or at least monitored, at the point-of-care, there’s more opportunity to facilitate and influence at the point-of-care. Modeling workflow contributes to understanding workflow, in my view an intrinsically valuable state of affairs. These workflow models can represent and compensate for interruptions to necessary care task handoffs. During workflow execution, “enactment” in BPM parlance, workflow state is made transparently visible. Finally, workflow data “exhaust” (particularly times-stamped evidence-based process maps) can be used to systematically find bottlenecks and plug care gaps.

In light of the fit between complex event processing detecting changes in patient state, and BPM’s automated, managed workflow at the point-of-care, I see no alternative to what I predicted in 2010. Regardless of whether it’s rebranded as care or healthcare process management, business process management is the most mature, practical, and scalable way to create the care coordination and population health management IT systems required by Accountable Care Organizations and the Affordable Care Act. A bit dramatically, I’d even say business process management’s royal road to healthcare runs through care coordination.

This was my fifth and final blog post in this series on healthcare and workflow technology

When you took a drivers education class, do you remember the importance of mental “awareness” to traffic safety? Continually monitor your environment, your car, and yourself. As in traffic flow, healthcare is full of work flow, and awareness of workflow is the key to patient safety.


First of all, the very act of creating a model of work to be done forces designers and users to very carefully think about and work through workflow “happy paths” and what to do when they’re fallen off. A happy path is a sequence of events that’s intended to happen, and, if all goes well, actually does happen most of the time. 

Departures from the Happy Path are called “exceptions” in computer programming parlance. Exceptions are “thrown”, “caught”, and “handled.” At the level of computer programming, an exception may occur when data is requested from a network resource, but the network is down. At the level of workflow, an exception might be a patient no-show, an abnormal lab value, or suddenly being called away by an emergency or higher priority circumstance.
 
Developing a model of work, variously called workflow/process definition or work plan forces workflow designers and workflow users to communicate at a level of abstraction that is much more natural and productive than either computer code or screen mockups.
 
Once a workflow model is created, it can be automatically analyzed for completeness and consistency. Similar to how a compiler can detect problems in code before it’s released, problems in workflow can be prevented. This sort of formal analysis is in its infancy, and is perhaps most advanced in healthcare in the design of medical devices.
 
When workflow engines execute models of work, work is performed. If this work would have otherwise necessarily been accomplished by humans, user workload is reduced. Recent research estimates a 7 percent increase in patient mortality for every additional patient increase in nurse workload. Decreasing workload should reduce patient mortality by a similar amount.
 
Another area of workflow technology that can increase patient safety is process mining. Process mining is similar, by analogy, to data mining, but the patterns it extracts from time stamped data are workflow models. These “process maps” are evidence-based representations of what really happens during use of an EHR or health IT system. 

Process maps can be quite different, and more eye opening, than process maps generated by asking participants questions about their workflows. Process maps can show what happens that shouldn’t, what doesn’t happen than should, and time-delays due to workflow bottlenecks. They are ideal tools to understand what happened during analysis of what may have caused a possibly system-precipitated medical error.
 
Yet another area of particular relevance of workflow tech to patient safety is the fascinating relationship between clinical pathways, guidelines, etc. and workflow and process definitions executed by workflow tech’s workflow engines. Clinical decision support, bringing the best, evidence-based medical knowledge to the point-of-care, must be seamless with clinical workflow. Otherwise, alert fatigue greatly reduces realization of the potential.
 
There’s considerable research into how to leverage and combine representations of clinical knowledge with clinical workflow. However, you really need a workflow system to take advantage of this intricate relationship. Hardcoded, workflow-oblivious systems? There’s no way to tweak alerts to workflow context: the who, what, why, when, where, and how of what the clinical is doing. Clinical decision support will not achieve wide spread success and acceptance until it can be intelligently customized and managed, during real-time clinical workflow execution. This, again, requires workflow tech at the point-of-care.
 
I’ve saved workflow tech’s most important contribution to patient safety until last: Interruptions.
 
An interruption–is there anything more dreaded than, just when you are beginning to experience optimal mental flow, a higher priority task interrupts your concentration. This is ironic, since so much of work-a-day ambulatory medicine is essentially interrupt-driven (to borrow from computer terminology). Unexpected higher priority tasks and emergencies *should* interrupt lower priority scheduled tasks. Though at the end of the day, ideally, you’ve accomplished all your tasks.
 
In one research study, over 50% of all healthcare errors were due to slips and lapses, such as not executing an intended action. In other words, good clinical intentions derailed by interruptions.
 
Workflow management systems provide environmental cues to remind clinical staff to resume interrupted tasks. They represent “stacks” of tasks so the entire care team works together to make sure that interrupted tasks are eventually and appropriately resumed. Workflow management technology can bring to clinical care many of the innovations we admire in the aviation domain, including well-defined steps, checklists, and workflow tools.

Stay tuned for my fifth, and final, guest blog post, in which I tackle Population Health Management with Business Process Management.

Workflow technology has a reputation, fortunately out of date, for trying to get rid of humans all together. Early on it was used for Straight-Through-Processing in which human stockbrokers were bypassed so stock trades happened in seconds instead of days. Business Process Management (BPM) can still do this. It can automate the logic and workflow that’d normally require a human to download something, check on a value and based on that value do something else useful, such as putting an item in a To-Do list. By automating low-level routine workflows, humans are freed to do more useful things that even workflow automation can’t automate.


But much of healthcare workflow requires human intervention. It is here that modern workflow technology really shines, by becoming an intelligent assistant proactively cooperating with human users to make their jobs easier. A decade ago, at MedInfo04 in San Francisco, I listed the five workflow usability principles that beg for workflow tech at the point-of-care.

Consider these major dimensions of workflow usability: naturalness, consistency, relevance, supportiveness, and flexibility. Workflow management concepts provide a useful bridge from usability concepts applied to single users to usability applied to users in teams. Each concept, realized correctly, contributes to shorter cycle time (encounter length) and increased throughput (patient volume).

Naturalness is the degree to which an application’s behavior matches task structure. In the case of workflow management, multiple task structures stretch across multiple EHR users in multiple roles. A patient visit to a medical practice office involves multiple interactions among patients, nurses, technicians, and physicians. Task analysis must therefore span all of these users and roles. Creation of a patient encounter process definition is an example of this kind of task analysis, and results in a machine executable (by the BPM workflow engine) representation of task structure.

Consistency is the degree to which an application reinforces and relies on user expectations. Process definitions enforce (and therefore reinforce) consistency of EHR user interactions with each other with respect to task goals and context. Over time, team members rely on this consistency to achieve highly automated and interleaved behavior. Consistent repetition leads to increased speed and accuracy.

Relevance is the degree to which extraneous input and output, which may confuse a user, is eliminated. Too much information can be as bad as not enough. Here, process definitions rely on EHR user roles (related sets of activities, responsibilities, and skills) to select appropriate screens, screen contents, and interaction behavior.

Supportiveness is the degree to which enough information is provided to a user to accomplish tasks. An application can support users by contributing to the shared mental model of system state that allows users to coordinate their activities with respect to each other. For example, since a EMR  workflow system represents and updates task status and responsibility in real time, this data can drive a display that gives all EHR users the big picture of who is waiting for what, for how long, and who is responsible.

Flexibility is the degree to which an application can accommodate user requirements, competencies, and preferences. This obviously relates back to each of the previous usability principles. Unnatural, inconsistent, irrelevant, and unsupportive behaviors (from the perspective of a specific user, task, and context) need to be flexibly changed to become natural, consistent, relevant, and supportive. Plus, different EHR users may require different BPM process definitions, or shared process definitions that can be parameterized to behave differently in different user task-contexts.

The ideal EHR/EMR should make the simple easy and fast, and the complex possible and practical. Then, the majority/minority rule applies. A majority of the time processing is simple, easy, and fast (generating the greatest output for the least input, thereby greatly increasing productivity). In the remaining minority of the time, the productivity increase may be less, but at least there are no showstoppers.

So, to summarize my five principles of workflow usability…


Workflow tech can more naturally match the task structure of a physician’s office through execution of workflow definitions. It can more consistently reinforce user expectations. Over time this leads to highly automated and interleaved team behavior. On a screen-by-screen basis, users encounter more relevant data and order entry options. 

Workflow tech can track pending tasks–which patients are waiting where, how long, for what, and who is responsible–and this data can be used to support a continually updated shared mental model among users. 

Finally, to the degree to which an EHR or health IT system is not natural, consistent, relevant, and supportive, the underlying flexibility of the workflow engine and process definitions can be used to mold workflow system behavior until it becomes natural, consistent, relevant, and supportive.

In the next blog post in the series, I’ll discuss workflow technology and patient safety.

Just in time for the 2016 AMIA Symposium, I’m delighted that Manish Sharma, the force behind @HCITExperts, is republishing my five-part series on workflow technology in healthcare. Thank you Manish!

If you pay any attention at all to interoperability discussion in healthcare and health IT, I’m sure you’ve heard of syntactic vs. semantic interoperability. Syntax and semantics are ideas from linguistics.


Syntax is the structure of a message. Semantics is its meaning. Think HL7’s pipes and hats (the characters “|” and “^” used as separators) vs. codes referring to drugs and lab results (the stuff between pipes and hats).

What you hardly every hear about is pragmatic interoperability, sometimes called workflow interoperability. We need not just syntactic and semantic interop, but pragmatic workflow interop too. In fact, interoperability based on workflow technology can strategically compensate for deficiencies in syntactic and semantic interoperability. By workflow technology, I mean Business Process Management (BPM).
 
Why do I highlight BPM’s relevance to health information interoperability? Take a look at this quote from Business Process Management: A Comprehensive Survey:
 

“WFM/BPM systems are often the “spider in the web” connecting different technologies. For example, the BPM system invokes applications to execute particular tasks, stores process-related information in a database, and integrates different legacy and web-based systems…. Business processes need to be executed in a partly uncontrollable environment where people and organizations may deviate and software components and communication infrastructures may malfunction. Therefore, the BPM system needs to be able to deal with failures and missing data.”

“Partly uncontrollable environment where people and organizations may deviate and software components and communication infrastructures may malfunction”? 

Sound familiar? That’s right. It should sound a lot like health IT.

What’s the solution? 
A “spider in the web” connecting different technologies… invoking applications to execute particular tasks, storing process-related information in a database, and integrates different legacy and web-based systems. Dealing with failures and missing data. Yes, healthcare needs a spider in the complicated web of complicate information systems that is today’s health information management infrastructure. Business process management is that spider in a technological web.

Let me show you now how BPM makes pragmatic interoperability possible.

I’ll start with another quote:

“Pragmatic interoperability (PI) is the compatibility between the intended versus the actual effect of message exchange.”

That’s a surprisingly simple definition for what you may have feared would be a tediously arcane topic. Pragmatic interoperability is simply whether the message you send achieves the goal you intended. That’s why it’s “pragmatic” interoperability. Linguistics pragmatics is the study of how we use language to achieve goals.

“Pragmatic interoperability is concerned with ensuring that the exchanged messages cause their intended effect. Often, the intended effect is achieved by sending and receiving multiple messages in specific order, defined in an interaction protocol.”

So, how does workflow technology tie into pragmatic interoperability? The key phrases linking workflow and pragmatics are “intended effect” and “specific order”.

A sequence of actions and messages — send a request to a specialist, track request status, ask about request status, receive result and do the right thing with it — that’s the “specific order” of conversation required to ensure the “intended effect” (the result). Interactions among EHR workflow systems, explicitly defined internal and cross-EHR workflows, hierarchies of automated and human handlers, and rules and schedules for escalation and expiration are necessary to achieve seamless coordination among EHR workflow systems. 

In other words, we need workflow management system technology to enable self-repairing conversations among EHR and other health IT systems. This is pragmatic interoperability. By the way, some early workflow systems were explicitly based on speech act theory, an area of pragmatics.

That’s my call to use workflow technology, especially Business Process Management, to help solve our healthcare information interoperability problems. Syntactic and semantic interoperability aren’t enough. Cool looking “marketectures” dissecting healthcare interoperability issues aren’t enough. Even APIs (Application Programming Interfaces) aren’t enough. Something has to combine all this stuff, in a scalable and flexible ways (by which I mean, not “hardcoded”) into usable workflows.

Which brings me to usability, tomorrow’s guest blog post topic.

Tune in!