Jerome Carter, MD  November 23, 2015

Part Two: Workflow Disruptions and the Interplay between Data and Processes
Workarounds and workflow disruptions are inefficiencies that at some point manifest as lower productivity, safety issues, care quality concerns, and where software or equipment is concerned, usability problems (1, 2, 3).   Getting to the root cause often requires a detailed workflow analysis.  However, initiating an analysis has to begin with a feeling for what has gone wrong i.e., symptoms of improper workflows.  Below are a few examples.

Diagnostic errors/safety issues/results management
Failure to manage test results appropriately can create serious safety problems.   Lack of timely follow-up can result in diagnostic misses or failures to make adjustments in treatment plans.   From a process perspective, managing results can be thought of as a means of assuring that ordering loops are completely closed.  For example, every test ordered should be tracked to assure its completion or cancellation, and the results of every completed test should be reviewed and either declared acceptable or flagged for follow-up.  Any test flagged for follow-up should then remain active until all follow-up actions are completed. Results management is a process, and the workflow for that process can be modeled.  Results management seems to be a nearly perfect application for workflow technology.

Missing Information at the point of care/decision support
The goal of decision support is influencing behavior by providing information while decisions are being made.   For clinical decision support, that information could be a range of types —evidence, patient information, clinical rules, research protocols—to name several.   The key is having that information available in a consumable form at the exact point in the process where it is  needed.  Detailed workflow models are needed to line up information needs with process steps.  Using workflow models, complex decision paths, whether with many branches or simple yes/no choices,  can be easily modeled.  Even better, these models can be executed using workflow technology (4, 5). 

Usability problems/software design
Usability issues come in many forms — hard-to-read screens, too many clicks, missing information, or having to access multiple systems to get all needed information (3).  For software users, usability problems usually result in workflow disruptions.  In process terms, usability issues can add steps.   Keeping a paper list because finding the same info using a computer takes too much time adds extra process steps.  Finding that same paper list when it is occasionally misplaced adds even more steps.   The interactions between user and software constitute a process making it possible to analyze usability issues using workflow analysis.  

Software design can also benefit from workflow analysis.  Use cases can be converted to workflows for detailed analysis, which, in turn, can be used to develop software requirements.  Using workflow technology to create clinical applications can make meeting functional requirements easier because workflow models can be executed directly.

Productivity losses/inefficient processes
Information technology is not the only cause of inefficiency in healthcare.  Paper-based organizations can be just as inefficient.  Incomplete or lost paper forms cause numerous problems.  Poorly-designed or improperly-implemented HIT can make paper inefficiencies worse.  Practice optimization is about making all processes as efficient and productive as possible.   Doing so requires a detailed understanding of how those processes work.

Process derangements, even small ones, may have significant downstream effects.  When workarounds and disruptions appear the underlying cause should be addressed expediently.   

Once the need for a workflow analysis becomes obvious, the next step is selecting the proper type of analysis to perform.  A brief overview of the various workflow analysis types appears below.

Types of workflow analyses **
Performing a high-quality workflow analysis begins with understanding the specific type needed.    Once the required type has been  determined, the intended use of the analysis determines both the type of model created and the level of detail required. For example, the level of workflow detail required to implement a new appointment process is different from that required to design a user interface for a prescription writer.  Here are six types. 

Searching for the cause of an unexpected or unsatisfactory process outcome is an example of a diagnostic workflow analysis.   For example, if key data are consistently wrong or missing, a diagnostic analysis would be used to determine which process steps were at fault.   Care process issues such as incorrect dosing errors or failure to follow-up abnormal results would also required a diagnostic analysis.   

As the name implies, documentary analyses are done to capture the current state of a process.  These analyses are often done prior to software implementation as the start of reengineering efforts.  They may also be a part of a policy/procedure update.   Gathering requirements for software design is another good reason to perform a documentary analysis.  

Prospective workflow analyses are actually plans for changing current processes or plans for implementing processes that do not currently exist.  They are blueprints for something yet to be built.  Obviously, some guess work is involved.   Modeking for a new workflow application requires a prospective analysis.  Since the anticipated workflows do not exist, prospective analyses may require frequent revisions to the model as assumptions change.  

When implementing new processes or significantly changing current ones, workflow analyses may be done to assure that recent changes are not causing unexpected/unsatisfactory process outcomes.   Adaptive workflow analyses are a way of monitoring the effects of process changes as they occur, rather than waiting for a symptom to appear.  When workflow technology is used to implement a process, monitoring can be accomplished via system information (logs) along with output assessments.  For example, an automated guideline system might have logs that show the number of times a path is taken based on the presence of specific patient data values. 

Software selection can be improved through the use of test scripts. Test scripts walk users through a simulated workflow as a means of determining how well the software supports a specific process.  Selective workflow analyses look at current workflows then tries to ascertain how they might change when software is used. As such, they are a type of prospective workflow analysis.   A test script is the main outcome of a selective analysis. 

While of the above workflow types may be used for clinical care, the “outcome” designation is reserved for those that target specific clinical outcomes. For example, a workflow designed to assure that all patients with a given diagnosis will receive one or more interventions is an outcomes workflow.   Designing a process to assure that HbA1c and eye exams are performed for patients newly-diagnosed with diabetes is an example of an outcomes-focused workflow analysis.  

Why the analysis type is important
All processes consists of steps. Steps produce and consume information, access resources, and may be performed by people or machines.   Performing an analysis may require assessment of one or all of these process aspects.  One can save time and headaches by knowing the type of analysis required, which helps one choose a proper starting point.   

If software selection is the goal, then software functionality that supports user processes must be assessed along with current user processes that will be impacted by the software. In addition, a test script that emulates normal user processes is a required deliverable.   If a new clinical service is planned (e.g., a disease-specific clinic), then a prospective analysis is in order.  Since the physical space may not be known, the workflow analysis may draw on site visits, clinical protocols, patient traits, and planned research.   In such situations, it would be foolish to attempt to drill down to specific steps.  A big picture approach that allows for easy revision is best.    If no one can agree on when the “green” form should be used, perhaps a documentary analysis, policy/procedure update, and an in-service presentation are the way to go. 

Once the symptoms have been identified and the type of analysis required is known, then the actual analysis can begin. 

Next week: Knowing what to analyze and why.

*  This classification of workflow types is drawn from my experiences of the last 15 years.   I was not able to locate any documents/articles that classify workflow analyses by high-level goals.   AHRQ does provide a taxonomy for CDS workflows, but it is care delivery-centric.

  1. Flanagan ME, Saleem JJ, Millitello LG, Russ AL, Doebbeling BN. Paper- and computer-based workarounds to electronic health record use at three benchmark institutions. J Am Med Inform Assoc. 2013 Jun;20(e1):e59-66. 
  2. Friedman A, Crosson JC, Howard J, Clark EC, Pellerano M, Karsh BT, Crabtree B, Jaén CR, Cohen DJ. A typology of electronic health record workarounds in small-to-medium size primary care practices. J Am Med Inform Assoc. 2014 Feb;21(e1):e78-83 
  3. Zahabi M, Kaber DB, Swangnetr M. Usability and Safety in Electronic Medical Records Interface Design: A Review of Recent Literature and Guideline Formulation. Hum Factors. 2015 Aug;57(5):805-34.
  4.  Huser V, Rasmussen LV, Oberg R, Starren JB. Implementation of workflow engine  technology to deliver basic clinical decision support functionality. BMC Med Res  Methodol. 2011 Apr 10;11:43.
  5. Lee J, Kim J, Cho I, Kim Y. Integration of workflow and rule engines for clinical decision support services. Stud Health Technol Inform. 2010;160(Pt 2):811-5