American Association for Clinical Chemistry
Better health through laboratory medicine
July 2011 Clinical Laboratory News: Designing Processes for Patient Safety

Designing Processes for Patient Safety 
The Key is Finding the Vulnerabilities and Fixing Them

Richard Gitomer, MD, MBA, is chief quality officer for Emory University Hospital Midtown and is a general internist by training. He has led numerous improvement projects related to access, efficiency, and innovation in the ambulatory setting and care of patients with chronic conditions. Dr. Gitomer is a recognized leader in process improvement through his collaborations with the Institute for Healthcare Improvement and the Association of American Medical Colleges.

This interview was conducted by Corinne Fantz, PhD.

What are the three most important elements of safe process design? 
The first is accounting for human factors. It is important to recognize that processes which rely on perfect human performance are incapable of high reliability. The second is standardization with limited discretionary variability (See “Examples,” below). Furthermore, reasons for appropriate variation should be captured and fed into the process design. The third is organizational culture. It is essential to have a positive reporting culture that exposes misses and near misses, because no process or procedure is failure-proof. Successful reporting cultures also identify key vulnerabilities.

Lab Examples of Standardization with Limited Variability

  • Standardized procedures across all sites
  • Standardized phlebotomy trays
  • One type of point-of-care glucometer in the health system
  • Automation that limits the care providers choice of tube types
  • Selection of a primary referral laboratory
  • Use of templates for interpretative reports

How do attitudes about safety in the laboratory affect process design?
All processes that rely on vigilance and hard work are incapable of performing at a high level of reliability. If leadership and/or the front line worker do not understand the limitations of human performance, then high levels of reliability will not be possible. Also, a positive reporting culture is essential. Punishment for human error will impede process improvement and make the organization less safe and less reliable.

Clinical laboratories routinely monitor key processes. Is it not enough to know what errors occur and why they occur?
Measurement alone does not result in improvement, but it is essential for identifying areas that need improvement and for monitoring improvement. There is a saying, “You don’t fatten the chicken by weighing it.” (See Figure, below).

Prothrombin Time
90% Turn Around Time

Measurement is important to quality improvement, but by itself will not significantly improve quality.

What opportunities and challenges do new technology bring to old processes?
New technologies increase the capability of systems. For example, autoverification in the lab is faster and more reproducible than manual verification of results. But technology comes with some inherent risks. For those technologies that still require human intervention, the interface between humans and the technology remains the most vulnerable point. These vulnerabilities include incorrect human operation, due either to human error or errors that result from interacting with the new technology. All systems will fail at some point. With older technology, we know those failure modes and have designed processes to limit their impact. We do not have the same understanding of new technologies. So, until those vulnerabilities are fully understood, there will be safety and reliability shortfalls.

Communication errors are a major contributor of preventable harm to patients. How can laboratory staff make change-of-shift hand-offs safer?
Handoffs, like all processes, benefit from standardization, which increases the likelihood that key information will not be missed. Standardization also creates an opportunity to measure and improve the process. The standard process should include a template, or checklist (See “Checklist,” below), which highlights key information that should be exchanged.

While every laboratory has a process for communicating critical values, could you elaborate on a few design elements that make one process safer than another?
Communicating critical lab values is a process that is heavily reliant on humans with all their inherent vulnerabilities. In addition, creating a first-level communication process that includes all the eventualities and exceptions necessary to respond to every situation would result in a process so complex that the staff would be unable to execute it in a reliable fashion. Using a reliable approach to design as described by Roger Resar, performance levels in the high 90 percentages can be achieved with a series of simple processes (See “Designing a Process,” below). 

The first step is to create a standardized process that effectively communicates at least 80% of the critical lab values on the first attempt. For all the failures, there is a second process that ensures appropriate communication. This second process usually is more resource-intensive, but it is able to handle the complexities not addressed in the first process. The benefit of this strategy is more effective use of resources. The risk is that failure of the initial standard process to function at an 80% level of reliability or better can overwhelm the redundant process. By design, the redundant process is more resource-intensive but has limited capacity.

Designing a Process for Handling Critical Values

Step 1

  • Standardize critical value list throughout the system
  • Employ templates for communication and documentation

Step 2

If critical value communication is unsuccessful after the first attempt:

  • Pass calls from busy technologist to call center with access to many other data feeds that help overcome problems, such as a patient being registered to wrong the physician.
  • Use a clear escalation plan, such as licensed practitioner, ordering provider, on-call physician, chief-of-service.
  • Call patient at home if an outpatient.

By design, the redundant process is more resource-intensive but has limited capacity.

As a physician consumer of laboratory services, could you provide an example of a laboratory process that directly impacts the safety of your patients?
I am a general internist who sees ambulatory patients. Reliable communication of critical lab results directly impacts the safety of my patients. By definition, these patients are mostly healthy, so the incidence of critical lab values is very low. Being an infrequent event makes the likelihood that any process designed to remedy the notification barriers has a greater likelihood of failing. Therefore, in my case, because receiving prompt and accurate notification of critical lab results is a rare event, it has a higher failure rate than a process that occurs regularly.

Positive patient identification is essential to ensuring patient safety. Do errors in patient identification always mean the process is bad?
Process failure is expected for all processes. Since identifying patients is a process carried out by humans, it will fail at some measurable rate. Based on the specifications of the clinical team, these error rates could be well below the needs of the users of the information and therefore considered a good process. But if the error rate is such that it significantly impedes the care of the patient, the process does not meet specifications and must be improved.

Would you provide examples of a few good questions to ask when assessing how well a process is working?
The first question to ask is: does the process meet the needs of the customer of the process? If it is not meeting the specifications of the customer, then there are three possibilities to assess. First, if there is no standard process to do the task, then the standard process must be developed. The second possibility is that there is a standard process, but it is not being followed. In this case, it is important to understand why the process is not being followed and redesign the process if necessary. Monitors have to be put in place to detect the decline in performance and allow for the appropriate response. Last, if the standard process is being followed, and customer need is not met, then the standard process must be improved.

How can laboratories design processes that integrate clinical teams and patients? 
Laboratories are frequently accused of optimizing their processes at the expense of the patient and the clinical team. Part of the problem is that we tend to measure and assess each section independently rather than the entire care process that results in healing the patient. There may be times, however, when efficiency in the lab may need to be subordinate to the efficiency of the entire process of care.


Grimm E. Shift-to-Shift communication: what can labs learn from NASA and other highly reliable organizations? Clinical Laboratory News 2011 (January).

Resar RK.  Making non-catastrophic health care processes more reliable:  learning to walk before running in creating high-reliability organizations. Health Serv Res 2006; 41:1677–1689.

Patient Safety Focus Editorial Board

Michael Astion, MD, PhD
Seattle Children's Hospital
Seattle, Washington

Peggy A. Ahlin, BS, MT(ASCP)
Salt Lake City, Utah 

Corinne Fantz, PhD
Emory University
Atlanta, Georgia

James S. Hernandez, MD, MS
 Mayo Clinic Arizona
Scottsdale and Phoenix

Brian R. Jackson
ARUP Laboratories
Salt Lake City, Utah

Sponsored by ARUP Laboratories, Inc.