Sentinel event leads to safety checklist

Quality project stresses input from frontline staff

A tragic sentinel event at a Michigan hospital prompted staff to develop a system to reduce the chances of it ever happening again. The result was an innovative quality improvement project that emphasized immediate results, input from the frontline staff, and ongoing refinement of the safety initiatives.

The process began in May 1999 at the Veterans Affairs Ann Arbor (MI) Healthcare System (VAAAHS), a medical center with four campuses and more than 26,000 patients. Like many sentinel events, the 1999 incident was the result of a combination of process problems and human error. A nurse at the medical center accidentally administered a bolus of regular insulin to a nondiabetic patient through an arterial line, causing brain death, explains Marcia Piotrowski, RN, MS, clinical risk manager in the office of the chief of staff. The staff’s root-cause analysis determined that the causative factors were improper mingling and storage of multidose vials on the top of medication carts in the intensive care units (ICUs), she says.

Devastated by the loss of life, the VAAAHS staff were eager to develop a quality improvement process that would address the problem. After some preliminary discussions with staff members involved with the root-cause analysis, Daniel B. Hinshaw, MD, VAAAHS chief of staff, suggested a safety checklist process that could improve the storage of medication and also proactively examine multiple safety elements in the ICUs. Hinshaw is now professor of surgery at the University of Michigan Medical School in Ann Arbor and a staff surgeon at VAAAHS.

The idea was that nurses in the ICUs could monitor compliance with safety standards established in the process improvement project, establishing a culture of safety that extends directly to the patient’s bedside, Hinshaw says.

To get the project going, both Hinshaw and Piotrowski met with key staff members, including unit-level nurse managers, ICU nurses, and management. The initial meetings focused on explaining the goals of the project and gaining the support of these key staff.

Hinshaw explained to the participants that he envisioned a safety checklist that could be used similar to the way a hotel manager might conduct a "white glove" inspection. "We wanted a system where people could step back from their daily concerns and look at the bigger picture, to try to spot things that they might block out during their work routine," he says.

Input from ICU nurses would be crucial

Actually writing the safety standards proved to be difficult, Piotrowski says. The team knew from the start that the standards needed to be brief and easily understood. Consistent application of the standards was a key goal. And because of the direct relation to patient safety, the team members felt that they must implement the improvements quickly. For that reason, they decided to forgo the typical period of baseline data collection and move straight to developing a draft of the safety checklist.

"We realized that this project was not a research study," Piotrowski says. "It was a quality management tool for enhancing safety, and we didn’t have the time to collect data in a leisurely way."

Everyone involved with the project agreed that the new safety standard must have input from the ICU nurses who use it. But how should they involve the nurses? Trying to work around their 24-hour schedules was too difficult and would take too long, even if the team wanted only a couple of ICU nurses to participate. So Piotrowski decided on another tactic. She camped out in an ICU break room for several hours one day, chatting with nurses as they stopped by on a coffee break. Over a few hours, she was able to gain useful input from several nurses.

Two days later, the process improvement team was ready with the first draft of the safety checklist. Covering a wide array of quality measures, from control of medication carts to use of restraints, the new safety checklist was intended to address more than just the medication error that led to the sentinel event. The checklist was arranged by topic, outlining a number of standards to be met, such as "medication carts locked" and "IV pump labeled with infusion name." Two ICU nurses used the checklist for a couple of days, keeping track of how well the standards on the checklist were met.

That information was used to revise the checklist, and then it was tested on the units again. Six versions of the form were created over the next five weeks. One of the key improvements, prompted by nurse feedback, was to include respiratory therapists and unit maintenance personnel in the safety checklist.

Realizing that the safety checklist would not be useful without compliance monitoring, the team established a system in which ICU nurses periodically check for compliance with all the standards on the list. The nurses initially examined 26 safety points, divided into three separate daily data collection forms.1 They would check off compliance with each standard for each patient room. Checking for compliance was divided among three nurses each day, checking each other’s patient rooms for compliance. Each one had to spend about 10 minutes checking and filling out the forms. That proved difficult on busy days, but the nurses almost always filled out the forms, Piotrowski says.

Most of the standards had a 90% compliance rate over the first year, she says. At that point, the nurses suggested revising the safety checklist to concentrate on fewer elements and reduce the time burden.

Four items were selected for daily measurement: intravenous pumps labeled with infusion name and solution in IV bag; bioclusive dressings dated and timed on arterial, central, and peripheral IV insertion sites, and each dressing changed within the past 72 hours; gauze dressings, dated and timed on arterial, central, and peripheral IV insertion sites, and each dressing changed within the past 24 hours; and an identification band on each patient’s wrist.

Those items would be checked daily for every patient, but another five points were labeled "intermittent" and checked only if they applied to a particular patient. They dealt with isolation and restraint protocols. The rest of the items on the original checklist were divided into four groups, and the ICU nurses checked a different group each quarter. This change significantly reduced the daily time burden on the nurses, Piotrowski says.

Program streamlined after two years

After about two years of success with the program, the system was refined further. Instead of daily forms, the nurses now use a weekly data collection tool posted in each patient room. At each shift change, twice daily, the incoming and outgoing nurses jointly document compliance with each standard.

This change has improved quality of care by reducing the chance of "confirmation bias" in which one nurse "sees what she expects to see," Piotrowski says. With two nurses checking at the same time, at least one is likely to spot any deviation from the standard. The procedure also reduces the time burden on the nurses because they can incorporate the checklist into their change of shift reports.

In addition to the standards that are to be checked for all patients, the intermittent standards are rotated onto the weekly checklists periodically to keep the nurses’ attention and to facilitate data collection.

Data management proved to be a challenge in the project, mostly because the team wanted it done quickly so patient safety could be improved without waiting for analysis. VAAAHS hired students from a nearby university to set up spreadsheets and transfer data from the forms. The results were compiled into bar charts that are posted in the ICUs to show compliance rates.

While the rapid evolution of the safety checklist kept it useful and encouraged nurses to use it, the many changes made data analysis difficult, Piotrowski says. The best measure of the program’s effectiveness can be found in patient safety outcomes, she says.

Piotrowski and Hinshaw decided that the project’s success could best be measured through a series of questions addressing safety concerns. 

Since the program was begun in 1999, no adverse events have been traced to improper storage of medications on top of carts, but there have been two near-misses that led to improvements in procedure. The biggest improvement, Piotrowski says, is the way nurses and other staff now are actively involved in improving processes and ensuring patient safety. Even when compliance with a particular safety standard falters, posting the rates in the ICU results in more attention from nurses, she says.

"Major adverse events with morbidity happen so seldom that it’s not a good benchmark of whether you’re successful," she says.

"That’s not to say that errors don’t happen in the ICU and we couldn’t have another sentinel event, but their investment in looking at their environment makes us think that the likelihood is less on those ICU units than in other places in the institution," Piotrowski says.

Program may be expanded to other areas

The success of the program has prompted Hinshaw and Piotrowski to look at expanding the quality improvement project to other departments. 

Likely candidates include inpatient psychiatry and the emergency department. Both are good candidates, Hinshaw says, because the clinicians use a team approach and are with patients for long periods. Physicians do not participate much in the ICU safety project because they tend to come and go quickly, unlike the nurses and other staff who are on the unit for long stretches.

"Besides, working with physicians is like herding cats," he says. "We thought it would be too ambitious to include them in the initial implementation of the project. We’re hoping to see more physician involvement when we take the idea to other areas."

Physician involvement may bring up other problems, however. The safety checklists are completed by peers in the ICU, and Hinshaw suspects "it will create war" in the emergency department if physicians use the checklists to critique a nurse’s work.

"Those are the kinds [of things] we’ll have to consider every time we take it to another department," he says. "It is absolutely critical that the system be owned and accepted by the people who will use it."

[For more information, contact:

  • Marcia Piotrowski, RN, MS, Clinical Risk Manager, Office of the Chief of Staff, VA Ann Arbor Healthcare System, 2215 Fuller Road, Ann Arbor, MI 48105. Telephone: (734) 769-7100. E-mail: Marcia.
  • Daniel B. Hinshaw, Staff Physician, Department of Surgery, VA Ann Arbor Healthcare System, 2215 Fuller Road, Ann Arbor, MI 48105. Telephone: (734) 769-7100.]


1. Piotrowski M, Hinshaw D. The Safety Checklist Program: Creating a Culture of Safety in Intensive Care Units. Journal on Quality Improvement 2002; 28:306-315.