Surviving Sepsis Campaign

ENHANCING RELIABILITY

Enhancing Reliability

How will you know that your changes as intended are being faithfully implemented on the wards? Where are the bottlenecks? What steps can be taken to make the new clinical processes you have implemented function more reliably?

The Model for Improvement will be used in concert with Enhancing Reliability, an active approach to clinical process improvement that embeds steps to prevent, identify, and mitigate failures directly into the process itself. Reliability is a scientific method that evaluates, calculates, and improves the overall reliability of a complex system. It is a goal of consistently producing appropriate outcomes and preventing adverse events. The Enhancing Reliability methodology has three levels, as described below.

1. Stabilization (Prevention):

It is impossible to sustain improvement of a chaotic process. In general, any process that fails 1 time in 10 (functions less than 90 percent of the time as intended) is chaotic and unreliable. Efforts to stabilize the system prevent failures.

The first step in stabilizing a process is the establishment of a standardized approach. Therefore, a standardized protocol customized to your institution will be necessary to implement the sepsis bundle.

Next, measure the baseline reliability of the clinical processes you have created in your protocol. These are process measures. Each process being measured can be improved individually.

In the event that standardization is already in place but a particular process is functioning at less than 90 percent reliability, more work is necessary to stabilize that process. Some examples include building decision aids and reminders into the system and making the desired action the default strategy rather than an option.

2. Redundancy (Identification):

Once a process functions reliably 90 percent of the time or better, it is reasonable to pursue improvement in the next PDSA cycle toward the next threshold of reliability - 1 failure per every 100 opportunities, or 99 percent reliability. Redundancy of procedures helps to achieve this goal by identifying more instances when the process should be applied.

No effort to pursue this level of improvement in the course of a PDSA cycle should be undertaken for an unstable process. This is so because redundant efforts are, by definition, resource intensive. Implementing a redundant procedure in an unstable system is wasteful.

Effective redundancy steps function independently of the normal mechanism that triggers use of the clinical process. For example, the laboratory may have a procedure to contact physicians directly for routine labs that suggest acidosis, such as very low bicarbonate levels, which may indicate sepsis. A successful redundancy step may bring the number of missed cases below 10 percent and advance the process toward a 1 percent failure rate.

3. Failure Modes Analysis (Mitigation):

Once sufficient recovery of cases can be established to approach better than 90 percent reliability, further steps involve identifying the causes of failure and mitigating their effect.

Specific impediments to the functioning of reliable processes will be identified in the course of implementation. These represent failure modes that need to be eliminated or in some other way circumvented in subsequent PDSA cycles.

The particular impediment detected will drive your team to create a new level of customization to prevent further failures and to achieve increasingly reliable processes.

Example of Enhancing Reliability Using the Plan-Do-Study-Act (PDSA) Cycle

Example: Use an insulin drip protocol to achieve tight glucose control.* [*Note: This PDSA cycle is a continuation of the first example in the Testing Changes section.]

After several PDSA cycles, a standardized protocol is ready for use across the hospital:

Plan

Implement the insulin drip protocol developed by staff to control the glucose in a hyperglycemic septic patient keeping it between 60 and 150 mg/dl as measured at the bedside.

Measure the reliability of the use of the protocol by staff. The reliability of the use in this case would be the total number of instances of use, divided by the total number of indicated cases.

Study

The reliability of use was measured to be 80 percent on the step-down unit.
The reliability of use was measured to be 92 percent in the ICU.

Act

In the step-down unit, stabilization steps were needed. Reminders to use the protocol were posted and physician staff were prompted by nurses to consider using the protocol each time insulin was ordered.
In the ICU, redundancy steps were now appropriate because the process was stable. Any time a sliding scale insulin order was written and faxed to pharmacy, the pharmacist checked the laboratory system for glucose values above 200. If present, a phone call was made to the admitting physician to inquire about the use of the glucose control protocol.
In the ICU, it was also identified that many patients were still on oral hypoglycemic medications and that physicians were not starting the glucose control protocol in those cases. This failure mode was mitigated by automatically discontinuing oral hypoglycemic medications on admission to the ICU.

Cycle 1

Reliability on the step-down unit was subsequently measured to be 90% of indicated cases. The redundancy step used in the ICU, independent pharmacy review of labs and physician notification, was now applied in the step-down unit.
Reliability in the ICU was subsequently measured to be 97% of indicated cases. It was identified that the moonlighting physicians were not using the protocol for admissions during the overnight shift. This failure mode was corrected with appropriate in-servicing.

Cycle 2

Reliability on the step-down unit was subsequently measured to be 95% of indicated cases. A failure mode was identified in the protocol itself because it did not allow for transitioning to home regimens necessary for transfer from step-down to the wards. The protocol was modified to allow for this change.
Reliability in the ICU reached 99%.

Cycle 3

Measurement of the reliability in the step-down unit now showed 99% adherence.