Don’t Underestimate the Power of the Pump
In April 2013, The Joint Commission published a Sentinel Event Alert describing a patient death that occurred secondary to a delayed response to a device alarm. This same article describes two of the most common contributing factors reported are alarm fatigue and alarm settings that are not customized to the individual patient or patient population. Recommendations made by both the Association for the Advancement of Medical Instrumentation (AAMI) and ECRI Institute included the need to “establish guidelines for tailoring alarm settings and limits for individual patients. The guidelines should address situations when limits can be modified to minimize alarm signals and the extent to which alarms can be modified to minimize alarm signals.” Additional recommendations included to define a process for continual improvement and constant optimizing of alarm systems policies and configurations. Intermountain Healthcare is a nonprofit healthcare system of 22 hospitals with a Medical Group of more than 185 physician clinics covering patients in Utah and southeastern Idaho. In July 2013, the quality and safety teams of Intermountain Healthcare followed the recommendations from the AAMI and ECRI organizations and convened a task force to evaluate the many devices in use within our facilities and analyze the alerts and alarms that correspond to those devices. One of these devices, the infusion pump, has allowed for a very accurate analysis of infusion related alerts and alarms that occur. This task force used the infusion pumps as a standard of measure to show that the quality and number of alerts that occur from the infusion pumps could be evaluated for effectiveness and adjusted to be more patient or workflow specific. The goal is to make each alert that occurs from an infusion pump an “actionable” alert.

Intermountain Healthcare currently uses the Sigma Spectrum Infusion System for all large volume infusions throughout the system and within 20 Medical Group clinics. A total of 3,630 infusion pumps were used to program 986,131 infusions during 2013. However, included within those infusions were 113,202 infusion pump alerts consisting of 56,825 hard limit alerts and 56,377 soft limit alerts. An alert reduction initiative plan was implemented to analyze these alerts and make the alerts that occurred from the infusion pump an actionable alert for the nurse instead of a false alert leading to increased alert fatigue.

Infusion Pump Safety Score Developed
In August 2012, the basic mode use across the system was 8.1%. Over the course of 2013 there were 117 new library additions or adjustments made to the drug library and basic mode use declined to a combined low of 1.8% among the 22 different hospital facilities representing 98.2% drug library compliance. This level of drug library compliance was a significant achievement however it only tells one part of the infusion pump story. Several of the Intermountain facilities achieved basic mode use as low 0.35% however, there were hard or soft limit alerts occurring nearly 40% of the time an infusion was programmed in the pump. To account for this and provide a visual alert for where alerts needed to be looked at first an infusion pump safety score was developed. This score was reported for each of the 22 hospital facilities and combined to offer a single system-wide score for repeat analysis. The score was originally developed with a minimum system goal of achieving 5% basic mode use. Hard and soft alert goals of 5% each were also arbitrarily assigned. A formula was developed to calculate a score out of a maximum of 5 points each for basic mode use and hard and soft limit alerts for a total of 15 points. The hard and soft limit alerts were compared as a percentage of total infusions for each individual facility. If a facility had no basic mode use they would get a score of 5 points.

The higher the percent basic mode, the lower the score and once a basic mode of 10% was passed the score became negative. The same formula was used for hard and soft limits. Meeting the minimum goals of 5% in each area would give a score of 2.5 points for each category so the system infusion pump safety score was set at a minimum of 7.5 points out of a total of 15. Facilities that had a safety score below 7.5 would be evaluated in more detail, whereas facilities with a score of greater than 7.5 would be evaluated in more depth at a later time. Figure 1 below gives an overview of one hospital within Intermountain Healthcare and how the safety score progressed over the course of 3 months.

Figure 1: In June the basic mode use was 2.01% which was well below the system goal of 5% and calculated a score of 3.99 out of a possible 5 points. Soft limit alerts represented 9.28% which gave 0.36 out of 5 possible points. Hard limit alerts were 7.5% which gave a score of 1.25 out of a possible 5 points. The total infusion pump safety score is the sum of 3.99, 0.36, and 1.25 for a total score of 5.6. This is below the minimum goal of 7.5 and required an in-depth review of this facility and their infusion pump use.

Alert Analysis Example Each month the total number of alerts is evaluated and the top ten alerting medications are reported in detail. These top ten alert associated infusions account for over 40% of the total alerts reported at Intermountain Healthcare. In the month of October 2013, magnesium infusions were the number one alerting infusions accounting for 762 alerts or 8% of the total alerts that were triggered during that month. Magnesium was programmed in the infusion pump library with a variable concentration, a lower soft limit infusion rate of 0.5 grams per hour and an upper hard rate of 6 grams/hour. There were no concentration associated limits for magnesium. An indepth analysis of magnesium infusions uncovered several different problems associated with the

infusions. First, many infusions were being entered in the pump with erroneous concentrations. Concentrations ranged from 0.02 grams/mL all the way up to 500 grams/mL. The concentration of straight magnesium is 0.5 grams/mL. Second, the infusion rates being entered ranged from 0.2 grams/hour all the way up to 95,000 grams/hr. The majority of the infusion rates were programmed for 1000 grams/hr. The total number of magnesium infusions showed an increase over the reported time frame. A close analysis of these alerts revealed that four hospital facilities had magnesium entered in their pharmacy database as milligrams as opposed to grams. This led to complications with both programming the correct concentration and the infusion rate into the pump. For example, since the pump was set up to reflect magnesium infusions and concentration in grams a facility reporting magnesium in milligrams would display on the pharmacy label an infusion as 2000 milligrams in 100 mLs of diluent. This would actually be programmed into the pump as 2000 grams/100 mL for a final concentration of 20 grams/mL. Administering this infusion over a four hour time frame would give an infusion rate of 500 grams/hour. Since there was not a concentration limit on the pump the concentration would be accepted and the infusion rate of 500 grams/hr would trigger a false alarm. This was remedied by making changes to the pharmacy database to report magnesium in grams and assigning an upper hard concentration limit of 0.5 grams/mL to the magnesium entry of the infusion pump. A recommended infusion rate of 0.5 grams/hr was also added as a default starting rate to the infusion pump. Individual departments within the top alerting facilities were also targeted with both nurse and pharmacist education being provided on appropriate magnesium infusion rates as well as an update on the system-wide effort to reduce the number of infusion pump alerts. A library update was performed in January 2014 to activate these changes.

Although a wireless update of the infusion pump takes about 10 days to reach each infusion pump throughout the system, both soft and hard limit alerts immediately experienced a significant decline in the month of January with continued declines in February and March. Hard limits alerts experienced an exponential decline dropping from 492 alerts in October to just 40 for March 2014. Figure 2 below demonstrates visually the results, specific to magnesium, of our alert reducing initiatives.

Figure 2: The top alerting infusion in October 2013 was magnesium. Hard limit alerts representing erroneous infusion

rates and concentrations accounted for the majority of the alerts. Education efforts and changes to the infusion pump library in January provided a significant reduction in overall magnesium infusion pump related alerts.

Since the infusion pump safety score has been implemented at Intermountain Healthcare the score has seen an improvement from 6.41 to 7.57 combined for all 22 hospital facilities. Individual facilities with scores less than 7.5 are able to be reviewed in detail to evaluate individual medications that may need to have infusion pump limits reviewed and updated as well as provide facility and department specific education regarding alert causing infusions. Similar to evaluating the limits and workflows around magnesium, these limits can be refined to decrease the amount of alerts that come from the infusion pump and similar efforts can be done to evaluate the alerts and alarms that come from other devices used within the hospital environment. Evaluating and making these changes will decrease the amount of false positive alerts that are seen and make each alert an actionable item.