Safe practices in pharmacy sterile compounding areas
From the June 2, 2011 issue
Problem: Errors during pharmacy preparation of parenteral products and admixtures may happen more often than you think. A five-hospital observational study on the accuracy of preparing small- and large-volume injectables, chemotherapy solutions, and parenteral nutrition showed a mean error rate of 9%, meaning almost 1 in 10 products was prepared incorrectly and then dispensed.1 Error rates for complex solutions such as parenteral nutrition were especially high—37% for manual preparation and 22% for preparations that were partly automated. More recently, a 2009 State of Pharmacy Compounding Survey showed that 30% of hospitals have experienced a patient event involving a compounding error in the past 5 years.2 While the causes of these errors are numerous and varied, two examples of pharmacy preparation errors follow, illustrating common problems that have been reported to ISMP.
IV admixture error
A premature infant admitted to a neonatal intensive care unit (NICU) was placed on a ventilator. Along with parenteral nutrition, arterial fluids were started with sodium acetate 80 mEq/L and heparin 1 unit/mL to infuse at 0.7 mL/hour to maintain use of an inline bedside blood gas and electrolyte monitoring device. When the infant’s sodium level began rising, the arterial fluids were changed to sodium chloride 40 mEq/L and heparin 1 unit/mL.
Technical problems occurred with the inline blood gas and electrolyte monitor. The monitoring device was replaced, but the readings showed an elevated sodium level that staff attributed to failure of the replacement device. The monitoring device was replaced a third time, but by then more than 12 hours had elapsed since the arterial fluids had been changed. At this point the inline monitor displayed a sodium level of more than 200 mEq/L. A confirmatory serum sodium level was ordered, and by the time the blood was collected and tested, the level was at 270 mEq/L. All parenteral fluids were taken down, sent for analysis, and replaced with new solutions. The infant’s condition deteriorated, and despite aggressive treatment, the baby died.
The following day, analysis of arterial fluids led to the discovery of an error that occurred when the solution was compounded. Several weeks before the event, the pharmacy had switched from glass bottles of sterile water to bags of sterile water but had not communicated this change to all staff, including the technician who was in the IV room on the day of the event. The technician pulled what he thought was a 250 mL bottle of sterile water, but it was actually a 250 mL bottle of 23.4% sodium chloride,** which he then used as the base solution.
Concentrated solutions of high-alert medications used for parenteral compounding—including bulk containers of 23.4% sodium chloride—were stored in the anteroom between the central pharmacy and IV clean room. These products were on shelves along with other solutions—including bulk containers of sterile water for injection. Of note, on the day of the event, the labels of the glass bottles of 23.4% sodium chloride were facing away from the technician in the IV room. The 250 mL glass bottles of 23.4% sodium chloride and the previously used 250 mL glass bottles of sterile water for injection both had aluminum caps. So they looked identical when viewed from the unlabeled sides of the bottles. The technician, who was accustomed to com-pounding arterial fluids in glass bottles, failed to notice that he had selected the wrong product.
The technician drew appropriate amounts of the heparin and sodium chloride additives into syringes to prepare the ordered solution and then asked the pharmacist to check the syringes. The pharmacist checked each additive but failed to identify the base solution. The additives were then injected into the glass bottle, a pharmacy label was placed over the manufacturer’s label, and the solution was dispensed to the patient care unit. Completely covering the manufacturer’s label with the pharmacy label did not allow nurses to detect the base solution error.
Small-volume injectable preparation error
A woman in labor who was receiving fentaNYL and bupivacaine epidural analgesia developed significant hypotension. A nurse removed a pharmacy-prepared syringe labeled ePHEDrine (5 mg/mL) from an anesthesia cart and administered 2 mL. During administration, the patient complained of a burning sensation at the IV site and said she also had neck pain. A few seconds later, the patient’s hands began to twitch, she developed respiratory distress, and she was unable to communicate with staff. A rapid-response team was summoned. An emergency C-section was planned, but the patient’s symptoms subsided by the time she reached the surgical suite.
The rapid-response team initially thought the patient was experiencing an allergic reaction to ePHEDrine. However, further investigation revealed that there had been a pharmacy labeling error. A nurse became suspicious when she noticed that the medication labeled as ePHEDrine was in a smaller syringe than pharmacy routinely used for that drug. Lab analysis of the remaining contents of the syringe revealed that it contained the neuromuscular blocking agent succinylcholine, not ePHEDrine.
Due to the short half-life of succinylcholine and the relatively small dose the patient received (40 mg), the patient was partially paralyzed for only a few minutes after the nurse injected the medication, which corresponded to the symptoms the patient exhibited. Symptoms completely resolved in about 10 minutes after administration of the drug. The baby was delivered without incident, and mother and baby experienced no permanent harm.
Concerns regarding stability had led the pharmacy staff to prepare three syringes each of succinylcholine (10 mL of 20 mg/mL) and ePHEDrine (10 mL of 5 mg/mL) every other day to stock in the labor and delivery anesthesia cart. The succinylcholine was drawn into a 10 mL syringe directly from the 10 mL vial. For ePHEDrine, normal saline was used to first dilute the 50 mg/mL, 1 mL ampul to a final concentration of 5 mg/mL. However, each 10 mL dose was prepared in a 20 mL syringe to help differentiate the two drugs.
Both drugs were often prepared and checked around the same time so they could be delivered together to the labor and delivery unit. In this case, the technician had both drugs under the laminar-airflow hood at the same time. The technician prepared both drugs correctly and placed the filled syringes into a divided bin, with each drug placed in its own compartment. The pharmacy labels were also placed in the bin, including “Paralyzing Agent” labels for the succinylcholine. The bin was handed to the pharmacist to check the products. The hospital where this error occurred determined that either the technician placed the syringe labels in the incorrect compartments of the bin, or the labels were in the correct compartments but the pharmacist placed them on the wrong syringes. The error occurred on a busy weekend when a scheduled pharmacist had called in sick.
Safe Practice Recommendations: While these two errors occurred under different sets of system issues and performance shaping factors, there are several key safety improvement strategies that should be employed in all settings to avoid these types of errors.
Limit admixtures. Hospital pharmacy preparation of injectables and parenteral solutions should be limited to products that are not available commercially. The study cited at the beginning of the article demonstrated a clear difference in pharmacy dispensing error rates with ready-to-use products (0.3%) and compounded admixtures (9%).1 Therefore, unnecessary preparation of products should be avoided.
For example, the succinylcholine syringes prepared by pharmacy in the latter error simply involved withdrawing the entire contents of the vial into a syringe and affixing a label. This had been instituted to help ensure availability of the drug during emergent intubations. However, stocking emergency kits with the 10 mL vial and a label and syringe may be all that is needed to ensure ready availability (and less waste). Some commonly used products that are prepared in batches also can be outsourced to a trusted IV compounding company (if allowed by pharmacy regulations) to reduce the workload on the parenteral admixture staff, who can then focus attention on individual patient-specific preparations.
Safe storage of supplies. Separate the storage of concentrated bulk solutions, particularly concentrated electrolytes, amino acids, and dextrose, from all other products that are directly dispensed to a patient care unit. Among admixture supplies, also sequester sterile water for injection to reduce the risk of inadvertent use. When possible, employ single-use vials of concentrated electrolytes for manual IV additives so that bulk bottles cannot be inadvertently used as base solutions or introduced back into the production line. In fact, it would be safest to use a vial size closest to the dose required to prepare the final product for any drug used in sterile compounding.
Pharmacist product selection. Whenever possible, have a pharmacist pull the necessary ingredients prior to preparation of the product and place them in a bin to give to technicians. Technicians should verify the ingredients and, if they don’t match what is expected, ask questions before preparation.
Maximize automation. Whenever possible, use automated compounding devices to prepare complex solutions; barcode verification of base solutions and additives; quality control systems that use computerized tracking of parenteral preparation and remote visualization and checking of product preparation; and end-product testing, and/or weighing of the solution to help verify accurate preparation.
One product under the hood. It is critically important for technicians and pharmacists to place the ingredients for only one product under a laminar-airflow hood at a time. Complete the preparation of that product and remove the supplies from the hood before beginning the next product. Only essential materials should be placed in the hood. If insulin is an additive, the product should be used under the hood alone and removed from under the hood immediately after use. Only one staff member should be working under a single hood at any given time.
Label placement on admixtures. Patient labels should be placed on the final product immediately after it has been checked. (Some hospitals have technicians partially affix the label to signify that the final product has not yet been checked.) The label normally should not cover the manufacturer’s label in its entirety. Wherever possible, a standard procedure for patient label placement should be created to allow independent checks of the base solution if a manufacturer’s container is used (also see the sidebar that follows). Educate nursing staff how to check these solutions and to recognize potential discrepancies.
Standardize and establish an effective checking process. Establish a standard operating procedure for how pharmacists should verify products prepared in the IV room. For additives checked before mixing (superior checking process, recommended when mixing high-alert medications), ensure that the process allows the pharmacist to clearly verify the source of the additive visually (e.g., vial, ampul, stock bottle) and the measured amount. For additives checked after mixing (inferior checking process, which involves containers placed alongside empty syringes with plungers pulled back to the volume measured [not recommended]), ensure that the method allows the pharmacist to make reasonable assumptions regarding the products and amounts added to base solutions.
In either situation, it’s also important to include verification of the base solution and any pre-mixed dilutions used to make the final preparation, as with neonatal/ pediatric preparations where an ingredient is first diluted and then used as a stock bottle during daily preparations. It’s also critical to require the checking pharmacist to verify the label and final preparation via comparison to the original order. The automated technologies mentioned above can make safer verification processes within reach of most hospitals.
Orderly and controlled checking area. Place each prepared product and associated supplies into a single bin/basket for checking, one at a time. Avoid bins/baskets with multiple compartments. Do not leave products on open counters for checking, even if they are well separated. Establish a system to communicate when prepared products are ready for checking, monitor the workflow, and provide assistance when necessary to avoid a backlog of products awaiting verification. Some hospitals use different colored bins to signify steps in the process (e.g., red bin for final preparations ready to be checked) or categories of solutions (i.e., neonatal, pediatric, adult). Ensure the checking area is uncluttered, of sufficient size, and well lighted.
Discard preparations in question. Ensure that products are discarded if there are any doubts, even if minor, about the preparation and storage of the product.
Given the frequency and potential severity of pharmacy parenteral drug preparation and admixture errors, ISMP plans to hold an invitational summit this October to explore the best interventions for safety associated with these practices. At the conclusion of the summit, we will be providing a draft set of comprehensive guidelines on the pharmacy admixture process for comment before we finalize them. We would like to hear from you regarding particular challenges you currently face with both manual and automated compounding and admixture—please contact us at: firstname.lastname@example.org. Also look for one or more surveys about admixture practices in upcoming newsletters this summer.
**This error occurred before the current shortage of 250 mL glass containers of 23.4% sodium chloride.
2) Pharmacy Purchasing & Products. State of pharmacy compounding 2009: survey findings. Pharmacy Purchasing & Products. April 2009;6(4):4-20.
1) Flynn EA, Pearson RE, Barker K. Observational study of accuracy in compounding i.v. admixtures at five hospitals. Am J Health-Syst Pharm. 1997;54(8):904-12.