After the Institute of Medicine report To Err is Human, healthcare organizations initiated new data collection and analysis tools to help resolve medication errors.1 This is the description of a successful medication error reduction initiative that evolved over an 8-year period. The purpose of this study was to inspect both the administration process and the human component of medication errors—not to look at the errors in isolation. Although multiple process improvements decreased the number of errors, the most significant improvement came with the change in our care delivery model. Since that change 5 years ago, there has been a sustained, significantly low medication error rate at our facility.
The study team consisted of RNs, licensed practical nurses (LPNs), pharmacists, unit secretaries, unit nurse managers, and the service line clinical director. Clinical nurses and clerical personnel were charged with reporting all real or potential errors via an incident reporting system (IRS). Nursing leadership, the nurse manager, and the clinical director performed a concurrent review of all errors reported through the IRS and determined the root cause for each error. Errors were assigned to one of three categories: system failure, process failure, or human failure. The subsequent investigation then enabled leadership to assign the error into subgroups to trend the data more precisely.
Results were dependent on 100% reporting of errors in a nonpunitive environment. Concurrent reporting and investigation of errors was utilized. IRS reports were reviewed daily and involved staff members were contacted to clarify the causative factors for the errors. This was essential because our IRS had “canned responses,” which staff could select rather than submitting an insightful response. A quantitative study of errors and outcomes, in addition to a retrospective analysis, were performed to establish baseline information. Medication errors were analyzed by process, procedure, and person. Data were entered into a spreadsheet. All data were shared with unit staff members monthly and presented biannually to the hospital's shared decision-making and performance improvement (PI) councils, the PI committee, and the hospital board.
The investigation begins
The first challenge to overcome was the underreporting of errors. Staff members didn't report errors for two specific reasons: fear of punishment and lack of consensus on what constituted an error. Encouraging staff members to report potential and actual errors was critical to the data collection accuracy. In order to achieve a goal of 100% reporting, it was imperative to establish a nonpunitive reporting environment. The leadership team concentrated on repair of the processes that failed rather than affixing blame. Staff members quickly realized that this initiative focused on providing a safe environment for patients and staff. They diligently reported potential and real errors following the philosophy of “if it looks wrong, report it.”
It fell to the management team to investigate and classify occurrences as an error or aberrance. This was done through chart review and discussions with the staff member(s) involved. The Just Culture model was employed for error analysis. This model initiates different resolutions dependent on the root cause of the error/problem. If the cause is determined to be the result of human error, the Just Culture model requires that management console, coach, or counsel. It was imperative to track staff member involvement to provide appropriate education and remediation as needed.
The second challenge was extrapolating the data from the IRS, which didn't categorize errors on a sufficiently detailed level. Further expansion on the reasons for the errors and input into the database were needed. It was considerable work to initiate but, once established, it provided more in-depth information that allowed for trending of specific error types. The task was accomplished utilizing a spreadsheet.
The third challenge to reducing medication errors was to determine if there was a correlation of errors to overstaffing or understaffing on the unit. A spreadsheet was created to examine staffing per shift, patient census, and errors reported. It was staff members' perception that most errors were made during periods of minimal staffing. The data were tracked to ascertain if this perception was factual.
A “whodunit” case
A retrospective analysis of our data demonstrated that in 2004 there were 168 medication errors (0.1%) on the nursing unit. This information was presented to the unit PI and clinical practice (CP) councils and at staff meetings. Staff members reviewed the data for gaps in the process that may have allowed the error to occur. These gaps ranged from material and programmatic issues to the human element of nonadherence with hospital procedures. Clinical nurses reviewed all of the protocols and determined where the process could create an error. This resulted in the development of new procedures.
After processes were changed to provide a safer medication administration practice, data analysis demonstrated a decrease in the error rate: 2006, 0.06; 2007, 0.04; 2008, 0.02; 2009, 0.02; 2010, 0.02; and 2011, 0.03. (See Figure 1.) The initiation of the electronic medical record (E-MAR) and a computerized provider order entry (CPOE) system resulted in no statistically significant change in the medication error rate. (See Figure 2.)
Reviewing the notes
All processes related to medication administration—simple to complex—were scrutinized to determine the potential points within the process where an error could happen. A minor process correction was made to the paper medication administration record (MAR). The MAR form was located in a centralized binder on the medication carts for all members of the healthcare team. The MARs consisted of multiple pages and, through the investigation, it was determined that errors occurred due to pages getting lost and then recreated or misplaced into the wrong section of the binder. Staff members initiated the process to use a notebook ring on each patient's MAR to ensure that the pages remained together. With the initiation of each new process step, staff members held each other accountable for knowing and adhering to the process. Failure to adhere resulted in reeducation and additional nonadherence placed employees into remediation and/or corrective action. Remediation was performed in a respectful manner to maintain the employee's dignity.
The second process improvement was the creation of a medication quiet zone. This consisted of the medication cart itself and the nurse. The idea was that the nurse wasn't to be disturbed for any reason when at the medication cart. A sign was affixed to the medication cart and hung from the pole when medication was being prepared for administration. Staff members adhered to the process and assisted each other in providing the quiet time to facilitate safe medication preparation.
The third process issue that came to light was that nurses weren't taking the medication card filing system with them during the medication pass. This set the staff up for failure to adhere to the five rights of medication administration: right patient, right drug, right dose, right route, and right time. The process was reviewed and each nurse was held accountable for following the policy and procedure of medication administration.
A retrospective study was performed to compare error time with unit staffing levels. There was no correlation between medication errors and perceived understaffing of RNs or LPNs. However, when medication error types were defined, it was apparent that approximately 40% were classified as transcription errors. On the oncology unit, the unit secretary transcribes the orders and a clinical nurse checks and verifies the orders for accuracy and completeness. Trending the error time we found that the majority of the transcription errors occurred on the evening shift (3 p.m. to 11 p.m.). Analysis of the activity level, admissions, discharges, and transfers revealed that our activity levels were highest during the evening shift. Clerical support was redistributed to this shift and the errors dropped precipitously.
Finally, after all of the task processes were rectified, the main process that became the focal point was the care delivery model. A team approach was utilized. Each patient had an RN as the team leader, a medication nurse, and unlicensed assistive personnel (UAP) to deliver patient care. The RN focused on the patient assessment and coordination of care with physicians and other disciplines; the LPN performed the role of medication nurse. The medication nurse would deliver medications for the team and, depending on the shift worked, could be assigned to more than one RN team leader.
Analysis of the medication errors revealed that errors resulted from (1) lack of communication between the RN and LPN, (2) LPN lack of understanding of the medication effects/interactions, and (3) RN inability to effectively coordinate care for seven to eight patients. The unit CP council met with the nursing leadership team to discuss the issue. The decision was made to revamp the care delivery model and create a primary nurse model with a 1:5 nurse-to-patient ratio.2,3 LPNs and UAP assumed the role of support personnel. Our research also demonstrated that adding LPNs into the care model didn't improve patient outcomes. Therefore, the nurse-to-patient ratio changed by converting LPN and UAP positions to RN positions through staff attrition and encouraging/supporting LPNs and UAP to return to nursing school. After graduation, each staff member was offered a position to remain on the unit as an RN. With the initiation of the new care delivery model, there was a dramatic drop in the number of medication errors. (See Figure 2.)
Upon closer examination
Staff members reported any deviation from policy as an error. During the course of the investigation, not all process errors were within the investigators' scope to fix and others required a higher level of authority to affect change or impacted other hospital departments. Staff frustration needed to be managed when a quick fix couldn't occur, as well as staff members' expectations when process changes required committee approval or other departments to change their part of the process. Staff member morale was enhanced by focusing on owning their practice and acknowledging that they were doing the right thing by exposing the areas of potential errors in other departments. Successful instillation of the nonpunitive reporting environment was evident when staff members reported their own errors and subsequent analysis with action plans to correct the process or behavior.
Discussions were held with the pharmacy department regarding timing and administration of medication, delivery methods of “stat” and routine medications, and safety indicators on medication labeling. Point people were chosen from pharmacy and oncology to serve as communication conduits. Errors and concerns were channeled through the point people. The pharmacy point person attended the practice committee meeting to discuss pertinent changes to practice and collaboration between the departments.
A fragmented care delivery system, high nurse-to-patient ratios, and a perceived punitive work environment contributed to medication errors. By defragmenting the care delivery model, lowering the number of patients per nurse, and educating staff on the Just Culture philosophy, the medication error rate was significantly lowered and maintained for over 5 years. This process is simple to replicate and may help other facilities that are experiencing a high rate of medication errors.
2. Aiken LH, Clarke SP, Sloane DM, Sochalski J, Silber JH. Hospital nurse staffing and patient mortality, nurse burnout and job dissatisfaction. JAMA
. 2002;288(16): 1987–1993.
3. Needleman J, Buerhaus P, Pankratz VS, Leibson CL, Stevens SR, Harris M. Nurse staffing and inpatient hospital mortality. N Engl J Med