Automated electronic reminders to prevent miscommunication among primary medical, surgical and anaesthesia providers: a root cause analysis
- 1Department of Anesthesiology, University of Michigan, Ann Arbor, Michigan, USA
- 2Department of Quality Improvement, University of Michigan, Ann Arbor, Michigan, USA
- 3Office of Risk Management, University of Michigan, Ann Arbor, Michigan, USA
- Correspondence to Dr Robert Edward Freundlich, Department of Anesthesiology, 1H247 Box 0048, University of Michigan Hospital, 1500 East Medical Center Drive, Ann Arbor, MI 48109, USA;
Contributors All authors have contributed to the data described in this article, and in the drafting and revision of this manuscript. All authors have approved the final version of the manuscript.
- Accepted 24 May 2012
- Published Online First 7 July 2012
In this case report, the authors present an adverse event possibly caused by miscommunication among three separate medical teams at their hospital. The authors then discuss the hospital's root cause analysis and its proposed solutions, focusing on the subsequent hospital-wide implementation of an automated electronic reminder for abnormal laboratory values that may have helped to prevent similar medical errors.
- Adverse events
- epidemiology and detection
- quality improvement
- root cause analysis
- decision support
- comparative effectiveness research
- information technology
A 1999 Institute of Medicine report attributed 98 000 patient deaths to preventable medical errors each year. In the same report, anaesthesiology was singled out to demonstrate that medical errors can be avoided through the creation of clinical standards and processes that address the root cause of common adverse events.1 The adoption of machine ‘check-outs’, monitoring standards and rigorous self-reporting has allowed anaesthesiology to become a leader in patient safety research and practice.
Errors in communication are a possibility in the perioperative environment2 and a documented cause of medical errors and patient deaths.3 ,4 We report an adverse event due to miscommunication among the inpatient neurology, surgery and anaesthesia teams. We then describe the hospital's root cause analysis (RCA) and proposed solutions, focusing on one intervention, an automated abnormal laboratory value electronic reminder system, designed to help prevent such events in the future.
A 51-year-old male patient with a history of testes cancer was transferred from an outside hospital with mental status changes and new-onset left-sided weakness. Neuroimaging could not differentiate among a primary tumour, metastasis or a new cerebrovascular infarct. The primary neurology team requested neurosurgical evaluation for a brain biopsy. The neurosurgery team reviewed the patient history, exam results and laboratory values and scheduled the patient for an elective stereotactic brain biopsy under general anaesthesia for diagnostic and therapeutic planning purposes. The brain biopsy was performed several days later without intraoperative complications. No abnormal bleeding was noted during the case. The patient was extubated in the operating room, observed for several hours in the postanaesthesia care unit and returned to the general care floor. Several hours after the operation, the patient was found unresponsive with agonal respirations. The patient was emergently intubated and underwent CT imaging, which revealed a massive intracranial haemorrhage with associated uncal and falcine herniation. As per the family's request, comfort care was instituted and the patient expired on postoperative day 1 (figure 1).
Although previously normal, the patient's partial thromboplastin time (PTT) was markedly elevated before surgery at 83.6 s (institutional normal range: 23.3–30.2 s). Neither the neurosurgery service nor the anaesthesiologists were aware of this abnormal value at the time of the operation (Monday). Each had started the operation aware of a previously normal PTT drawn several days before surgery. The patient was evaluated and deemed fit for surgery based upon the normal PTT of 24.6 s. Additionally, the neurology service did not discontinue the patient's prophylactic subcutaneous heparin as requested by the neurosurgery service, although this would not explain the elevated PTT.
A sentinel event review process, including RCA, was undertaken by the Office of Clinical Affairs, Office of Risk Management, Department of Quality Improvement, Department of Neurosurgery and Department of Anesthesiology. Based upon this sentinel event, a comprehensive risk reduction strategy, including the entire perioperative process, ranging from surgical providers to nursing and anaesthesiology, was undertaken (table 1). Interventions included improved documentation of sign-out in the electronic medical record, standardised order sets in the electronic medical record and expanded preoperative checklists, among others.
In addition to the other interventions described in table 1, an automated electronic reminder system was implemented. This decision represented a group consensus among those involved with the review process, although some members expressed concern about the possibility of alert fatigue. The automated reminder system uses the hospital alphanumeric paging system to alert the scheduled surgical and anaesthesiology providers of any markedly abnormal laboratory values. Five common laboratory values were chosen for review and alerting in a hospital-wide pilot study: partial thromboplastin time, prothrombin time (international normalised ratio), serum potassium, serum magnesium and haemoglobin (table 2). These values were chosen because they require additional preoperative preparation such as obtaining blood products, electrolyte supplementation or haemodialysis. In extreme cases, they may even warrant a case delay while investigative or corrective actions are undertaken. Based on the overwhelmingly positive feedback received, the alerting system is still in place at this time.
At our institution, all laboratory values and operative information are integrated into a perioperative clinical information system (Centricity, General Electric Healthcare, Waukesha, Wisconsin, USA). Each day at 13:00 h, the next day's surgical schedule is considered finalised and preoperative preparations are undertaken by anaesthesiology and nursing staff. For Monday cases, the schedule is finalised on Friday.
Each day at 15:00 h, the automated reminder system reviews all patients scheduled for the next day. The most recent laboratory value within the previous 30 days is compared with high and low thresholds established for the alert system (table 2). These thresholds are based upon supporting literature when available and guidance from the hospital's operating room management committee. If a value exceeds the thresholds, an alphanumeric page containing the patient name, laboratory test, laboratory date and laboratory value is sent to the scheduled anaesthesiology attending, anaesthesiology resident or certified registered nurse anaesthetist, surgical attending and surgical resident. If a given patient has more than one laboratory test out of bounds, a single alphanumeric pager message contains all the information. The neurosurgery service also requests an alerting system which reviews the laboratory values at 07:00 h for patients scheduled for surgery that day. Other surgical services at the hospital were given the option of having this additional alert but felt that it did not add any incremental benefit. A review of the paging history demonstrates that approximately 190 cases are reviewed each day by the automated reminder system, resulting in 14 pages per day (table 2).
The aforementioned serious adverse event and subsequent RCA uncovered several potential problems in our clinical care processes. Even when structured properly, an RCA may produce non-actionable solutions, often for a variety of reasons.5 As described in table 1, the RCA for this incident produced a set of recommendations that have since been completely implemented.
In this case report, we have described one particular intervention in greater detail, an automated laboratory reminder system. The value of routine preoperative laboratory testing has been under scrutiny for some time.6–8 Nevertheless, clinicians often order a battery of laboratory tests prior to elective surgery. Although the practice of ordering unnecessary exams may be wasteful at best, failing to identify truly abnormal results and act on them may have medico-legal and clinical consequences. Clearly, clinicians must review laboratory values prior to proceeding with an operative case. However, the patient safety literature has demonstrated that any process that depends solely on non-automated, human intervention will remain prone to error.3 ,9 ,10 Automated reminders and alerting systems have been proposed to serve as an additional layer of defence that complements existing human vigilance, conscientiousness and decision-making.1 ,11–13 Specifically, automated alphanumeric paging of abnormal laboratory values has been reported in the general medicine literature as a means of improving patient safety.11 ,14 However, we are unaware of any automated reminder systems used to inform perioperative providers of markedly abnormal preoperative laboratory values. There are reports of reminder systems utilised to improve perioperative documentation completeness.9 ,15
Provider feedback to the alerting system has been largely positive and the system is still in place at this time. In fact, the clinical community has since requested two additional laboratory values for alerting inclusion, platelets and a positive antibody screen. Based upon provider feedback, serum magnesium has been removed from the alerting system due to its limited perioperative safety impact (table 2). A comprehensive review of adverse events since the implementation of the alerting system is beyond the scope of this case report; no similar events have occurred but, given the rarity of this adverse event, this is not necessarily due to the post-RCA interventions. However, clinicians have provided anecdotes demonstrating that the alerting system did result in increased vigilance and investigation and have expressed universally positive feedback.
There are several limitations associated with the alerting system. Advanced functionality such as thresholds tailored to patient age, surgical service or comorbidities has not been implemented due to software programming resource limitations. More importantly, some clinicians may find the alerts to be a redundant nuisance rather than a helpful safety net. The interruptions caused by the alphanumeric page itself could be a source of distraction and increase the likelihood of a medical error.16 In addition, the system is dependent upon accurate scheduling and personnel documentation in order to alert the correct providers. As a result, abnormal values can easily ‘fall through the cracks’ if a case is added to the schedule after 15:00 h or a laboratory value is not available when the automated reminder system performs its check. Finally, users may become increasingly complacent and dependent upon the reminder system.
This intervention was not the only solution we proposed following our RCA and, in fact, it may not have even prevented this serious adverse event in and of itself, as the elevated partial thromboplastin time may have been a spurious value. We believe, however, that it highlights the challenges of RCA processes, particularly the difficulty of using patient outcomes as a metric for RCA effectiveness. For example, it is difficult to establish the quantitative patient safety impact of the laboratory reminder system. It is essentially impossible to attribute trends in case cancellation, procedure deferral, electrolyte supplementation or blood product requests to a specific intervention such as the one described in this manuscript. Given these difficulties, we would argue that a process outcome, the successful implementation of all actionable RCA recommendations, is a more appropriate measure of the effectiveness of our RCA.
In summary, we report an adverse event associated with miscommunication and decreased vigilance. We have described the RCA process and discussed the subsequent interventions to illustrate some of the challenges hospitals face in implementing the recommendations of an RCA.
Funding This work was supported by internal funding supplied by the Department of Anesthesiology, University of Michigan.
Competing interests None.
Ethics approval Ethics approval was provided by University of Michigan IRB.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement There are no additional unpublished data; however, we would be happy to provide additional information about automated electronic laboratory reminders upon request.