A 60-year-old woman with peripheral artery disease and chronic mesenteric ischemia was admitted for management of inferior mesenteric artery (IMA) stenosis. She underwent IMA and right renal artery stenting by vascular surgery and was transferred to an internal medicine teaching service for ongoing management of hyponatremia and hypertension during the second week of hospitalization.
On hospital day 18, the patient had sudden nausea and abdominal distension. The overnight cross-cover intern evaluated the patient and ordered an abdominal radiograph. Although the patient was uncomfortable, her vital signs were stable, and the intern did not observe an acute abdomen on examination. The intern read the radiograph as an ileus and attributed patient's symptoms to this; tube feeds were held. While the intern intended to contact radiology to confirm her interpretation, another patient acutely deteriorated (requiring a Code Blue activation), and she forgot to follow up on the radiograph. Early the next morning, the vascular surgery patient developed delirium, hypotension, and hypoxemia, and she was emergently transferred to the intensive care unit.
The radiograph was formally read by a radiology attending at approximately 8:00 AM. She immediately recognized portal venous gas—a sign of acute mesenteric ischemia. She contacted the primary team, who ordered a CT angiogram, which revealed internal mesenteric artery dissection, diffuse bowel ischemia, and infarction of the liver and spleen. After discussion between surgery, the primary team, and the family, the decision was made not to pursue further surgery as the expected mortality was extremely high. The patient was transitioned to comfort care and died that evening.
The case was formally reviewed by the medical service and by radiology. The reviews concluded that the delay in diagnosis of internal mesenteric artery dissection and bowel infarction represented a preventable adverse event. The major contributing factors were considered to be the workload of the overnight cross-cover intern and the lack of a formal system for obtaining radiology interpretations of radiographs performed overnight. At this hospital, the overnight staffing on the medicine service consisted of two hospitalists, two senior residents, and one intern, but the intern was "first call" for all acute issues overnight. The radiology department had two residents in house overnight, but provided interpretation only for emergency department studies, CT scans, and MRIs performed overnight. Regular radiographs were not formally read until the morning, unless specifically requested.
by Clinton J. Coil, MD, MPH, and Mallory D. Witt, MD
The case presented here illustrates two common errors occurring in sequence and contributing to a poor outcome. First, the intern failed to appreciate a subtle radiographic sign. Second, she neglected to complete a planned consultation with radiology in the course of a busy call night. To say that the intern should have interpreted the image correctly while simultaneously confirming the finding with a radiologist is all too easy. The next step in the "blame and train" cycle is typically to recommend additional education.
However, blaming the clinician associated with an adverse event is generally an ineffective strategy for improving patient safety. A more fruitful approach requires that we accept that well-intentioned humans will occasionally make errors, while modifying our systems of care to protect patients from harm. The overlooked radiographic finding is relatively subtle, and teaching all interns the skills required to identify such a finding is unrealistic. In fact, relying on the efficacy of such training would predictably lead to repetition of similar events. Instead, a more robust solution would involve both timely interpretation by an experienced radiologist and a failsafe mechanism to transmit this interpretation to the covering physician.
Distractions can derail even a well-conceived care plan.(1) In this case, the Code Blue event took priority, preventing the intern from consulting radiology as planned as she attended to the code and distracting her from recalling her plan to complete the consult. Creating a system to ensure expert interpretation of radiographs, independent of trainees' workload, is an important strategy. A variety of approaches can be considered, and combining multiple strategies is likely to yield the best outcome.
One strategy for reducing harm associated with distractions is to include other physicians in care processes as a matter of routine so that if a step is forgotten another team member will notice and intervene. In this case, though hospitalists and senior residents were on duty at the time of the error, they do not seem to have been involved in overseeing the care provided by this intern.
While most training programs now make available overnight physicians with greater training and experience to support trainees, expecting a trainee to know when to ask for help is problematic. Inexperienced trainees may not have the knowledge or experience base to realize that they are in over their heads ("you don't know what you don't know"). In some cases, the organizational culture may discourage seeking help, with the covert implication that a request is a sign of weakness. Many training programs have created "must call attending" lists to define triggers that mandate consultation with a supervisor, an approach that appears to be effective.(2) Such a list might include any unplanned intubation, an unplanned transfer to ICU, transfusion of more than two units of blood products, etc. Another approach is to schedule supervision in a defined timeframe, such as setting rounds at midnight and 4:00 AM, or using tools such as scheduling briefings to update the team on new admissions or key events on existing patients. In this case, a scheduled briefing might have allowed another team member to catch the intern's lapse.
Redundant work processes can be another effective systems-based strategy to prevent errors. Having a dedicated radiologist read all images the following workday ultimately led to this patient's correct diagnosis. It is likely that hospital leadership's decision to require only certain studies be reviewed by a radiologist was an effort to conserve scarce resources. However, as this case illustrates, it is neither safe nor realistic to assume that plain films lack critical, time-sensitive findings. Although plain films may seem less critical than three-dimensional images such as CT and MRI, subtle radiographic signs can be the only indication of a life-threatening emergency, and a high level of expertise may be required for their correct interpretation. The assumption that plain films obtained at night are unlikely to yield important findings unless ordered in the ED is risky.
This question of radiology procedures at night is only one of many safety concerns that relate to off-hours care. Differences in the structures and processes governing hospital workflows during off-hours are common. Fewer physicians may be covering a larger number of patients, diagnostic studies may be unavailable or available only in an emergency, and certain treatments may be inaccessible. In addition, a less-experienced person may provide care in the after-hours setting, especially in teaching hospitals. The so-called weekend effect (3-5) describes increased mortality and adverse events during periods of reduced staffing such as nights, weekends, and holidays. Since its description in 2001 (6), the weekend effect has been the subject of more than 100 studies in different patient populations.(7) While some studies have focused exclusively on weekends versus weekdays, others have looked more globally at times of decreased staffing, including nights. The weekend effect has been observed in a variety of clinical settings, from acute myocardial infarction and stroke to routine cosmetic surgery (8) to palliative care.(9) Not only may complications be more common during weekends, but a patient who experiences a serious adverse event on a weekend is more likely to die than one experiencing a similar event during the week.(10) This suggests that the response to a deteriorating patient may be less robust outside of regular working hours.
On top of the issues related to staffing and workflow on weekends and nights, it is worth noting that the cognitive errors in this case occurred during an overnight shift, when human cognitive performance tends to be at its worst. Some studies indicate that the weekend effect may be reduced or even eliminated in settings such as trauma centers or emergency departments, where attending level supervision is typically present around the clock.(3) With this in mind, efforts to establish support systems to prevent adverse events at night should be a high priority.
Teaching hospitals must implement reliable systems that balance the obligation to provide safe, evidence-based, and cost-effective care with the need to allow trainees to gain experience and confidence as they progress toward independent practice.(11) To be effective, these systems cannot rely upon the knowledge and vigilance of trainees, but they should ensure consistent supervision regardless of time of day.
- Hospitalized patients may be at higher risk of dying during off-hours. The underlying cause of the weekend effect (which may extend to nighttime and other off-hours care) is not entirely clear, but reduced staffing and supervision, as well as limited availability of diagnostic tests and therapies may contribute.
- Systems of care that rely on the performance of a single individual will predictably fail. Creating safeguards and redundancies in provision of care can help protect patients from harm.
- Accurate interpretation of plain radiographs may require as much expertise as CT scans and MRIs. Correct interpretation may provide critical information in life-threatening emergencies.
- Rather than relying on trainees to seek help, establishing systems to ensure that supervision is built into the structure of care delivery may enhance patient safety.
Clinton J. Coil, MD, MPH Chief Quality Officer Associate Clinical Professor of Emergency Medicine Harbor-UCLA Medical Center Torrance, California
Mallory D. Witt, MD Professor and Vice Chair Department of Medicine Harbor-UCLA Medical Center Torrance, California
1. Rivera-Rodriguez AJ, Karsh BT. Interruptions and distractions in healthcare: review and reappraisal. Qual Saf Health Care. 2010;19:304-312. [go to PubMed]
2. Silverman AT, Goldfarb MA, Baker T. When should a surgical resident call an attending surgeon? J Surg Educ. 2008;65:206-212. [go to PubMed]
3. Metcalfe D, Perry DC, Bouamra O, et al. Is there a 'weekend effect' in major trauma? Emerg Med J. 2016;33:836-842. [go to PubMed]
4. Lilford RJ, Chen YF. The ubiquitous weekend effect: moving past proving it exists to clarifying what causes it. BMJ Qual Saf. 2015;24:480-482. [go to PubMed]
5. Concha OP, Gallego B, Hillman K, Delaney GP, Coiera E. Do variations in hospital mortality patterns after weekend admission reflect reduced quality of care or different patient cohorts? A population-based study. BMJ Qual Saf. 2014;23:215-222. [go to PubMed]
6. Bell CM, Redelmeier DA. Mortality among patients admitted to hospitals on weekends as compared with weekdays. N Engl J Med. 2001;345:663-668. [go to PubMed]
7. Freemantle N, Ray D, McNulty D, et al. Increased mortality associated with weekend hospital admission: a case for expanded seven day services? BMJ. 2015;351:h4596. [go to PubMed]
8. Tadisina KK, Chopra K, Singh DP. The "weekend effect" in plastic surgery: analyzing weekday versus weekend admissions in body contouring procedures from 2000 to 2010. Aesthet Surg J. 2015;35:995-998. [go to PubMed]
9. Voltz R, Kamps R, Greinwald R, et al. Silent night: retrospective database study assessing possibility of "weekend effect" in palliative care. BMJ. 2014;349:g7370. [go to PubMed]
10. Ricciardi D, Nelson J, Francone T, et al. Do patient safety indicators explain increased weekend mortality? J Surg Res. 2016;200:164-170. [go to PubMed]
11. Haber LA, Lau CY, Sharpe BA, et al. Effects of increased overnight supervision on resident education, decision-making, and autonomy. J Hosp Med. 2012;7:606-610. [go to PubMed]