A 66 year old man with a history of prostate cancer treated with radiation presented to his primary care physician for a routine follow up appointment. He complained of neck pain and spasms for a few weeks. An antispasmodic was prescribed and no further history and examination were documented. Three months later he presented for a second clinic visit. He now complained of worsening pain in the lower part of the neck with radiation down both arms that was accompanied by arm weakness. The notes documented that he had difficulty carrying grocery bags and dated this to several months previously. A neurological examination showed weakness of both upper extremities. A diagnosis of cervical radiculopathy was made and cervical spine radiographs were ordered. The patient was advised to return for follow up in 3 weeks.
Complaints of neck pain are common in the outpatient setting and the majority are related to benign conditions such as muscular strain and arthritis. To arrive at a diagnosis, physicians often use heuristics—which are mental shortcuts in clinical decision making that often are correct.7,23
In this case the physician initially used a pattern recognition heuristic and assumed a benign condition of the neck at the first visit. Given that the patient had a history of prostate cancer (a malignancy that often metastasizes to the spine), the physician should have documented additional information about the history and examination.24
Our evaluation of this case is limited by having only the medical record to review, but applying the SA model to the first clinic visit illustrates a failure in level 1 SA. It appears that the physician did not perceive all the relevant information. This level 1 failure affects all subsequent levels of SA.
Ideally, the best time to assess a clinician's decision making would be soon after the decision is made, thereby lessening recall or hindsight bias.25
The tool that Endsley suggests as the best method to directly measure SA is called the Situation Awareness Global Assessment Technique (SAGAT).26
SAGAT provides an objective measure of SA by directly comparing the subject's self‐reported SA with “reality”, as determined by the observer/investigator. This measurement is carried out in a simulated setting where a simulation is “frozen” at several points in time and subjects are asked specific questions to assess their level 1, level 2, and level 3 SA. This procedure overcomes the problems associated with hindsight bias and could potentially be applied to better understand the making of a diagnosis as it unfolds. Measures of SA have correlated with performance in aviation,27
and similar association will need to be demonstrated in medicine before its applicability.
Returning to the case, at the second clinic visit level 1 SA appeared compromised as indicated by the clinician's failure to record information about bowel/bladder symptoms and other neurological findings such as sensory level and reflexes. The clinician had a greater degree of level 1 SA than in the previous visit because the relevant clinical information about the prostate cancer history and some of the neurological symptoms and signs were documented in the progress note. However, the integration of this information led him to believe that this was a case of radiculopathy, a compression of a nerve root in the cervical spine, rather than to suspect spinal cord compression by a tumor. The clinician presumably did not fully comprehend and integrate all the available data and therefore did not consider cord compression due to metastatic prostate cancer. This is an example of a loss of level 2 SA.
Because of a loss of SA at level 2, forecasting future events and dynamics (consequences of a metastatic process leading to spinal cord compression syndrome in a patient with cancer who presents with back pain and neurological symptoms) was clearly lacking and thus SA at level 3 was also lost. This resulted in an incorrect resolution path which involved ordering radiographs and arranging for a future outpatient appointment rather than ordering an MRI scan, perhaps with admission to hospital.
A strength of the SA model is that it also includes system factors (see fig 1). For example, the loss of SA levels 1 and 2 may have been impacted by lack of time, excessive workload, interruptions, and information overload. The clinician in this case was working in an office that typically sees 9–10 patients during a half day session, and most of the patients have multiple medical problems.
The case continued …
A cervical radiograph was requested and the reason given by the primary care physician to the radiologist on the request form was “pain”. The radiologist, who was unaware of the other clinical findings of the patient, interpreted the radiograph as increased density in C6 (cervical vertebra) with a possibility of blastic bone metastases, and recommended a nuclear bone scan correlation. He personally communicated these findings to the primary care physician on the following day, who then called the patient to return for a scan.
Information given to the radiologist on the radiograph request form was only that of “pain”, which by itself conveys ambiguous information about a patient who has overt neurological findings. In order to provide optimal interpretations of radiological studies, radiologists must receive correct and pertinent clinical information. Although the test of choice in this case was an MRI scan, the clinician followed the radiologist's interpretation (which was not based on the clinical findings) to obtain a bone scan, a test perhaps appropriate to confirm metastasis only if there was no ongoing neurological compromise. The radiologist's effort to communicate the findings directly is laudable given the seriousness of the report, but he attained only partial level 1 SA. He never really understood the situation about ongoing neurological compromise. This could be due partly to poor radiology ordering systems leading to poor communication and information transfer, another example of the impact of systems factors on SA and diagnostic errors. Errors in communication (here due to inadequate communication regarding the past medical history and indication for the imaging procedure), which are well known as causes of mishaps in aviation, are also a very common reason for preventable adverse events and medical errors.28
The case continued …
When the patient returned he was noted to have worsening leg weakness and unsteadiness. Concerned about metastatic prostate cancer, the primary care physician communicated with a consulting urologist. Although it is not clear what information was relayed to the consultant, the advice was to follow up in the urology clinic in 3 days. The patient was then sent for the bone scan. However, while having the scan, he developed worsening neck pain with left arm and leg numbness and paralysis while lying on the table with his neck extended. He was sent to the emergency room (ER) for immediate evaluation. Documentation by the ER physician revealed weakness, absence of bowel and bladder incontinence, and acknowledged the prostate cancer. After review of the case, an impression of “C spine radiculopathy possibly from blastic lesions” was made by the ER physician. He discussed the case with a neurosurgical consultant over the phone and a decision was made for an outpatient MRI scan and follow up in the neurosurgery clinic in 7 days. He was then discharged from the ER.
Using cognitive science to explain these chains of events illustrates another shortcut in reasoning called the “framing effect”. Using this effect, clinicians can arrive at different decisions depending on how the information was presented, or framed.7
For example, the primary care physician most likely presented the patient to other physicians as a case of “radiculopathy” rather than spinal cord compression due to metastasis. All the physicians framed their diagnosis into mental models (mental representations of external reality that play a major part in cognition) without considering alternative explanations. Once again, the aviation framework can be applied to this situation. Achieving SA depends on the use of mental models and pattern matching using cues from the environment. In addition to the primary care physician, three other physicians experienced problems with gaining SA because they were fixated on one cue or diagnosis (cervical radiculopathy) and ignored others (cord compression). Because they were unable to integrate adequately a correct comprehension of the situation (level 2), their prediction of possible outcomes and the resultant treatment path was incorrect (levels 3 and 4). In this instance, a lack of gaining SA (versus losing it) resulted in an inability to appreciate the severity of the patient's problem and an alternative and incorrect path to approach the case. A lack of gaining SA could be due to workload issues in surgery and ER medicine, communication failures, time pressures, and other system issues. However, other cognitive and affective influences such as bias, conscious or unconscious complacency with a false sense of security, negative affective states from fatigue and burn out from being on call could all potentially cause failure to gain SA.
SA can also be applied to medical teams,20
including those in ambulatory care where the team concept traverses several disciplines and locations and is hence more “distributed”. Team dynamics played a critical role in this case where the entire physician team lost their SA. Effective team SA depends on team members developing accurate expectations for team performance through “shared mental models” that allow for better team process and performance among them.29,30
SA has been identified as an important team skill that can be used to decrease human error.31
To gain SA, each team member must adopt an attitude called “aggressive scepticism” in order to constantly evaluate the environment.32
In this case the use of aggressive scepticism would have led to more independent evaluations of the case by each consultant, or more information sharing and questioning among the consultants. Although none of the consultants personally evaluated the patient (which by itself was detrimental to their SA), enough information seemed available to them to interject with their own perspective. In fact, combining the perspectives of all team members may be the most reliable way to remain fully aware of the environment (and thus of a patient's condition), leading to a high level of team SA. If one of the clinicians had raised a red flag for cord compression and alerted the others, both team and individual SA would have been recovered. Breakdown in team SA in this scenario could be partly responsible for the delayed diagnosis. This was probably due to both cognitive breakdowns and lack of good communication systems in the healthcare setting, in a background of other system inadequacies.33
The case continued …
The next day the patient was brought in a wheelchair by the family with worsening weakness and was noted to have bladder incontinence. He was admitted to hospital and had an MRI scan which revealed that he had metastatic spinal cord compression at multiple cervical levels.