There is a plethora of evidence that suggests, as physicians we over-diagnose, over-treat and over-investigate illness, often with little benefit to patients. The idea of rationale resource utilization is beginning to gain momentum, with campaigns such as Choosing Wisely, advocating for physicians to use the best evidence available in their practice.
In order to appreciate inappropriate resource utilization, physicians first need to be aware of why they’re over-investigating and any inherent biases they may have, in order to correct this behaviour.
Here, we seek to examine some of the rationale for why physicians over-utilize resources, and address some myths and solutions around these issues. Before you read any further though, take a moment and consider why physicians may over-investigate patients, and in particular in the Emergency Department (ED) (go ahead, take a second, we’ll wait)…
One of the most classically cited reasons for over-investigating or over-treating patients, is the fear of medicolegal consequences 1. However, this is probably an misappropriated justification for resource utilization in Canada.
In 2014, the Canadian Medical Protective Association (CMPA) had 91, 000 members – of these 1092 (1.2%) had reported legal actions, resulting in 26 plaintiff judgements (0.1%), and 393 (36%) of these were settled2. Firstly, one can conclude that it is rare for physicians in Canada to be sued (and sued successfully). Secondly, the more important component here is that the majority of legal actions taken against physicians were not as a result of negligent diagnoses or treatments, but rather due to issues surrounding communication, unprofessional behaviour, sexual misconduct and billing3.
CMPA experts suggest that it is acceptable to miss a diagnosis, you’re just not allowed to have not thought about it 21. From a practice standpoint, it is very reasonable to document your thought process as to why a patient may or may not have a particular disease – rather than over-investigating an unlikely entity, to alleviate a medicolegal concern.
A physician’s low tolerance for diagnostic uncertainty has been associated with increased resource utilization4, and ultimately I suspect this is one of the main reasons for medical over-testing. Intuitively, we can rationalize where this comes from – in medicine, we’re trained to seek out and diagnose disease, and it is therefore inherent within our training to obtain an specific diagnosis. When none is discovered, or uncertainty exists, this makes us uncomfortable. There is a cultural expectation within medicine that it is unacceptable to miss things (just consider the last time you, or a colleague missed a diagnosis – there is always an negative associated connotation).
In their attempts to increase diagnostic certainty, physicians often naturally drift towards increased testing. However, there is no evidence to suggest increased testing actually leads to improved patient outcomes5. Historically, excellent physicians were thought to be the most astute, and able to identify rare disease – learners are rewarded for being thorough rather than efficient. Ultimately, the excellent clinician needs to be comfortable with an element of diagnostic uncertainty6.
All physicians have a standard spiel for chest pain or abdominal pain, when no particular disease process is identified. Part of this description to patients includes the idea that sometimes we don’t identify an particular pathology, but we have ruled out the potentially dangerous or treatable causes, and beyond that we’re not sure. Disease should be considered as an evolutionary process, and that if the patient has something significant, their disease will progress and they will get worse – highlighting the importance of thorough and understandable discharge instructions.
In conjunction with diagnostic uncertainty, risk tolerance has been cited as another reason for increased resource utilization. Studies have demonstrated that risk averse physicians order statistically more tests than their colleagues, with no differences in adverse outcomes7.
In particular, three studies examining this, correlated diagnostic testing in the ED with physician risk tolerance (as determined by utilizing three validated risk tolerance personality scores). The first study examined CT ordering rates for patient’s with abdominal pain – and found that more risk averse physicians ordered 15% more CT scans on patients (50% vs 35% of patients)4. In patient’s presenting with chest pain to the ED, 78% were admitted by more risk averse physicians, and 68% by the less risk averse8 (American literature, so their rate of admissions for chest pain is significant higher than in Canada). A third study looked at patient’s with dizzyness, and found more risk averse physicians ordered statistically more CT scans (22% vs 13%)9.
The biggest take home point from all of these studies, is that these patient’s had no differences in outcomes – suggesting that the increased testing did not result in an benefit for patients.
Fear of missing a diagnosis
Fear of missing a low probability diagnosis likely has a role in conjunction with diagnosis uncertainty and risk tolerance. In the literature, the fear of missing a diagnosis is often cited as an significant driver for over-investigating.
An interesting study prospectively examined physician practice habits over an one-year span for patient’s presenting with suspected PE. Physicians had to fill out a study form, including their pre-test probability and rationale for test ordering (i.e.: elevated D-Dimer, positive Well’s, etc). Ultimately, 55% of physicians cited an ‘fear of missing an PE’ (even when the pre-test probability was low) as a rationale for ordering the CT10.
Interestingly, the authors found that the ‘fear of missing an PE’ was actually negatively associated with having an PE (unsurprisingly, a high pre-test probability had the greatest association).
As previously discussed, there is no evidence to suggest that this increased imaging results in improved patient outcomes, and looking at CT scan data from the US from 2001-2005, the rate of abdominal CT scan use more than doubled (and ultrasound utilization also went up during this time frame), but there were no increased rates of diagnosis of appendicitis, cholecystitis or diverticulitis, and no resultant increased in hospital admission rates5.
It makes me feel better
I think the combination of diagnostic uncertainty, risk tolerance and the fear of missing a low probability diagnosis are perhaps the biggest drivers of over-investigating, and if one recognizes that they fall into this category I would suggest considering whether or not investigations are being done for the benefit of the patient, or to make yourself feel better. Because if you’re doing it for self-reassurance, perhaps it is a test that does not need to be done, or shared decision making needs to be employed.
Sometimes, especially when the department is busy it is easier to order a test than it is to think through the appropriateness of that test. Consider a patient who you suspect may have pneumonia clinically, we’ve all, at one point or another, ordered an CXR without listening to the patient’s chest.
It is often quicker and easier for a physician to order a test than it is to explain to a patient why you are not doing a particular test. Tackling such a practical barrier is critical for minimizing test overuse. We’ve become so reliant upon testing without truly appreciating the harms of false positive and negatives that are associated with them.
Another problem to consider when practicing ‘easy medicine’, is the often ignored economic consequences. With excessive test ordering, physicians incur significant costs to the system, there is no data to have truly looked at this, but it would be interesting to observe if behaviour would be different if real tangible costs were felt by the physician or patient, for each test or treatment provided.
Patient expectations can also lead to over-testing, as many patients expect blood work, or imaging (i.e.: x-rays for back pain). Physicians may often worry that patient satisfaction, or belief in a clinical diagnosis hinges upon these tests.
However, ordering tests that aren’t indicated represents the difference between good and easy medicine. Just like it is inappropriate (yet still commonly done) to provide antibiotics for a viral illness, ordering tests that are not indicated for a patient’s presentation is just as potentially harmful.
It is important to appreciate that patient’s expectations about their healthcare is often based on their past experiences with healthcare providers (i.e.: they previously received antibiotics for their sinus infection, and so they expect it now). However, there is some excellent evidence that shows patient’s are very responsive to education regarding their healthcare and decision making11–13.
Another important (and often under-appreciated) concept is the idea of Baysian theory, an idea predicated upon two main principles. The first is that the probability of an event is based upon conditions related to that event (for example, an 23 year old male is unlikely to have cardiac disease, but an 76 year old diabetic, hypertensive male is). This context, is what is the most important thing for determining your pre-test probability.
The second factor is that tests have false positives and false negatives, so a positive (or negative) test doesn’t tell you if a disease is present or absent, it merely adjusts the probability that a disease may be there. How much this probability changes really depends on your pre-test probability (or, your context), and the accuracy of the test14–16.
Illusion of control (the Therapeutic illusion)
As a species, we tend to overestimate the effect our actions have, and tend to infer causality where none exists. Classically, one may think of a gambler believing that their actions may influence a slot machine or poker game. In medicine, this concept certainly exists – and is known as the therapeutic illusion.
When we believe particular treatments are more efficacious than they actually are, it can result in inappropriate and unnecessary care for patients. Take for example, prescribing antibiotics for a viral illness – after a few days, the patient is better, and both the physician and patient think this is a result of the antibiotic. It is more likely, however, that the patient had a viral illness, and the natural course was improvement in a few days. The big concern with the therapeutic illusion is that it is enhanced by the confirmation bias in which we tend to seek out evidence to support our erroneous conclusions.
Dogma and Pseudoaxioms
Change in medicine can be difficult, which is why historical thinking, dogma and pseudoaxioms have a tendency to persist. We often accept information handed to us from professors, lecturers and mentors without considering the evidence behind it. Take for example, the necessity of treating pharyngitis (viral or bacterial) with antibiotics – commonly done, despite a lack of evidence to support this practice.
Anecdotal evidence is commonly utilized in the practice of medicine, while it may have benefit in some circumstances, it has the potential to cause significant harms in the form of over-utilization of resources15,17. As an heuristic method of thinking, it is subject to various biases, the largest of which is the availability bias. When an infrequent event is very easily or vividly recalled (think, terrorist attacks or natural disasters), the power and frequency of those events tend to be over-estimated. An excellent example of this in emergency medicine is aortic dissection – an relatively rare entity with variable presentations and poor decision rules. Physicians often tell stories of doing an CT-dissection on an ‘hunch’ for a patient, and diagnosing an dissection. This then becomes rooted in their mind, and they continue to investigate with a fervour for dissection, ignoring their many negative investigations.
Often times, one of the main contributors or features that allows for continued propagation of over-investigation is that we are not aware our practice differs from our colleagues. Providing physician metrics is a way of improving upon this, but this is still a relatively novel concept that many departments are only beginning to do. Early evidence, however, suggests that a large proportion of test outliers are not aware of the fact that they are outliers in regards to their practice18–20.
There are numerous reasons for why we often over-investigate in emergency medicine. It is important that we begin to acknowledge and address this as a problem, because we are doing harm to patients in often subtle – but very real ways. There is a large amount of physician variation in regards to test ordering, and attempting to minimize this variation by recognizing correctable reasons for resource over-utilization by physicians will have significant and measurable impacts to patient care in the ED.