Recently, I posted on the surgical relevance of Daniel Kahnemann's well-known book, Thinking Fast and Slow and even more recently on his concept of "objective ignorance" in making predictions of treatment outcomes.
Now let's look at his last major book before his death one year ago by assisted suicide in Nunningen, Switzerland, choosing to end his life while still in relatively good health to avoid the potential suffering associated with aging.
He pointed out that where there is judgment, there is error.
Two contributors to error can be illustrated by the kicker's judgment exercised when trying for a field goal.
On the left we see Bias, a predictable, consistent deviation in a particular direction - perhaps due to wind direction, or way the holder positions the ball.
On the right we see Noise: imprecision, inconsistency, unpredictability, the lack of reproducibility.
An important difference between these two sources of error is that bias can be identified and corrected (by directing the kick a bit windward or changing the holder's positioning of the ball). The inconsistency of noise makes it harder to fix (why does the ball go too far to the left sometimes and too far to the right on other occasions?).
BIAS
What about bias in orthopaedic practice? Here are some possible examples:
*financial bias: a surgeon may consistently recommend surgery more often than non-operative care because surgery pays better.
*practice bias: surgeons working in the Kaiser system, a Veterans' Administration hospital, a private practice, a Public Health Service hospital and an academic medical center may be subject to consistent but different influences on how they practice. This type of bias may be further influenced by ownership of a surgery center, MRI, or PT facility and by the incentive system of the practice (whether based on dollars collected or on Relative Value Units (RVU)).
*selection bias: a surgeon may consistently favor different treatments depending on the patients' age, sex, ethnicity, or insurance (private, workers' compensation, medicaid, medicare, self-pay).
*training bias: surgeons having completed a sports medicine bias have been shown to be more likely to perform reverse total shoulders for cuff intact glenohumeral arthritis than surgeons having completed a shoulder fellowship.
*familiarity bias: surgeons may be more comfortable with an open bone block for instability associated with glenoid deficiency than with arthroscopic iliac crest grafting.
*availability bias: the tendency to treat a patient based on their similarity to a recent case rather than on the surgeon's overall experience.
*hindsight bias: distortion of judgment so that outcomes that could not have been anticipated appear easily "foreseeable" in retrospect.
There are circumstances in which bias is appropriate and necessary. It seems best of these biases can be stated in terms of rules. Rules can shared with patients so they do not feel that decisions are being made randomly. Here are some possible examples:
*Degenerative cuff tears are not considered for surgery until after 6 weeks of a defined therapy program
*Elective arthroplasty is not performed within 3 months of an intra-articular corticosteriod injection, or on patients who are actively smoking or on patients that do not have an adequate post discharge support system.
NOISE
A certain amount of noise may be acceptable. While this kicker's performance shows some noise, each kick landed in between the uprights.
Kahnemann refers to this as the "valley of the normal", meaning that outcomes within the valley are OK but those outside the valley are undesirable. Some examples: insertion of an implant within 4 degrees of the desired position can lead to good function; however when the implant is in 20 degrees of varus or valgus, the outcomes can be disastrous.
Importantly, the effects of noise do not cancel out. The average position of the six kicks is right in the middle of the uprights, but only 2 of the 6 kicks resulted in points for the kicker's team.
Varus and valgus errors do not cancel each other out either. 10 too tight shoulders and 10 too loose shoulders are not perfect on the average.
Kahnemann manages this phenomenon by using the mean of squared errors (MSE). The magnitude of each error is squared. This has two benefits: (1) the direction of the error doesn't matter (an error of +5 degrees gets the same weight as an error of -5 degrees) and (2) larger errors have a much greater effect on the mean than small errors.
Consider two surgeons. Note that they both have average errors in implant position of zero. However, because of the small magnitude of Surgeon A's errors, the mean of her squared errors is almost a tenth of the MSE for Surgeon B (who brags that his implants are, on average, in perfect position). Who would you have your relative see?
We can only conclude that noise is never good and that the ill effects of noise rise rapidly with the amount of individual deviation from the target.
Kahneman points out several types of noise
System noise – total amount of imprecision – the degree to which the treatment decisions vary – either within a surgeon (day 1 vs day 2) or among surgeons (A vs B)
Pattern noise – variations among the treatments different surgeons select for a given diagnosis. When a patient solicits a second opinion from a surgeon who is unaware of the first surgeon's opinion, the two opinions are unlikely to be the same. Anatomic vs reverse, PT vs surgery, ORIF vs arthroplasty, single row vs double row, biologics vs none, and so on.
Occasion noise – random differences in the treatments a surgeon selects for a diagnosis under varying circumstances (his mood, the time of day, fatigue, hunger, a recent fight with domestic partner, or the fiscal health of practice). Consider the following chart regarding judgments by two surgeons in the treatment of osteoarthritis with an intact cuff, cuff tear arthropathy with pseudoparalysis, cuff tear arthropathy without pseudoparalysis, massive irreparable cuff tear with pseudoparalysis and massive irreparable cuff tear without pseudoparalysis.
1 Note that there is substantial system noise (patients with the same diagnosis are getting different treatments).
2 Note that the patterns of the two surgeons are level but different.
3 Note that on different occasions an individual surgeon selects different treatments for the diagnosis.
It is understandable that this noise causes confusion on what to do for whom.
Our American Academy of Orthopaedic Surgeons has tried to come up with practice guidelines or "Appropriate Use Criteria (AUC)", but the results are not always helpful because of noise.
Here are three examples regarding the treatment of rotator cuff tears.
(1) Use of biologics (e.g., PRP) to improve healing: Limited and conflicting evidence; no strong recommendation for or against.
(2) Early vs delayed surgery in chronic tears: Evidence is limited; clinical judgment is important.
(3) Tendon transfers (e.g., latissimus dorsi, lower trapezius) for massive irreparable tears in younger active patients: Evidence limited; decision individualized.
One final point, not only does noise complicate decisions made by patients and by surgeons, it also complicates clinical research: if each case is a "one of" receiving "individualized" management based on their particular surgeon's judgment, how can we learn what works best for which diagnoses in which patients?
| | |
Let me end this noisy post with a final salute to the mothers of the world. Here's a photo of a red tailed hawk mom tending her chicks that I took on a recent trip to the Malheur National Wildlife Refuge in Oregon.
Here are some additional thoughts occasioned by a re-reading of this treasure. You can support cutting edge shoulder research that is leading to better care for patients with shoulder problems, click on this link
Follow on twitter/X: https://x.com/RickMatsen
Follow on facebook: https://www.facebook.com/shoulder.arthritis
Follow on LinkedIn: https://www.linkedin.com/in/rick-matsen-88b1a8133/
Here are some videos that are of shoulder interest
Shoulder arthritis - what you need to know (see this link).
How to x-ray the shoulder (see this link).
The ream and run procedure (see this link).
The total shoulder arthroplasty (see this link).
The cuff tear arthropathy arthroplasty (see this link).
The reverse total shoulder arthroplasty (see this link).
The smooth and move procedure for irreparable rotator cuff tears (see this link)
Shoulder rehabilitation exercises (see this link).
|
