On the making and taking of recommendations
Every day we hear about some new medical recommendation in the popular press. Usually what we get is an oversimplified report of the workings of a prestigious and thoughtful group, the United States Preventive Services Task Force (USPSTF). This august group has the incredibly challenging mandate of making preventive medical recommendations to hundreds of millions of strangers.
"Created in 1984, the U.S. Preventive Services Task Force is an independent, volunteer panel of national experts in prevention and evidence-based medicine. The Task Force works to improve the health of all Americans by making evidence-based recommendations about clinical preventive services such as screenings, counseling services, and preventive medications." The key to their scope is prevention. Put another way, these are the things that you should do BEFORE something really bad happens to stop that bad thing from happening. It sounds like an easy task, but it turns out prevention is a tricky thing.
First and foremost, all preventive steps have some element of cost. Costs look different to different parties but can be broken down into:
$: financial cost for a service
Risk of badness: every medical test has an element of risk from anxiety whilst awaiting results through death from complications of a procedure.
Patient time
Provider time that could be spent doing something else (we could argue that based on the average paperwork load of a primary care doctor, we value their time close to zero).
Health system resources to be used elsewhere-- there are only so many colonoscopes and so many colonoscopy suites and so many colonoscopists to go around
But what is the upside of prevention? It turns out that prevention, in the clinical sense, is only usefully applied to conditions that have the following characteristics:
High prevalence disease
A prolonged, asymptomatic phase
Effective treatment
Treatment is more effective before developing signs and symptoms
In other words, only common conditions that you can have without knowing, that we can treat, and treat early, fall under the category of preventive recommendations. Similarly preventive tests are only useful when they:
can detect a high proportion of disease in said asymptomatic phase
are safe, relatively inexpensive and widely available
lead to demonstrated improvements in health (as opposed to lead time bias)
It turns out that the USPSTF ends up recommending fairly few things, and they are worth reviewing to understand what the medical world means when they talk about prevention, as opposed to what the general public means (diet, exercise, seat belts, helmets, etc.). I find in my own practice that different definitions of prevention lead to tremendous stress and disappointment on the part of my patients.
As stated earlier, the USPSTF must make recommendations to strangers, that is to say, it must advise doctors and patients whose individual circumstances cannot be known. This burden is a hard one, and it leads to considerable conservatism on the part of the organization. There are other similar organizations recommending guidelines, standards and evidence review for all sorts of conditions.
The proliferation of guidelines (and payment attached thereto) has been variously hailed and lamented as the future of medicine. Coming of age in the era of evidence-based medicine, I have long struggled with the pros and cons of guideline-based medicine. Rather than try and resolve the answer, I'd like to come at the problem orthogonally considering just how recommendations are used and useful based on the recommender and the recomendee.
Population level recommendations
These are the province of organizations like the USPSTF. The recommender is often expert, reviewing reams of evidence of variable quality. The recommender NEVER knows the individual facts or case of the recommendee, who is effectively the union of patient and provider. The recommendations ought to be conservative because of their scale and anonymity. The Ecological Fallacy comes in to play as well-- observations about a group cannot be attributed back to an individual in the group.
At this level, recommendations are exactly that-- recommendations. They are incredibly valuable so that we do no harm but will always need to modified by individual providers and patients for individual circumstances. Without population level guidelines, we would do really, really stupid things to our patients, and/or miss out on really obvious things.
All of the bristling at population level guidelines come when measurement and its cousin, payment, become attached. Trying to ensure that their populations get the best possible care, insurers have attempted to tie payment to providers to test completion levels. While well-intentioned, this can be incredibly dangerous as discussed by Hartzband and Groopman above. One way to circumvent this issue is to not measure testing rates, but rather measure shared decision rates. That is, how many patients were offered a real choice in the matter. Whatever patients ultimately decide, they have to live with the consequences. While this may be harder to capture than straight mammography rates in an eligible population, if we are committed to excellent care, then it is worth the challenge.
Panel recommendations
We now pass into the realm from an organization recommending testing to strangers into the realm of relationships. A panel is a group of patients attributable to an individual provider (or clinical team). It is a list of individuals who are theoretically known to said clinical care team. As you can see, we have just moved from the unknown into the known. Many organizations will begin to recommend treatment pathways for their specific populations that take the best of available evidence and match it with the characteristics of a given population. For instance, foot ulcers are rare in the general population, but relatively common in people with diabetes. Therefore, where looking at everyone's feet may not make sense, looking at diabetic individual's feet makes a ton of sense.
It is at the level of panel recommendations that clinical organizations can combine the best of available evidence with intimate knowledge of their own patients. Here organizations can be a bit more proscriptive to ensure that proper care happens quickly. Additionally, through the creation of great protocols/pathways/internal guidelines, a larger care team (beyond physicians and other front line providers) can be brought to bear to ensure excellent care.
The owning of recommendations to specialized sub populations may be the greatest area for growth in clinical medicine, and the promise of all of this data we seem to be collecting across the country.
Individual recommendations
Traditionally the stock and trade of physicians and other providers, individual recommendations are the apex of great clinical care. The basic truth of all of the evidence out there is that your patient just does not meet the specific criteria. Studies of diabetics may exclude those with other conditions, heart disease may cut off at age 65. Just what are we to do? And the common the rallying cry of beleaguered providers-- "My patients are sicker!"
The individual recommendation remains firmly within the Art of medicine. Here, patient and provider must together combine everything that is known about diseases, treatment, all of the evidence assembled, the individual combination of life and health that make up that patient, and actually make a plan that can work. In my opinion, this can be the most fun part of medicine because it is where the rubber begins to meet the road. We can also try things, measure if they work, and try other things. Where one medicine fails, another may succeed. Where one sleep apnea mask causes agony, another may bring much needed restful sleep.
Individual knowledge and recommendations are the safeguard against the errors inherent with amassing giant pools of data and distilling specific behaviors. Nevertheless, to work on individual judgement alone is to forsake great care for our patients. Similarly, we are professionals, bound by ethics and law to recommend things that are justifiable given clinical circumstances. We are limited in our ability to make certain recommendations especially when the evidence gets slim and/or the risk to patients becomes low.
Exercise and diet are great examples of this challenge. How many times have you had a provider recommend changing your diet and/or exercising. Beyond that, they may recommend one of a few well-studied diets. There is a problem here, and it is not your doctor's fault.
First, diets are hard to study. Second, the risk of trying to eat differently are very low. This intersection of hard to study and low risk in individual experimentation fall out of the bottom of what providers can comfortably recommend. Fundamentally, we don't know, because knowing is hard, and finding out on your own is not particularly risky. Painful, annoying, time-wasting, but not risky. And remember, we first do no harm.
Individual Action
This brings us to the ultimate level of recommendation--what each of us decides to do. Some panel in Washington may recommend screening for HIV in all individuals once aged 18-65. Our doctor may be urged by their employer and guided by their own beliefs to recommend the same. However, we each must act and make the choice to do or not do what is recommended. It is in this responsibility that our cynicism protects us from the errors of the above recommendations. It is also in this responsibility that our inaction causes us to miss out on what could be life-saving interventions.
Each of us has the agency, the authority to decide to do or do not with any recommendations we receive. We also have the power to experiment on our own with exercise, with diets, with life hacks that may make our life better. In the end, no matter who recommends what, in the immortal words of Steve Winwood, "While you see a chance take it ... Because it's all on you."