January 3rd, 2014

2013: The Year of the Guideline

“I always pass on good advice. It is the only thing to do with it. It is never of any use to oneself.” — Oscar Wilde

This year culminated in an exciting flurry of academic activity. Long awaited guidelines for the treatment of cholesterol to reduce the risk of atherosclerotic cardiovascular disease and for the management of hypertension were finally released, both after agonizing delays.

While both guidelines have been mired in controversy, regrettably being undermined by some experts, they represent a sea-change for the field of preventive cardiology. Both place primary emphasis on evidence from randomized controlled clinical trials. This threshold means that the evidence base for clinical guidelines has never been stronger. The guideline writers should be commended for their efforts. Indeed, significant improvements in the health and well-being of our patients (and the population as a whole) are likely to result. For example, the use of evidence-based statin therapy should increase dramatically and reductions in atherosclerotic cardiovascular disease-related mortality are likely to ensue.

However, the guidelines have also inflicted significant collateral damage. The loss of LDL-Cholesterol targets and the increase of systolic blood pressure therapeutic thresholds (to 150mmHg for many patients) have shaken the very foundations on which many of us stand. One can only imagine what forefathers like Nikolaj N. Anitschkow would think of these new developments!

While full details of pros and cons of the new guidelines are outside the scope of this piece, it is disconcerting to realize that we appear, as a medical community, to have abandoned the idea of incorporating the full breadth of evidence-based medicine (including observation studies) into our decision-making armamentarium. When Francis Bacon first developed the scientific method, he proposed that causality can be approximated by interpreting the results of one’s observations. Thus, the very foundations of science were laid on the wide breadth of observation, not the narrow focus of randomization.

Admittedly, one must acknowledge that requiring randomized evidence as the basis for guidelines is the most robust strategy to deal with ever-present confounding and to ensure that recommendations are fully compatible with scientific truth. For example, no longer will physicians feel compelled to get patients to an arbitrary LDL-Cholesterol “target” with medicines that may have deleterious “off-target” effects.

However, multiple problems also arise with this thinking. For instance, at the most basic level, requiring results from randomized trials as the evidence base for guidelines is a problem when most clinical questions have no randomized evidence base in the first place. Thus, it appears that we will need many more randomized trials in the future. Indeed, the whole clinical research enterprise will need a major overhaul. Massively scalable mechanisms to conduct simple and varied randomized studies in “usual care” environments, as part of routine clinical workflow, are now urgently needed. This is the model of the learning health-care system.

For example, methods for embedding randomization within the clinical workflow of providers who use electronic health record systems are already being developed. However, besides cost, multiple hurdles obstruct progress in this field. These hurdles are significant and will require time and effort to overcome. Examples include difficulty acquiring streamlined consent given the current regulatory environment (which has not adapted to modern technology or comparative effectiveness research), poor physician motivation to sacrifice ever-precious time and effort to research as part of “usual care”, poor perception of research amongst patients, and a research hierarchy that does not engender inclusiveness (particularly with those non-academic providers who will be asked to contribute to this effort).

It is clear that new ideas and disruptive innovations will be necessary to make this learning health-care system a reality. However, once realized, it is possible that this reality could generate the volume of randomized evidence now necessary in the current era of cardiology guidelines. Of course, it would be folly to presume that we can perform randomized clinical trials for every single clinical scenario. No matter how much we think we can know, it is also unlikely we will ever have generate enough evidence such that there is no room left for different interpretations of this evidence. Disagreement will continue, indeed science depends on it. Such is the nature of inherently subjective human interpretations of data. Thus, randomization is not the panacea for our incomplete understanding of scientific truth, and we need to be careful not to become too inflexible to other (non-randomized) sources of data.

Personally, I think our current fixation on “hard facts” (evidence) from randomized clinical trials at times borders on utilitarian. I am reminded of Mr. Gradgrind, the headmaster from Charles Dickens’ Hard Times. In a famous passage, a visiting official asks Gradgrind’s students:

“Suppose you were going to carpet a room. Would you use a carpet having a representation of flowers upon it?”

The character Sissy Jupe replies, ingenuously, that she would because, “If you please, sir, I am very fond of flowers.”

“And is that why you would put tables and chairs upon them, and have people walking over them with heavy boots?” the official retorts.

“It wouldn’t hurt them, sir. They wouldn’t crush and wither, if you please, sir. They would be the pictures of what was very pretty and pleasant, and I would fancy -“

“Ay, ay, ay! But you mustn’t fancy,” cried the gentleman, quite elated by coming so happily to his point. “That’s it! You are never to fancy.”

“You are not, Cecilia Jupe,” Thomas Gradgrind solemnly repeated, “to do anything of that kind.”

“Fact, fact, fact!” said the gentleman. And “Fact, fact, fact!” repeated Thomas Gradgrind.

In conclusion, I hope that in 2014 we will not be consumed by the need for randomized “facts”, that the full breadth of evidence-based medicine will continue to be valued, and that guidelines can be seen as scientifically sound “advice.” And like all good advice, guidelines should be passed on and utilized wherever possible, as Mr. Wilde advises above. However, it is worth reminding ourselves that guidelines are not absolute rules. Thus, if the fully informed physician feels this “advice” does not apply to (or is of no use for) a specific patient, then they should continue to feel free to use clinical reasoning (inclusive of observational evidence, but with appropriate caution) to manage their patient.

4 Responses to “2013: The Year of the Guideline”

  1. I found this blog to be dispiriting because, despite its erudition – and the clear intellect of the writer – it is so off target (IMHO). The guidelines are not about rejecting the cholesterol hypothesis – that is secure (Anitschkow would not be bothered) – or the indifference of these experts to various forms of evidence – it is in response to evidence that has failed to show that lower is always better that has produced recommendations that are more anchored in evidence. It is not the absence of the evidence – but the evidence itself. And the recognition that any medication has unpredictable effects likely because it has such a spectrum of influence on physiological processes beyond a single set of biomarkers. What this post demonstrates to me is that there remains considerable confusion about why the last three guidelines from the ACC/AHA – and now the KDIGO guidelines from an international group – have abandoned the idea of targets.

    And I just have to share this story with someone….I talked with an 84 year old patient yesterday who asked my opinion about a prescription he received from his primary care doc for simvastatin – he has no hx of vascular disease, no risk factors, his LDL is 130 and HDL is 45 – and his doc told him that he should have an LDL less than 100.

  2. John W McEvoy, MB BCh BAO says:

    Many thanks to Dr Krumholz for contributing his considerable insight, intellect, and experience to this debate. Unfortunately, my piece was really not about LDL targets at all. I see now that the reference to Anitschkow may have led one down such a path, but as I state above “full details of the pros and cons of the new guidelines are outside the scope of this piece”. My intention with Anitschkow was to highlight how much the new guidelines shook the field of lipidology (I think it is fair to say that they did).

    However, the primary intention of this piece was to discuss how the new guideline focus on randomized evidence will require new innovation in the Clinical Trial Enterprise and is likely to challenge long-held inculcated perceptions of truth that have historically been based on very compelling observational data (data that I think should remain of value). I want to reiterate that do not think randomized studies can answer every question and we should not exclude non-randomized data ‘a priori’.

    In summary, my piece is meant to be primarily about 1) the increasing exclusion of non-randomized evidence (no matter how compelling) from guideline recommendations, 2) the consequent need to generate more randomized data and ways to do it, and 3) the fact that clinical reasoning (inclusive of compelling observational data) should still take precedent in the right scenario.

  3. Joel Wolkowicz, MDCM says:

    Dr. Krumholz – at least your patient didn’t get ezetimibe.

  4. Joel: thanks – that made me smile.
    And Bill – thanks for your response – we are in total agreement about the need to learn from different types of evidence. Thanks for engaging in the discussion.