The Case Against Clinical Guidelines

Talal Hilal, MD


November 12, 2019

Jim was a patient I treated in 2017. (Names in this article have been changed to protect patients' privacy.) He had hepatocellular cancer that was progressing despite the best available therapies. The National Comprehensive Cancer Network (NCCN) guidelines were updated around that time, and an additional option became available—nivolumab, an immune checkpoint inhibitor.

After discussing this treatment option, I offered Jim this new drug, which provided new hope. He tolerated the first infusion but was hospitalized with colitis a few weeks later, a development attributed to nivolumab-related autoimmune inflammation. By the time his colitis improved, his liver disease had decompensated. The drug hadn't worked; Jim died soon thereafter.

The NCCN guidelines often indicate drugs in situations that are not supported by RCT data.

I questioned whether Jim would have been better served with supportive care instead of cancer-directed therapy. Although it had been demonstrated that nivolumab shrank some tumors in a minority of patients in a single-arm trial, whether it provided a benefit in terms of survival or quality of life for patients with hepatocellular cancer remained unanswered.

Evidence Too Late

Two years later, a larger, controlled trial failed to show improved outcomes with nivolumab for patients with hepatocellular cancer compared with sorafenib in the front-line setting. Would I have offered Jim nivolumab if that study finding had been available? Maybe. At the time, the benefit of nivolumab in hepatocellular cancer was questionable, but like many other drugs, it was a guideline-suggested option based on limited evidence, and its uptake was probably greater owing to its inclusion in the NCCN guidelines.

I couldn't help but wonder how often clinicians deviate from guidelines in their decision-making. Would I have deviated if I knew the data better?

The situation left me wondering whether propagating standardized clinical practice with little variability was truly in our patient's best interest.

Why Guidelines?

In the 1990s, studies of the effect of clinical guidelines on practice showed that guidelines improved compliance (ie, process of care) and, in some instances, outcomes of care.[1] However, the magnitude of benefit in outcomes was small.[2] Nevertheless, it was clear that clinical practice guidelines improved efficiency and value for healthcare systems across medical disciplines.[3,4]

In short, guidelines were benefiting policymakers, service organizations, payers, and regulatory authorities, but their benefit to patients remained suspect. However, with the expanding arsenal of new drugs and the need to summarize large amounts of information, it also became harder to stay informed on the latest, and perhaps best, approaches to management. Physicians seeing patients with multiple diseases found guidelines to be a helpful resource that presented all the information they were seeking in one concise document.

In oncology, the most widely used cancer guidelines come from the NCCN. Since the first publications in November 1996 on acute leukemia and breast, colon, lung, ovarian, prostate, rectal and pediatric cancers, NCCN guidelines now exist for virtually all types of cancer. Their impact on patient outcomes is less clear, owing to the heterogeneous nature of cancer. Some studies suggest that adhering to guidelines improves outcomes in specific cancers, such as gastric cancer,[5] whereas others have failed to show an improvement in outcomes in breast or lung cancers, for example.[6,7] Nevertheless, the push for standardization in order to reduce costs has resulted in the NCCN guidelines serving as one of five compendiums that private health insurers, and the Centers for Medicare & Medicaid Services (CMS), use to make coverage decisions.[8]

Given the effort by regulators, medical societies, and insurance companies to standardize practice, it is not surprising that guidelines have been widely adopted by practicing oncologists. But what has been lost in the process?

The Case Against Guidelines

Despite efforts by medical societies to maintain physician autonomy, clinical practice guidelines fundamentally limit individual physician autonomy; in essence, a few guideline authors dictate the decisions of the many. These recommendations are often viewed as the best way to practice. However, there are caveats that come with adhering to clinical practice guidelines in general, and to those in oncology in particular.

The fundamental mission of medicine is to care for the patient in front of you.

First, if you concede that randomized controlled trials (RCTs) are the gold standard for establishing a given intervention's efficacy, then you would expect that guidelines would recommend only those interventions tested in RCTs. After all, any trained physician can use their clinical judgment and the individual patient's circumstances in the absence of RCT data.

However, that is not the case with NCCN guidelines, which often indicate drugs in situations that are not supported by RCT data. In one study looking at 69 US Food and Drug Administration (FDA)-approved drug indications (mostly based on RCTs) between 2011 and 2015, the NCCN listed 37 additional indications that were not supported by RCT data.[9] These additional indications may be viewed as giving oncologists greater latitude when choosing a therapy. Indeed, CMS, and many private insurers, will not cover a drug indication that is not listed in the NCCN guidelines.

The downside to listing an indication not supported by RCT data is that it discourages further research. There is no incentive for sponsors to conduct additional studies if their drugs have already made it into national guidelines and are covered by payers.

Today's best practice may be abandoned tomorrow.

Second, by their nature, guidelines are typically filled with blanket recommendations that may be appropriate for the "average" patient. This undermines the complexity of medical decision-making, clinical judgment, and the unique needs of patients by introducing algorithms that limit care to a finite set of options. The tradeoff that comes with uniformity and standardization is a lack of individualized care. As oncologists, we can find loopholes to provide this individualized care, but why should we have to do that when the fundamental mission of medicine is to care for the patient in front of you?

Guidelines convey a false sense of certainty. In science, there is no absolute certainty. Today's best practice may be abandoned tomorrow. Guidelines that propagate treatments without acknowledging their uncertainty can be harmful as in Jim's case. This was also the case with Sarah, a patient with metastatic soft-tissue sarcoma whom I treated with olaratumab per guideline recommendations before the phase 3 RCT failed to show improvement in survival compared with standard chemotherapy. Similarly, we used bevacizumab for metastatic breast cancer before the FDA reversed their approval of that drug.

Bias in Guidelines

There is a degree of bias in any interpretation of scientific evidence, and there will always be data gaps in many areas of medicine. These factors make it difficult to develop guidelines that reflect these nuances.

Another layer of bias comes from the fact that oncology guidelines are often written by physicians with financial conflicts. A study in 2016 reviewed 125 guideline authors and found that 86% had at least one financial conflict of interest. These ranged from general payments for consulting, meals, and travel (average amount of $10,000) to research payments from industry averaging $236,000.[10] If we task writers with determining standard-of-care therapy for the nation, shouldn't we insist that they have no such conflicts?

The 'art of medicine' is increasingly being pushed out of the clinic room and replaced with algorithm-driven medicine.

This brings me to the last point. Despite the push for standardization, much of oncology seems to be conducted outside of guideline recommendations. Whether you are a proponent of guidelines based on the highest level of evidence (ie, RCT data only), or those that include lower-level evidence and consensus opinions, there is little doubt that at some point you will have to practice outside the scope of the guidelines. A quick poll on Twitter with approximately 330 votes from a sample of physicians in various medical specialties found that 52% deviate from clinical practice guidelines 10%-30% of the time, whereas 20% judged their deviation rate to be 30%-50%. At the further ends of the spectrum, approximately 15% deviate only rarely (< 10%), and 10% deviate commonly (> 50%).

On the basis of my practice and that of my mentors, I would judge the deviation rate to be between 30% and 50%. Often our initial therapy choice matches guideline recommendations, but this is because of our familiarity with published evidence, not the fact that it was recommended in the guidelines. Decisions for subsequent lines of therapy as well as dose modifications, treatment breaks, and supportive therapy often do not align with guideline recommendations.

What Should Be Done?

The best physicians know the evidence base for the guidelines and are able to judge that information and tailor it for their patients.

Guidelines might then be best viewed as a shortcut or a good starting point. They should not, however, replace careful, thoughtful medical decisions between patients and their physicians.

Standardization has made it difficult to deviate from guidelines because physicians are often viewed as providing substandard care when their decisions fall outside the scope of guidelines. The "art of medicine" is increasingly being replaced with algorithm-driven medicine, and the needs of individual patients are being overlooked when they fall outside of these standard treatments.

Perhaps disassembling the industrial-medical complex starts by viewing deviations from guidelines not as an irregularity, but as individualized and human-oriented medicine.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: