Skip to main content

Why are we (doctors) deviating from guidelines?

Dr. Marc Wysocki
Medical Director and Intensivist - LCS Digital

Dr. Marc Wysocki

GE Healthcare

Though considering myself as curious and keen to adopt new technologies, I must confess that I was initially reluctant to use a GPS (Global Positioning System) when driving around in my area.

After some brief introspection (subject to my own memory bias) a few of reasons for my reluctance came to mind:

  • I believed that by growing up in that region, travelling by any means available (e.g. bike, car, public transportation, etc.) and still living in that region, my gestalt would make me an expert! After all, I must have enough tips, tricks, and shortcuts to perform better than a piece of software.
  • At that time, the user interface for such GPS were very basic. You had to stop the car to enter the final destination step by step and repeat this process several times before it would recognize the address! You also had to upload the map upfront.
  • At that time, GPSs did not provide contextual information like traffic jams, accidents, speed control, alternatives routes, or driving safety alerts. And, alternative routes with comparative times and distances were indeed not provided.

So, when they were first introduced, GPS solutions were not offering obvious additional value compared to my own expertise. Added to that, they were also quite expensive.

What made me change my mind and begin using a GPS, even for going to the bakery at the next corner (the most important place to go – almost twice a day – for a French citizen)? It had to do with advances in technology. But that wasn’t all.

  • GPSs today not only tell me where the bakery is, they also tell me also if the bakery is open, or where alternative bakeries are located
  • GPSs not only show me where I am on the map, they also tell me about traffic jams, accidents, and weather conditions. Additionally, they offer some alternative routes, and understand my preferences (time, distance, toll pay, etc.)
  • GPSs actualize the map much faster than previous generations
  • There is a community of users which help to improve real-time information (again: traffic jams, accidents, work on the roads, etc.)
  • GPSs today work on my smartphone, which means that I don’t have to manage another “box”
  • GPSs work virtually worldwide and acts as an invaluable guide in countries I’ve never been to before
  • It’s free of charge, displaying advertisements only when I’m waiting at the traffic light

Analogy never works perfectly well, but let’s try using that one to understand why doctors don’t like guidance.

The Guidelines…

In the early 1990s, the Institute of Medicine issued several reports on clinical practice guidelines. These have proliferated enormously in the ensuing years to the point that the Guidelines International Network’s database is listing today more than 6,700 guidelines (GIN 2015). 

According to the Institute of Medicine (2011) definition, clinical practice guidelines are statements that include recommendations intended to optimize patient care that are informed by a systematic review of evidence, and an assessment of the benefits and harms of alternative care options (IOM 2011).

To have clinical guidelines be even more efficient in improving clinical outcomes, the Institute for Healthcare Improvement recently created the “bundles,” which are defined as “A small set of evidence-based interventions for a defined patient segment/population and care setting that, when implemented together, will result in significantly better outcomes than when implemented individually” (Resar 2012).

So, bundles can be viewed as “super-powerful mini-guidelines” to boost patients’ outcomes when implemented and applied to individual patients.

Recommendations (IOM 2011) have also been given to make guidelines trustworthy. We can assume that they would make bundles trustworthy as well. Guidelines should:

  • be based on a systematic review of existing evidence;
  • be developed by a knowledgeable, multidisciplinary panel of experts and representatives from key affected groups;
  • consider important patient subgroups and patient preferences, as appropriate;
  • be based on an explicit and transparent process that minimizes distortions, biases, and conflicts of interest;
  • provide a clear explanation of the logical relationships between alternative care options and health outcomes, and provide ratings of both the quality of evidence and the strength of the recommendations;
  • be reconsidered and revised as appropriate when important new evidence warrants modifications of recommendations.

Assuming that the above “trustworthiness” is achieved, all physicians should have enthusiastically embraced the existing guidelines and bundles. Almost all patients, with few exceptions, should be managed per guidelines and bundles.


The Reality… (How much are we using guidelines?)

Among many studies published on this topic, (Garcin 2010, Leone 2012, Sevranski 2015) we will just pick a recent one (Pun 2019) which evaluated the relationship between bundle performance, defined as a patient-day in which every eligible element of the bundle was performed, as well as patient outcomes in the Intensive Care Unit (ICU).

This was a prospective U.S. multicenter cohort study that included 68 academic, community, and federal ICUs and over 15,000 ICU patients. Nearly 50,000 patient-days of data.

Good news from that study:

Patients with complete bundle performance (e.g. the seven elements of that bundle being performed) had a higher likelihood of ICU discharge and hospital discharge, and a lower likelihood of death at any given time compared with patients who did not receive the complete bundle.

It confirms that bundles are a super powerful tool to improve patient outcomes.  Well… at least when fully completed.

And that’s where the bad news comes in:

Only 8% of the patient-days had a complete bundle performance! That’s not a typo. The “0” is not missing. That’s not “80%”! You read it properly, 8% (eight percent)!!!

Which means that less than one in every 10 patients eligible for a bundle had that bundle completed.

When individual elements of the bundle are analyzed, the percentage of completion for each element ranged from 29% to 77%.

Fortunately, there’s a dose-response relationship here. Every increase in individual element completion independently predicted an improvement in the patient’s outcome, which means that 50% completion is better than 40% which is better than 30% in terms of clinical outcomes.

In terms of our GPS analogy: we have now at our disposal a super powerful solution ensuring being on time when driving from point A to B, with the optimal distance-timing combination, with all information needed regarding traffic jam, accident, etc… all this being proved by evidence, free of charge. Yet, for some reasons, very few (only 8%) are using it!

Why are we deviating from clinical guidelines and what it tells us?

Deviation can be described as a failure to follow, intentionally or not, all or part of the recommendations included in the guidelines.

Barriers to following guideline recommendations are multifactorial, related to the guideline itself when items of the above “trustworthy” list are not 100% fulfilled, but also related to the clinicians themselves.

Over 20 years ago a study was already investigating the barriers to physician adherence to clinical practice guidelines (Cabana 1999). Based on a systematic review, including 76 articles, they found plenty of “good reasons” for clinicians not following guidelines. For example, a lack of awareness (“I didn’t know that was a guideline!”), lack of familiarity (“I don’t know what’s inside the guideline”), or lack of agreement regarding the guidelines (“I’m not too sure it’s going to improve my patient’s outcome”).

Deviation from guidelines may also have to do with clinicians’ interaction within the organization. More precisely, it might be an unconscious way for the clinician to tell the organization that something is going wrong.

Recently a PhD thesis (Lépront 2019) was investigating that approach from an organization/management perspective, and using surgeon’s deviations in the OR as a use case.

Freely adapted from that document, there might be four types of deviations with different meanings:

  • The Functional Deviation
    The functional deviation allows the clinician to mention that there’s a gap between the guideline and the real world. For example, “This guideline is telling me to complete a scan within 60 minutes, but I don’t have a scanner available right away and within that time frame.” That’s probably a common reason to deviate from a guideline. Guidelines must be tailored to the local organization, workflow, resources, etc. Most of the time this isn’t occurring. The guideline is being imposed with a top down approach “as is” (“one rule fits all”), which allows the organization to check the box, “Yes, we have that guideline in place.”

  • The Cultural Deviation
    The cultural deviation allows the clinician to mention that there’s a gap between their cultural and professional values, and those of the guideline. For example, “This guideline is asking me to shorten an antibiotic duration to reduce the cost. My very first concern is about the patient and not about cost…” This is about education and values. To stay on that example, significant efforts have been made in the last decades to make clinicians more accountable for expenses and to apply the most cost-effective solutions. However, if asking a young doctor why they chose to study for 15 years to be an intensivist, they will not mention “cost reduction” as the very first reason.

  • The Political Deviation
    The political deviation allows the clinician to say that they want to maintain their own decision making and autonomy. And not only for the sake of keeping some form of political power. For example, “I’m the only one able to see the full picture for that patient, including clinical, functional, relational, cultural, and educational aspects. All this will drive the right decision.”

    With guidelines, and these days with sophisticated artificial intelligence on board, we try capturing as many variables as we need to provide guidance. However, even with the best I’ve seen so far, there’s always something missing. Something that has to do with human interaction. Something we call in medicine “in colloquium unique” between the patient and the doctor. All the information you get from that “colloquium” is difficult to translate into guidelines and equations.

  • The Economic Deviation
    The economic deviation, when the clinician wants to maximize their own personal interest and pleasure. For example, “I don’t have time to do this, because I need more time for doing that (which is more interesting for me).” What else can be more human than this? As an example, one of the most boring and time-consuming activities is EMR documentation. Doctors may believe that documentation is not worth losing time for patient investigation. Interestingly, to improve documentation completion and maybe guidelines adherence, new jobs have been created, like Medical Records Clerks (Greiver 2011).

Better understanding the reasons behind guideline deviations may help us investigate adequate strategies to reduce deviation. A “deviation rate” might also be used as a key performance indicator, or as a wellbeing/well working indicator for intra- or inter-organization benchmarking.

Functional, Cultural, Political and Economic deviations from guidelines
Better understanding the reasons behind guideline deviations may help us investigate adequate strategies to reduce deviation
Freely adapted from Lépront (2019). PhD thesis available on Last access Oct 12, 2020

Additional provocative thoughts…

Should we promote deviations from guidelines?

A paper from the Institute for Healthcare Improvement, recently published as a view-point in the JAMA (Berwick 2017), had a striking title: “Breaking the Rules for Better Care.”

The paper was not supporting deviations from guidelines, but it clearly suggested that some rules (a rule being somehow more than a guideline) may generate unwanted burdens and complexity, wasting time and ultimately detracting physician’stime (and morale) from patients.

The paper reported the efforts undertaken by a network of 40 North American hospitals and clinics to identify rules with little or no value to patients and staff. In the top 10 identified (among the 342) some were having a direct impact on the patient condition, such as visiting hours, sleep disruption, and patient mobility. Contrary to the initial expectations, almost 80% of such wasteful rules were fully within the administrative control of health care executives and managers to change. And some of them finally did change.

This to say that rules and guidelines are not carved on stone. They should be revised, revisited, and re-discussed in a sort of lean approach to improve patient satisfaction and care outcomes while reducing costs (NEJM Catalyst 2018).

Deep investigation on guideline deviations may help improve guideline operability. And who knows, maybe also generate some unexpected innovation.

How can we make guidelines even more trustworthy?

In line with a relatively isolated/ignored paper (Kavanagh and Nurok 2016), we have to admit that guidelines (as opposed to drugs or other interventions) are seldom subjected to rigorous testing. Instead, they are often implemented on the basis of belief or the results of simplistic “before and after” studies. The authors described two concepts,

  1. Protocol misalignment, or a mismatch between the context in which a protocol is developed and the context in which it is implemented, and
  2. Protocol misattribution, or a mismatch between the proposed—versus actual—reasons offered to explain how a protocol resulted in improved outcomes.

They suggested applying guidelines to the same standards of proof as other interventions to increase insight and help ensure “true” patient benefits.


Deviation from guidelines can tell us about the “real life.” More precisely, they can tell us about human interaction within complex organizations like healthcare systems. It tells us that even a sophisticated guideline can hardly capture all the complexity of a patient, of healthcare working conditions and of the local environment.

Guidelines, even those that are well designed, and follow all the best practices and recommendations (guidelines to design guidelines?), might not be enough alone to change practices, to improve care consistency and to ensure the best therapeutic strategy for a given patient, just like GPSs started offering contextual information that increased their value, guidelines can become more valuable if delivered through tools that can incorporate clinical insights and provide situational awareness. Imagine the clinical equivalent of alerting a clinician to a “alternate route” or even a “traffic jam.” Interestingly, in the 2011 publication (IOM 2011) the Committee on Standards for Developing Trustworthy Clinical Practice Guidelines (CPGs) had a dedicated chapter on computer-aided clinical decision support, often based on the translation of guidelines to facilitate a more personalized and timely form of guideline-based care. I will quote here the committee recommendation on this topic:

The committee recommends that guideline developers and implementers take the following actions to advance this aim. Guideline developers should structure the format, vocabulary, and content of CPGs (e.g., specific statements of evidence, the target population) to facilitate ready implementation of electronic clinical decision support (CDS) by end-users. CPG developers, CPG implementers, and CDS designers should collaborate in an effort to align their needs with one another.

One would say that patients and hospital organizations are much more complex than a GPS and real time traffic information. Well, I’m not so sure. Funny enough, we may even have some interactions between the two systems, like when an ambulance driver needs to get to the hospital. The driver needs guidance to reach the hospital as fast as possible, in a safe and un-bumping route, while the paramedics inside the ambulance need guidance to start the right therapeutic strategy for that patient.  

Finally, we should also change the collective perception of being guided. Being guided should not be a sign of weakness or frailness, but should be seen more as a tool for being successful, a tool to better manage patients collectively, and somehow a strength. It should be seen as everyone on the team guiding each other to reach the best outcome for the patient. Guidelines should be a companion that are simply part of the “team.”

That’s a deep cultural and even philosophical paradigm that needs to be challenged every day. Collectively we are always stronger than as individuals. To quote this Kenyan proverb:

“Sticks in a bundle are unbreakable.”


• Institute of Medicine 2011. Clinical Practice Guidelines We Can Trust. Washington, DC: The National Academies Press.
• GIN 2015 report available from Last access: Oct 12, 2020
• Resar R ad al. IHI Innovation Series white paper 2012. Available from Last access oct 12, 2020
• Garcin and al. Non-adherence to guidelines: an avoidable cause of failure of empirical antimicrobial therapy in the presence of difficult-to-treat bacteria. Intensive Care Med 2010 ; 36: 75-82.
• Leone and al. Variable compliance with clinical practice guidelines identified in a 1-day audit at 66 French adult intensive care units. Crit Care Med 2012; 40: 3189-95.
• Sevranski and al. Protocols and Hospital Mortality in Critically Ill Patients: The United States Critical Illness and Injury Trials Group Critical Illness Outcomes Study. Crit Care Med 2015; 43: 2076-84.
• Pun and al. Caring for Critically Ill Patients with the ABCDEF Bundle: Results of the ICU Liberation Collaborative in Over 15,000 Adults. Crit Care Med 2019; 47: 3-14.
• Cabana and al. Why don’t physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999; 282: 1458-65.
• Lépront. PhD thesis available on Last access Oct 12, 2020
• Greiver and al. Using a data entry clerk to improve data quality in primary care electronic medical records: a pilot study. Inform Prim Care 2011; 19: 241-50
• vYan and al. A Quality Improvement Project to Increase Adherence to a Pain, Agitation, and Delirium Protocol in the Intensive Care Unit. Dimens Crit Care Nurs. 2019; 38: 174-181.
• Ebben and al. Effectiveness of implementation strategies for the improvement of guideline and protocol adherence in emergency care: a systematic review. BMJ Open 2018 ; 8: e017572.
• Trogrlic and al. Prospective multicentre multifaceted before-after implementation study of ICU delirium guidelines: a process evaluation. BMJ Open Qual. 2020; 9: e000871.
• Trogrlic and al. Improved Guideline Adherence and Reduced Brain Dysfunction After a Multicenter Multifaceted Implementation of ICU Delirium Guidelines in 3,930 Patients. Crit Care Med 2019; 47: 419-427.
• Berwick. Breaking the Rules for Better Care. 2017; 317: 2161-2162.
• NEJM Catalyst April 27, 2018 available on Last access Oct 12, 2020
• Kavanagh and Nurok. Standardized Intensive Care. Protocol Misalignment and Impact Misattribution. Am J Respir Crit Care Med 2016; 193: 17-22.