The trouble with guidelines

A friend of mine took her 98 year old, 98 lb mother to the doctor recently. Additional diabetic medication was prescribed because her A1c (also known as glycated hemoglobin -a proxy measure of average blood sugar levels over the previous months) was “not at target”. When my friend questioned the safety of yet more pills for this legally blind elderly woman, the doctor replied “I’m just going by the guideline!”

I have trouble with guidelines. Although many are sensible and many (not always the same ones) are used to establish public policy, it seems that their use too often leads us to overdiagnose and overtreat. It is hard for the average family doctor to feel confident that the agenda driving a guideline is purely what is best for the patient in front of her. Frequently the experts who write them are tied to industry, so immersed in their own culture that their recommendations are self justifying and self serving. Following their recommendations can lead to testing and treating too many, too often.

Studies like the United Kingdom Prospective Diabetes Study, which showed modest improvement in some of the complications of Type 2 diabetes over 10 years in a particular subset of patients (all between the ages of 48 and 64 at the beginning of the study) are used to justify medicalization of a far larger population. The glucometer makers, and pharmaceutical companies are happy to sponsor educational events, presentations to government etc to make sure no possibly diabetic person is missed or undertreated.

Even when it becomes clear that overtreating is not associated with improved outcomes, guidelines can be slow to reflect change in knowledge. The Canadian Diabetes Association guidelines still advocate for an A1c target of 7.0% although there is evidence from the ACCORD trial, the diabetes and aging study and subsequent metanalyses that a higher level of 7.5% is safer especially in the geriatric age group.

Family doctors are often berated by specialists, industry and the media for “not following guidelines” and it makes us feel badly. There’s an Ontario Health Insurance Plan payment for regular monitoring of people with diabetes (not yet one for achieving target numbers, but we have all been sent report cards). Despite increasing her risk of complications, having this woman spend a significant amount of her remaining life testing her sugars, taking more medication and going for the recommended visits to the clinic and laboratory, will directly financially benefit her health care team. And, when the team is reviewed for compliance with guidelines, they will look good. Of course she should have the  pills!

The comments section is closed.

  • Dr. Gaines says:

    Is there any wonder why students shun primary care? It’s because of this kind of stuff.

    We all went into medicine to use our brains to do the right thing. Being reduced to a robot whose sole purpose is to reach “targets” based on the “guidelines” is not what anyone signed up for. Many of our experiences on family medicine rotations as medical students echo what the author of the article struggles with: an expectation for blind adherence to guidelines for guidelines’ sake.

    I also suspect that there is an element of defensive medicine at play. If everyone follows the “guidlines”, they become the standard of care, yet if someone goes against them, and there is a complication, the lawyers could very well use that against them.

    There are really only a few fields in medicine that have their autonomy preserved so that patient advocacy is still possible. Family medicine is not one of them.

  • Mary Diwell says:

    Thank you Shelagh McRae! Now I know why my doctor has tried to scare me into becoming a diabetes patient. Money what else!

    My doctor, the drug industry and the labs all win. Who Loses? Me!

    You speak of harm to the doctor patient relationship. That is an understatement.

  • Jonathan Marcus says:

    Bingo! While reading your blog, I was planning on commenting on the financial interest that doctors will soon have to produce numbers that show they are meeting targets. Then you commented on that in your last paragraph.

    I want to emphasize that as soon as doctors have a financial interest in reaching a particular goal, this can (and I believe will) have the unintended consequence of acting in their own interest in conflict with the patient’s. Treating healthcare providers as automatons who’s aim it is to ‘reach targets’ could damage the doctor-patient relationship, which is at the heart of how doctors help influence patient’s health for the better. It’s taken thousands of years to build this universal relationship. I’m worried that this could be seriously damaged in just a few short years.

    • Andrew Holt says:

      I agree with the concerns Jonathan expresses that we need to pay more attention to how financial incentives for clinicians – (doctors, nurses, physiotherapists, ….. ) impact clinical and resource management decision making. The array of conflicting incentives currently in place does little to encourage evidence based decision making at the bedside and leads to considerable ethical peril for all … who’s interests are being served by day to day patient care decisions is fundamental to ‘the health of the health system’. In my opinion the further we stray from the compassionate application of best evidence in a trusted patient-clinician relationship the poorer the quality of health care we collectively provide. We do not support this well as a health system. Often this results in a lot of wasted effort, miscommunication, finger pointing and less productive behaviors which further erodes our collective ability to truly provide proper health care. My hope for all is that we can step back from the daily pressures, recognize this trend and systematically rebuild the capacity of the health care system to meet patients needs long into the future.

  • Chris says:

    For what it is worth….there were a few papers that addressed primary care adherence to guidelines for ADHD management a few years ago. I was tracking them for a project I did, and have added a few non-ADHD citations to this collection:


    In case useful/of interest to anyone here.

  • Andrew Holt says:

    Clearly the appropriate use of rigorously developed Guidelines by highly qualified clinicians with sufficient time and support to do this will result in the best clinical judgement and fewest errors. Not only are these errors avoidable clinically they cost a huge amount of money for unnecessarily extended care, repair or long term supports – in addition to the human toll on everyone involved.

    In an era of emerging austerity, I wonder what the value to patients, overall savings to the health care system and society would be if we consciously slowed the frantic pace that most clinicians and managers are forced to work at, systematically invested in rigorously developed guidelines where they will have the largest impact and ensured clinicians, leaders at all levels, boards and government policies aligned decision making to ensure wide adoption. An idea what the ROI would be?

  • Laurence Moseley says:

    The problem has two ends. At one end, the problems we have to face and the decisions that we have to make are much more complex than is commonly imagined. A relatively simple nursing decision (interpreting the Bristol Activities of Daily Living Scale) for example involves a search space of over 67 million decision points. Decisions are typically of the order of 10^6 or 10^7, but in one case with which I have dealt it was of the order or 10^29.

    At the other end, our unaided mental equipment for solving such problems, or taking such decisions, is limited by the capacity of our short-term memory to store, recall, and manipulate items. That has been measured many times as Miller’s Magica Number (7 plus or minus 2) i.e. is of the order of 10^0.

    This huge imbalance between the problem and our unaided mental solutions means that we are faced with a difficulty. Rather than using algorithms we instead choose heuristics – and fallible heuristics at that. In other words, we make mistakes. In hospitals, those have been well documented by the annual adverse incident data, in both the UK and the USA. These millons of misjudgements exemplify the imbalance – and recent retrospective clinical reviews suggest that the published patient safety data probably represent at most 12% of the real incidents

    One solution has been to use expert-systems software. These have been shown to produce statistical and clinical improvements in patient outcomes in over 90% of the cases where they have been used. That is because they use algoriths rather than heuristics and because they rely on computer memory and processing rather than the human variant. Overall, the results have been that of 136 studies comparing intuition/clinical judgement/heuristics with algorithms, only 8 found any advantage to the heuristic approach.

    Although the use of such systems improves practice in part by ensuring that all the relevant questions, but only the relevant questions, have been asked, their main advantage is that they handle the problem that there are often many confounding variables involved, and our mental weaknesses mean that we cannot give appropriate weights to the influential variables. ain the research is clear. Clinical data combined clinically makes the most mistakes. Mechanical data combined mechanically does much better. However (and this is the saving factor for “human” approaches) using both clinical and mechanical data but combining them mechanically produces the best outcomes. In other words, the clinical/intuitive data are valuable, but only when they are (a) combined with mechanical (hard) data and (b) only when they are combined algorithmically.

    The problems with most guidelines is that they are at one level only and variables do not interact. You cannot easily encode a search space of millions of items on a single sheet of paper. Guidelines should be automated wherever possible. Computers do not forget things and do not accidentally forget any special cases which have been encoded, however complicated those special cases are. It is now over 25 years since the acute abdominal diagnosis study showed a reduction of 50% in both false positive and false negatives.

    To reduce either is easy. To reduce both shows progress.

    The lesson is “Use your intuition but if you have access to some rule-based software, use that too. Put the two together and accept the recommendation of the software unless you have a good reason not to. The good reason will normally mean that you will put into the equation data items which were not encoded in the software.” T

  • Dmitri says:

    A very important issue. It is a disturbing trend indeed when doctors start relying on guidelines in a less that reflective, shall we say, manner. Guidelines are not, so far as I can surmise, universal laws and need to be applied contextually. Even if the guidelines are written with the best of intentions, and they rarely are, this does not mean that they necessarily apply in every single case. Unfortunately, its often easier to use them as crutches.


Shelagh McRae


Republish this article

Republish this article on your website under the creative commons licence.

Learn more