Sunday, April 20, 2014

Overprescribing Bubble Diagram Explained

This post is intended to explain the bubble diagram that I used in the last post about overprescribing.  It focuses on the physician rather than a number of extraneous factors that are the speculative causes of overprescribing that do not make sense.  The literature on this is spotty.  There are a few references that include diagrams and purported mechanisms of overprescribing.  Before I consider those, I want to put down what I know.  The bubble by bubble explanation follows starting at the top left of the diagram - but each bubble is easily identifiable:

1.  Unconscious Motivations:  This is one of the areas that I consider to be the primary cause.  It encompasses both unconscious and preconscious causes.  The best way to look at it is with a few examples.  Overidentification with the patient seems to be a fairly common cause.  For example, a physician might have attention deficit hyperactivity disorder and realize that they were able to do much better academically.  They have a lower threshold for seeing the problem in others based on their experience.  In their medical group they are seeing many more ADHD patients than their colleagues and prescribing far more stimulant medications.  That is really a special case of the physician wanting to do something positive for the patient.  That can lead to a reduced awareness of other causes such as substance abuse or a person seeking cognitive enhancement and drive the prescription process.  The wish to be appreciated and have a positive relationship with the patient is another issue.  Consider the patient in clinic an uncomplicated upper respiratory infection.  Near the end of the clinical encounter, the patient senses that there will be no prescription and starts to get annoyed.  The patient talks about his or her past experience and how an antibiotic always seems to clear up their cold symptoms.  They put the physician on the defensive about the antibiotic issue and rather than argue with the patient - the physician hands over the prescription.  All of these scenarios and more reflect unconscious factors involved in prescribing and as far as I can tell this important area is untouched in the overprescribing literature.

2.  Medical knowledge:  This occurs at two levels.  The first is just the rote declarative knowledge that involves the memorization of factual knowledge about medications.  This seems like a fairly basic skill, but the amount of information we are talking about is encyclopedic.  To illustrate this all one has to do is to pull up recent FDA approved package inserts for drugs that are advertised on television.  All of the pharmacological details are presented in a neutral and highly detailed manner.  There is some prioritization in terms of  contraindications and precautions, but that is still a lot to memorize.  Experienced clinicians will still read these documents repeatedly and use summary references until they gain enough experience with the medication.  The second level is the patterns that become familiar when using the drug.  For example, after I have prescribed a drug enough times, I can tell anyone taking it what my experience with the medication is and what the experience of my patients has been.  That information is really not available anywhere else and it also can lead directly to additional testing.

3.  External validation:  The best objective source of external validation is prescribing medications according to the FDA approved indications and as instructed in those documents.  Off labels prescriptions should be backed up by some data. I typically research all off label prescribing by doing Medline searches and looking for actual clinical trials looking at that application.  Example would include looking at the use of gabapentin for chronic pain and treatment of alcohol withdrawal.  Even though there are no FDA approved applications, there is ample evidence about how to use these medications, including very specific dosing recommendations.

4.  Conservative prescribing bias: My goal has always been to treat the primary problem but also eliminate or minimize problems related to the medication.  That is not the goal of some physicians.  I heard an expert in the treatment of anxiety disorders suggesting that patients needed to be treated to the point of toxicity with a medication and then the medication could be reduced.  That is not necessary as far as I can tell.  Another issue here is exceeding the FDA approved range for medication use.  Quality markers for dose range and multiple medications from the same class are nothing new and date back at least 20 years. This information that is also part of informed consent and the patient should hear about it and other options at the time the change is being considered.  I have spent a lot of time correcting unnecessary high doses and polypharmacy, generally to the benefit of the patient involved.

5.  Social context:  A group of interested physicians who keep up on the research and analyze prescribing patterns is a robust safeguard against overprescribing. There are some setting for example where benzodiazepines or benzodiazepine related sleep medications are not prescribed in any circumstance.  Stimulant medications are used only by review with a colleague and the medical director and documentation on a signed form.  Such a review procedure is available to any group has utility in preventing variation in prescribing patterns and practice variation.  Practically every major clinic and hospital has a program that targets antibiotic overprescribing especially the overuse of expensive, toxic or antibiotics that are the last defense against bacteria with multiple drug resistance.  This social factor is a powerful safeguard against overprescribing.

6.   Physician as a source of information and teaching:   If the physician is in a position to teach patients about a medication they generally have a more nuanced knowledge about the medication.  They avoid stereotypes like: "When I see this problem I always prescribe this drug."  They are also aware of the likelihood of the patient recovering without the drug and the fact that any drug effect make be weak.  A classic example is low back pain in adults.  Patients generally lack knowledge about the importance of avoiding deconditioning and the need for ongoing physical therapy and back specific exercises.  They are generally surprised that conditioning provides a significant amount of pain relief.

These are a few brief examples about what can be done.  All of these dimensions occur in the context of political and regulatory influences.  For example, the FDA initiatives to approve potent opioid medications that are essentially not much different in terms of risk from current medications during an opioid epidemic.  Guidelines suggested by outside sources may not be very useful if there is insufficient evidence and the information is broadly disseminated to clinicians.  The bubbles listed on the diagram are potent factors in countering any trend that may lead to oversubscribing and they do not involve penalizing or threatening physicians - typical actions used by law enforcement or politicians that may have caused the problem in the first place.

The reader will note that I am not concerned about diagnostic proliferation.  I don't think that overprescribers do this based on the availability of new diagnoses.  I base this on my observations of the patterns that exist independently of new diagnoses and can provide a number of examples.  Physicians are trained to "do" something medical or surgical and over the course of my career that bias has not panned out in a number of areas.  Targeting that tendency will be more productive than complaining about pharmaceutical companies, diagnostic manuals, or guidelines especially when our major regulatory agency has the bias to get a drug out there for general use based on highly variable scientific evidence.  As recently noted the FDA can also ignore the recommendations of their own scientific committee.

In the end, the best assurance of adequate medical treatment and minimal risk from the treatment depends on the expertise and judgment of a physician.

George Dawson, MD, DFAPA  

Supplementary 1:  After staring at the diagram for a while I think I need to add a 7th Bubble that I would probably call "Heuristics".  It would focus on the cognitive and emotional biases that exist in prescribing.  For example,  everyone in their practice may have a person with an addiction who was subsequently able to take an addicting medication without relapsing to their primary addiction.  Is that a reason for making that a rule for prescribing?  There are a number of ways to look at potential biases in prescribing that can improve the evaluative structure of that process.


  1. My only change would be "medical knowledge" should be replaced with "medical wisdom". The best and the brightest KOLs in some of the most prestigious medical societies in the last two decades have been a big part of the problem.

  2. thanks for posting on this topic

    an unasked for edit that i suggest improves your argument

    change ....The need to be appreciated and have a positive relationship with the patient is another issue.

    revised text would read... The wish to be appreciated and have a positive relationship with the patient is another issue.

    i think both of your overidentification examples are "a wish to be appreciated and have a positive relationship with the patient.."