Wednesday, March 17, 2010

A Working Definition for Self Medication

(See: Hien, et al, Do Treatment Improvements in PTSD Severity Affect Substance Use Outcomes? A Secondary Analysis From a Randomized Clinical Trial in NIDA’s Clinical Trials Network Am J Psychiatry 2010; 167:95-101)

Most discussions of self medication fail to address whether it is use or disordered use that is supposed to “treat” an underlying illness. One notion of self-medication would suggest that effective treatment of the illness would lead to abstinence. An alternative notion suggests that effective treatment of the illness might, for example, leave an alcoholic capable of controlled, rather than uncontrolled, drinking. No generally accepted rigorous definition of self-medication exists currently.

It is not necessary to invoke a self medication hypothesis or model to explain the findings in this study. And in fact the findings also support the opposite conclusion. If high intensity of symptoms occur during times when there is heavy substance use, this suggests that at best the substance fails to treat the symptoms, and perhaps that it exacerbates the symptoms.

The authors present no data regarding the relationship of the date of trauma to the onset of substance use or disordered substance use. If substance use (disorder) predates the trauma, it is more likely that the substance use contributed to exposure to trauma.

It is challenging for an addict or alcoholic to achieve abstinence even in the absence of mental disorder. The study supports the idea that effective treatment of co-occurring mental disorder should enhance recovery from SUD. However, the study does not address ethical or logistical considerations in attempting to treat a mental disorder in a patient who regularly uses substances of abuse, particularly where those substances may interact adversely with pharmaceuticals or where they might seem likely to exacerbate symptoms of the mental disorder. It should be noted that of all the drugs listed as used by participants in the study only “sedatives” (and barbiturates, which are themselves sedatives) would seem likely to provide any benefit to the symptoms of PTSD, and these were among the least used of the drugs listed. Also, cocaine, among the most used substances, would seem likely to exacerbate any anxiety disorder, including PTSD. There is certainly no evidence from any clinical trial of which I am aware that cocaine might have any therapeutic effect for treating PTSD. These two facts combined with the fact that sedatives might be more easily obtained than cocaine, weigh strongly against a self-medication model.

The authors of this paper like many other authors use the term self medication without defining it. The usual implied definition seeks to explain the patient’s “reason” for using the drug. However, we know that addiction is about how drugs are used, not why, and that the purpose of use, e.g. to get high, may have little to do with the rationalization, e.g., to relax.

Khantzian’s original self-medication hypothesis related not to Axis I, but to Axis II, and not even necessarily to a disorder. He suggested that the choice of drug of someone who was addicted might be explained as compensating for undesired personality traits. This hypothesis did not address how the drug was used. Furthermore, this hypothesis only made sense for those addicted to a single substance, the mythical drug of choice.

Proposed operational definition

I propose first the “SEE” test. The degree to which the substance use in question fulfills these criteria determines the degree to which self medication has occurred:
  •     Specificity: The drug’s beneficial effect is specific to the symptom. For example general anesthesia will remove awareness of almost any symptom, but does not meet the specificity test. (Rarely a characteristic of drugs of abuse which produce a state of euphoria rather than removing a symptom.)
  •     Efficacy: The drug attenuates or removes symptoms (Rarely a characteristic of drugs of abuse, many of which cause or exacerbate symptoms, depending on dose.)
  •     Economy: The drug is used only after symptoms have started, and only in amounts necessary to control symptoms (not usually a characteristic of disordered substance use). Persistent drug use in the absence of symptoms fails this arm of the test.
Next we apply this SEE test to what might be called types or stages of self medication:

    Primary (na├»ve) self-medication: The patient has never used the drug before, but based on information from another source, or simply because no other treatment is available, chooses to experiment. For example, Joe develops a headache and finds only a bottle of fluoxetine in the medicine cabinet, so he tries one tablet. The headache gets slightly worse. He never uses the drug again.
    This type is of little relevance to disordered substance use.

    Secondary self-medication: The patient has used the drug before with some benefit and decides to try it a second time based on that experience. For example, Judy develops a headache. The last time she experienced a headache she tried acetaminophen and the headache improved markedly after about 20 minutes. This second use of the drug produces a similar result.
    This type again is of little interest related to substance use.

    Tertiary self-medication: The patient has used the drug multiple times in the past and begins to use it regularly. Now we must consider three subtypes:

a)    S(+) E(+) E(+): A patient with recurrent back pain has an old bottle of oxycodone used several times in the past for a variety of painful conditions or injuries. It was first prescribed after a dental procedure. She takes the drug for several days according to dosing directions on the bottle. She finds the side effects mildly unpleasant and stops taking the drug when she realizes the pain has remitted.

This patient medicated herself in a way that one could argue was appropriate even though it was not recommended by a physician. Is this disordered use? It certainly was not used to get high. To call this self medication may be valid but is trivial.

b)    S(+) E(+) E(-): Another patient with chronic back pain like the patient above begins taking oxycodone from a bottle prescribed for her father before he died of cancer. After taking the drug regularly for weeks she notices diminished effect so she increases the dose. She too dislikes the side effects, especially after she begins to experience withdrawal symptoms when she tries to do without the drug, even though the pain has remitted. She flushes the remaining pills down the toilet resolving never to use them again, and does well after several days of intense, but diminishing, withdrawal discomfort. Several months later an oral surgeon prescribes the same drug after a procedure. After the first dose, recalling her previous negative experience, she switches to ibuprophen. This patient was physiologically dependent on the drug but not addicted.

This is the first case in which use of the drug after resolution of symptoms serves the purpose of “self medicating” not a separate illness or symptom, but rather symptoms that directly result from absence of the drug (which we call withdrawal. Again this use of the term seems trivial.

c)    S(-) E(-) E(-): Our next patient suffers from chronic nausea. He looks through his friend’s medicine cabinet, finding only an old bottle of oxycodone. He recalls the last time he used the drug he became mildly nauseated, but what stands out in his mind is the recollection of intense euphoria that dominated any negative effect. He takes the drug, and despite worsening nausea, and once more experiencing euphoria, continues to take it regularly. Like the patient in b) he soon finds he must increase the dose of the drug but never quite experiences the level of euphoria of the first few uses. He also experiences withdrawal symptoms so intensely that when he exhausts his supply he visits a local emergency room where he feigns severe headache in order to obtain another prescription. He was physiologically dependent and addicted.

This patient like patient b) has reverted to secondary self medication of withdrawal symptoms, but his purpose in using is only to reduce withdrawal symptoms or in the hope of attaining euphoria. This clearly meets the definition of disordered use, regardless of whether the term self medication applies. Because the drug causes nausea it is illogical to say he is medicating that symptom.

It is critical to understand that in the case of “tertiary self-medication” any combination of a, b, and c is possible, that is a patient may be “treating” real symptoms effectively, becoming physiologically dependent, and also seeking a euphoric effect. I have not proposed an S(+) E(-) E(-) case because specificity requires efficacy.

To judge the extent to which application of the term self medication to the study we would need to know more about the patterns of use and the effect of the drugs used. In general, however, it appears that the use described fails the SEE test. There is no evidence for Specificity as the patients used a variety of drugs with varied effects, but all can produce a high. There is no evidence for Efficacy for any of the drugs in treating post-traumatic stress disorder. In fact it is more likely that use of the drugs exacerbated the illness. There is no evidence for Economy as the authors failed to demonstrate that use was restricted to the amount necessary to control the symptoms.

What this study does demonstrate is that it may not be necessary to stop substance use before treating PTSD. The study does not support a self medication model with any useful definition. It does not support the notion that treating the underlying illness can lead to controlled substance use or that the SUD is not a separate disorder. Mention of self medication in this study was gratuitous. Pending a more rigorous definition the term is either trivial or meaningless and should be abandoned in discussions of disordered substance use.

Jack London knew almost 100 years ago the irrelevance of the "reason" for using to the addiction itself:

"It is the way of John Barleycorn. When good fortune comes, they drink. When they have no fortune they drink to the hope of good fortune. If fortune be ill, they drink to forget it. If they meet a friend, they drink. If they quarrel with a friend and lose him, they drink. If their lovemaking be crowned with success, they are so happy they needs must drink. If they be jilted, they drink for the contrary reason. And if they haven't anything to do at all, why they take a drink, secure in the knowledge that when they have taken a sufficient number of drinks the maggots will start crawling in their brain and they will have their hands full with things to do. When they are sober they want to drink; and when they have drunk they want to drink more.
Jack London, John Barleycorn or Alcoholic Memoirs


  1. I look at self-medicating as simply a conditioned learned behavior. It can be normal and helpful, it can become part of "chemical coping" mechanism more or less helpful, it can also progress to abuse and outright addiction. As a concept I believe it has value though I agree we are to be careful about how we define and interpret it. Context is obviously important as the above examples note.

  2. But what definition do you propose?

  3. Drink vodka, get drunk, the voices and paranoia don't matter anymore...fall asleep. Drink vodka, get drunk, the voices and paranoia don't matter any more...fall asleep. Drink vodka, get drunk, the voices and paranoia don't matter anymore...fall asleep. Self-medication...get it?

    I self-medicated with alcohol before I started taking medication for paranoid schizophrenia. That's all it was. I used alcohol strictly to deal with the voices and paranoia. Once I was properly treated with prescription medication, I no longer felt a need to use alcohol, so I stopped using it. I'd have an occasional margarita, but that was it. Now I don't drink at all because my meds and alcohol don't play well together.