Spotlight on PMDD -- A Closer Look at the Diagnostic Criteria for PMDD: An Expert Interview With Dr. David R. Rubinow

David R. Rubinow, MD; Peggy Keen, PhD, FNP


February 12, 2008

Editor's Note:

Menstrual cycle-related symptoms have been identified in the medical literature as early as in the 1930s. Currently, premenstrual syndrome (PMS) is considered a spectrum of conditions, with premenstrual dysphoric disorder (PMDD) at the end of the spectrum -- reflecting the most severe form. Although there is an organizationally proposed definition, PMS remains rather loosely defined at this point in time.

Even the definition proposed by American College of Obstetrics and Gynecology does not specify much with respect to symptoms -- beyond requiring that symptoms be related to the menstrual cycle (occurring premenstrually and disappearing following cessation of menses). In contrast, the diagnosis of PMDD has fairly rigid criteria as specified in the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV).

Medscape recently had an opportunity to speak with David R. Rubinow, MD, Meymandi Distinguished Professor and Chair of Psychiatry, University of North Carolina School of Medicine, Chapel Hill, about diagnostic criteria for PMS and PMDD. Dr. Rubinow's research interests have focused largely on the neurobehavioral effects of gonadal steroids. Past research has involved use of hormone superagonists and receptor blockers to manipulate the menstrual cycle in order to identify the central effects of gonadal steroids in isolation. He has been a leading investigator in the area of menstrual cycle-associated symptoms and conditions for more than 25 years.

Medscape: Could you tell us about your beliefs related to the PMS spectrum and, in particular, about the current diagnostic criteria for PMDD?

Dr. Rubinow: Sure. In order to best understand either of the disorders that are grouped under PMS -- particularly the condition now called PMDD -- one needs to recognize at the outset that these are time-oriented rather than symptom-oriented diagnoses, by which I mean the symptoms, whatever they may be, need to occur during the luteal phase of the menstrual cycle -- following ovulation -- and need to disappear during the follicular phase, sometime after the onset of menses. This is what defines this group of disorders and distinguishes them from a whole variety of other disorders, rather than any particular symptom or group of symptoms.

I would conclude, therefore, that the looseness of the definitions of PMS and PMDD, as you have described them, has less to do with particular symptoms and more to do with the failure of clinicians to ensure that the fundamental criterion for the disorders exists -- that is, symptoms appear during the luteal phase and disappear during or at the onset of the menses. If that criterion is met for a particular symptom -- let's say for sadness or irritability or anxiety -- but is not met for any other symptom, there are no data to suggest that this particular disorder is in any way different from what is called PMDD in terms of history, in terms of subsequent course, or in terms of treatment response characteristics. It is the exclusive appearance of symptoms during and confinement to the luteal phase that really defines the disorder, be it called PMS or PMDD.

Medscape: How were particular types and a specific number of symptoms identified in order to meet the criteria for PMDD diagnosis? In the current diagnosis, if a patient only meets 4 criteria, they do not have PMDD. If they have 5, the PMDD diagnosis is "official."

Dr. Rubinow: As a preface to answering your question, it's important to look at the old PMS literature.

Medscape: That would be literature from the '80s or from the '70s?

Dr. Rubinow: Actually, the literature dates back to at least the 1930s. PMS was particularly popularized during the '50s, '60s, and '70s. In the mid '80s, there were 2 significant events that occurred. The first was a conference convened with a goal of trying to reach some uniformity of procedures for diagnosing PMS. Second, I convened a National Institute of Mental Health (NIMH) workshop for the purpose of developing criteria for PMS, so that it could be operationally defined. This was to ensure that findings across different studies would be comparable. The criteria that came out of the NIMH workshop formed at least the conceptual basis for the DSM workgroup that later established the first formal diagnostic criteria for a form of PMS, which was called late luteal phase dysphoric disorder (LLPDD). That particular disorder had the same symptoms that had been referred to as PMS and are now referred to as PMDD.

To directly answer your question, the particular formulation of meeting 5 out of 9 criteria simply follows a "recipe" that was adopted for most of the psychiatric disorders. You will see the same format for depression, other mood disorders, or personality disorders. Although this type of "recipe" approach, if you will, certainly offers advantages over the lack of any diagnostic criteria when one is attempting to diagnose a psychiatric disorder, the downside is reifying a particular constellation of symptoms as having diagnostic validity in contrast to an alternative constellation of symptoms.

I think that the criteria for PMDD arose organically from a template approach that the DSM workgroups applied, which was informed by the findings of investigators who had previously described symptoms that typically occurred in women having what had been commonly called PMS.

Now one could say, "Well, what if somebody has only headaches that are confined to the late luteal phase? Is that PMS?" And what I would say is, "No, that is not what is generally meant by PMS." It is not what was described in the literature for many years. However, one could argue that there are premenstrual syndromes that present largely with affective and behavioral symptoms, there are those that present with both affective and somatic symptoms, and there are undoubtedly some that present exclusively with somatic symptoms (though this is uncommon).

A caveat I need to state: one could always offer the counterargument that while PMDD focuses on mood and affective symptoms, PMS could refer to any symptom that occurs only premenstrually. This may be true. But the reality is that what historically has been known as PMS was always associated with dramatic behavioral and mood symptoms.

Also, PMS has always included, by definition, the presence of impairment -- because women would not go to their healthcare provider and request treatment for symptoms unless they felt those symptoms, to some extent, interfered with their ability to function. Currently, impairment is a criterion required for diagnosis of PMDD. However, I do not consider the inclusion of the criterion of impairment in the PMDD diagnosis as a major conceptual or diagnostic breakthrough.

Medscape: Would you advocate that "timing is timing is timing" and that whether a woman had 1 or 10 symptoms and no matter what the mixture was between affective and somatic symptoms, they should be treated for PMDD if symptom(s) are severe enough to bring them requesting medical treatment?

Dr. Rubinow: I think that is a fair distillate. Obviously, the nature of symptoms might make a clinician suspicious that there were other processes resulting in a particular constellation of symptoms.

I can give you several examples of this. We, for example, have seen women with hypothyroidism who presented with a cyclical mood disorder but also with profound fatigue. When we treated their thyroid disorder, all of the affective symptoms went away -- and these were cyclical affective symptoms. Bud Keye described, years ago, several women with meningiomas who presented with cyclical, classic PMS symptoms, presumably because (although this is an inference) fluid shifts allowed the meningiomas to be more symptomatic during the luteal phase of the menstrual cycle.

There are also seizure disorders that are more prominent during the luteal phase. If the most prominent presenting symptom was headaches and additionally the patient complained of "feeling terrible," rather than just saying, "oh yes, this is PMS," I would want to know why the patient was having headaches. But if someone presents with menstrual cycle-entrained symptoms and wants treatment, and it can be documented that the symptoms are menstrually entrained, do they deserve treatment after they have been clinically evaluated? Absolutely.

And what about the patient who says, "I have such profound irritability premenstrually that I am in danger of losing my job"? You follow her symptoms, and she doesn't have symptoms of depression and doesn't have symptoms of anxiety and doesn't have mood swings -- she just becomes incredibly irritable during the premenstrual time period. Does it make sense to say this person does not deserve to be treated because she doesn't have 5 symptoms? Of course it doesn't because she has a disorder that can be documented and has presented requesting to have her symptoms alleviated.

Medscape: So, essentially, you always evaluate symptoms with respect to the menstrual cycle and rule out other conditions as indicated. But if the symptoms only occurred premenstrually and other problems were ruled out, you would advocate that treatment not be based on the number or severity of symptoms?

Dr. Rubinow: The number of symptoms, absolutely not. Severity of symptoms -- yes, it does depend upon this to some extent. I have had patients in whom I have been able to document clearly, with longitudinal ratings, the presence of a menstrual cycle-related syndrome. When I discussed the syndrome with them, they said, "You know, I still have these symptoms but they really don't bother me that much anymore because I sort of understand them now." Their symptoms have been validated, and they don't feel like it is all in their head and don't feel like they're going crazy. Now, they can deal with the symptoms by sort of adjusting their life schedule. I am not going to treat persons like these because they are telling me that they don't need treatment, although symptoms are still there.

So, yes, severity is definitely an issue. But it should be recognized as well that the DSM-IV manual does not provide any guidelines for determining severity.

What you would want to see is a change in the symptom scores during the month. But, again, in the vast majority of cases, when a patient says, "This is a problem for which I need evaluation and treatment," their symptoms are significant enough to warrant treatment. With rating scales, keep in mind that one person's moderate is another person's severe.

Also, people respond differently to the demand characteristics of a rating scale. For example, if you ask "Tell me, from 1 to 10, how bad is your symptom right now" -- and the patient is crying and obviously distraught, but she rates it a 5, you'd say, "What is up with that?" She might respond, "Well, a 10 for me is when I had just had lost my father and I was ready to jump off the roof." She will not want to go back there and will avoid saying anything is a 10.

These numbers, particularly when patients are rating themselves, can take on symbolic importance. This may result in somebody not wanting to rate themselves a particular number because it indicates they were doing so much worse than they want to be doing.

That's why you simply can't go purely on the basis of numbers. But what I have always advocated, as others have as well, is that there needs to be a substantial change in the ratings of symptoms in the luteal phase. This was originally defined as a 30% change in a patient's ratings, after adjustment for the range of the scale that they used, and taking into account the point that I just mentioned. If someone's ratings only occupy half of the scale, then you don't want to impose the same kind of change on that individual as you would for somebody whose ratings traverse the entire scale.

So, for example, if somebody only used half of our 100-mm line scale, we would say 30% times the amount on the scale used, which would be 50%. This would mean you would need to see a 15-point increase in the mean symptom ratings during the luteal phase compared with the follicular phase.

Let's say you had a 6-point scale, and let's say that somebody's asymptomatic ratings were all rated 1. So that means to get a 50% increase, their symptomatic ratings could be 1.5 on a 6-point scale. Well I wouldn't want to treat somebody whose symptomatic ratings were only a 1.5 on a 6-point scale. So you need both: you need to see a change and you need to see the surpassing of a particular threshold.

Medscape: It sounds as if you're in favor of fewer criteria and less rigidity in PMDD diagnosis. If this were the case, it would certainly follow that more patients would be eligible for pharmacologic therapy. Could you talk a bit about this?

Dr. Rubinow: The main implication for me is that one ought to adopt a fairly practical approach to the evaluation of complaints of menstrual cycle-related mood alterations. There are lots of daily rating instruments that are available.

But the easiest thing to do is to find out from each individual what she feels are her 3 or 4 most difficult symptoms. Then have her rate them for a couple of months and see what you've got at the end of that time. As long as she indicates when she is menstruating, you don't need a computer program to look at the ratings and see whether the ratings uniformly go up during the premenstruum and disappear within a few days of menses or whether they are all over the place.

If they are all over the place, that's useful diagnostic information for you because then you can say, "You know, I can see that you've got a lot of these symptoms" and then either what else is going on, or, you might say, "Maybe you should talk to somebody about this," however committed one is to going down the road to understanding the source of the difficulties that the patient is experiencing.

But aren't you always way ahead of the game if you know what your patient's symptomatic profile really looks like? The fact is, with these disorders, you have to have corroborative, prospective information. This is because the recall bias is just so large that you can't really rely on the history.

Medscape: Even though many of the criteria for diagnosis could be somewhat arbitrary, you would still support using some kind of symptom scale that the patient would fill out for a certain period of time.

Dr. Rubinow: Absolutely, because then you know what you have. And, at the end of the day, do you have a symptomatic condition in which the symptoms are confined to the postovulatory period and disappear or are virtually absent during the follicular phase or the postmenstrual phase? If the answer to that is yes, and those symptoms were brought to your attention as a clinician, then it doesn't make any difference how many symptoms there are. The patient has, by definition, a menstrual cycle phase entrained disorder that is causing enough distress that they want help for it -- and they deserve help.

Medscape: Would this patient be, at least theoretically, responsive to treatment with the typical medications that have been approved for PMDD only at this point?

Dr. Rubinow: Yes, that's right. That's exactly right, again, with all of the usual caveats about differential diagnosis and making sure that you are not missing something and have gotten a good psychiatric and medicine history. Then, yes, one could very reasonably infer, on the basis of all other things being equal, that this is somebody who would be a very good candidate for the usual kinds of treatments for PMDD.

Medscape: Is this the approach that you have been using in your clinical practice? And, if it is, could you talk a little bit about experiences you've had with different types of medication or different patient profiles?

Dr. Rubinow: Yes. Most of what I have done prior to the last 2 years has been in the context of my work at the NIH -- so there it is less practice and more research. We have seen hundreds of patients in research studies, but, as you pointed out, they were not necessarily the same cohort one would see in a private practice.

There is no question that selective serotonin reuptake inhibitors (SSRIs) are very effective for some of these disorders. The literature suggests that SSRIs are good for about two thirds of women with PMDD or PMS. Actually, the original studies on the efficacy of SSRIs were antecedent to the development of PMDD as a diagnosis.

These are very effective medications. Another form of treatment that we have used as part of our research portfolio -- one that is effective, again, in about the same percentage of women -- is ovarian suppression. This approach is not something that I would do as a first-line treatment. But it is a way of determining whether there is a hormonal etiology of, or contribution to, the disorder. One is able to demonstrate that, after ovarian steroid reduction, symptoms go away and stay away until hormones are reintroduced in about two thirds of women.

Medscape: Would the oral contraceptive (OC) approved for treatment of PMDD fall into the category of ovarian suppression?

Dr. Rubinow: Ovarian suppression and hormones stabilization -- yes. We are about to test this hypothesis a bit more extensively because the OC approved at this point is effective only with an altered regimen. That is, it is not given in the usual 21-days-on, 7-days-off regimen; it is 24 days on and 4 days off.

So the question is, what happens with a reduced pill-free interval? Does that reduced pill-free interval prevent the kind of hormonal changes that one would see with a more standard regimen? And does it have different effects on follicular development in the ovary than that seen with a more extensive pill-free interval?

So, is the form of treatment ovarian suppression or, rather, suppression of ovulation? Or is the treatment related to stabilization of steroids, or does it have something to do with the unique components or the unique progestin in that particular compound?

Medscape: It seems this agent doesn't have a category yet other than hormonal.

Dr. Rubinow: Yes, I think that is right. But we will know more about that soon. We will be able to distinguish among those 3 mechanistic hypotheses.

Medscape: Is this going to be determined via a 3-prong study using different treatment modalities?

Dr. Rubinow: It is a study in which continuous OC administration is compared with discontinued administration and with no administration (or administration of placebo). We are testing both the efficacy of altered regimens of administering the hormonal agent and also the influence on neurosteroid synthesis. When you get even minor fluctuations in hormones, you can actually create changes in neurosteroids that, in turn, have been found to trigger behavioral syndromes. We are trying to understand exactly how this occurs.

We have been able to clearly demonstrate in our studies that it is the change in the hormones that is the trigger. In studies that Peter Schmidt and I did at the NIH, we demonstrated that if you add back hormones once somebody has had their ovaries suppressed, the woman becomes depressed. But if hormones are continued without interruption for the next several months, there is no return of symptoms after that -- there are symptoms for several weeks after the hormones are reintroduced and then no further symptoms. So it really is a change in hormones that is the trigger.

Medscape: Thank you for taking time to talk with us. Do you have any concluding remarks?

Dr. Rubinow: The take-home message is that independent of the advances that we have made in understanding the pathogenics of menstrual cycle-related mood disorders, these syndromes --irrespective of the name applied -- are easily diagnosed and very worthy of treatment.

This activity was funded in part by Bayer


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.