• Depression Types
  • Etiology

What kinds of depression are recognized?

You will run into many adjectives describing depression itself. Here are some of the most common variations, in a narrative I wrote for patients:

  • acute or chronic
  • situational or unprovoked ("exogenous" or "endogenous")
  • prolonged grief
  • due to medical condition
  • "atypical depression"

"Acute" means severe, and generally also means sudden, implying that prior to this severe depression the person was much less depressed, perhaps not at all. The opposite is "chronic", meaning the person has had this for years. The usual DSM term for this is "dysthymia" (technically, depressed more days than not, for more than two years).

Situational depressions have clear causes (or at least people think there is such a connection). Unprovoked, or "endogenous" depressions come "out of the blue": no clear event is associated with the start. Some people will call this "chemical" depression, meaning there was an internal cause, not an external one.

Most people experience sadness when they lose something to which they were emotionally attached. When the loss is great, such as the death of a family member, people can experience a deep grief. In many respects this looks like "depression", but most people will feel themselves gradually coming out of it, starting within a few days or sometimes a week or two. Grief that continues beyond about 1 month is generally considered "prolonged" and may need some help to resolve.

Depression can be associated with medical conditions. Strokes and Parkinson’s disease are among the most common such causes, but several common medications, and many other diseases (problems with the immune system , heart, and especially hormones such as thyroid) can be associated with depression. The list is so long that "ruling out" all these other diseases is very impractical. You can end up with a lot of tests that are painful and even carry some risk, such as a heart catheterization or using a scope to examine your stomach or intestines. After a blood test looking at least at thyroid function, it is not routine to go looking for potential medical causes for depression. Rather, the doctor looks for signs and symptoms of these illnesses to see if they might be the cause. Without such signs and symptoms, it is very unlikely that there is some "medical" basis for your depression.

Finally, in this brief survey, there is "atypical depression". Technically this refers to a group of patients who have depressive symptoms but also have other features: "mood reactivity" (intense emotional reactions), and sleeping too much and/or eating too much, thus gaining weight. This symptom complex has been shown to identify a group of people who respond better to an MAOI (monoamine oxidase inhibitor) than to imipramine, another older antidepressant. Other symptoms associated with the "atypical" label include "leaden paralysis" (extreme lack of energy), and "rejection sensitivity" (over-sensitive to perceived slights), although these were recently shown not to predict imipramine non-response as well.Sotsky All these symptoms have tremendous overlap with bipolar symptoms.  In fact, interpersonal sensitivity is specifically associated with bipolar II.Benazzi Moreover, bipolar disorder tends not to respond well to the older antidepressants, and may respond best to the MAOI’s. Are these really different diseases, or just different aspects of a complex syndrome? If a person does well on an MAOI, great. If the response is not enough, or fades away, consider bipolar disorder as an alternative explanation for this symptom pattern.

What is treatment resistant depression?

This subset of patients has a differential of its own. Substantial controversy exists, even amongst psychiatrists, about the boundaries of some of these "diagnoses" (especially Bipolar II and Borderline Personality Disorder; for more on this particular differential, see the section I wrote for patients addressing that issue). Treatment options can be simplified as follows:

  1. more antidepressant trials
  2. more psychotherapy (some, more, different type)

Recent research in psychotherapy has shown efficacy for interpersonal and cognitive/behavioral therapy equal in many cases (e.g. acute treatment of mild to moderate depression) to medications, and possibly with greater long term effectiveness.Frank and Thase. Eventually, after multiple trials of A and B, however, there is a truly "treatment resistant" group. What then? Perhaps even long before then, one should apply the old rule of medicine: if the patient does not respond to a routine treatment, always reconsider the diagnosis. Has something previously been ruled out that now needs to be reconsidered? Has something been omitted previously that should now be included?

As you attempt to handle your patients with treatment resistant depression, you may have positive results with mood stabilizers. Since these medications may also be useful in multiple other conditions related to mood disorders (including, from my experience: dissociative disorder, impulse control disorders, borderline personality disorders, PMS, IBS, migraine; and as adjuncts (especially lithium) even in presumably unipolar depression), you may see good results even if your patient does not have a bipolar variant per se. This leads to an argument that these medications are being used indiscriminately, or that the diagnosis is being stretched so far as to become meaninglessSobo. For a discussion of this stretch, click here.

What if anxiety is also prominent?

Types of anxiety

First, you must recognize the distinct types of anxiety that have highly effective treatments:

  • panic disorder
  • obsessive compulsive disorder
  • social phobia

Each can have up to an 80% response to treatment, without medications! For panic disorder, David Barlow and colleagues developed a cognitive-behavioral therapy that has the best track record, substantially better than medications at the 6-months-after-treatment mark (see the landmark 8 year randomized trial results from Barlow et al in JAMA). Several current self-help books capture much of this approach, though perhaps not enough. Barlow’s 12-session (or less) method is still the "gold standard" of this technique. I have used his treatment manual with over 50 patients with results similar to his research — and it’s actually rather fun! To access Barlow’s approach, call all the psychologists in your area and ask if they are using cognitive behavioral therapy for treatment of panic disorder. If they actually use Barlow’s manual (Mastery of Your Anxiety and Panic: MAP; not available in bookstores, as the creators intended it for use with a therapist), all the better.

For obsessive-compulsive disorder, several research teams have defined a cognitive-behavioral therapy that can also work as well or better than medication. An easily read summary is Brain Lock, by the head of the UCLA treatment program. Lee Baer and Edna Foa are other major contributors to this field who each have their own paperbacks. To access this treatment in your area, use the same "call ‘em all" method and ask if the therapist is doing "exposure and response prevention", or "cognitive-behavioral therapy", for OCD.

People with social phobia experience intense anxiety, very similar to or including panic attacks, but only in social situations. When they are by themselves, they do not feel anxious. They can go to a mall if they don’t have to interact with anyone (as opposed to panic disorder patients, who tend to fear the enclosed, "can’t escape" feeling of a crowded public place). Social phobia ranges from very mild performance anxiety to an inability to handle interactions with any but the most familiar people, and can be extremely debilitating. You will hear about Paxil for social phobia as it’s manufacturer was the first to jump through the hoops to gain FDA approval for this indication, but other SSRI’s may be equally effective. Moreover, there is a cognitive/behavioral therapy similar to that for panic disorder that has efficacy equal to medications and may have better long term results.Liebowitz

Post-traumatic stress disorder can be much harder to treat. Similar cognitive methods can help some people, and several classes of medications likewise (antidepressants, antipsychotics, and mood stabilizers), but many symptoms can remain.

Generalized Anxiety Disorder

Lastly among the chronic anxiety diagnoses there is "Generalized Anxiety Disorder" (GAD). Here are the symptoms:

Generalized Anxiety Disorder
Disorder (DSM-IV)
  • worry
  • difficulty concentrating


  • keyed up, on edge
  • restlessness, tension
  • easily fatigued
  • difficulty falling/staying asleep


  • irritability

There is no treatment with response rates comparable to the treatments for panic or OCD. Several groups have tried to adapt cognitive therapy to this condition, with limited success. Medication treatment is even more disappointing. Benzodiazepines (Valium, Librium, Klonopin, Ativan, Restoril, Xanax) usually help a lot, almost perfectly, for a short while. But they lose their effect in most patients, and the dose must be increased to get symptom control again. Then the pattern repeats, the dose growing steadily higher. In some patients some benefit is maintained at the higher doses, but many physicians are uncomfortable with long term use of these drugs. Eventually some doctor will simply refuse to continue them, and then the person is exposed to her/his anxiety again. If the medication is stopped quickly, she/he has "withdrawal" anxiety (or even more serious physical withdrawal symptoms, which can be life-threatening) in addition to the original symptoms.

Buspirone (U.S.: Buspar) is similar to benzodiazepines but without the dose increase/withdrawal risk. It generally has has little or no effect, in my experience. Certainly if there were patients who did very well with it, I wouldn’t see them! But usually the good news filters to me eventually, and currently there is very limited enthusiasm for this medication, based on what I hear from colleagues.

Antidepressants can provide substantial relief of symptoms. Recently venlafaxine (U.S.:Effexor) has been approved by the FDA as having effectiveness, as other antidepressants have shown before.Feighner The problem with this approach is that it clearly does not help enough people, as most patients who reach me have already had trials of multiple antidepressants with limited improvement.

In my experience it is very clear that some people who are experiencing anxiety, who often have been diagnosed with GAD, respond to the mood stabilizer approaches described below. Could they have had a form of bipolar disorder?

In my experience two symptoms best identify people who are going to respond well to mood stabilizers: profound insomnia and difficulty concentrating. These are two of the GAD symptoms; I’ve put them in brackets in the right column below. Add the symptoms of bipolar II (using the broad criteria of the head of the mood disorders clinic at UC San Diego, Dr. Akiskal), and you’ll see we have basically duplicated the DSM rules for GAD:

Generalized Anxiety Disorder
Bipolar II
(Akiskal + Phelps)
  • worry
  • difficulty concentrating


  • keyed up, on edge
  • restlessness, tension
  • easily fatigued
  • difficulty falling/staying asleep


  • irritability
  • free floating anxiety
  • [difficulty concentrating]


  • motor agitation
  • restlessness
  • extreme fatigue
  • [profound insomnia]


  • dysphoria, irritability

When you look at these virtually identical lists, I hope it makes sense to you than anyone with GAD symptoms who has not responded well to antidepressants should consider a trial of mood stabilizers. Remember, not everyone will respond. Raised hopes carry risk of dashed hopes later, so be cautious and wait for the "proof in the pudding".

Are we stretching the diagnosis too far? Is this a bipolar "bandwagon"? (revised Apr 2000)

Even some psychiatrists think this bipolar way of thinking has gone too far.Sobo They are concerned that excitement about this diagnosis and the treatment options it opens has become a "bandwagon". Yet, how can we tell the difference between a useful new way of thinking and a bandwagon? Won’t they look similar, perhaps identical, at first? Surely we already know the answer to this puzzle: "the proof is in the pudding"! Patient outcomes will tell us how valuable a new approach really is.

Granted there will be some placebo value initially for almost any "new" treatment that comes along. That’s why data from "open trials" of a new medication (no control group, no blinding of patients or providers) almost always overrate effectiveness. Only later when the controlled trials are done do we get a clearer sense of what the medication offers relative to placebo.

Gabapentin (U.S. Neurontin) is a great example of this: there are multiple open trials reporting very positive resultsErfurth, Letterman , but there are no controlled trials showing efficacy as a mood stabilizer. The main multi-center study showed efficacy slightly less than placebo, and the results were not published.Sachs(a),Zarate   A second randomized trial by The National Institute of Mental Health (small sample, but well-controlled) found no difference between gabapentin and placebo.Frye  Perhaps the most conclusive evidence of lack of efficacy comes from the manufacturer itself, as it has chosen not to pursue any more studies of gabapentin as a mood stabilizerZarate And yet, gabapentin clearly helps some people. The results can be dramatic. It looks like a great antidepressant that reduces anxiety at the same time. So if some people respond so well, and the overall data show it no better than placebo — some people must be getting worse! I think I’ve seen a lot of them. I stopped using gabapentin, except with great caution, about 8-12 months after we first began using it. Dr. Sachs at Harvard has the same hesitation.Sachs(b) Thus I believe we have more of a problem of "bandwagon" treatments for which little supportive data exists, than with bandwagon diagnoses where multiple mood experts share the same view.

Bandwagons aside, if a patient with depression happens also to have irritability, or insomnia, or anxiety — is he/she "bipolar", or is that stretching the diagnosis too far?

First, remember the new "spectrum" way of thinking about this. We should not be asking a black-and-white "bipolar or not?" question. Rather, ask "how much of this bipolar-like trait might be present?"

Secondly, we should think of "diagnosis" in these cases as a heuristic process. I love this term. It describes exactly where we are in this dilemma. Per Webster’s dictionary, heuristic means "valuable for research but unproved or incapable of proof". We let go of the need for certainty in favor of assessing usefulness.

A heuristic approach to this problem asks not "does this patient really have bipolar disorder?" but rather: "If I try mood stabilizers, what is the likelihood of benefit? What is the likelihood of risk? How effective are the alternatives, and what is the risk they pose?"

A brief trial of valproate presents low risk in people who have normal liver tests. Long-term treatment is different. Weight gain is very common and must be managed cautiously, as there are numerous health risks associated with obesity. One of these is a change in hormonal balances.

There is some concern about valproate causing a hormone imbalance called "polycystic ovarian syndrome (PCOS)."Isojarvi Whether this is from the medication, or associated with the conditions for which the medication is commonly used, remains debated.Herzog (c). A recent review concluded that the evidence was suggestive but not conclusiveChappell. The magnitude of a possible effect was much smaller overall than implied by the data of Isojarvi et al. Weight gain might be cause of PCOS with valproateIsojarvi et al(b), if the medication does indeed cause PCOS at all. If true, then management of the medication to avoid weight gain would also protect patients against this problem.

When patients understand that we will switch medications if they begin to gain weight, they seem to find the potential for benefit outweighs the risk, even when benefits are of low likelihood — that is, when they have few clear symptoms to suggest "bipolar disorder". If there is no benefit but no side effects at low dose (e.g. 750mg of valproate; see Depakote details), a brief trial at standard doses of 1000-1500mg poses little additional risk.

Even if the risks were greater, such as with carbamazepine, many patients still conclude the risk is justified because they have reached a stage where "no treatment", or continued trials of antidepressants, or even quality psychotherapy, is far less preferable.

For a similar but more historical and elegant discussion of this "bandwagon" issue, see this essay by Dr. Ghaemi, a research psychiatrist in the Psychopharmacology Program at Cambridge Hospital, and an instructor in psychiatry at Harvard Medical School.  

Normal? mentally ill? Where’s the line between them?

In general, people who ask this question are really worried about their own sanity, or about how they will be perceived by others, especially if they are labeled as "bipolar". However, the question should be resisted, in favor of a new "model" of the illness. Again, think in terms of a spectrum, or continuous line. You will remember the "mood spectrum" from above [link to "What is the depression spectrum" in Diagnosis section]:

plain depression------------------------------manic-depressive
("unipolar")                                  (bipolar I)

A similar spectrum exists, from completely mentally healthy, to severely impaired by mental symptoms. As stated in the milestone document "Mental Health: Report of the Surgeon General" (1999), mental health and mental illness are merely extremes on a continuum:

Mental Health----------------------------------Mental Illness

"Ill" or "healthy" sets up a yes-or-no, black-or-white distinction that is really a problem in bipolar disorder, where people can have long phases with no symptoms. Do they still have a mental illness? Imagine a man who has had a heart attack. Perhaps he has worked hard at physical activity after his attack and no longer has any limitations. Does he have a cardiac illness? What if he instead has no limitations because his medications work very well? Does he still have a cardiac illness? (Perhaps even more to the point, have you ever heard anyone ask those kinds of questions!?) Our society’s tendency to think in black-or-white terms creates this labeling problem.

For example, consider the following headline: "Kinkel unlike the other shooters: mental illness set Springfield teen apart from other youths who terrorized schoolmates". Kip Kinkel killed multiple classmates in a Springfield, Oregon high school. He was later said to have an "urge to kill". This was regarded as evidence of a mental illness, and somehow different from the presumed motives of the several other gun-wielding adolescents who killed their classmates in other schools. These other youths were completely "normal"?

Or as a less extreme example, consider one of my patients with bipolar disorder, a college professor. When she applied for a new drivers’ license, there was the question: "Do you have a mental illness?" What’s she supposed to say, when her symptoms have been 80% controlled for several years, and she has Bipolar II anyway, which has not been recognized as impairing driving safety --? The question reflects the expectation: yes, or no? Do you, or do you not? The professor was very uncomfortable with being forced into this black-or-white position.

Mental events of other kinds can also be spread on a spectrum from completely unremarkable ("normal") to very unusual. Fanaticism, for example, demonstrates a continuous spectrum. It extends from people with no particular intense interests; to strongly held beliefs; to extreme beliefs as manifest in some members of Greenpeace or the NRA; to complete loss of perspective as in followers of suicidal religious cults, or Timothy McVeigh (bomber of the U.S. Federal Building in Oklahoma City) ; to overt paranoia such as seems common in the statements of Militant Federalist members (mysterious black helicopters, elaborate conspiracy theories) or the Y2K fearful (no need to stock up on supplies, as the world is coming to an end); and finally to clearly delusional beliefs such as those in the "Unabomber" Manifesto, or Kip Kinkel’s "need to kill".

But a "fanatacism spectrum" makes sane/insane questions much more difficult. Our legal system is predicated upon "black-or-white" distinctions: right/wrong, impulsive/premeditated, sane/insane. Society can handle black and white, but struggles with continuous spectrums.

Yet in the process it leaves patients with bipolar disorder in yes/no dilemmas that have no answer as such. Part of "destigmatizing" mental illness will eventually require recognizing "shades of gray" of mental dysfunction. We should all resist the question "mentally ill?", as such. It is just another limitation set upon people with mental health symptoms.

Childhood bipolar: Does it look different?  

We know this is a genetic disorder.  So anyone who has this illness is going to wonder if their kids (born or yet to be created) could get it.  If the current view that early treatment can decrease an entire lifetime of symptoms holds up, detecting early signs of the illness will be crucial.  So the question of how bipolar disorders show up in kids is crucial too.  

Unfortunately, the "diagnosis" puzzle is even harder in kids than it is in adults.  But, as one researcher put it, the most important step in diagnosing bipolar disorder is to suspect it in the first place.  The worst error is not to consider it.  Even if the diagnosis is held until fairly certain, that's better than missing it entirely, which is the case all too often.  

Fortunately, though, there is a wonderful website to help you keep track of the thinking on this question, and to which you can refer patients.  It's run by the Child & Adolescent Bipolar Foundation.  It has the full text of an expert summary from 1997 about how to diagnose and treat kids. See their "About Early Onset Bipolar Disorder" in the Learning Center as a start, but go back for their Reference Center as well.  It includes full text of a wonderful collection of articles, many pretty technical, but very well selected.  This site is a real gift to parents.   


What about hormonal effects on mood?

When the DSM included premenstrual syndrome ("PMS") as a psychiatric disorder, there were protests in the streets of San Francisco (at the American Psychiatric Association meeting) by women’s groups opposed to "pathologizing" women’s mood experiences. This point is well taken, and at the same time raises the question again regarding the term "mental illness". Meanwhile, growing evidence suggests a need to critically evaluate the role of steroid hormones on mood, as follows.

Research physicians have reported several patients whose anxiety was successfully treated with medications altering steroid metabolism.Jacobs, Herzog(a) DSM "mixed states" and "rapid cycling" are widely recognized as occurring far more often in women than in men.Kilzieh Women with bipolar disorder have a 50% risk of depression after childbirth.Leibenluft Newer antidepressants affect the production of a key progesterone metabolite.Griffin And steroid medications are recognized as capable of precipitating manic symptoms in both Bipolar I and IIBrown and Suppes.

Andrew Herzog, chief of neuroendocrinology at Harvard, has performed or reviewed much research which indicates that reproductive hormones powerfully modulate mood and anxiety.Herzog(b) Estrogen increases the action of an "excitatory" neurotransmitter called glutamate, which is so powerful it can damage cells with its excitation effects if not properly balanced. The "yin" to glutamate’s "yang" is GABA, an "inhibitory" neurotransmitter, the activity of which is increased by progesteroneGriffin. Estrogen acts like an antidepressant, and progesterone roughly like Valium or Xanax. But too much estrogen can cause anxiety (just like antidepressants do in bipolar disorder), and too much progesterone relative to estrogen can leave a woman sedated and low-energy. There is a complex balance between these hormones, and that balance varies throughout the menstrual cycle.

Recently an excellent summary of what Psychiatry currently knows about hormones and mood was published by Deborah Sichel, M.D. and Jeanne Watson Driscoll, M.S., R.N., called Women's Moods.  They review what is known about premenstrual symptoms, pregnancy, postpartum (just after childbirth) changes, and menopause -- examining how mood is affected by the hormonal changes of each phase.  Although this website will continue to collect and present further advances in our understanding of these mood changes -- for a single reference source, easily read, in the most understanding and supportive language you could ask for your patients, Dr. Sichel and Ms. Driscoll's book is the place to go.  

Continued research on these hormonal influences will shed more light on the mood variations discussed here as "bipolar disorders". Perhaps soon we will reclassify some women’s mood variability as a hormonal disturbance, rather than a mood disturbance — but the views of the San Francisco protesters should not be forgotten.

What’s the latest on the etiology of bipolar disorder? 
(updated 12/2005 on the patient site)

This is an area of intense research.  I have updated this section of the "patient" version of this site and refer you there. 

Use your "back" button to return, or use the navigation buttons at left, or go: 

On to Treatment -->