Major revision, January 2008

Thank you for your interest in this resource.  As you may know, I have been adding most of my updates to the version of this website written for patients and families, as they represent the vast majority of readers.

Having just made a major revision in the Treatment section of the patient version, rather than replicate those changes here, I'm going to advise you to use the patient version yourself, from this point forward. Since this is the version your patients may be reading, you won't have to read two different versions to know what they know.

Here is a link to the patient version of the Treatment section of this website.

Again, thanks for getting here.  Let me know if there is information you need that you cannot find.

Jim Phelps, M.D.


Here is the old material in case you're looking for something you used to find here: 

Patients and families are welcome here. This version is a trimmed down, action-oriented discussion of a treatment approach for primary care physicians who can't refer their patients to a psychiatrist.  Doctors who want to know as much as patients using this site may be better off using the patient version of "Treatment", which has far more information overall, more links, more details.   

Patients should not use this information to guide their own treatment without consulting their physician.  There are risks in doing so beyond what you can imagine (that's why we went to medical school:  to learn about drug interaction chemistry; changes in liver metabolism from medications; etc.).  Dr. Phelps specifically instructs you not to use this information on your own.   

What are the basic principles of bipolar disorder treatment?

Rule out thyroid with your baseline labs (if it's >3, read about thyroid and bipolar), and other potential organic bases with your exam, and youíre ready to apply the two basic principles of treating bipolar disorder:

  1. rely on mood stabilizers; and
  2. beware of antidepressants.
    (note that the patient version includes a third principle: get bipolar-specific psychotherapy.)

Antidepressants are widely recognized as having the capacity to make bipolar disorder worse.  In bipolar I (full manic episodes, including delusions), depressed patients given antidepressants switch into full mania 35% of the time.Altshuler  In bipolar II, antidepressants are associated with precipitating mixed states and promoting rapid cyclingKilzieh, Altshuler  Mood experts recommend using antidepressants only with caution and stopping them as soon as possible.Ghaemi, Sachs(b)

Moreover, there is preliminary evidence suggesting that antidepressants can permanently alter a patientís life course: the Stanford bipolar research team has shown that patients with bipolar II who have never been given an antidepressant respond better to low dose valproate than those who have previous exposure.Winsberg   Although this was a small study, yet to be replicated or enlarged, it does suggest a possible major problem.  There may be a substantial risk using antidepressant medication in the "bipolar spectrum" patient: antidepressants may worsen a problem which might have been milder and more treatment responsive had that type of medication never been used.Ghaemi   Just how much risk antidepressants pose is controversial.  I cannot emphasize enough the importance of understanding the antidepressant controversies (a summary page I keep updated frequently). 

Antidepressants That Aren't Antidepressants
Because of those risks of antidepressants (or possible risks, depending on how you interpret the available data), I emphasize to patients that we are fortunate to have at least 8 alternatives which have antidepressant effects, but are not associated with cycle acceleration or long-term destabilization. There is a full-page essay on these 8, with this table and additional comments:  

No Pills

Risk or Problems

"Natural" pills

Risk or Problems

Mood Stabilizers With
 Antidepressant Effects

Possible Risks (not all get them)



omega-3 fatty acids 
(fish oil)

Lots of pills

Lamictal (lamotrigine)

Severe skin problems 1/3000

Cognitive-Behavioral Therapy

Cost, Time; Finding a Good Therapist

optimize thyroid

Blood tests, hyperthyroidism

Seroquel (quetiapine)

Weight gain, diabetes

Light manipulations 
(dark therapy, light box, dawn simulator)

Cost; Routines Are Limiting



Zyprexa (olanzapine)

Weight gain, diabetes

These eight agents represent alternatives to which you can turn instead of using antidepressants, when the main problem a patient with bipolar disorder is experiencing is depression.  

What are the basic mood stabilizer options? (revised 11/2003)

Once you have decided to try a mood stabilizer, your principle choices used to be lithium and valproate.  Now we have a large array of options, as shown in this table based on the data supporting their use.  For a full explication of this table, and links to several expert guidelines which are consistent with this presentation of options, see the Mood Stabilizer page. 

Randomized trials positive Randomized trials negative Open trials suggestive Case reports/series

(Lithobid, Eskalith)

thyroid hormone (T4) thyroid hormone (T3/T4)

zonisamide (Zonegran)
E.M. Power Plus
(40 vitamins/minerals)
(Carbatrol, Tegretol)
tiagabine (Gabatril) clozapine  acetazolamide
  levitiracetam (Keppra) ketogenic diet
(negative, but pt. not ketotic)
    phenytoin (Dilantin)
(N= 23, controlled)



omega-3 fatty acids (fish oil)      
(the trade name is just too smarmy, sorry; it has to go in small print:  Abilify)

 But, how does one decide among them?  Remember, until about 2002, lithium and Depakote were clearly recognized as the first-line agents to use.  Only since then have we enjoyed a dramatic increase in options.  Primary care doctors can begin by using just lithium and Depakote, turning to the newer agents only when these well-established approaches are not effective (maybe by that time you'll be able to get the patient in to see a psychiatrist!).  There are visit-by-visit guidelines for these agents below. 

However, there are three other medications that are no more difficult to use, each with specific advantages, per the table below.  Weight effects are specifically noted as this is becoming a serious health issue unto itself and significantly complicates bipolar treatment.  (Similarly, all medications below are proscribed during pregnancy; you must refer to a specialist if the patient wishes to become pregnant, and she must not become pregnant while on these treatments.)

Medication Advantages Disadvantages Weight effect
full dose lithium Full antidepressant power 
(blood levels 0.7 to 1.0)
Tremor, diarrhea; toxicity risk Can gain slowly; monitor levels and TSH
low dose lithium Good augmenting antidepressant, even in unipolar; 300-600 mg works, few side effects (occasionally not tolerated: nausea, lethargy) Can gain slowly; initial level, follow TSH, yearly level
Depakote Fast, good antimanic, well known risk profile Little inherent antidepressant effect compared to lithium Rapid gains if appetite threshold crossed; PCOS risk?
Lamictal Antidepressant, perhaps better than lithium Slow titration required to avoid rash Weight neutral
Seroquel Anti-anxiety, pro-sleep, and some antidepressant effects Daytime sleepiness during titration, risk of weight gain (less than Zyprexa) Intermediate weight gain risk vs. Zyprexa (more) and Geodon/aripiprazole (less)
Zyprexa Superb when need rapid onset, antimanic Profound weight gain, worse than all other comparable options, common (Cases of diabetes even without weight gain)

So again, how do you decide?  Sometimes there are some obvious advantages you just must have:  for a severely depressed patient, lithium is quite fast and has clear evidence that it reduces suicide risk.Goodwin  Or for a really agitated patient whom you'd like to help right now, Zyprexa can be extremely useful (as long as you warn the patient about the risks if it ends up becoming a longer-term treatment; and Depakote can be loaded up quickly to as much as 1500 mg in 24 hours, with benefit within 24-48 hours, at a much lower cost).  For someone whose predominant symptom is depression, which happens to be the significant majority of patients with bipolar disorder, lamotrigine now has solid data supporting its use, as long as the patient can afford to wait a few weeks for the benefit to show up. For such a patient who also has substantial anxiety and severe insomnia, Seroquel may be worth jumping earlier in the progression (weighing against this advantage the greater risk, long term, of metabolic shifts; start with 25 mg and let the patient decide how fast to go up, and you'll have fewer saying "I can't take this stuff"). Here is a diagram I use to explain this to patient/families: 

I almost always start by showing a list of all options/pro's/con's to patients.  I like them to know that we have multiple alternatives in case the first one is not working out.  And I watch their faces as we go down the list of advantages and disadvantages:  that will tell you a lot about which medications are simply not an option, and which are looking favorable. 

There are non-pharmacologic modalities for the treatment of bipolar disorder of which you should be aware (patients will ask).  Five bipolar-specific psychotherapies have randomized trial data supporting their efficacy as adjuncts to medications; these are reviewed for you here, though there are few therapists who know about them so far.   Other modalities, including exercise (that link takes you to my not-the-usual-exhortation) and light therapies, are reviewed for patients on their version of this site under Mood stabilizing without medications.  Fish oil, as a means of delivering omega-3 fatty acids, has one preliminary study with robust dataStoll  and several more recent studies, summarized here.  

Finally, there are adjunctive medications as well. These include benzodiazepines, which generally should be limited to brief use due to their potential for dose escalation and tolerance (though some patients demonstrate sustained benefit without dose increases, perhaps particularly with clonazepam). Generally antipsychotics are not required nor useful in bipolar II, though among them olanzapine stands out as above; here is a discussion of the rest of the newer ones.  Except for the expense, Ambien (zolpidem) is an excellent adjunct for sleep when benzo's are not wise.   

Visit by visit plans for primary care treatment of bipolar variations

Here is a general outline of sequential office visits in which you begin treatment of bipolar II. Details for the use of lithium, valproate, lamotrigine and oxcarbazepine follow in the next section.

Visit "0" History (presumably accumulating over multiple previous visits)

Visit 1: Explain bipolar II diagnosis

Visit 2: Start mood stabilizer

Visit 3: Mid-course corrections ("The Art of Medicine")

Visit 4: Further adjustments

Specific guidelines for lithium and valproate

Unless you are quite familiar with these agents, I recommend you have a look at my patient education handouts for lithium and valproate). The following section will give you more detail on how to use them.


Tell your patient: the goal is 100% symptom control, with 0% side effects! There is tremendous individual variation in susceptibility to side effects with lithium: some people can tolerate blood levels above 1.2 mmol/L without any side effects at all (e.g. doses over 1500 mg); others will have intolerable effects with 300mg alone. If your patient has side effects, reduce the dose to a tolerable level and move on down the step-by-step approach.

Update 2005: For your PAR, you'll need to mention thyroid, kidney and weight gain risks. Details on all these risks, from a 2005 review of same, can be found on the Lithium Risks page. You probably know most of this information, but take that link if you need a refresher. For a more basic review of what you'll want to discuss with patients about lithium, try my lithium-basics handout

Start with 300mg slow-release lithium (U.S.: "Lithobid"): it can be cut in half and thus titrated by 150mg increments if needed; and it is generally much better tolerated than lithium carbonate. When your patient is doing well, you can try switching to a generic lithium to reduce costs. Starting with the generic runs the risk of "giving lithium a bad name" if your patient has a bad reaction that could have been avoided with a slow-release form. It really makes quite a difference!

Most patients will not reach the toxic range (usually associated with blood levels of 1.1mmol/L and above) on 900 mg. However, you should alter the handout instructions and check a level at 600mg if your patient is:

  1. hesitant about lithium, or hesitant about medications or side effects in general (or if you are!);
  2. already on valproate (because the lower lithium dose may well be sufficient);
  3. taking a medication that may elevate lithium: NSAIDís, ACE inhibitors, diuretics.

Emphasize the instructions: "do not increase the dose if you are already having troublesome side effects". Your patient must understand the toxicity risks (e.g. see handout: "too high a dose"...).

Instructing your patient when to have a lithium level can be complicated for them (see handout for how I do this). If in doubt, stop at 600mg and have them check a level 4 days or more later. Generally, if the level on 600 is less than 0.7 mmol/L, severe side effects are unlikely with a 300mg step up. Minor side effects such as dry mouth and urinary frequency (including nocturia) are common at almost any level, and generally increase with each dose increase, but most patients can handle moderate levels of these effects.

There are three side effects that commonly limit lithium dosing before toxicity sets in:

  1. loose stool, progressing to diarrhea
  2. tremor
  3. mental dulling

None of these is likely to decrease with time: the patient will have to reduce the dose (try bid or even tid dosing, and make sure youíre using slow-release, before going down, though).

Lithium is contraindicated if the patient has renal compromise: the levels may rise unpredictably to the toxic range. A patient who uses NSAIDís, ACE inhibitors or diuretics intermittently is also at risk of unpredicted toxicity. These medications are not an absolute contraindication but do raise the patientís risk of renal damage, requiring additional caution to insure that levels remain in the therapeutic range. To my knowledge, COX-II selective NSAIDís have not been studied re: their effects on lithium levels or renal risk.  Finally, and very important: repeat a TSH every six months at minimum, to catch the very common hypothyroidism lithium will induce (on the order of 1 in 10).  


Remind your patient: the goal is 100% symptom control, with 0% side effects! In general, with slow-release valproate (U.S.: "Depakote") there are very few side effect problems. About 1/10 will have some nausea when starting, reduced if the medication is taken with meals. About 1/30 (in my experience) will have severe nausea. Many in this group can eventually tolerate the medication if the 125mg "sprinkles" are used and titrated up by one pill per week or so, but you may have difficulty getting the patient to accept the necessity of that approach. Other than nausea, however, other side effects are very uncommon -- except weight gain, discussed in detail below.

The handout instructs patients to start quite slowly.  For inpatients with severe bipolar I, a "loading dose" of 20mg/kg is used, thus doses of >1500mg in the first 24 hours!McElroy et al   So you are already using a very slow titration if you follow the handout. For outpatients with severe symptoms you can start at 1000mg ER, instead of 500mg ER.   Instruct patients to stop increasing if benefits are seen, emphasizing weight gain risk as the primary reason for keeping the dose as low as possible.

Many male patients can tolerate 1500 or even 2000mg of valproate and not gain weight. Few women can, though, and most patients seem to hit a "weight gain threshold" somewhere around 1000mg.  In my experience, using the old slow release Depakote as opposed to the ER version, more than 50% of women will gain weight at 1250mg or above (and that is a conservative estimate). Is this an appetite increase, as patients almost universally experience when gaining weight? Or is there some metabolic shift? The basis for this problem is still unknown, but data is accumulating to suggest substantial hormonal shifts even before weight gain sets in (see the PCOS/Depakote story, which I keep updated).  

Fortunately, the marked appetite increase serves effectively as an indicator for this problem. I donít think Iíve seen a patient gain weight who did not experience the appetite increase (there probably is one somewhere). When patients lower their dose, they can detect when they fall below the "threshold" of this effect: their appetite returns to normal, and they do not seem to gain weight. Thus, the obvious strategy: if patients experience an abnormal appetite, they should lower their dose until their appetite returns to normal. The "threshold" seems to lie between 750mg and 1000mg per day for most patients on the old version.  With the new ER version, most patients can take 1000mg without appetite increase, and many can even get to 1500 although then it's time to watch closely. 

I donít think Iíve seen a patient who experienced weight gain at 500mg per day (there probably is one somewhere), if they werenít taking other medications that raise valproate levels (e.g. Paxil and other "SSRIís").

Valproate at 500mg/day is not generally enough for symptom control, but when combined with low-dose lithium, it can be a very effective medication. And, not all patients will experience the weight gain problem. Hair loss is also common when patients hit the weight gain range, but the dose reductions required for the latter problem generally take patients out of the "hair loss range" as well; it has not been an independent problem in my experience. You will see selenium and zinc touted for valproate-induced hair loss, but I have not seen this vitamin approach (e.g. Centrum Silver has these minerals) have much impact, especially compared to dose reduction.

Lamotrigine and quetiapine
For basics on these, I'll refer you to the brief introductions I wrote for patients (lamotrigine, quetiapine).  Using them is pretty straightforward, as you'll gather from the PAR outlines

Does one really have to minimize antidepressants? (revised 10/2005)

There is strong consensus that antidepressants, especially without mood stabilizers, carry risk of precipitating manic symptoms in clearly diagnosed bipolar I patients. The depression gets better, but the "manic" side symptoms (remember, this can include sleep problems, anxiety/agitation, irritability, and difficulty concentrating) get worse. The whole mess can start to "cycle" more frequently, even though the depression is better. In some patients, eventually a full depression episode occurs again, despite being on an antidepressant, even one that "worked" before!  How often does this occur?  How risky are antidepressants for someone who has only subtle "hypomanic" symptoms -- e.g. for a person whom some might not even diagnose as "bipolar"?

This is an area of controversy.  We have very limited data on which to proceed.  As I see it, clinicians with lots of direct experience in mood disorders are among the most cautious about using antidepressants.  I have met psychiatrists all over the country evolving this same view through clinical experience. For my personal view (same as Fred Goodwin, Nassir Ghaemi, and numerous other well-respected bipolar experts reviewed on that Controversies page), derived from 5 years of experience with patients referred by PCP's because they were not getting better on antidepressants or therapy:  I find antidepressants are definitely a risk for many patients.  Indeed, for some patients, you can even say that the antidepressant is causing depression, by making the cycling continue, including cycling into depression.

In that case, even if a mood stabilizer does not have antidepressant effects by itself (valproate and carbamazepine/oxcarbazepine appear to lack direct antidepressant effects) it can work as an "antidepressant" very well, by stopping the cycling.  I have seen this happen many, many times -- so many, that I routinely rely on the mood stabilizers to help these referred patients, and taper off their antidepressants, even while they are depressed.  Many times itís the only way out of the problem (though usually Iíll start the mood stabilizer with the antidepressant they were already taking still going, then taper off the antidepressant when itís clear the person is getting better with the mood stabilizer).  For a further explication of this concept, see the Antidepressants That Aren't Antidepressants page. 

For your purposes in primary care, suffice it to say that antidepressants should be used with caution in patients who might have a "bipolar component" to their depression, and tapered if a person is clearly rapid cycling.  Beyond that, stay tuned to the summary of this controversy (see revision date at top; also noted in "What's New" on the home page here).

Congratulations, thatís it for the basics of treatment. From here you can review, or you can select from the following appendices for more treatment details.

On to Treatment Details -->