Depression That Keeps Coming Back: When It Is Actually Bipolar II, and Why That Changes Everything
🔄 A familiar history: four antidepressants over eight years. Each one worked, properly, for a few months. Then it stopped, the dose went up, it stopped again, and the next one was tried. Somewhere in the file the words treatment-resistant depression appeared. In roughly one case in five, that history is not resistant depression at all. It is bipolar II, and it has been treated with the one class of drug that makes it worse.
📌 That last number explains the whole problem. In bipolar II the depressed weeks outnumber the elevated weeks by something like twenty to one, so people arrive at the appointment depressed, describe depression, and are treated for depression. The four good days last spring never come up, because they did not feel like an illness. They felt like being well.
📊 How often this gets missed, and why
📉 This is not a rare technicality. It is one of the most common diagnostic errors in all of mental health, and the mechanism behind it is entirely logical.
💡 The single most useful thing you can do before the next appointment. Ask a partner, a parent or an old friend one question: have you ever seen a stretch of several days where I barely slept, talked faster than usual, started things, and seemed somehow too much? Their answer is frequently the diagnosis. People close to you remember those weeks vividly; the person who lived them often does not register them at all.
🎚️ The spectrum, and where the lines are drawn
📐 Four patterns, distinguished by how high the highs go and how long they last. The clinical question is never "is this bipolar or not" but "which pattern is this", because the treatment differs between them.
| 🔎 Pattern | 📝 What defines it, and how common it is |
|---|---|
| 🔵 Unipolar depression | Depressive episodes only, nothing above normal mood ever. Antidepressants are the mainstay. Affects a large share of the population at some point |
| 🟣 Bipolar II | At least one hypomanic episode lasting four days or more, plus depressive episodes. Never a full manic episode. Affects roughly 1 in 100 people, and it is the one that gets called treatment-resistant depression |
| 🔴 Bipolar I | At least one manic episode lasting a week, or any mania needing hospital admission. Depression is common but not required for the diagnosis. Affects roughly 1 in 100 |
| 🟠 Cyclothymia | Two years or more of mood swinging up and down without either side reaching full criteria. Frequently mistaken for a personality trait rather than a treatable condition |
| ⚪ Mixed features | Depressed and activated at the same time — flat and hopeless but wired, agitated and unable to settle. This carries the highest risk of the lot, and antidepressants are particularly unhelpful here |
⬆️ What hypomania actually feels like
📘 The popular image is euphoria, which is why so many people say no when asked. A large share of hypomania is irritable, driven and unpleasant rather than elated, and the person experiencing it usually attributes it to circumstances.
| 🎯 What is actually happening | 💭 How people describe it at the time |
|---|---|
| 😴 Needing far less sleep, with no tiredness | The most specific sign there is. Four or five hours and genuinely fine the next day. People call it "a productive patch" — and this is different from insomnia, where you are exhausted |
| ⚡ Energy and output going up sharply | "I finally got on top of everything." Three projects started in a week, the garage cleared at two in the morning |
| 😡 Irritability rather than elation | "Everyone was being slow and stupid." The version most often missed, because it does not look like a high from the outside either |
| 🗣️ Talking faster, harder to interrupt | Others notice first. "You were going a mile a minute" is a report worth taking seriously |
| 🌀 Thoughts arriving faster than you can use them | "My brain was finally working properly." Jumping between topics, ideas feeling unusually connected |
| 💳 Decisions you would not normally make | Spending, a sudden resignation, uncharacteristic risk, driving faster. Judged afterwards as out of character, which is exactly the point |
| 🌟 Unusual confidence | Certainty about a plan that looks thin a month later. Not grandiose delusion — just markedly more sure than usual |
| 🎯 Starting a great deal and finishing little | Activity up, completion down. The pattern is visible in a bank statement or a to-do list long after the memory has faded |
🌙 The sleep question, asked properly. There is one distinction worth getting right, because it separates hypomania from ordinary insomnia and almost nobody is asked it clearly.
🔹 Insomnia: you want to sleep, you cannot, and you are exhausted the next day. This belongs to depression and anxiety.
🔹 Reduced need for sleep: you do not want to sleep, you do not feel you need to, and you are fine the next day. This belongs to hypomania and is close to diagnostic on its own.
🔹 So the question is not "did you sleep badly" but "did you need less sleep and still feel good". Those get completely different answers from the same person.
🚩 The clues hiding in a depression history
🔎 Even with no reported hypomania at all, certain features in the depressive history shift the odds substantially, and several of them are in every file already.
Early onset is one of the strongest single pointers
The strongest risk factor of all. A first-degree relative raises risk several-fold
Many short episodes fits bipolar better than unipolar
Hypersomnia, increased appetite and limbs that feel like lead
Postpartum onset carries a notably higher bipolar rate
Agitation, no sleep, racing thoughts on starting one. Not simply a side effect
Good response for months, then loss of effect, repeatedly, across different drugs
Episodes arriving at the same time of year, year after year
📌 No single item on that list proves anything. Three or four together, in someone being treated for recurrent depression, is a reason to ask the hypomania questions properly rather than reach for a fifth antidepressant.
🧪 Screening tools, and their honest limits
📋 Two short questionnaires are used, both self-completed in a few minutes. They are good at flagging people who need a proper assessment and poor at confirming anything, and treating a positive screen as a diagnosis is its own error.
🔹 The Mood Disorder Questionnaire. Thirteen yes-or-no questions about lifetime periods of elevated mood, energy and activity, plus two follow-ups: did several of these happen in the same period, and did they cause problems. Both follow-ups matter — thirteen scattered yes answers across twenty years mean little.
🔹 The Hypomania Checklist. Thirty-two items, more sensitive to the milder bipolar II pattern, and better at catching the irritable rather than euphoric version.
🔹 What they cannot do. Both produce a fair number of false positives, particularly where there is ADHD, a personality disorder, or ongoing substance use. A positive result starts a conversation; it does not end one.
🔹 What helps most and is almost never done. Bring someone who has known you ten years. A collateral history outperforms every questionnaire available.
⚠️ What antidepressants do when the diagnosis is wrong
📊 This is the reason any of it matters. An antidepressant given alone to someone with bipolar disorder can push them into hypomania or mania, and can make the pattern of episodes faster and harder to treat. The risk varies by drug class, and it drops sharply when a mood stabiliser is already in place.
🔁 There is a slower version of the same problem that causes more damage overall. Rapid cycling means four or more distinct episodes in twelve months, and long-term antidepressant use is one of the recognised drivers of it, alongside an underactive thyroid. It emerges over years, so the connection to the medication is rarely obvious to anyone involved. The first move is usually to withdraw the antidepressant, not to add another.
📊 Where the burden of bipolar II actually sits
⏳ Worth seeing, because it reframes what treatment is for. People expect bipolar treatment to be about controlling highs. In bipolar II it is overwhelmingly about the lows.
💊 What gets used instead
🎯 The principle is a drug that holds both poles rather than lifting one. Something stabilising goes in first, and anything else is added on top of it.
| 💊 Medicine | 🎯 Its place, and the practical detail |
|---|---|
| Lamictal (Lamotrigine) | The one aimed at the depressive pole, which is the pole that dominates bipolar II. Strong for preventing depressive relapse, weaker for treating an episode already underway. The titration is slow and non-negotiable because of a serious rash risk — see the next section. Little weight gain and no blood level monitoring, which is why people stay on it |
| Lithobid (Lithium) | Still the reference standard after seventy years, and the only one with solid evidence for reducing suicide. Covers both poles. Requires blood levels, kidney and thyroid monitoring, and real care with dehydration, anti-inflammatory painkillers and some blood pressure drugs |
| Seroquel (Quetiapine) | One of the few treatments with direct evidence in bipolar depression itself, and useful where sleep is wrecked. Sedation, appetite and weight are the trade-offs, with glucose and lipids monitored |
| Depakote (Divalproex) | Effective for the manic side. One absolute restriction: it must not be used in women who could become pregnant unless there is no alternative and a strict programme is in place, because of serious harm to a developing baby. Weight gain and tremor are common |
| Zyprexa (Olanzapine) | Reliable in acute mania and in mixed states. The metabolic cost is the highest of the group, so it is often a drug for getting control rather than for staying on for a decade |
| Tegretol (Carbamazepine) | An older option, still useful, with more drug interactions than any of the others. It speeds up the breakdown of many medicines, hormonal contraception included |
| Effexor (Venlafaxine) and other antidepressants | Not used alone in bipolar disorder. Where one is used, it sits on top of a mood stabiliser, at the lowest effective dose, and it gets reviewed rather than left running for years. What to expect in the first weeks of any of them is in why week two feels worse |
📈 Lamotrigine: why the slow start is not caution theatre
⚠️ Lamotrigine carries a small risk of a severe skin reaction, and that risk is directly tied to how fast the dose goes up. Introduced slowly it is one of the better tolerated medicines in psychiatry. Rushed, it is dangerous. The schedule exists for exactly that reason.
🚨 Three lamotrigine rules worth memorising.
🔹 Any new rash, particularly with fever, mouth or eye involvement, or blistering, means stop and seek advice today. The serious reaction almost always appears in the first eight weeks, and it is treatable when caught early.
🔹 Missing more than about five days means restarting the titration from the beginning. Resuming at the old dose after a gap reintroduces the risk. This catches people after a holiday or a lost prescription.
🔹 Combined hormonal contraception lowers lamotrigine levels substantially, and stopping the pill raises them again. Starting or stopping either one means the other needs reviewing.
🧪 Lithium, and the monitoring that makes it safe
📊 Lithium has a narrow margin between the level that works and the level that harms, which is the whole reason for the blood tests. Within that margin it remains the most effective mood stabiliser available, and it is the only one that reduces suicide risk in its own right.
⏳ How quickly any of it works
📅 Expectations get set badly here, and a drug abandoned at week three was never going to have declared itself by week three.
📒 Mood charting, which is the cheapest useful thing available
📋 Thirty seconds a day, and it answers questions no consultation can. It is also the only way to catch an episode while it is still small, which is when it is easiest to interrupt.
| 📝 Record daily | 🎯 What it tells you later |
|---|---|
| 😴 Hours slept, and whether you needed them | The most valuable single column. Sleep dropping while energy holds is the earliest reliable sign of an episode starting upward |
| 🎚️ Mood on a scale from minus five to plus five | Puts the swings on paper. Patterns invisible in memory become obvious over six months |
| ⚡ Energy and activity, separately from mood | They come apart in mixed states, which is the combination that carries the highest risk |
| 💊 Medication actually taken | Shows whether an apparent treatment failure was a treatment that was not being taken |
| 🍷 Alcohol and caffeine | Both destabilise, and both are badly underestimated by the person using them |
| 📅 Life events, in a few words | Travel across time zones, shift changes, nights without sleep. These precede episodes often enough to be worth tracking |
🛏️ What actually holds the ground between episodes
🧑⚕️ Monitoring, pregnancy and the practical admin
📋 Three things that need arranging rather than remembering, and that get neglected once things are stable.
Lithium level, kidney and thyroid function on lithium. Weight, glucose and lipids on quetiapine or olanzapine. Neither is optional, and both get skipped during good years
Valproate is out. Lithium and lamotrigine need specialist review before conception rather than after. Stopping everything unsupervised carries a high relapse risk, which is its own danger in pregnancy
Anti-inflammatory painkillers with lithium. Hormonal contraception with lamotrigine and carbamazepine. Worth mentioning the diagnosis at every pharmacy counter
🗿 Six things people get told that are wrong
| 💭 The belief | ✔️ What is actually the case |
|---|---|
| "Bipolar means wild mood swings day to day." | Episodes last days to months, not hours. Mood changing several times within a day points elsewhere, which is one reason bipolar is also over-diagnosed in some settings |
| "If I had been manic I would know." | Bipolar II involves hypomania, which is milder, often irritable rather than euphoric, and routinely experienced as finally feeling well |
| "Bipolar II is the mild one." | It involves more total time unwell than bipolar I and carries a comparable suicide risk. Less visible is not the same as less serious |
| "Antidepressants cannot hurt, so why not try another." | Given alone in bipolar disorder they can trigger an episode and speed up cycling. That is the single most consequential point on this page |
| "Lithium is an old drug that has been superseded." | It remains the most effective mood stabiliser available and the only one with evidence for reducing suicide. It needs monitoring, not replacing |
| "I am stable, so I can stop the medication." | Stopping, particularly abruptly, is the commonest precipitant of relapse, and abrupt lithium withdrawal carries a specific risk of a rebound episode |
🚨 When to act, and how fast
🚨 Now — emergency services or a crisis line
Thoughts of ending your life, or a plan · depressed and agitated at the same time, unable to sit still · not sleeping at all for several nights with rising energy · losing touch with reality, hearing or believing things others do not · spending or risk-taking that is about to cause serious harm · signs of lithium toxicity: vomiting, coarse tremor, slurred speech, unsteadiness · any rash while starting lamotrigine
🟡 This week
Sleep need dropping while energy rises · feeling unusually good and unusually certain after a long depression · a new antidepressant making you wired, agitated or sleepless · four or more mood episodes in the past year · your written early warning signs appearing · a missed week of lamotrigine · no lithium level in over six months
🟢 Book a proper reassessment
Three or more antidepressants that each worked and then stopped · depression starting before age 25, or after childbirth · a close relative with bipolar disorder · anyone who has ever described a four-day stretch of you as too much · depression with oversleeping and overeating · planning a pregnancy on a mood stabiliser · no blood monitoring in the past year
📋 What to bring to the appointment
🗣️ Five things that change the diagnosis, and that almost never get said.
🔹 Every antidepressant you have had, with dates, doses and what happened. "Each worked for about four months then stopped" is a sentence that redirects the whole assessment
🔹 Any period of four days or more with less sleep and more energy — including the ones you would describe as a good patch, a productive spell, or finally being yourself
🔹 Whether anyone in the family has bipolar disorder, lithium, or an unexplained psychiatric admission. Family history is the strongest single pointer available
🔹 Someone who has known you for years, in the room. This outperforms every questionnaire, and it is the one thing people never think to arrange
🔹 Your age at the first episode, and roughly how many there have been. Early onset and many short episodes both shift the odds
❓ And one question worth asking: given this history, has bipolar been formally considered and ruled out, or has it simply not come up?
❓ Frequently asked questions
How do I know if it is depression or bipolar II?
The difference is whether you have ever had four or more days of needing less sleep with more energy and activity. That period, not the depression, is what makes it bipolar.
What does hypomania feel like?
Often not euphoric. Needing less sleep without tiredness, faster speech and thoughts, more projects started, unusual certainty, and frequently irritability rather than elation.
Can antidepressants cause mania?
Yes, in people with underlying bipolar disorder. Risk runs from around 3% on an SSRI to 10 to 15% on a tricyclic, and falls below 3% when a mood stabiliser is also prescribed.
Why did every antidepressant stop working?
Repeated loss of response across different drugs is one of the recognised pointers towards bipolar disorder rather than resistant depression. It is worth a formal reassessment.
Is lamotrigine good for bipolar depression?
It is strong for preventing depressive relapse and weaker for treating an episode already underway. The titration must be slow, over six weeks, because of a serious rash risk.
What should I do if I get a rash on lamotrigine?
Stop the tablets and get medical advice the same day, especially with fever, blistering, or mouth and eye involvement. Serious reactions almost always appear in the first eight weeks.
Is bipolar II milder than bipolar I?
No. It involves more total time spent unwell and carries a comparable suicide risk. The highs are smaller, which makes it less visible rather than less serious.
Why does lithium need blood tests?
The gap between the effective and the harmful level is narrow. Maintenance usually targets 0.6 to 0.8, measured 12 hours after the dose, with kidney and thyroid checks too.
Does a positive screening questionnaire mean I have bipolar?
No. These tools flag people who need assessment and produce a fair number of false positives, particularly alongside ADHD or substance use. A positive result starts a conversation.
Why does sleep matter so much in bipolar disorder?
Irregular sleep timing is one of the most reliable triggers of an episode, and a single lost night can start hypomania. Consistent bed and wake times act as treatment, not advice.
📑 Sources and editorial
- 📘 Diagnostic criteria for bipolar I disorder, bipolar II disorder, cyclothymic disorder and the mixed-features specifier, including the four-day minimum duration for a hypomanic episode and the seven-day or hospitalisation criterion for mania
- 📊 Epidemiological estimates of bipolar spectrum prevalence, and studies of the proportion of patients presenting with unipolar depression who screen positive for or are later diagnosed with bipolar disorder
- ⏳ Studies of diagnostic delay in bipolar disorder, including mean years from first episode to correct diagnosis and the number of clinicians consulted beforehand
- 🔎 Research on clinical features of depressive episodes that predict a bipolar course, including age at onset, episode frequency, family history, atypical features, postpartum onset, psychotic features, seasonality and antidepressant-induced activation
- 😴 Literature on reduced need for sleep as a discriminating feature of hypomania, and its distinction from insomnia in unipolar depression
- 🧪 Validation studies of the Mood Disorder Questionnaire and the Hypomania Checklist, including sensitivity, specificity, false-positive rates, and the importance of the clustering and impairment items
- ⚠️ Meta-analyses of treatment-emergent mania and hypomania by antidepressant class, including tricyclics, serotonin-noradrenaline reuptake inhibitors and selective serotonin reuptake inhibitors, and the effect of concurrent mood stabiliser treatment
- 🔄 Reviews of rapid cycling bipolar disorder, including its association with long-term antidepressant exposure and with thyroid dysfunction, and guidance on withdrawing antidepressants as a first step
- 📈 Prospective studies of the proportion of weeks spent with depressive, hypomanic and euthymic symptoms in bipolar II disorder, and comparative data on total time unwell and suicide risk between bipolar I and bipolar II
- 💊 Clinical guidelines on the pharmacological management of bipolar disorder, including the position against antidepressant monotherapy and the recommended first-line options for acute mania, bipolar depression and maintenance
- 📉 Trial evidence for lamotrigine in the prevention of depressive relapse in bipolar disorder, its limited efficacy in acute bipolar depression, and its established maintenance role
- 🚨 Prescribing information and regulatory safety guidance for lamotrigine, including the mandated titration schedule, the relationship between titration rate and serious cutaneous reactions, dose adjustment with valproate and enzyme inducers, re-titration after an interruption, and the interaction with combined hormonal contraception
- 🧪 Guidance on lithium therapy, including target serum concentrations for maintenance and acute treatment, 12-hour post-dose sampling, monitoring intervals for serum level, renal and thyroid function, signs of toxicity, and interactions with non-steroidal anti-inflammatory drugs, diuretics and renin-angiotensin blockers
- 🩸 Evidence for the specific anti-suicidal effect of lithium in mood disorders, and studies of rebound episodes following abrupt lithium discontinuation
- 🌙 Randomised trial evidence for quetiapine in bipolar depression, and for olanzapine and other second-generation antipsychotics in acute mania and mixed states, with associated metabolic monitoring requirements
- 📜 Trial evidence and prescribing restrictions for valproate, including the regulatory position on use in women of childbearing potential, and for carbamazepine, including its enzyme-inducing interactions and cutaneous reaction risk
- 🤰 Guidance on the management of bipolar disorder before and during pregnancy, including the contraindication of valproate, specialist review of lithium and lamotrigine, and the relapse risk associated with unsupervised discontinuation
- 🕰️ Research on circadian rhythm and social rhythm regularity in bipolar disorder, on sleep deprivation as a trigger for hypomania, and on interpersonal and social rhythm therapy
- 📒 Trial evidence for structured psychoeducation, prodrome recognition and mood charting in reducing relapse rates, and for the role of collateral informants in diagnosis
- 🍷 Studies on alcohol and stimulant use as destabilising factors in bipolar disorder
- 📖 Related reading: how long depression treatment should last, antidepressant side effects and why week two feels worse, apathy in depression, inside the worry loop in generalized anxiety, insomnia medication compared
- 🛒 Related products: Lamictal (Lamotrigine), Lithobid (Lithium), Seroquel (Quetiapine), Zyprexa (Olanzapine), Depakote (Divalproex), Tegretol (Carbamazepine), anticonvulsant range, antidepressant range
- 👨⚕️ RXshop Editorial Team — reviewed by Thomas Walsh, MD — Psychiatrist and Mental Health Specialist
Medical Disclaimer: The information in this article is for educational and informational purposes only and is not a substitute for professional medical advice, diagnosis, or treatment. Always seek guidance from a qualified healthcare provider with any questions you may have regarding a medical condition, and before starting, stopping or changing any medication.