Is George a vulnerable narcissist, borderline, or something more complicated? Does Jerry’s obsession with cleanliness and order actually resemble OCD or an obsessive-compulsive personality style? Could Kramer meet criteria for antisocial personality disorder despite being charming, affectionate, and largely nonviolent? And is Elaine secretly the healthiest person in the group?

Along the way, we explore flanderization, narcissism, borderline personality organization, obsessive-compulsive personality traits, psychopathy versus antisocial personality disorder, gambling disorder, and the DSM-5 Alternative Model for Personality Disorders.

All of ADHD is an unbranded campaign aimed at prescribers. No drug appears anywhere on it. Its message: ADHD is more than inattention, hyperactivity, and impulsivity. Clinicians should be looking for executive dysfunction, emotional dysregulation, trouble starting and finishing tasks, poor sleep, low motivation. It cites emerging science implicating dopamine, norepinephrine, and serotonin, and notes that current treatments don’t address all of this for every patient.

The site was made by Otsuka a month ago…

Big surprise…. yesterday the FDA approved Otsuka’s centanafadine, the first triple reuptake inhibitor for ADHD. It works on norepinephrine, serotonin, and dopamine. In its trials, the outcomes that improved included emotional dysregulation and executive function.

Line the two up and you can see what actually happened. The campaign didn’t describe ADHD and then find a drug that fit. It started from the drug and worked backward. The trials measured emotional dysregulation, so emotional dysregulation became an underrecognized face of ADHD. The molecule hits three neurotransmitters, so ADHD became a three-neurotransmitter illness, and the old dopamine model became an outdated oversimplification. Even the name gives it away. All of ADHD. All the symptoms, all the neurotransmitters, all reachable by one product.

This is supposed to run in the other direction. Nature makes illnesses. Nosology is our attempt to describe what nature made: which symptoms cluster together, how they run in families, what course they follow, what they respond to. It’s slow, contested, imperfect work, and it belongs to clinicians and researchers watching actual patients over actual years.

What I watched this month was that work being done by a marketing agency. Someone in a conference room decided which symptoms belong to ADHD, and the deciding criterion was a receptor profile. The boundaries of a psychiatric illness were drawn to match a label.

Symptoms don’t come tagged with their diagnosis. Emotional dysregulation lives in borderline personality, in mood disorders, in trauma, in ordinary misery. Poor sleep and low motivation belong to half of psychiatry. Figuring out where a symptom actually comes from is the job. It’s most of the job. When an ad campaign does that sorting in advance, the public gets the message that every one of those symptoms defaults to ADHD, and the diagnoses underneath stop getting made. Any clinician doing this work has watched personality pathology settle in comfortably behind a cleaner label, often with the patient’s full cooperation, because “I have ADHD” is an easier thing to carry than the alternative.

None of this minimizes ADHD. I diagnose it and treat it every week, and I’ve watched the untreated version produce every consequence that website describes. The symptoms are real... what’s being manufactured is the degree of attribution to a particular diagnosis.

Dr. Phelps’ YouTube: https://www.youtube.com/@psycheducation

That’s because, with one critical exception, antipsychotics are remarkably similar in efficacy.

The CATIE trial spent $40 million to deliver an inconvenient truth: a first generation drug from 1957, perphenazine, performed comparably to the expensive second generation agents that displaced it. The differences that actually drive prescribing decisions are not about whether the drug works on positive symptoms. Almost all of them do. The differences are about side effects, the specific syndrome you’re treating, formulation, cost, and adherence.

This article is organized the way you actually think at the moment of decision. It is not a drug by drug encyclopedia. It is a framework for answering the only question that matters at the bedside:

Which one, and why?

The First Principle: Efficacy Is Mostly a Tie

Except Clozapine

For positive symptoms of schizophrenia, the antipsychotics are roughly interchangeable in efficacy. The exceptions you must commit to memory:

1. Clozapine is in a class of its own for treatment resistance and is the only antipsychotic that reduces all cause mortality.

2. Olanzapine and risperidone edge out the field modestly for raw positive symptom efficacy. Olanzapine had the longest time to discontinuation in CATIE.

3. Cariprazine has the best evidence for negative symptoms.

4. Everything else competes on tolerability, indication specific data, formulation, and price, not on whether it quiets hallucinations.

Once efficacy is off the table as a differentiator, side effects become the primary decision axis.

Step 1: Let Side Effect Avoidance Drive the Choice

Because efficacy is largely a tie, the fastest way to a good choice is to ask:

What side effect can this particular patient least afford?

Then eliminate the offenders.

The Metabolic Axis

Weight Gain

This is the dominant long term tolerability problem and the one patients most often discontinue over.

High risk: Clozapine, Olanzapine

Moderate risk: Risperidone, Quetiapine, Paliperidone

Low or near neutral risk: Ziprasidone, Aripiprazole, Cariprazine, Brexpiprazole, Lurasidone, Lumateperone, Xanomeline/trospium

Olanzapine averages 4 to 7 kg of weight gain and has the worst metabolic syndrome burden short of clozapine.

If your patient has obesity, prediabetes, or strong cardiometabolic risk, and you don’t have a compelling efficacy reason to reach for olanzapine, start somewhere in the low or near neutral group.

Don’t Just Avoid. Prevent.

If you do choose olanzapine or clozapine, start metabolic protection proactively rather than reactively.

Metformin: Start alongside olanzapine or clozapine from the outset. Otherwise, initiate once weight gain exceeds 3% in the first year. Dose 500 to 2000 mg/day. Expect roughly 3 to 4 kg attenuation.

GLP 1 agonists: These are becoming standard of care. Semaglutide produced a 13.88% body weight reduction vs. 0.42% on placebo over 36 weeks in clozapine treated patients. This is the most effective pharmacologic weapon we have for antipsychotic associated weight gain.

Topiramate: 25 to 300 mg/day is an option, but uptitrate no faster than 25 mg/week to limit cognitive dulling.

The Prolactin Axis

High risk: Risperidone, then Paliperidone, then Amisulpride

Modest risk: Olanzapine

Minimal or no risk: Quetiapine, Aripiprazole, Cariprazine, Lurasidone, Lumateperone, Xanomeline/trospium

Risperidone elevates prolactin in up to 70% of patients: galactorrhea, gynecomastia, amenorrhea, sexual dysfunction, and long term bone loss.

In a young woman, a man worried about sexual function, or anyone with osteoporosis risk, this is a real reason to look elsewhere.

Management by Baseline Prolactin Level

Greater than 100 ng/mL: Add aripiprazole 5 mg. This is the only reliably effective strategy, with an NNT near 1. Aripiprazole’s partial agonism normalizes prolactin even while you keep the offending agent on board.

50 to 100 ng/mL: Switch to aripiprazole, augment with aripiprazole 5 to 10 mg, or trial vitamin B6 200 to 1200 mg.

Less than 50 ng/mL: No strategy is convincingly effective. Reassure if asymptomatic.

Always rule out a prolactinoma before blaming the drug for markedly elevated levels.

The Akathisia Axis

Akathisia is the most underrecognized cause of:

“The medication is making me worse.”

Patients describe inner restlessness, can’t sit still, and it is frequently misread as agitation or worsening psychosis. That can lead to the catastrophic move of raising the dose.

Highest Risk Agents

Lurasidone: dose dependent, climbs sharply

Aripiprazole: especially at higher doses

Cariprazine: doubled vs. placebo

Treatment, Ranked by Effect Size

From a 15 RCT meta analysis:

1. Mirtazapine 15 mg: best effect size, with bonus benefits for mood and negative symptoms. Stay at 30 mg or lower. Higher doses can cause akathisia.

2. Vitamin B6 200 to 600 mg: best risk benefit ratio, excellent tolerability.

3. Trazodone

4. Propranolol 20 to 40 mg BID to TID: works, but the above edge it out.

5. Benzodiazepines: no better than placebo.

6. Benztropine: does not help akathisia and worsens TD. Do not use it for this.

Sedation, QT, and Anticholinergic Burden

Sedation: Quetiapine and olanzapine are the heavy hitters, which is exactly why quetiapine is so abused as a sleep aid. Useful when you want sedation. A liability when you don’t.

QT prolongation: Ziprasidone, quetiapine, and haloperidol warrant baseline ECG. Ziprasidone is contraindicated with other QT prolonging drugs.

Anticholinergic load: Clozapine, olanzapine, and quetiapine. Avoid in the elderly and in anyone where cognition is the treatment target.

Bottom Line

Pick the side effect your patient can least tolerate, cross off the agents that cause it, and you’ve usually narrowed ten drugs to three.

Step 2: Match the Drug to the Syndrome

Once tolerability has trimmed the list, the specific clinical syndrome usually picks the winner.

Schizophrenia

First Episode

First episode patients are exquisitely sensitive to side effects, and these early experiences shape lifelong adherence.

Favor tolerability.

First line, per the Harvard algorithm: aripiprazole, risperidone, ziprasidone. Add cariprazine if cost is acceptable.

Don’t dismiss perphenazine: CATIE showed it comparable to the SGAs at a fraction of the cost.

Violence or aggression history: olanzapine first line; clozapine if severe.

Negative Symptoms

First, rule out secondary negative symptoms:

1. Depression

2. Untreated positive symptoms

3. Substance use

4. Emotional flattening caused by excessive dopamine blockade itself

Often the fix is less antipsychotic, not more.

For primary negative symptoms:

1. Cariprazine: best evidence of any drug. NNT 9, effect size 0.31. Modest, but it’s the leader.

2. Clozapine: effect size 0.6 during acute psychosis; drives functional recovery.

3. Amisulpride in Europe, then olanzapine, asenapine, perphenazine.

4. Augment with antidepressants: duloxetine, vortioxetine, mirtazapine.

Cognitive Impairment

Lurasidone is the only antipsychotic with RCT evidence of cognitive improvement, and it beat quetiapine head to head on working memory testing.

Conversely, avoid quetiapine and all anticholinergics, including benztropine and trihexyphenidyl, when cognition matters.

Bipolar I

Acute Mania

Here efficacy genuinely separates the agents.

Effect sizes from meta analysis:

1. Risperidone: 0.82

2. Haloperidol: 0.74

3. Paliperidone: 0.71

4. Cariprazine: 0.57

5. Aripiprazole: 0.53

6. Olanzapine: 0.49

7. Quetiapine: 0.42

8. Asenapine: 0.37

9. Ziprasidone: 0.33

10. Brexpiprazole: 0.12

Placebo: 0.05

Risperidone is the most effective antimanic agent we have.

Haloperidol is nearly as good but can precipitate post manic depression. Use it for acute control, not maintenance.

Avoid brexpiprazole for mania. At 0.12, it is barely distinguishable from placebo.

Paliperidone works but failed FDA approval for mania due to inconsistent trial results.

Mania Dosing Is Not Schizophrenia Dosing

Some agents need more, some less.

Quetiapine: mania optimal 618 mg; schizophrenia optimal 482 mg

Aripiprazole: mania optimal 24.6 mg; schizophrenia optimal 21 mg

Risperidone: mania optimal 4.8 mg; schizophrenia optimal 5.7 mg

Haloperidol: mania optimal 10.1 mg; schizophrenia optimal 11 mg

Bipolar Depression

The depressive pole is where bipolar patients spend most of their disabled time, and antipsychotics are central.

FDA approved options, all with NNT around 4 except where noted:

Lurasidone: 20 to 120 mg; optimal 40 to 60 mg. NNT 4. Take with food. Cognitive benefits. Now generic, roughly $10 to $20/month.

Lumateperone: 42 mg QHS fixed. NNT 4. No titration. Novel mechanism. Weight neutral.

Olanzapine fluoxetine: NNT 4. Works, but worst metabolic burden.

Quetiapine: 300 mg. NNT 4. Anxiolytic, but sedation plus weight.

Cariprazine: 1.5 to 3 mg. NNT about 11. Weakest evidence in class.

Lurasidone is the default first choice: effective, cognitively favorable, weight neutral, and now dirt cheap.

Remember: it must be taken with a full 350 calorie meal or more. A partial meal makes absorption unpredictable. An empty stomach can render it subtherapeutic.

It is not for mania, never tested, and is a poor monotherapy choice for bipolar I if mania control is the goal.

Depression With Mixed Features

1. Lurasidone

2. Lumateperone

3. Cariprazine

MDD Augmentation

For unipolar depression that hasn’t remitted, dopamine partial agonists are the best studied add ons.

Across 16 RCTs and roughly 7,200 patients:

1. Aripiprazole 5.5 mg: SMD 0.38. Best evidence, generic, cheap. Start here.

2. Brexpiprazole 1.6 mg: SMD 0.31. If aripiprazole fails.

3. Cariprazine 1.5 mg: SMD 0.12 to 0.13. Very weak signal. Or pivot to esketamine for a different mechanism.

Critical Dosing Pearl

All three plateau at these low doses.

Pushing aripiprazole to 15 mg for depression augmentation buys you akathisia, not efficacy.

A reality check on the esketamine hype: head to head, esketamine produced 27.1% remission vs. quetiapine’s 17.6% in TRD. Statistically real, but the roughly 2.8 point gap on a 60 point scale is arguably not clinically meaningful, and esketamine costs vastly more with REMS burden.

Where Antipsychotics Disappoint

Be Honest

Delirium: Haloperidol did not improve any meaningful outcome in a 1,000 patient ICU RCT. Nonpharmacologic management is first line. Skip antipsychotics for mild to moderate delirium.

Dementia agitation: Brexpiprazole earned the first ever FDA approval here in May 2023, but the effect is clinically thin: a 5 point reduction on a 174 point scale when the threshold for clinical significance is around 17 points, against a black box mortality signal. Nonpharmacologic approaches first, always.

Dementia with Lewy bodies: Antipsychotics are largely contraindicated due to NMS, delirium, and death.

PTSD: Brexpiprazole failed FDA approval. Only one of two pivotal trials was positive, and only in women. Second line at best.

Bottom Line

The syndrome usually names the drug:

Risperidone for mania

Lurasidone for bipolar depression

Cariprazine for negative symptoms

Aripiprazole for MDD augmentation

When it doesn’t, you’re back to tolerability and cost.

Step 3: Know When to Stop Cycling and Reach for Clozapine

This is the single most important, and most neglected, decision in psychiatry.

Only about 5% of US schizophrenia patients are on clozapine, versus roughly 30% in many other countries.

That gap represents preventable death.

Why Clozapine Is Different

1. It is the only antipsychotic with proven mortality benefit: roughly 50% lower all cause mortality across 24 studies and 200,000 plus life years.

2. Response in treatment resistance: 40% to 60% overall.

3. It uniquely reduces substance use disorder risk in schizophrenia.

4. It carries no TD risk and minimal EPS or akathisia.

5. It is FDA approved for suicidality in schizophrenia even without any prior antipsychotic failure.

The Timing Imperative

Clozapine’s efficacy is time sensitive. This is the fact that should change your practice:

Started within 3 years of the first antipsychotic failure: 82% response

Delayed beyond 3 years: response drops to 32%

Each additional failed trial: response falls 8% to 11%

So the conventional ritual of cycling endlessly through SGA after SGA is actively harming patients.

After two adequate antipsychotic trials fail, go to clozapine, and do it sooner rather than later.

Dismantling the Excuses

The barriers to clozapine are mostly myths.

“Patients won’t tolerate it.” Patients actually stay on clozapine longer than other antipsychotics.

“The monitoring is too burdensome.” The FDA removed the REMS program. And when surveyed, patients rank tolerability and side effects far above monitoring burden as concerns. Monitoring is rarely the real barrier. Clinician discomfort is.

“You have to admit them to start it.” Outpatient titration is entirely feasible on a slower schedule.

Dosing and Monitoring

Dose: 200 to 600 mg/day divided, up to 900 mg. Titrate slowly.

ANC monitoring: weekly for 6 months, then every 2 weeks for 6 months, then monthly thereafter.

Baseline: CBC, ECG, metabolic panel, prolactin, weight.

The Side Effects That Actually Kill

Agranulocytosis, around 1% and mostly in the first 18 weeks, gets the attention.

But the two leading causes of clozapine death are mundane and preventable:

1. Constipation: up to 60%, progressing to ileus. Start a bowel regimen prophylactically. Treat it as the life threatening complication it is.

2. Sialorrhea: can lead to aspiration pneumonia. Don’t dismiss drooling as cosmetic.

Also watch for:

1. Myocarditis: weeks 6 to 8

2. Seizures: 3% at 300 to 600 mg, 5% above 600 mg

3. Orthostasis or syncope: early

4. OCD symptoms: roughly 10%

5. Nocturnal enuresis

Interactions That Move Levels

Smoking: CYP1A2 induction drops clozapine levels by roughly 50%. When a patient quits, or is hospitalized on a smoke free unit, levels can spike dangerously. Monitor at any change in smoking.

CYP2D6 or 3A4 inhibitors: Fluoxetine, paroxetine, fluvoxamine, and sertraline 150 mg or above raise levels. Titrate slower.

Bottom Line

Two failed trials equals clozapine, now.

Every year you wait, you are trading away response rate. The monitoring is not the obstacle you think it is.

Step 4: Will the Patient Actually Take It?

LAIs and Adherence

The best antipsychotic is the one that reaches the bloodstream.

Nonadherence is the leading driver of relapse, and long acting injectables are underused.

Available LAIs

Risperdal Consta: every 2 weeks

Invega Sustenna: monthly

Invega Trinza: quarterly

Abilify Maintena: monthly

Aristada: monthly or quarterly

LAIs delay relapse in both schizophrenia and bipolar disorder. Aripiprazole and risperidone LAIs delay relapse when patients stop taking oral meds.

Aripiprazole LAI is the most forgiving: its terminal half life of about 46.5 days means a missed injection is far less catastrophic than with shorter acting formulations.

Don’t reserve LAIs for the “noncompliant.” Offer them early, including in first episode patients, where they may meaningfully alter the disease trajectory.

Step 5: Adjust for Special Populations

The Elderly

All antipsychotics carry a black box mortality warning in dementia. Mortality is roughly doubled. Reserve for genuine need.

TD risk is 5% per year regardless of drug class. Age is an independent risk factor that eclipses the FGA vs. SGA distinction. Lower doses don’t protect.

Preferred: aripiprazole for clean profile; quetiapine when sedation or anxiolysis is desired.

Watch orthostasis, falls, and anticholinergic load. Start low, go slow.

Pregnancy

Untreated psychosis or mania usually poses greater risk than the medication.

SGAs show low teratogenicity. Olanzapine, risperidone, and aripiprazole are reasonable.

Postpartum psychosis: antipsychotics are first line.

Substance Use Comorbidity

Clozapine uniquely reduces substance use disorder in schizophrenia, a reason to consider it earlier in this population.

How to Switch Antipsychotics

When you do change agents, how you switch matters.

The best supported method, from a 6 RCT meta analysis:

Add the new agent, continue the old one for 1 to 2 weeks, then taper the old one over 1 to 2 weeks.

That is, cross taper.

Abrupt switching carries 1.6 times higher dropout.

And for clozapine specifically: never stop it abruptly. Rebound psychosis is severe. Always cross taper.

Tardive Dyskinesia

Screening and the Modern Algorithm

TD is permanent often enough that prevention and early detection are non negotiable.

Incidence: FGAs 5.5% per year, SGAs 2.6% per year. Lifetime risk is 20% to 30% after a decade.

Over age 50: both classes converge at around 5% per year.

Top risk factors: age above 50, mood disorders, substance use, brain injury, diabetes, prior EPS.

Screen: AIMS at baseline and every 6 to 12 months. The free AI based TDScreen app reaches 85% to 98% accuracy.

Management Algorithm

1. Taper or discontinue the offending agent if feasible. About 30% to 50% resolve. Expect transient worsening: withdrawal dyskinesias can last a few months.

2. Switch to: clozapine, then aripiprazole, then olanzapine.

3. Do not switch to quetiapine. Counterintuitively, meta analysis gives it the highest TD risk among common SGAs despite low D2 occupancy. This surprises people. Remember it.

4. Add a VMAT2 inhibitor: valbenazine 40 to 80 mg/day or deutetrabenazine 12 to 48 mg BID with food. TD returns within a month of stopping. These are maintenance therapies.

5. Second line: amantadine, ginkgo biloba, levetiracetam.

For drug induced parkinsonism, prefer dose reduction or a switch over benztropine. Anticholinergics worsen TD and impair cognition.

The New Frontier: Xanomeline/Trospium

Cobenfy

For the first time in 70 plus years, we have an antipsychotic with a genuinely new mechanism:

Muscarinic M1/M4 agonism, with xanomeline paired with peripheral anticholinergic trospium to blunt GI side effects, and no dopamine blockade whatsoever.

Efficacy: schizophrenia effect size 0.56, comparable to standard agents.

The headline: because there is no D2 blockade, there is no akathisia, no EPS, no TD, no hyperprolactinemia, no weight gain, no metabolic burden, no sedation, no orthostasis, and no QTc prolongation.

The tradeoff: cholinergic GI effects, including nausea, vomiting, constipation, dry mouth; hypertension; elevated triglycerides; urinary retention, especially in the elderly; and rare hepatotoxicity or angioedema.

Dosing: start 50/20 mg BID, titrate to 125/30 mg BID. Short half life mandates BID.

Interactions: CYP2D6 inhibitors raise levels, so slow the titration. Do not combine with anticholinergic antipsychotics such as clozapine or olanzapine.

What We Don’t Know

It has not yet been tested in treatment resistant schizophrenia, though a trial is underway, and it is brand name expensive.

But for a patient who cannot tolerate metabolic or movement side effects, this is a legitimately new tool, not just another me too.

Monitoring: The Non Negotiables

Weight, BMI, waist: baseline yes; ongoing every visit, with BMI every 3 to 6 months.

Fasting glucose or HbA1c: baseline yes; ongoing at 3 months, then annually.

Lipid panel: baseline yes; ongoing at 3 months, then annually.

Prolactin: baseline if using risperidone or paliperidone; ongoing only if symptomatic.

ECG or QTc: baseline if QT concern, including ziprasidone, quetiapine, or haloperidol; repeat if dose increases, symptoms develop, or a new QT drug is added.

AIMS or TDScreen: baseline yes; ongoing every 6 to 12 months.

BP with orthostatics: baseline yes; ongoing as indicated.

Clozapine ANC: baseline yes; weekly for 6 months, then every 2 weeks for 6 months, then monthly.

The One Page Decision Framework

1. Start with side effects. Efficacy is mostly a tie. Identify the side effect this patient can least afford and eliminate the offenders.

2. Match to syndrome. Risperidone for mania. Lurasidone, with food, for bipolar depression. Cariprazine for negative symptoms. Aripiprazole 5.5 mg for MDD augmentation. Olanzapine for violence.

3. Two failures equals clozapine. Now, not later. Within 3 years buys you 82% response vs. 32%. The monitoring is not the real barrier.

4. Ask about adherence. Offer an LAI early. Aripiprazole LAI is the most forgiving.

5. Adjust for the population. Black box mortality in dementia; age independent TD risk in elders; olanzapine, risperidone, or aripiprazole in pregnancy.

6. Switch by cross taper, never abruptly. And never abruptly off clozapine.

7. Screen for TD every 6 to 12 months. If it appears, don’t switch to quetiapine. Do reach for clozapine, aripiprazole, or a VMAT2 inhibitor.