Depression & Anxiety
The most prevalent treatable conditions in primary care, and the internist manages most of them without a psychiatrist. Four decisions carry the weight: screen and score it, exclude bipolar disease and the medical mimics before prescribing, pick the first drug by side-effect profile rather than efficacy, and give every trial an adequate dose for an adequate time before calling it a failure. Ask about suicide directly; asking does not plant the idea.
🔍 Overview
The Diagnoses in Brief
- Major depressive disorder: ≥ 2 weeks of depressed mood or anhedonia (one of the two is required) plus enough of: sleep change, guilt or worthlessness, low energy, poor concentration, appetite change, psychomotor change, suicidal ideation -with functional impairment. Anhedonia matters because patients who deny "feeling sad" will still report that nothing is enjoyable.
- Generalized anxiety disorder: ≥ 6 months of excessive, hard-to-control worry across domains, with restlessness, fatigue, poor concentration, irritability, muscle tension, or sleep disturbance.
- Panic disorder: recurrent unexpected panic attacks plus a month of worry about further attacks or avoidance behavior. The attacks peak within minutes -a "panic attack" building over hours is something else.
- Adjustment disorder: symptoms tied to an identifiable stressor, out of proportion but not meeting MDD criteria -support and monitoring rather than reflexive prescribing.
- They travel together: half of depressed patients have significant anxiety, and comorbid presentations respond to the same first-line drugs, which is why SSRIs anchor both algorithms.
Medical Mimics and Contributors (look before labeling)
- Hypothyroidism, B12 deficiency, anemia, OSA -the treatable somatic causes of fatigue, poor concentration and low mood. TSH and CBC are cheap; snoring plus daytime somnolence earns a sleep study.
- Substances and medications: alcohol (both cause and self-treatment), stimulant withdrawal, glucocorticoids, interferon, isotretinoin. Ask about alcohol quantitatively -it is the most commonly missed driver.
- The cardiac masquerade runs both directions: panic attacks get worked up as ACS, and real hyperthyroidism, arrhythmia, PE or hypoglycemia get labeled "anxiety." First presentations of "panic" in older patients deserve the organic workup.
Screen for bipolar disorder BEFORE starting any antidepressant. Ask directly about prior episodes of decreased need for sleep with increased energy, racing thoughts, pressured speech, grandiosity, or impulsive spending and sexual behavior -or use the MDQ. Antidepressant monotherapy in unrecognized bipolar disorder can precipitate mania, and a bipolar patient's first presentation is usually a depressive episode, which is exactly when the wrong prescription gets written. Family history of bipolar disease raises the stakes further.
Ask about suicide directly, every time. "Are you having thoughts of killing yourself?" does not plant the idea -that fear is disproven and it is the main reason clinicians under-ask. Assess ideation, plan, intent, means and protective factors (Columbia protocol), and ask specifically about firearms in the home: means restriction is one of the few interventions with strong evidence for preventing suicide deaths. Active ideation with plan and intent is an emergency, not a referral.
🧪 Screening & Workup
Who Gets Screened (USPSTF 2023)
- Depression: all adults, including pregnant and postpartum patients (Grade B). Use the PHQ-2 as the gate (a positive is ≥ 3), then the PHQ-9 to confirm and grade.
- Anxiety: adults 64 and younger (Grade B), with evidence rated insufficient at 65+. GAD-7 is the standard instrument.
- Screening only works with a follow-through system: a positive score needs a diagnostic conversation, not just a number in the chart.
Scoring and What It Buys You
- PHQ-9: 5 mild, 10 moderate, 15 moderately severe, 20 severe. ≥ 10 is the usual treatment threshold, and the score tracks response over time -remission is < 5, response is a ≥ 50% drop. Re-administer at every visit; treating without re-measuring is like titrating insulin without glucose values.
- GAD-7: same 5/10/15 banding; ≥ 10 warrants treatment discussion.
- Item 9 (thoughts of self-harm) is a trigger, not a verdict -any positive response gets the direct suicide conversation the same day.
Baseline Workup
- TSH, CBC, CMP, B12 where the picture includes fatigue, cognitive complaints, or age -the same cheap treatable list as the dementia workup, for the same reason.
- Alcohol and substance history taken quantitatively (AUDIT-C), sleep history including OSA screening, and a medication review for contributors.
- ECG before or shortly after starting citalopram in patients with cardiac disease or other QT drugs -citalopram carries an FDA dose cap of 40 mg/day (20 mg in the elderly and hepatic impairment) for dose-dependent QT prolongation.
- Document the episode history: first episode vs recurrent changes treatment duration; prior response to a specific drug predicts response again and is the single best selection clue.
🚨 Depression Treatment
Matching Treatment to Severity
- Mild (PHQ-9 5-9): psychotherapy (CBT has the best evidence), behavioral activation, exercise (a genuine, underprescribed antidepressant), and watchful waiting with re-scoring -medication is optional here, not obligatory.
- Moderate (10-14): psychotherapy or an antidepressant; patient preference is a legitimate tiebreaker because efficacy is comparable and adherence follows preference.
- Moderately severe to severe (≥ 15): medication, ideally combined with psychotherapy -the combination beats either alone in moderate-to-severe disease.
- Psychotic features, catatonia, refusal to eat or drink, or high acute suicide risk: psychiatry now, and remember ECT is the most effective treatment in the field for exactly these presentations -it is a modern, safe therapy, not a historical one.
Choosing the First Drug (by side-effect profile, not efficacy)
| Drug | Dose (start → usual) | Why choose it / Watch for |
|---|---|---|
| Sertraline (Zoloft) FIRST-LINE | 25-50 mg → 50-200 mg daily | Well tolerated, safe in cardiac disease, the default in pregnancy and lactation. GI upset early; sexual dysfunction common to the class. |
| Escitalopram (Lexapro) FIRST-LINE | 5-10 mg → 10-20 mg daily | Among the best tolerated; few interactions. Mild dose-dependent QT effect (much less than citalopram). |
| Citalopram (Celexa) | 20 mg → max 40 mg (20 mg if elderly or hepatic impairment) | FDA QT dose cap -check ECG with cardiac disease or other QT drugs. Otherwise similar to escitalopram. |
| Fluoxetine (Prozac) | 10-20 mg → 20-60 mg daily | Long half-life self-tapers, so best for patients who miss doses; activating (helpful for hypersomnia and low energy, harder in anxiety). Most CYP interactions of the class. |
| Paroxetine (Paxil) | 10-20 mg → 20-50 mg daily | Most anticholinergic, most weight gain, worst discontinuation syndrome, avoided in pregnancy -generally the last SSRI to reach for despite equal efficacy. |
| Venlafaxine XR (Effexor) | 37.5-75 mg → 75-225 mg daily | SNRI; effective in depression and anxiety. Dose-dependent hypertension (check BP at titration) and a harsh discontinuation syndrome -taper always. |
| Duloxetine (Cymbalta) | 30 mg → 60 mg daily | SNRI that also treats neuropathic pain and fibromyalgia -the rational pick when depression and chronic pain coexist. Avoid in heavy alcohol use and hepatic disease. |
| Bupropion (Wellbutrin) | 150 mg → 300-450 mg daily (XL) | No sexual dysfunction, no weight gain, activating -the switch or add-on when SSRIs cause either. Aids smoking cessation. Contraindicated in seizure disorders and active eating disorders. Not for prominent anxiety. |
| Mirtazapine (Remeron) | 7.5-15 mg → 15-45 mg qHS | Sedation and appetite stimulation are the therapy in the depressed elderly patient with insomnia and weight loss. More sedating at LOWER doses. Minimal sexual dysfunction. |
| Trazodone | 25-100 mg qHS | At low dose it is a sleep aid, not an adequate antidepressant -using it as monotherapy for depression undertreats the depression. |
Running the Trial Properly
- Start low, titrate to a therapeutic dose, and judge nothing before 4-6 weeks at that dose (full effect 8-12 weeks). Most "failed" antidepressants were underdosed, under-timed, or not taken -confirm all three before switching.
- Expect several attempts, and say so at the start. STAR*D, 2006 found remission in only about 37% on the first antidepressant, with each subsequent step yielding less (roughly 31%, then 14%, then 13%). Framing this honestly at the outset protects the alliance, because a patient who was promised the first drug would work and is now on their third reasonably concludes that treatment is failing, when in fact this is the expected course.
- Warn about the first two weeks: GI upset, jitteriness and sometimes worse anxiety come before benefit; patients who expect this stay on the drug. Energy can return before mood lifts, which is part of why early weeks carry a boxed warning for emergent suicidality in patients under 25 -schedule early follow-up.
- Ask about sexual dysfunction at follow-up -patients rarely volunteer it and it is a leading cause of silent nonadherence. Bupropion (switch or add-on) is the usual answer.
- Inadequate response after an adequate trial: optimize the dose, then switch (within class or to an SNRI/bupropion/mirtazapine) or augment (bupropion or mirtazapine added, or an atypical antipsychotic such as aripiprazole). Two adequate failed trials = treatment-resistant depression: refer, where esketamine and ECT live.
- Continue 6-12 months after remission for a first episode -stopping at "I feel better" roughly doubles relapse. Two or more prior episodes, or a severe episode: discuss maintenance therapy long-term.
- Stopping: taper over weeks (except fluoxetine). Discontinuation syndrome -dizziness, "brain zaps," flu-like symptoms, irritability -is worst with paroxetine and venlafaxine and is routinely misread as relapse.
Safety Small Print
- Hyponatremia (SIADH) in the elderly, usually in the first weeks -check sodium if confusion, falls or headache appear after starting.
- Bleeding risk rises with NSAIDs, antiplatelets and anticoagulants (platelet serotonin depletion) -relevant, not prohibitive; consider GI protection in the stacked patient.
- Serotonin syndrome with MAOIs (contraindicated), linezolid, tramadol, and stacked serotonergic agents -see the NMS vs serotonin syndrome topic for the recognition grid.
- Peripartum: untreated depression carries real fetal and maternal risk; sertraline is the usual choice in pregnancy and lactation, paroxetine the one to avoid. Treatment is usually the safer path, decided jointly.
💭 Anxiety Treatment
First-Line for GAD and Panic Disorder
- SSRIs and SNRIs are first-line for both, at the same target doses as depression -but start at half the usual starting dose, because anxious patients are exquisitely sensitive to the early activation and a bad first week ends the treatment. Sertraline 12.5-25 mg or escitalopram 5 mg are typical openings.
- Set the timeline out loud: benefit begins at 2-4 weeks and matures over 8-12. Anxious patients abandon treatment fastest, and the expectation-setting is itself therapeutic.
- CBT is as effective as medication for GAD and panic disorder, with more durable benefit after stopping -offer it to everyone, combined for moderate-to-severe disease.
- Buspirone -adjunct or alternative for GAD (not panic): no dependence, no sedation, needs 2-4 weeks and BID-TID dosing. Useful where benzodiazepines must be avoided.
- Panic disorder specifics: name the diagnosis and explain the physiology -"the attack is a false alarm, it cannot hurt you, and it will peak and pass within minutes" reduces attack frequency by itself. Interoceptive-exposure CBT is the definitive therapy.
- Propranolol 10-40 mg before performances for situational performance anxiety -it treats tremor and tachycardia, not worry, and has no role in GAD.
Benzodiazepines are a bridge at most, and usually not even that. They work immediately, which is exactly why they entrench: avoidance is reinforced, tolerance and dependence develop within weeks, and in the elderly they add falls, fractures and cognitive impairment. If used at all, define the duration in the first prescription (2-4 weeks while an SSRI takes hold), avoid PRN dosing for panic (it teaches the patient the pill ended an attack that would have ended anyway), and do not start them in patients with substance use history or on opioids -the combination carries a boxed warning for fatal respiratory depression. Long-term users need a slow taper, not an abrupt stop (withdrawal seizures).
🏥 Rounds
Pimp Questions
❓ Why must you screen for bipolar disorder before starting an antidepressant?
Antidepressant monotherapy in unrecognized bipolar disorder can precipitate mania or induce rapid cycling. The trap is structural: bipolar disease usually presents first as a depressive episode, so the patient in front of you asking for help with depression is exactly the patient who might be bipolar. Ask directly about past episodes of decreased need for sleep with increased energy, racing thoughts, pressured speech, and impulsive behavior, plus family history -or use the MDQ. If positive, the treatment is a mood stabilizer pathway with psychiatry, not an SSRI.
❓ A patient stopped venlafaxine three days ago and now has dizziness, electric "brain zaps," nausea and irritability. Relapse?
No -discontinuation syndrome. Venlafaxine and paroxetine have the shortest half-lives of the commonly used agents, so abrupt cessation produces dizziness, paresthesias and "brain zaps," flu-like symptoms, insomnia and irritability within days -too fast for relapse, which takes weeks and reproduces the original mood symptoms rather than neurologic ones. Management: restart and taper over weeks. Fluoxetine rarely causes this because its long half-life self-tapers, which also makes it a bridge option for hard tapers.
❓ Which antidepressant would you pick for a depressed elderly patient with insomnia and weight loss, and why?
Mirtazapine. Its two dominant side effects -sedation and appetite stimulation -are precisely the two symptoms that need treating, so the adverse-effect profile becomes the therapy. Dose at bedtime; note it is more sedating at lower doses (antihistamine effect dominates before noradrenergic effects arrive). It also causes minimal sexual dysfunction. Watch sodium in the elderly as with any antidepressant, and reassess weight long-term.
❓ Why is citalopram the one SSRI with a hard dose ceiling?
Dose-dependent QT prolongation led the FDA to cap citalopram at 40 mg/day, and 20 mg/day in patients over 60 or with hepatic impairment. Escitalopram (its active enantiomer) has a much smaller effect. Practical consequences: check an ECG in patients with cardiac disease, electrolyte disturbances or other QT-prolonging drugs, correct potassium and magnesium, and if a patient "failed citalopram 40 mg," the next move is a different agent -not more citalopram.
❓ Why are benzodiazepines discouraged as primary therapy for panic disorder when they stop attacks so effectively?
Because their immediacy is the mechanism of harm. A PRN benzodiazepine taken during an attack teaches the brain that the pill ended an event that would have self-terminated in minutes anyway -reinforcing both the fear of attacks and the dependence on rescue. Tolerance and physiologic dependence follow within weeks, elderly patients add falls and cognitive impairment, and combination with opioids carries a boxed warning for fatal respiratory depression. First-line is an SSRI started low plus CBT; if a benzodiazepine is used as a bridge, it is scheduled, short, and has an end date written into the first prescription.
Sample Presentation
Ms. Trần is a 34-year-old woman presenting with 3 months of low mood, early-morning awakening, poor concentration at work, and loss of interest in running, which she previously loved. PHQ-9 scores 16; item 9 positive for passive thoughts ("better off not waking up") without plan or intent. GAD-7 is 12. She drinks 2-3 glasses of wine nightly "to sleep." TSH and CBC normal. No prior manic symptoms on direct questioning; no family history of bipolar disorder. No firearms at home.
Key Points: (1) Moderately severe MDD with comorbid anxiety -the plan is an SSRI plus CBT, and with her anxiety, start sertraline at 25 mg with the first-two-weeks warning before titrating to 50-100 mg. (2) The item-9 answer required today's direct conversation: passive ideation without plan, intent or means access, with protective factors -document the assessment, safety-plan, and see her back in 1-2 weeks (she is under 25? No -but early follow-up still, given ideation). (3) The nightly wine is doing real work here -both depressogenic and the reason for the 3 AM awakenings -so it is named as part of the prescription: cut it, because alcohol will quietly defeat the SSRI. Re-score the PHQ-9 at every visit; success is defined as < 5, not "somewhat better."
Clinic Checklist
- PHQ-9/GAD-7 scored today and trended -treating without re-measuring is titrating blind.
- Suicide asked about directly; any item-9 positive gets plan/intent/means assessment and a safety plan. Firearms asked about specifically.
- Bipolar screen documented before the first antidepressant.
- Alcohol quantified, sleep apnea considered, TSH/B12 checked where the picture fits.
- Dose is therapeutic and the trial is 4-6+ weeks old before any switch; sexual side effects asked about, not waited for.
- Continuation plan stated: 6-12 months past remission for a first episode; maintenance discussed after two.
- Any benzodiazepine has a written end date, and none started with opioids or substance-use history.
- Under-25 patients seen or contacted within 1-2 weeks of starting (boxed warning window).
📋 Summary
Summary
Screen Everyone, Score Everything
USPSTF 2023: screen all adults for depression (including pregnancy and postpartum) and adults up to 64 for anxiety. PHQ-2 gates to PHQ-9; thresholds run 5/10/15/20, with ≥ 10 the usual treatment trigger, remission < 5. Re-score at every visit -treating depression without trending the PHQ-9 is titrating insulin without glucose values.
Two Questions Before the Prescription
Bipolar screen first -antidepressant monotherapy in unrecognized bipolar disease can precipitate mania, and bipolar illness usually debuts as depression. Suicide asked directly, every time -asking does not plant the idea. Ideation gets plan, intent and means assessment; firearms get asked about by name, because means restriction is one of the few interventions proven to prevent deaths.
Rule Out the Mimics
Hypothyroidism, B12 deficiency, anemia, OSA, and alcohol -the treatable somatic story behind fatigue, poor concentration and low mood. Quantify the drinking: it is the most commonly missed driver and it quietly defeats the antidepressant. In older patients, a first "panic attack" deserves the organic workup (hyperthyroidism, arrhythmia, PE, hypoglycemia) before the label.
Pick the Drug by Its Side Effects
Efficacy is comparable across agents, so the profile is the selection tool: sertraline or escitalopram as defaults; duloxetine when chronic pain coexists; bupropion when sexual dysfunction, weight gain or low energy dominate (never in seizures or eating disorders); mirtazapine for the elderly patient with insomnia and weight loss; fluoxetine for the dose-misser (self-tapers); paroxetine rarely. Citalopram is capped at 40 mg (20 mg elderly) for QT. Prior response predicts response -ask what worked before.
Run an Adequate Trial or You Learn Nothing
Therapeutic dose for 4-6 weeks minimum (full effect 8-12) before judging. Most "failures" were underdosed, under-timed or untaken -confirm all three, and ask about sexual dysfunction, the leading silent adherence killer. Then switch or augment (bupropion, mirtazapine, or an atypical). Two adequate failures = treatment-resistant: refer, where esketamine and ECT live -and ECT remains the most effective treatment for severe, psychotic or catatonic depression.
The Calendar Is Part of the Prescription
Continue 6-12 months after remission for a first episode -stopping at "I feel better" roughly doubles relapse. Two or more episodes: discuss maintenance. Taper on the way out (worst offenders venlafaxine and paroxetine; fluoxetine self-tapers) and teach the difference between discontinuation symptoms (days, brain zaps, dizziness) and relapse (weeks, the original symptoms).
Anxiety: Same Drugs, Gentler Start
SSRIs/SNRIs first-line for GAD and panic, started at half-dose because early activation is what makes anxious patients quit in week one -say the timeline out loud. CBT equals medication with more durable benefit. Buspirone adjuncts GAD (not panic); propranolol covers performance situations only. For panic, naming the physiology ("it peaks and passes; it cannot hurt you") is itself treatment.
Benzodiazepines: The Immediacy Is the Trap
They work instantly, which is how they entrench: PRN use teaches the patient the pill ended an attack that would have ended itself, avoidance is reinforced, dependence follows within weeks, and the elderly add falls and cognitive harm. Bridge use only, with an end date in the first prescription; never with opioids (boxed warning) or substance-use history; long-term users taper slowly (withdrawal seizures).
