The seasoned approach
Refractory insomnia management beyond the standard sleep-medicine algorithm, in the longevity-psychiatry frame. The discipline is to find and treat the underlying contributor, then layer the lowest-cognitive-cost pharmacology that produces restorative sleep — not the fastest sedative.
Layer 1 — First — confirm CBT-I has actually been done
Most patients labeled "refractory insomnia" have not actually had CBT-I. Confirm the patient has completed a structured course (6–8 sessions, in-person or via Somryst/Sleepio/SHUTi). If not, refer; the behavioral foundation must be addressed before declaring refractoriness.
Layer 2 — Workup for the treatable contributor
Sleep study (PSG or home test) to rule out OSA, PLMD, REM behavior disorder. Hormonal evaluation in perimenopausal women. Depression and anxiety reassessment. Medication audit (caffeine, late stimulants, alcohol, exercise timing). Pain assessment. Thyroid panel. Iron studies if RLS suspected. The workup frequently identifies the contributor that explains the refractoriness.
Layer 3 — DORAs as first-line refractory pharmacology
Suvorexant 10–20mg, lemborexant 5–10mg, daridorexant 25–50mg. Orexin antagonists with favorable safety profile in older adults, less fall risk than benzodiazepines/Z-drugs, less cognitive impact. Insurance often requires step therapy; document prior failures. The most evidence-based newer option.
Layer 4 — Low-dose doxepin for sleep maintenance
3–6mg qhs. Highly H1-selective at low doses, minimal anticholinergic burden at these doses (distinct from doxepin at antidepressant doses). FDA-approved for sleep maintenance specifically. Useful in older adults and patients with prominent middle-night awakening. Compounded generic substantially cheaper than brand Silenor.
Layer 5 — Mirtazapine combination when depression or anxiety contributes
7.5–15mg qhs. Useful when comorbid depression, anxiety, or appetite issues are present. Dose paradox: lower dose more sedating (H1 dominance until NE release overtakes at higher doses). Watch weight gain, morning hangover, RLS.
Layer 6 — Hormonal intervention in perimenopausal sleep disruption
When hot flash or perimenopausal pattern is contributing, hormonal interventions (Stage 19) may resolve refractoriness that pure sleep pharmacology will not. Consider in collaboration with gynecology or hormone-specialist clinician.
Layer 7 — When indefinite hypnotic use is appropriate
Rare scenarios: severe refractory insomnia, all other interventions adequately tried and failed, functional impact substantial, informed-consent conversation about long-term cognitive considerations. Prefer DORA or low-dose doxepin over benzodiazepines/Z-drugs in this scenario. Reassess annually.
Special situations
- Sleep maintenance insomnia at 3am: Low-dose doxepin specifically addresses sleep maintenance better than other agents. DORAs also helpful. Avoid short-acting agents that wear off mid-night.
- Insomnia with prominent anxiety hyperarousal: Mirtazapine combination or hydroxyzine 25–50mg qhs; address autonomic substrate with HRV training (Stage 11.2/22.1) and paced breathing as part of sleep-onset routine.
- Older adult with chronic Z-drug or benzodiazepine use: Concurrent taper plan (Stage 11.4) while building alternative regimen. Do not just add new agents on top of existing problematic medications.
- Shift worker: Circadian-specific approach — melatonin 0.3–3mg at desired sleep onset, strict light hygiene, possibly tasimelteon for non-24 patterns. Standard hypnotics often inadequate for shift work.
Generally avoid
- Adding chronic benzodiazepines or Z-drugs for refractory insomnia — these are not the longevity-psychiatry answer regardless of how reliably they produce sleep. The cognitive cost compounds over years.
- Low-dose quetiapine for sleep without psychiatric indication — metabolic and anticholinergic costs, limited evidence at sleep-dose ranges. The widespread practice is not evidence-based.
- OTC anticholinergics (diphenhydramine, doxylamine) — chronic anticholinergic burden contributes to dementia risk per Beers Criteria and the anticholinergic dementia literature.
- Declaring refractoriness without sleep study — undiagnosed OSA or RLS frequently explains "refractory insomnia" and changes the treatment entirely.
The chief-resident note: Most "refractory insomnia" is treatable when the workup finds the missed contributor. The patient on a chronic Z-drug declared refractory has often not had CBT-I, has not had a sleep study, has not had hormonal evaluation if perimenopausal, has not had iron studies if RLS suspected, and has not had the antidepressant or hormonal adjustment that might resolve sleep without indefinite hypnotic. Build the practice for the workup, not just the prescription.
Refractory insomnia — chronic insomnia that has not responded to first-line behavioral and pharmacological treatment — is one of the highest-leverage clinical opportunities in longevity psychiatry. Sleep is the single largest modifiable factor in cognitive trajectory across decades, and the patient with unresolved chronic insomnia is accumulating cognitive cost while consuming substantial clinical attention. The discipline is to find the contributor that explains the refractoriness, treat it appropriately, and layer the lowest-cognitive-cost pharmacology that produces restorative sleep when behavioral and contributor-targeted treatment is not sufficient.
The "refractory" label often masks inadequate first-line treatment. Most patients labeled treatment-resistant for insomnia have not actually completed CBT-I, the strongest evidence-based first-line intervention. CBT-I requires 6–8 sessions delivered in-person or through validated digital platforms (Somryst, Sleepio, SHUTi); the access barrier is real but the digital platforms have substantially closed the gap. Before declaring refractoriness, confirm that CBT-I has been actually delivered, not just recommended. The patient who has had multiple medication trials but no structured CBT-I is not refractory; they are inadequately treated.
The workup for the treatable contributor is the second key step. Sleep study (in-lab polysomnography or home sleep apnea test) identifies obstructive sleep apnea, periodic limb movement disorder, and REM behavior disorder — each of which produces a sleep pattern labeled as insomnia but requiring different treatment. Hormonal evaluation in perimenopausal women identifies the hormonal contribution that pure sleep pharmacology cannot address. Iron studies if RLS is suspected. Medication audit for caffeine, late-day stimulants, alcohol, and exercise timing. Depression and anxiety reassessment, since residual mood or anxiety symptoms frequently maintain insomnia. The workup is clinical action — done right, it identifies the contributor that resolves the refractoriness.
DORAs (dual orexin receptor antagonists) are the most clinically meaningful pharmacological advance. Suvorexant, lemborexant, and daridorexant block orexin signaling at the wake-promoting nucleus rather than potentiating GABA. The mechanism produces sleep without the broad CNS depression of GABAergic agents, with substantially better safety profile in older adults — less fall risk, less cognitive impairment, less dependency liability. Insurance often requires step therapy, but the documentation of prior failed agents typically justifies the change. For refractory insomnia in patients over 50, DORAs are increasingly first-line pharmacology.
Low-dose doxepin and mirtazapine combination address specific clinical pictures. Doxepin 3–6mg is highly H1-selective at this dose range, minimal anticholinergic burden, FDA-approved for sleep maintenance specifically — useful for patients with middle-night awakening. The cost-versus-effect calculation depends on brand versus compounded generic. Mirtazapine 7.5–15mg is useful when comorbid depression, anxiety, or appetite issues are present; the dose paradox (more sedating at lower doses) reflects H1 dominance at low doses with NE release overtaking at higher doses. Both are reasonable next-line agents with lower cognitive cost than benzodiazepines or Z-drugs.
The discipline avoids the cognitively-costly defaults. Chronic benzodiazepines and Z-drugs for sleep in adults over 50, low-dose quetiapine for sleep without psychiatric indication, and chronic OTC anticholinergic use are widely practiced and inconsistent with longevity-psychiatry principles. The cognitive cost compounds over years. The patient on long-term zolpidem who has not had CBT-I, has not had a sleep study, and has not been offered DORAs or low-dose doxepin is on suboptimal treatment from the longevity-psychiatry standpoint, and the audit-and-transition work is some of the most meaningful clinical care available. Refractory insomnia is rarely actually refractory; most patients can be moved to durable restorative sleep through systematic workup, behavioral foundation, and cognitively-friendly pharmacology.