CBT-I Is Guideline First-Line for Insomnia. Almost No One Gets It First.
Two patients, same diagnosis, same guideline recommending therapy before drugs. What actually determines whether either of them gets offered that choice has less to do with the evidence than with what’s reachable from where they’re sitting.
D.K., a 41-year-old woman, made partner at her firm eighteen months ago, a milestone she had worked toward for a decade and genuinely wanted — and it roughly doubled her caseload overnight, moving her from long hours to something closer to constant availability, with a phone that rarely stops through dinner. She has been a light, somewhat effortful sleeper for as long as she can remember, the kind of person who has always needed quiet and a consistent routine to fall asleep easily, but she describes what has happened since the promotion as categorically different: real difficulty falling asleep most nights, a mind that won’t stop running through the next day’s filings, and, more recently, waking at 1 or 2 a.m. and lying awake for an hour or more before drifting off again.
She is otherwise healthy — no psychiatric history, no chronic medical conditions, and no medications — and had, until recently, an evening glass of wine most nights, which she cut out on her own two months ago after reading it might be worsening her sleep-maintenance problem specifically; she noticed no real change either way. Her internist walked her through standard sleep-hygiene measures eight months ago — a fixed wake time, screens out of the bedroom, no caffeine after noon — which she followed carefully for several weeks without lasting benefit before her sleep gradually reverted to its current pattern. Her Insomnia Severity Index score today is 19, moderate-to-severe, and she estimates she is averaging four and a half hours of genuinely restorative sleep on a typical weeknight, functioning through the day on caffeine and, by her own description, a fair amount of professional adrenaline.
She has excellent insurance through her firm and lives in a city with two academic sleep centers, and when her internist mentioned CBT-I as an option she did what she does with most unfamiliar recommendations — read the primary literature herself, found a therapist who specializes in it with a six-week opening, and arrived today having already done the legwork, asking specifically whether she should start there and wait, or get a prescription now and treat therapy as a parallel track rather than a sequential one.
For her, this isn’t really a debate — it’s a scheduling decision. The AASM’s 2021 clinical practice guideline and the ACP’s 2016 guideline both name CBT-I as the recommended first-line treatment for chronic insomnia in adults, ahead of any medication, and she has exactly the access most patients don’t: a specialist opening inside six weeks and coverage that will pay for it.
The evidence isn’t close, either — CBT-I’s effect sizes for sleep-onset latency and wake-after-sleep-onset are comparable to hypnotic medication in the short term and durable well past the point most drug trials stop measuring. Starting a hypnotic now, for a patient who can get real therapy in six weeks, risks anchoring her on medication before she’s even tried the treatment guidelines actually recommend first.
I agree with the guideline, but six weeks is six weeks of a 41-year-old attorney averaging four and a half hours of sleep a night while managing a doubled caseload — that’s not a trivial bridge, and there’s nothing in the guideline that says she can’t do both.
A short, time-limited course of zolpidem while she waits for her CBT-I intake isn’t abandoning the guideline—it’s buying her functional sleep during the gap the guideline doesn’t actually address. The real risk isn’t combining the two; it’s her deciding after a good night on medication that she no longer needs the harder work of therapy.
Agreed: a hard 6-week stop date on zolpidem, timed to end when her CBT-I sessions begin, discussed with her explicitly so the bridge doesn’t quietly become the plan.
M.T., a 58-year-old man, has driven long-haul freight routes for over thirty years, most of them out of the same small town three hours from the nearest city with a sleep specialist. He married young, raised two kids largely around a rotating trip schedule, and describes himself, without much self-pity, as someone who has “never really slept right” for almost as long as he’s been driving professionally — difficulty both falling and staying asleep that he traces back nearly a decade, worsening gradually rather than arriving all at once. He has type 2 diabetes managed with metformin and hypertension on amlodipine, both stable and well-controlled on his last two visits, and no psychiatric history. For most of that decade he treated his sleep, when he treated it at all, with over-the-counter diphenhydramine taken irregularly before overnight runs, never discussing it with a physician because it was, in his words, “just part of the job.”
His regular physician — the single provider at a rural clinic he sees twice a year for his diabetes and blood pressure — raised CBT-I with him at his last visit, and he liked the idea enough to ask real questions about it. What he found afterward was that the nearest sleep-medicine practice offering it in person is a four-hour round trip from his home, not reachable inside a driving schedule that already eats most of his week, and that his house — several miles outside town, at the edge of reliable cell coverage — doesn’t have the kind of stable broadband a video-based version would need. He isn’t declining the idea; he genuinely has nowhere to take it.
His ISI score today is 21, severe range, and he is here specifically because a recent Department of Transportation commercial driving physical flagged his sleep history as a safety concern serious enough that his examiner told him directly it needed real treatment before his next certification, not another few years of diphenhydramine and hoping it holds.
For D.K., the guideline and reality lined up. For M.T., they don’t — there is no realistic path to specialist-delivered CBT-I for a man who can’t make a four-hour round trip or hold a stable video connection, and telling him to wait for access that doesn’t exist isn’t a first-line recommendation, it’s a non-recommendation dressed up as one.
I’d start low-dose trazodone now, given his diabetes and the general preference to avoid a Z-drug or benzodiazepine in a commercial driver whose job depends on next-day alertness, and treat medication as his actual first-line therapy — not a bridge to something he can’t reach.
I’d push back gently before conceding the point. Self-guided digital CBT-I programs — structured, app- or workbook-based courses that don’t require broadband video, some usable from a phone with intermittent signal — have shown real, guideline-acknowledged efficacy, meaningfully below full specialist-delivered therapy but well above no behavioral treatment at all. That’s not nothing for a man with a decade of untreated insomnia.
You’re right that specialist access genuinely isn’t there for him — I’m not disputing that — but I don’t think the honest choice is medication instead of CBT-I. It’s medication now, alongside the lowest-access-barrier version of CBT-I that can actually reach him, rather than writing off behavioral treatment entirely because the specialist version is out of reach.
Agreed: trazodone started now, with a self-guided digital CBT-I program offered as a genuinely lower-barrier behavioral option rather than deferred entirely. Not agreed: whether digital CBT-I is a real substitute for specialist-delivered therapy or a stopgap that still leaves him under-treated by the guideline’s actual standard — left open, revisited at his next DOT physical.