The Starting Dose Your Doctor Writes for Anxiety Is Already in the Danger Zone

If you were started on alprazolam for panic attacks at the standard opening dose your prescriber wrote down, you were not started on a “low” dose. You may have assumed you were — most people do, because a first prescription feels cautious by definition. A new study says otherwise, and the gap between what patients assume and what the dosing chart actually shows is the story here.

The dose doctors call a normal starting point for anxiety is, statistically, already the dose where the danger curve bends upward.

Researchers led by Heidi Särkilä, working with Finland’s national health insurance register, followed 48,124 people who started a new benzodiazepine or “Z-drug” (the newer sleep medications built on similar chemistry) with no prior use, tracking them for five years. It’s published this month in Acta Psychiatrica Scandinavica, one of psychiatry’s older and more rigorously peer-reviewed journals, and it is the kind of population-scale dosing study that individual clinics never have the numbers to run themselves.

What “dose” actually measured, and why the categories matter

The team didn’t just ask whether someone was prescribed a benzodiazepine. They tracked actual dispensing records over time and converted everything into Defined Daily Doses, a standardized pharmacology unit that lets you compare, say, diazepam to alprazolam to zolpidem on the same scale. They split people into three bands: low dose (under 1.0 DDD per day), medium-to-high (1.0 to just under 3.0 DDD per day), and very high (3.0 DDD per day or more). Translated into the diazepam-equivalent numbers a pharmacist would recognize, medium-to-high covers roughly 10 to 29 milligrams a day, and very high starts at 30.

Here’s the part that should change how you read your own prescription bottle: standard starting doses for alprazolam, one of the most commonly prescribed anti-anxiety medications in the country, land at roughly 0.75 to 1.5 milligrams a day — a number that, once converted to diazepam-equivalent terms, falls inside the “medium-high” category, not the low one. That’s not a dose someone escalated to over years of tolerance. That’s week one.

The numbers, and what they mean plainly

Across the full cohort, being in a period of active benzodiazepine use — versus a period of not using — carried a 28% higher risk of death from any cause. That number alone doesn’t tell you much; a lot of people are prescribed these drugs precisely because they’re already in crisis, which can confound a simple before-and-after comparison. The dose-response pattern underneath it is what makes the finding hard to wave away.

At low doses, the study found no meaningful increase in overall mortality — though even there, deaths from external causes (accidents, overdose, suicide) ran 17% higher than in non-use periods. At medium-to-high doses — the band that ordinary starting prescriptions often fall into — overall mortality rose 58%, overdose deaths tripled (a 3.34-times increase), and suicide deaths were roughly two and a half times more likely (2.43x). At very high doses, the numbers stop looking like a gradual slope and start looking like a cliff: 2.68 times the overall mortality risk, 6.18 times the overdose risk, and 4.46 times the suicide risk, compared to not using.

Layer another prescription on top and the risk compounds faster than simple addition would suggest. Taking one additional medication alongside a benzodiazepine raised mortality risk 81%. Three or more additional medications raised it 212%. Combining a benzodiazepine with a Z-drug sleep medication — a pairing many patients assume is safer than “two benzos,” because the drugs have different brand names and different marketing — raised risk 86%. Pharmacologically, Z-drugs act on close to the same receptor system benzodiazepines do. The body doesn’t much care what the box says.

What the study can’t tell you, and why that caveat matters

This is a population-level association study, not a controlled trial, and the researchers themselves would be the first to say it can’t prove the dose caused every excess death by itself — people prescribed higher doses may also be dealing with more severe underlying anxiety, more chronic pain, or more co-occurring conditions that independently raise mortality risk. What the study can do, because of its size and its dose-by-dose granularity, is show that the risk climbs in lockstep with the dose in a way that’s difficult to explain away as coincidence, and that the climb starts earlier — at doses many prescribers and patients still think of as conservative — than the field has generally assumed.

There’s a second limitation worth naming plainly: this is a Finnish cohort, inside Finland’s specific prescribing patterns and universal health register. American prescribing habits, insurance-driven dose adjustments, and the sheer prevalence of illicit fentanyl in the U.S. drug supply — which synergizes catastrophically with benzodiazepines by compounding respiratory depression — mean the raw multipliers here may not transfer exactly. The direction of the finding, and the fact that “starting dose” and “safe dose” are not the same category, almost certainly does.

drug supply — which synergizes catastrophically with benzodiazepines by compounding respiratory depression — mean the raw multipliers here may not transfer exactly.

This isn’t the first time the dosing chart got ahead of the warning label

Benzodiazepines have been here before, and the field’s memory of it has faded faster than the drugs themselves have. Through the 1960s and 70s, Valium and its chemical siblings were prescribed almost reflexively — marketed as safe, non-habit-forming alternatives to barbiturates, handed out for garden-variety anxiety with none of the caution now applied to opioids. It took roughly two decades of accumulating dependence cases and withdrawal reports before regulators moved. In 1988, the UK’s Committee on Safety of Medicines issued formal guidance stating plainly that benzodiazepines should be used for only two to four weeks, and only for anxiety severe enough to be disabling — an implicit admission that the preceding two decades of prescribing had drifted far past what the pharmacology actually supported. Prescribing fell afterward, but never to zero, and it climbed back over the following decades as new formulations arrived with the same reassuring packaging. The Särkilä study is this generation’s version of that 1988 correction — except this time the warning starts at the first prescription, not two decades in.

What this means for a prescriber, this week

If you prescribe benzodiazepines, the actionable version of this study isn’t “stop prescribing them” — for severe, disabling anxiety or acute alcohol withdrawal, they remain genuinely necessary medicine, and abrupt discontinuation carries its own real dangers, including seizure risk. The actionable version is narrower and more immediate: before writing a starting dose, run it through a diazepam-equivalent conversion rather than trusting that “starting dose” and “low-risk dose” are the same thing, because for several of the most commonly prescribed benzodiazepines, they are not. And check the medication list for a second sedative — a Z-drug, an opioid, another benzodiazepine from a different prescriber — before the patient leaves the office, not after.

If you’re the one holding the prescription bottle: this isn’t a reason to panic or to stop a medication on your own, which carries its own withdrawal risks that should be managed with a clinician, not alone. It’s a reason to ask a specific question at your next appointment — not “is this addictive,” which most people already assume the answer to, but “where does my actual dose fall on the risk curve, and is there a lower one that still works.” That question, asked plainly, is the one this study just gave you the standing to ask.