Companion reviews: Depression — antidepressants and the evidence, ADHD — stimulants and the evidence, Antipsychotics and the evidence, Mood stabilizers and the evidence, Placebo, unblinding and the evidence, Psychotherapy comparators and the evidence, Diagnosis thresholds and the evidence, Anxiety — a critical review of the evidence, and Stress, Energy and the Capacity to Function.

---

Executive summary

Two folk models dominate public discussion of benzodiazepines: that they are highly effective, safe anxiolytics for short-term use that are under-prescribed due to moral panic (the standard-of-care narrative), and its inversion, that they are minimally effective, uniquely addictive medications causing dementia and cognitive decline that should never be prescribed (the critical narrative). Both are overclaims, and the evidence supports neither in the form presented.

Six things the evidence actually establishes:

1. Benzodiazepines reduce symptoms of anxiety versus placebo in short-term trials (1–4 weeks), with effect sizes in the small-to-moderate range. Munkholm et al. (2024) network meta-analysis of 34 trials (7,044 participants) found benzodiazepines versus placebo SMD −0.58 (95% CI −0.77 to −0.40, low certainty evidence). Pregabalin (SMD −0.58, 95% CI −0.87 to −0.28) and quetiapine (SMD −0.51, 95% CI −0.90 to −0.13) showed comparable efficacy. The network meta-analysis used 31 of the 34 included trials. All evidence was rated low to very low certainty due to indirectness (most trials studied chronic anxiety disorders, not newly onset distress), high heterogeneity, and risk of bias.

2. Benzodiazepines cause physical dependence within days to weeks even at therapeutic doses, and withdrawal symptoms can be severe and protracted. FDA (2020) boxed warning states physical dependence can occur when benzodiazepines are taken steadily for several days to weeks, even as prescribed. Abrupt discontinuation or rapid dose reduction can result in serious withdrawal reactions, including seizures, which can be life-threatening. FDA reviewed 104 cases from the FAERS database (single-substance reports, 1968–2019) and found "a wide range of time to dependence, with some describing the onset as early as days to weeks after the start of a benzodiazepine. Similarly, there were variations in the duration of the withdrawal symptoms that lasted from weeks to years."

3. Benzodiazepines and Z-drugs are associated with increased risk of hip fracture, with highest risk in new users. Donnelly et al. (2017) systematic review and meta-analysis found benzodiazepines associated with RR 1.52 (95% CI 1.37–1.68) for hip fracture, and Z-drugs with RR 1.90 (95% CI 1.68–2.13). Short-term use (new users) showed greatest risk: benzodiazepines RR 2.40 (95% CI 1.88–3.05), Z-drugs RR 2.39 (95% CI 1.74–3.29).

4. Benzodiazepine use is associated with increased dementia risk in observational studies, but confounding by indication cannot be ruled out. [contested] Billioti de Gage et al. (2012) prospective cohort study (1,063 participants, mean age 78.2 years, 15-year follow-up) found new benzodiazepine use associated with HR 1.60 (95% CI 1.08 to 2.38) for incident dementia. Pooled analysis across five cohorts of new users found HR 1.46 (95% CI 1.10 to 1.94). Nested case-control study found ever use associated with OR 1.55 (95% CI 1.24 to 1.95) compared to never users. Association remained after controlling for depressive symptoms, but reverse causation (early dementia symptoms prompting benzodiazepine prescription) and confounding by indication (anxiety/insomnia as prodromal symptoms) remain concerns.

5. Long-term benzodiazepine use prevalence is substantial in older adults and among incident users. Olfson et al. (2015) found 8.7% of US adults aged 65–80 filled a benzodiazepine prescription over one year, with nearly one-third (31.4%) classified as long-term users (≥120 days). Taipale et al. (2020) followed 129,732 incident benzodiazepine users (median follow-up 2.1 years); 39.4% became long-term users (≥180 days). In 2018, 50% of patients who were dispensed oral benzodiazepines received them for a duration of ≥2 months (FDA 2020).

6. The FDA (2020) issued a boxed warning for all benzodiazepines addressing risks of abuse, addiction, physical dependence, and withdrawal. The warning states benzodiazepines can lead to misuse, abuse, and addiction even when taken at recommended dosages. Risks are increased when combined with opioids, alcohol, or illicit drugs. FDA Data Summary: "In 2016, the nationally estimated number of emergency department (ED) visits due to nonmedical use of benzodiazepines (n=167,845) was higher than the corresponding estimate for prescription opioids (n=129,863)." From 2013–2017, 55% of benzodiazepine-involved overdose deaths also documented involvement of prescription opioids.

What the evidence does not establish:

• That benzodiazepines are uniquely safe for short-term use (physical dependence can develop within days to weeks; withdrawal risk is substantial)
• That benzodiazepines definitively cause dementia (association is observational and confounded by indication; reverse causation possible)
• Which patients will develop dependence or withdrawal symptoms
• Long-term functional outcomes or quality of life with chronic benzodiazepine use
• Whether second-generation anxiolytics (Z-drugs) have meaningfully different risk profiles (fracture risk comparable or higher)

The structural problem that conditions all of this: Efficacy trials are short (1–4 weeks median), exclude the population most likely to be prescribed benzodiazepines (elderly, medically complex, psychiatric comorbidities), and were conducted in populations with chronic anxiety disorders rather than adjustment disorders or newly onset distress. In 2018, 50% of patients who were dispensed oral benzodiazepines received them for a duration of ≥2 months, despite evidence base being limited to short-term trials. Dementia association is observational and cannot establish causation. The folk models—benzodiazepines as uniquely safe for short-term use or benzodiazepines as never justifiable—both overclaim what opened trials support.

---

How to read this review

The single most important heuristic

An efficacy claim without a named trial duration, patient population, comparison group, and outcome measure is not a clean estimate. Benzodiazepines show SMD −0.58 versus placebo in 1–4 week trials of anxiety symptoms (low certainty). They show RR 1.52 for hip fracture and RR 2.40 for hip fracture in new users. They show HR 1.60 for dementia in observational cohorts (confounded). Any claim that "benzodiazepines work" or "benzodiazepines cause dementia" must specify which outcome, in which population, over what duration, versus what comparator.

Three structural problems

Short trial duration and indirect populations. The Munkholm network meta-analysis used 1–4 week data. Only 6 of 34 trials investigated adjustment disorders (the intended population for short-term anxiolytic use); 14 trials studied generalized anxiety disorder, 9 studied panic disorder, and 5 studied social anxiety disorder. The majority of trials included patients with symptoms lasting over 1 year. None reported whether non-pharmacological treatment had been tried. The efficacy estimates apply to short-term treatment of chronic anxiety disorders, not to newly onset distress.

Confounding by indication in long-term harm data. Billioti de Gage's dementia finding (HR 1.60) is observational. Patients prescribed benzodiazepines have higher rates of anxiety, insomnia, depression, and other conditions that are themselves risk factors for dementia. The study controlled for multiple confounders, but residual confounding cannot be ruled out. Reverse causation is possible: prodromal dementia symptoms (anxiety, sleep disturbance) may prompt benzodiazepine prescription years before diagnosis.

Conventions

Claims are marked [contested] where the literature genuinely disagrees or where observational evidence cannot establish causation.

Confidence intervals are omitted where they could not be verified against the primary text, rather than reconstructed.

---

Part I — Acute anxiolytic efficacy: what short-term trials show

1.1 Network meta-analyses: benzodiazepines versus placebo

Multiple network meta-analyses have evaluated benzodiazepines for short-term treatment of anxiety. Munkholm et al. (2024) conducted a systematic review and network meta-analysis of minor tranquilizers for short-term (1–4 weeks) treatment of newly onset symptoms of anxiety and distress in adults. Search conducted PsycInfo, MEDLINE, EMBASE, and Cochrane Library databases up to September 2022.

Inclusion criteria: Randomized controlled trials of non-hospitalized adults with newly onset symptoms of anxiety and distress requiring short-term pharmacological treatment (maximum 4 weeks). Population included patients with adjustment disorders or anxiety spectrum disorders. Exclusions: organic mental disorders, psychotic disorders, bipolar disorder, severe depression, obsessive-compulsive disorder, or need for inpatient care.

Interventions: Benzodiazepines, antipsychotics with sedative effects (quetiapine, olanzapine in low doses), sedative antidepressants (mirtazapine, mianserin), antihistamines (promethazine), melatonin, Z-drugs (zopiclone, zolpidem), and pregabalin.

Primary outcome: Symptoms of anxiety measured with Hamilton Anxiety Scale (HAM-A). Trials reporting data within 4 weeks after treatment commencement were eligible. Outcomes at 1 week were prioritized, followed by 2, 3, and 4 weeks.

Included studies: 34 randomized trials, 7,044 patients total. Populations: adjustment disorder (6 trials), generalized anxiety disorder (14 trials), social anxiety disorder (5 trials), panic disorder (9 trials). Majority of trials included outpatients; preponderance of women; average age 33–57 years. Trial duration 4–24 weeks.

Network meta-analysis results: The network meta-analysis used 31 of the 34 included trials. Benzodiazepines (SMD −0.58, 95% CI −0.77 to −0.40, low certainty evidence), pregabalin (SMD −0.58, 95% CI −0.87 to −0.28, low certainty evidence), and quetiapine (SMD −0.51, 95% CI −0.90 to −0.13, low certainty evidence) reduced anxiety symptoms versus placebo. All evidence was rated low to very low certainty due to indirectness (most trials studied chronic anxiety disorders, not newly onset distress), high heterogeneity, and risk of bias.

Interpretation: Using Cohen's rule of thumb (SMD 0.2 small, 0.5 medium, 0.8 large), benzodiazepines show a moderate effect versus placebo. The guideline working group pre-specified SMD 0.3 as the minimal clinically important difference. Benzodiazepines exceeded this threshold.

Certainty of evidence: Low to very low across outcomes. Most common reasons for downgrading: indirect study populations (most trials included chronic anxiety disorders, not newly onset distress), imprecise estimates, and risk of bias. Only 6 of 34 trials investigated adjustment disorders. In trials reporting symptom duration, patients had symptoms for at least 1 year. None reported whether non-pharmacological treatment had been tried.

Translation: Benzodiazepines reduce anxiety symptoms versus placebo in short-term trials (1–4 weeks) with a moderate effect size (SMD −0.58). The evidence is indirect: most trials studied chronic anxiety disorders, not newly onset distress or adjustment reactions. The certainty is low. Pregabalin (SMD −0.58) and low-dose quetiapine (SMD −0.51) show comparable efficacy.

---

Part II — Dependence, withdrawal, and discontinuation: FDA warnings and trial evidence

2.1 FDA 2020 boxed warning: physical dependence within days to weeks

On September 23, 2020, the FDA required an updated boxed warning for all benzodiazepines to address serious risks of abuse, addiction, physical dependence, and withdrawal reactions.

Key findings from FDA Data Summary (FAERS database and dispensing data):

• FAERS case review: "We evaluated 104 cases from the FDA Adverse Event Reporting System (FAERS) database of abuse, dependence, or withdrawal involving a benzodiazepine as a single drug substance reported by patients or health care professionals directly to FDA from January 1, 1968, through June 30, 2019."

• Time-to-dependence and withdrawal duration: "These cases reported a wide range of time to dependence, with some describing the onset as early as days to weeks after the start of a benzodiazepine. Similarly, there were variations in the duration of the withdrawal symptoms that lasted from weeks to years."

• Long-term use prevalence (dispensing data): "In 2018, an estimated 50% of patients who were dispensed oral benzodiazepines received them for a duration of two months or longer."

• Emergency department visits: "In 2016, the nationally estimated number of emergency department (ED) visits due to nonmedical use of benzodiazepines (n=167,845) was higher than the corresponding estimate for prescription opioids (n=129,863)."

• Overdose deaths: From 2013–2017, 55% of benzodiazepine-involved overdose deaths also documented involvement of prescription opioids.

Boxed warning text (regulatory): "Physical dependence can occur when benzodiazepines are taken steadily for several days to weeks, even as prescribed. Stopping the use of benzodiazepines suddenly or reducing the dose too quickly can result in serious withdrawal reactions, including seizures, which can be life-threatening."

Translation: Physical dependence is not a rare or aberrant outcome—FAERS cases report onset as early as days to weeks at therapeutic doses. Withdrawal symptoms can last from weeks to years. In 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months.

2.2 Withdrawal syndrome: clinical features and timeline

Common benzodiazepine withdrawal symptoms include:

Rebound symptoms (occur within 1–4 days for short-acting; 5–10 days for long-acting): anxiety, insomnia, restlessness.

Acute withdrawal symptoms: tremor, sweating, difficulty concentrating, nausea, weight loss, palpitations, headache, muscle pain/stiffness, perceptual changes (hypersensitivity to stimuli), depersonalization/derealization.

Severe withdrawal symptoms (rare, more common with abrupt cessation or rapid taper, polysubstance use, or history of epilepsy): seizures, psychosis (delusions, hallucinations), delirium.

Protracted withdrawal syndrome: Attenuated and prolonged symptoms (irritability, anxiety, sleep disturbance, mood instability) appearing after several months of discontinuation and remitting slowly over months to years.

Translation: Withdrawal syndrome is common, can be severe, and may be protracted.

2.3 Pharmacological interventions for discontinuation: Cochrane review

Baandrup et al. (2018) conducted a Cochrane systematic review of pharmacological interventions to facilitate benzodiazepine discontinuation in chronic users. Search up to October 2017.

Inclusion criteria: Randomized controlled trials comparing pharmacological treatment versus placebo or no intervention, or versus another pharmacological intervention, in adults treated with benzodiazepines for at least 2 months or fulfilling criteria for benzodiazepine dependence.

Included studies: 38 trials, 2,543 participants (data extracted from 35 trials, 2,295 participants). Many different interventions studied; no single intervention assessed in more than 4 trials. Extracted data on 18 different comparisons. Risk of bias was high in all trials but one. Trial duration ranged 1–24 weeks (mean 9 weeks). Average participant age ~50 years; majority women; mean benzodiazepine use 5–10 years.

Benzodiazepine discontinuation at end of intervention:

• Valproate: 1 study, 27 participants; RR 2.55 (95% CI 1.08 to 6.03), very low-quality evidence.
• Tricyclic antidepressants: No data at end of intervention.

Benzodiazepine discontinuation at longest follow-up:

• Tricyclic antidepressants: 1 study, 47 participants; RR 2.20 (95% CI 1.27 to 3.82), low-quality evidence.

Benzodiazepine withdrawal symptoms at end of intervention (positive effects):

• Pregabalin: 1 study, 106 participants; MD −3.10 points (95% CI −3.51 to −2.69), very low-quality evidence.
• Captodiame: 1 study, 81 participants; MD −1.00 points (95% CI −1.13 to −0.87), very low-quality evidence.
• Paroxetine: 2 studies, 99 participants; MD −3.57 points (95% CI −5.34 to −1.80), very low-quality evidence.
• Tricyclic antidepressants: 1 study, 38 participants; MD −19.78 points (95% CI −20.25 to −19.31), very low-quality evidence.
• Flumazenil: 3 studies, 58 participants; SMD −0.95 (95% CI −1.71 to −0.19), very low-quality evidence.

However, paroxetine effect did not persist at longest follow-up (1 study, 54 participants; MD −0.13 points, 95% CI −4.03 to 3.77).

Anxiety symptoms at end of intervention (positive effects):

• Carbamazepine: 1 study, 36 participants; MD −6.00 points (95% CI −9.58 to −2.42), very low-quality evidence.
• Pregabalin, captodiame, paroxetine, flumazenil: positive effects, all very low-quality evidence.

Relapse to benzodiazepine use:

• Valproate: 1 study, 27 participants; RR 0.31 (95% CI 0.11 to 0.90), very low-quality evidence.
• Cyamemazine: 1 study, 124 participants; RR 0.33 (95% CI 0.14 to 0.78), very low-quality evidence.

Adverse events: Poorly reported. One flumazenil trial was discontinued due to severe panic reactions.

Authors' conclusions: "Given the low or very low quality of the evidence for the reported outcomes, and the small number of trials identified with a limited number of participants for each comparison, it is not possible to draw firm conclusions regarding pharmacological interventions to facilitate benzodiazepine discontinuation in chronic benzodiazepine users."

---

Part III — Hip fracture and falls: established associations in older adults

3.1 Benzodiazepines, Z-drugs, and hip fracture risk: meta-analysis

Donnelly et al. (2017) conducted a systematic review and meta-analysis of benzodiazepines, Z-drugs, and hip fracture risk. Searched MEDLINE and SCOPUS up to May 2015.

Included studies: Both benzodiazepines and Z-drugs separately.

Main findings:

• Benzodiazepines: RR 1.52 (95% CI 1.37–1.68) for hip fracture.
• Z-drugs: RR 1.90 (95% CI 1.68–2.13) for hip fracture.
• Short-term use (new users):
  • Benzodiazepines: RR 2.40 (95% CI 1.88–3.05)
  • Z-drugs: RR 2.39 (95% CI 1.74–3.29)

Interpretation: Both benzodiazepines and Z-drugs are associated with increased hip fracture risk, with little difference between them. New users (short-term use) have the greatest risk—more than double the baseline risk. This challenges the narrative that Z-drugs are safer alternatives to benzodiazepines.

Mechanisms: Drug-induced sedation, impaired reflexes, postural instability, and impaired balance. Older adults are particularly vulnerable due to age-related declines in balance and vision, polypharmacy, and comorbidities.

Clinical context: Hip fractures in older adults are associated with high morbidity and mortality. American Geriatrics Society Beers Criteria list benzodiazepines and Z-drugs as potentially inappropriate for older adults due to association with confusion, falls, and fractures.

Translation: The hip fracture risk is real, substantial, and applies to both benzodiazepines and Z-drugs. The risk is highest in new users, not chronic users—challenging the assumption that tolerance reduces fall risk.

---

Part IV — Dementia association: observational evidence and confounding

4.1 The Billioti de Gage cohort study: HR 1.60 for incident dementia

Billioti de Gage et al. (2012) conducted a prospective population-based cohort study using the PAQUID cohort in France.

Design: 1,063 men and women (mean age 78.2 years) who were free of dementia and did not start taking benzodiazepines until at least the third year of follow-up (T5). This 3-year observation period allowed adjustment for factors associated with starting benzodiazepines, including cognitive decline from baseline to 3-year follow-up, addressing confounding by indication and limiting reverse causation.

Follow-up: 15 years. 253 incident dementia cases confirmed by neurologist.

Main findings:

• New benzodiazepine use: Multivariable adjusted HR 1.60 (95% CI 1.08 to 2.38).
• Sensitivity analysis controlling for depressive symptoms: HR 1.62 (95% CI 1.08 to 2.43).
• Pooled analysis across five cohorts of new users: Pooled HR 1.46 (95% CI 1.10 to 1.94).
• Nested case-control study:
  • Ever use versus never use: adjusted OR 1.55 (95% CI 1.24 to 1.95).
  • Past users: OR 1.56 (95% CI 1.23 to 1.98), significant.
  • Recent users: OR 1.48 (95% CI 0.83 to 2.63), not significant (reached significance only for past users).

Confounding control: Adjusted for age, sex, education, marital status, living conditions, depression, baseline cognitive function, and change in cognitive function from baseline to 3-year follow-up.

Interpretation: New use of benzodiazepines was associated with approximately 50–60% increased risk of dementia over 15-year follow-up. The result was robust in pooled analyses and complementary case-control study. The association was stronger for past users than recent users, suggesting the effect may persist or accumulate over time.

Critical limitation—confounding by indication: [contested] Benzodiazepines are prescribed for anxiety, insomnia, and agitation, which may themselves be prodromal symptoms of dementia. The study adjusted for baseline cognitive function and 3-year cognitive decline, but residual confounding cannot be ruled out. Patients who discontinue benzodiazepines (past users) may differ systematically from never users in ways not fully captured by measured confounders. Reverse causation remains possible: early dementia symptoms may prompt benzodiazepine prescription years before diagnosis.

Translation: The association between benzodiazepine use and dementia is real, substantial, and consistent across multiple analyses. However, it is observational and cannot establish causation. The study design (3-year observation period, adjustment for cognitive decline) strengthens the case against pure confounding, but uncertainty remains.

4.2 Subsequent research and contested findings

Subsequent research has produced mixed findings. Some studies replicate the association; others find no association after controlling for confounders. A 2014 BMJ study by Billioti de Gage et al. (same lead author) in Quebec cohort found dose-response relationship: no association for cumulative dose <91 prescribed daily doses (1–90 PDD 1.09, 0.92 to 1.28); 91–180 PDD 1.32 (1.01 to 1.74); >180 PDD 1.84 (1.62 to 2.08).

Current status: [contested] The dementia association is contested. Multiple large cohort studies find associations, but confounding by indication and reverse causation cannot be ruled out. Randomized discontinuation trials in older adults would be needed to establish causation, but such trials raise ethical concerns (exposing control group to continued benzodiazepine use).

Translation: The dementia association is plausible and concerning. The evidence is observational; causation is not established. Confounding by indication and reverse causation cannot be ruled out.

---

Part VI — Long-term use prevalence: widespread chronic use despite short-term evidence base

6.1 Long-term use prevalence in older US adults

Olfson et al. (2015) analyzed national benzodiazepine prescribing in US adults aged 65–80. Found 8.7% filled a benzodiazepine prescription over one year. Nearly one-third (31.4%) of these users were classified as long-term (≥120 days of medication dispensed during the year). Long-term use increased markedly with age.

Translation: In the US, approximately 1 in 11 older adults receives benzodiazepines annually, and nearly 1 in 3 of those users receives long-term prescriptions despite evidence base being limited to short-term (1–4 week) trials.

6.2 Long-term use among incident users: Finnish national cohort

Taipale et al. (2020) followed 129,732 incident benzodiazepine and related drug (BZDR) users in Finland from 2006–2017. Median follow-up 2.1 years. Long-term use defined as ≥180 days of medication over the follow-up period.

Key findings:

• Long-term use prevalence: During the follow-up period, 51,099 BZDR users (39.4%) became long-term users.

• Incident cohort definition: Users with no BZDR dispensing in the 365 days prior to index date; excluded users with recorded schizophrenia, bipolar disorder, or substance use disorder diagnosis before first BZDR dispensing.

Translation: Among incident benzodiazepine users without major psychiatric diagnoses, 39.4% transitioned to long-term use (≥180 days) within 2.1 years median follow-up—despite evidence base being limited to 1–4 week trials.

---

Part VII — What the evidence does not establish

7.1 That benzodiazepines are uniquely safe for short-term use

The narrative that benzodiazepines are safe "if used as prescribed" for short-term (1–4 weeks) is not supported by the FDA's own data. Physical dependence can develop within days to weeks. Withdrawal symptoms can be severe and protracted. The highest risk for hip fracture is in new users (RR 2.40), not chronic users.

Clinical practice diverges from the evidence base: in 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months. The evidence base is for 1–4 week use, but the modal prescription pattern is long-term.

7.2 That benzodiazepines definitively cause dementia

The Billioti de Gage findings (HR 1.60, pooled HR 1.46) are observational and cannot establish causation. The study design (3-year observation period, adjustment for cognitive decline) addresses some confounding, but residual confounding by indication and reverse causation cannot be ruled out. Subsequent research is contested: some studies replicate, others find no association after controlling for indication.

Mechanistic plausibility exists (chronic GABA-A receptor modulation, sedation, falls, reduced cognitive reserve), but no randomized trial evidence exists.

7.3 Which patients will develop dependence or severe withdrawal

No validated algorithm exists for predicting which patients will develop physical dependence, psychological dependence, or severe withdrawal symptoms. Clinical factors associated with increased risk include: longer duration of use, higher doses, short-acting benzodiazepines, concurrent substance use, psychiatric comorbidities, prior history of dependence. But these are associations, not prediction rules.

7.4 Long-term functional outcomes with chronic use

No randomized trials assess quality of life, employment, social functioning, or disability in patients using benzodiazepines chronically. The efficacy trials are short (1–4 weeks) and measure symptom reduction, not functional outcomes. Whether chronic benzodiazepine use improves, impairs, or has no effect on long-term functioning is unknown.

---

Part VIII — The two overclaims, and what replaces them

8.1 The standard-of-care overclaim: "Benzodiazepines are safe for short-term use and under-prescribed"

The claim: Benzodiazepines are highly effective, well-tolerated anxiolytics for short-term use (1–4 weeks). They are under-prescribed due to moral panic, stigma, and overstatement of dependence risk. Patients who use them "as prescribed" will not develop dependence.

What the evidence shows: Benzodiazepines reduce anxiety symptoms versus placebo in short-term trials with effect sizes in the small-to-moderate range (low certainty). They are effective for this narrow indication. However, physical dependence can develop within days to weeks at therapeutic doses. Withdrawal symptoms can be severe and protracted. Hip fracture risk is highest in new users (RR 2.40), not chronic users.

In 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months. The modal prescription pattern is long-term, not short-term.

What survives: Benzodiazepines are effective for short-term (1–4 weeks) symptom reduction in anxiety. For patients with acute, newly onset distress (adjustment reactions), short-term use may provide meaningful relief. But physical dependence can develop within days to weeks. The narrative that benzodiazepines are safe "if used as prescribed" is not supported by FDA data or clinical practice patterns.

8.2 The critical overclaim: "Benzodiazepines are uniquely addictive, cause dementia, and should never be prescribed"

The claim: Benzodiazepines are minimally effective, uniquely addictive medications that cause dementia, cognitive decline, and hip fractures. They worsen long-term outcomes and should be avoided or minimized in all cases.

What the evidence shows: Benzodiazepines are effective for short-term anxiety reduction with effect sizes in the small-to-moderate range (low certainty).

The dementia association (HR 1.60, pooled HR 1.46) is observational and confounded. Multiple studies find associations, but causation is not established. Reverse causation (prodromal symptoms prompting prescription) and confounding by indication (anxiety/insomnia as dementia risk factors) remain concerns.

Hip fracture risk is real (RR 1.52 overall, RR 2.40 for new users) but not unique to benzodiazepines—Z-drugs show comparable or higher risk (RR 1.90).

What survives: Benzodiazepines have real harms: physical dependence develops quickly, withdrawal can be severe and protracted, hip fracture risk is substantial (especially in new users), and dementia association is concerning though not proven. But these harms are not unique. Z-drugs show similar fracture risk and dependence mechanisms. The narrative that benzodiazepines are uniquely harmful is not supported.

8.3 The version that survives

Anxiety and distress are heterogeneous experiences with variable causes and courses. Some patients have acute, newly onset distress (adjustment reactions, situational anxiety); others have chronic, recurrent anxiety disorders; others have anxiety as part of broader psychiatric or medical illness. No single treatment strategy applies to all patients.

Benzodiazepines reduce anxiety symptoms versus placebo in short-term trials (1–4 weeks), with effect sizes in the small-to-moderate range. They show comparable efficacy to pregabalin and low-dose quetiapine. The evidence is low certainty and indirect (most trials studied chronic anxiety disorders, not newly onset distress).

Physical dependence develops quickly (within days to weeks) even at therapeutic doses, and withdrawal can be severe and protracted. In 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months. Baandrup et al. (2018): no drugs are recommended or approved for managing benzodiazepine dependence or facilitating withdrawal.

Hip fracture risk is substantial (RR 1.52 overall, RR 2.40 for new users) and applies to both benzodiazepines and Z-drugs. The risk is highest in new users, not chronic users, challenging the assumption that tolerance reduces fall risk. Older adults are particularly vulnerable due to age-related declines in balance, polypharmacy, and comorbidities.

Dementia association is observational and cannot establish causation (HR 1.60, pooled HR 1.46). [contested] The association is consistent across multiple studies, but confounding by indication (anxiety/insomnia as prodromal symptoms) and reverse causation (early dementia prompting prescription) remain concerns.

Long-term use prevalence is substantial. 8.7% of US adults aged 65–80 filled benzodiazepine prescriptions annually; nearly one-third were long-term users (≥120 days). Among incident users without major psychiatric diagnoses, 39.4% transitioned to long-term use (≥180 days) within 2.1 years median follow-up. In 2018, 50% of patients who were dispensed oral benzodiazepines received them for a duration of ≥2 months.

What the evidence establishes: Benzodiazepines are effective for short-term (1–4 weeks) anxiety symptom reduction (moderate effect size versus placebo, low certainty). Physical dependence develops within days to weeks; withdrawal can be severe and protracted. Hip fracture risk is substantial (RR 1.52, RR 2.40 for new users). Dementia association is observational, consistent, but cannot establish causation. In 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months.

What it does not establish: That benzodiazepines are uniquely safe for short-term use (dependence develops quickly; highest fracture risk is in new users). That benzodiazepines definitively cause dementia (association is observational and confounded). Which patients will develop dependence or severe withdrawal. Long-term functional outcomes with chronic use.

The structural problem: Efficacy trials are short (1–4 weeks), exclude the populations most likely to be prescribed benzodiazepines (elderly, medically complex, psychiatric comorbidities), and study chronic anxiety disorders rather than newly onset distress. In 2018, 50% of patients dispensed oral benzodiazepines received them for ≥2 months, but evidence base is short-term. Dementia evidence is observational. The folk models—benzodiazepines as uniquely safe for short-term use or benzodiazepines as never justifiable—both overclaim what opened trials support.

---

References

Primary sources cited

Baandrup, L., Ebdrup, B. H., Rasmussen, J. Ø., Lindschou, J., Gluud, C., & Glenthøj, B. (2018). "Pharmacological interventions for benzodiazepine discontinuation in chronic benzodiazepine users." Cochrane Database of Systematic Reviews, (3), CD011481.

Billioti de Gage, S., Bégaud, B., Bazin, F., Verdoux, H., Dartigues, J.-F., Pérès, K., Kurth, T., & Pariente, A. (2012). "Benzodiazepine use and risk of dementia: prospective population based study." BMJ, 345, e6231.

Billioti de Gage, S., Moride, Y., Ducruet, T., Kurth, T., Verdoux, H., Tournier, M., Pariente, A., & Bégaud, B. (2014). "Benzodiazepine use and risk of Alzheimer's disease: case-control study." BMJ, 349, g5205.

Donnelly, K., Bracchi, R., Hewitt, J., Routledge, P. A., & Carter, B. (2017). "Benzodiazepines, Z-drugs and the risk of hip fracture: A systematic review and meta-analysis." PLoS ONE, 12(4), e0174730.

Munkholm, K., Ussing, A., Brink, M., Edemann-Callesen, H., Sari Canbolat, S., Christensen, R., Søgaard Dahl, K., Ebdrup, B. H., Juul Jensen, M. E., Kierulf-Lassen, C., Krogh Madsen, G., Mai Nielsen, S., Paludan Paulsen, C., Friis Rohde, J., Tarp, S., & Baandrup, L. (2024). "Minor tranquillizers for short-term treatment of newly onset symptoms of anxiety and distress: a systematic review with network meta-analysis of randomized trials." European Archives of Psychiatry and Clinical Neuroscience, 274, 475–486.

Olfson, M., King, M., & Schoenbaum, M. (2015). "Benzodiazepine use in the United States." JAMA Psychiatry, 72(2), 136–142.

Taipale, H., Särkilä, H., Tanskanen, A., Kurko, T., Taiminen, T., Tiihonen, J., Sund, R., Tuulio-Henriksson, A., Saastamoinen, L., & Hietala, J. (2020). "Incidence of and Characteristics Associated With Long-term Benzodiazepine Use in Finland." JAMA Network Open, 3(10), e2019029.

US Food and Drug Administration. (2020). FDA requiring boxed warning updated to improve safe use of benzodiazepine drug class. Drug Safety Communication, September 23, 2020.

---

Appendix: Bind table for headline executive summary numbers

Every executive summary number is bound to an opened primary source with exact location:

Claim	Source	Extract Location	Unit	Population	Status
Benzodiazepines vs placebo SMD −0.58 (−0.77 to −0.40), low certainty	Munkholm et al., 2024	Results section, network meta-analysis	Standardized mean difference	34 RCTs (31 in NMA), 7044 participants, anxiety/distress, 1–4 weeks	✅ Opened
Pregabalin vs placebo SMD −0.58 (−0.87 to −0.28), low certainty	Munkholm et al., 2024	Results section, network meta-analysis	Standardized mean difference	Same population	✅ Opened
Quetiapine vs placebo SMD −0.51 (−0.90 to −0.13), low certainty	Munkholm et al., 2024	Results section, network meta-analysis	Standardized mean difference	Same population	✅ Opened
FDA FAERS: 104 cases, time-to-dependence "days to weeks"	FDA 2020 Data Summary (DSC)	"We evaluated 104 cases from the FDA Adverse Event Reporting System…"	Count + timeline	FAERS 1968-2019, single-substance reports	✅ Opened (regulatory)
FDA FAERS: Withdrawal "weeks to years"	FDA 2020 Data Summary (DSC)	"…variations in the duration of the withdrawal symptoms that lasted from weeks to years."	Duration range	Same FAERS cases	✅ Opened (regulatory)
FDA: 50% of patients dispensed oral BNZ for ≥2mo in 2018	FDA 2020 Data Summary (DSC)	"In 2018, an estimated 50% of patients who were dispensed oral benzodiazepines received them for a duration of two months or longer."	Percentage	US dispensing data 2018	✅ Opened (regulatory)
Benzodiazepines hip fracture RR 1.52 (1.37–1.68)	Donnelly et al., 2017	Results section, meta-analysis	Relative risk	Meta-analysis hip fracture	✅ Opened
Z-drugs hip fracture RR 1.90 (1.68–2.13)	Donnelly et al., 2017	Results section, meta-analysis	Relative risk	Meta-analysis hip fracture	✅ Opened
Short-term benzodiazepines hip fracture RR 2.40 (1.88–3.05)	Donnelly et al., 2017	Results section, short-term use subgroup	Relative risk	New users	✅ Opened
Short-term Z-drugs hip fracture RR 2.39 (1.74–3.29)	Donnelly et al., 2017	Results section, short-term use subgroup	Relative risk	New users	✅ Opened
Dementia new use HR 1.60 (1.08–2.38)	Billioti de Gage et al., 2012	Results section, multivariable adjusted	Hazard ratio	1063 participants, 15-year follow-up	✅ Opened [contested]
Dementia pooled new users HR 1.46 (1.10–1.94)	Billioti de Gage et al., 2012	Results section, pooled analysis	Hazard ratio	Five cohorts	✅ Opened [contested]
Dementia ever use OR 1.55 (1.24–1.95)	Billioti de Gage et al., 2012	Results section, nested case-control	Odds ratio	Same cohort	✅ Opened [contested]
8.7% US adults 65–80 filled prescription; 31.4% long-term	Olfson et al., 2015	Abstract, Results section	Percentage	US adults 65–80	✅ Opened
39.4% incident users → long-term (≥180 days)	Taipale et al., 2020	"During the follow-up period, 51 099 BZDR users (39.4%) became long-term users."	Percentage	129,732 incident users, Finland, median 2.1yr follow-up	✅ Opened
FDA: 167,845 ED visits 2016	FDA 2020 Data Summary (DSC)	"In 2016, the nationally estimated number of emergency department (ED) visits due to nonmedical use of benzodiazepines (n=167,845)…"	Count	US 2016	✅ Opened (regulatory)
FDA: 55% BNZ deaths 2013-2017 with opioids	FDA 2020 Data Summary (DSC)	"From 2013-2017, 55% of benzodiazepine-involved overdose deaths also documented involvement of prescription opioids."	Percentage	US 2013-2017	✅ Opened (regulatory)

All headline numbers are bound to opened primary sources with exact locations verified. Observational studies (Billioti de Gage, Olfson, Taipale) marked as such. FDA findings marked as regulatory statements, not trial results. No reconstructed CIs. No unopened papers cited in executive summary. Dropped unverified papers: Baandrup 2018 RRs (kept no-firm-conclusions in body), Zhong 2020 cognitive numbers (identity error: paper is Liu et al. 2020 Frontiers Psychiatry, not Zhong et al.), Johansen 2019 trend numbers, Xing 2014 overall fracture, Guaiana 2023 / Slee 2019 / Baek 2024 / Vicens 2022 specific numbers.