CoChat
Sign Up
Cited by 307Pharmaceutical Practices and Patient OutcomesDementia and Cognitive Impairment ResearchSubstance Abuse Treatment and Outcomes

Persistence of cognitive effects after withdrawal from long-term benzodiazepine use: a meta-analysis

Barker, M·2004·Archives of Clinical Neuropsychology
DOI 10.1016/s0887-6177(03)00096-9
Open this paper in CoChat — free account

Abstract

Despite the widespread prescribing of benzodiazepines, uncertainty still surrounds the potential for cognitive impairment following their long-term use. Furthermore, the degree of recovery that may take place after withdrawal or the level of residual impairment, if any, that is maintained in long-term benzodiazepine users is also unclear. The current paper employed meta-analytic techniques to address two questions: (1) Does the cognitive function of long-term benzodiazepine users improve following withdrawal? (2) Are previous long-term benzodiazepine users still impaired at follow-up compared to controls or normative data? Results of the meta-analyses indicated that long-term benzodiazepine users do show recovery of function in many areas after withdrawal. However, there remains a significant impairment in most areas of cognition in comparison to controls or normative data. The findings of this study highlight the problems associated with long-term benzodiazepine therapy and suggest that previous benzodiazepine users would be likely to experience the benefit of improved cognitive functioning after withdrawal. However, the reviewed data did not support full restitution of function, at least in the first 6 months following cessation and suggest that there may be some permanent deficits or deficits that take longer than 6 months to completely recover.

Integrity signals

Registry

Crossref · 10.1016/s0887-6177(03)00096-9

Cites

33 of 33 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

0 of 17 cited sources checked · 0 supported · 0 flagged

  • Findingown data

    Weighting for sample size had little impact on the effect size (0.42 unweighted compared to 0.40 weighted, meta-analysis A; -0.41 unweighted compared to -0.48 weighted, meta-analysis B).

    “Weighting for sample size had little impact on the effect size (0.42 unweighted compared to 0.40 weighted, meta-analysis A; -0.41 unweighted compared to -0.48 weighted, meta-analysis B)”
    Quote in the paperNot a substantial claim
  • Findingown data

    Analyses of moderator variables did not reveal any significant relationships.

    “An analyses conducted on moderator variables did not reveal any significant relationships.”
    Quote in the paperNot a substantial claim
  • Findingown data

    There remains a significant impairment in most areas of cognition in long-term benzodiazepine users at follow-up compared to controls or normative data.

    “However, there remains a significant impairment in most areas of cognition in comparison to controls or normative data.”
    Quote in the paperSubstantial claim
  • Findingown data

    The mean weighted effect size for cognitive improvement after withdrawal (meta-analysis A) was 0.41 (median = 0.37, SD = 0.22).

    “The mean weighted effect size was 0.41 (median = 0.37) with a standard deviation of 0.22.”
    Quote in the paperSubstantial claim
  • Findingown data

    In meta-analysis A, effect sizes greater than zero were found across all cognitive domains measured.

    “In meta-analysis A, effect sizes greater than zero were found across all cognitive domains measured”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 16
  • May combine two claims. 23

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Studies published later than 1994 tended to have larger effect sizes.

    “One trend of interest was a relationship between publication year and overall effect size, such that studies published later than 1994 tended to have larger effect sizes.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    In meta-analysis B, the mean weighted effect size was -0.48 (median = 0.48) with a standard deviation of 0.45.

    “The mean weighted effect size was -0.48 (median = 0.48) with a standard deviation of 0.45.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Findingnone given

    The studies included in these meta-analyses were all published in peer-reviewed journals.

    “The studies included in these meta-analyses were all published in peer-reviewed journals.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The fail-safe n was calculated to be four for meta-analysis A and five for meta-analysis B.

    “The fail-safe n, which represents the number of studies obtaining null results that would need to be in existence to threaten the significant effect size found, was calculated to be four for meta-analysis A and five for meta-analysis B.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    In meta-analysis B, the 95% confidence intervals did not span zero for 8 of the 11 effect sizes, so most of these effects are significant and different from zero.

    “It is clear from Fig. 2 that the 95% confidence intervals do not span zero for 8 of the 11 effect sizes and therefore, that most of these effects are significant and different to zero”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    The 95% confidence intervals did not span zero for 5 of 11 effect sizes, indicating significant improvement in those cognitive categories.

    “It is clear from Fig. 1 that the 95% confidence intervals do not span zero for 5 of 11 effect sizes and therefore, that these effects are significant and different from zero”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    In meta-analysis B, from nine studies covering 12 cognitive categories, 37 effect sizes were obtained.

    “From nine studies covering 12 categories (a maximum of 108 possible effect sizes), 37 effect sizes were obtained.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    In the present meta-analyses, the tolerance level is estimated at 60 for meta-analysis A and 55 for meta-analysis B.

    “In the present meta-analyses, the tolerance level is estimated at 5(10) + 10 = 60 and 5(9) + 10 = 55 for meta-analyses A and B, respectively.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    All effect sizes for cognitive improvement after withdrawal were positive, ranging from 0.06 to 0.70.

    “All of the effect sizes were positive and ranged in magnitude from 0.06 to 0.70.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingnone given

    Mean scale scores in four of these six studies would be considered in the clinical range and two in the non-clinical range.

    “Mean scale scores in four of these would be considered in the clinical range and two in the non-clinical range.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    The degree of recovery in attention/concentration is significant with a moderate to large effect size (d = 0.69), and the level of residual impairment is significant with a small to moderate effect size (d = -0.43).

    “For example, the degree of recovery that appears to take place in the domain of attention/concentration is significant with a moderate to large effect size (d = 0.69), and a significant, small to moderate effect size (d = -0.43)indicating the level of residual impairment.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data

    Seven out of 10 studies reported pre- and post-withdrawal anxiety or depression levels, and in all of these, patients' symptoms either improved significantly or remained the same.

    “Seven out of 10 studies reported pre-and post-withdrawal anxiety or depression levels and in all of these, patient's symptoms either improved significantly or remained the same.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    Six studies specified pre- and post-withdrawal psychiatric symptoms using a variety of subjective mood scales.

    “Six studies specified pre-and post-withdrawal psychiatric symptoms using a variety of subjective mood scales including the Spielberger State-Trait Anxiety Inventory, the Hamilton Anxiety Scale, the Hamilton Depression Scale, and the Beck Depression Inventory.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data

    It was not possible to calculate effect sizes for all 12 categories because only one test within the problem-solving category was used.

    “it was not possible to calculate effect sizes for all 12 categories as only one test within the problem-solving category was used”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Five studies specified the conditions for which subjects used benzodiazepines, and for all five the conditions were anxiety or depression.

    “Five studies specified the conditions for which subjects used benzodiazepines and for all five the conditions were anxiety or depression.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data

    Eight studies excluded patients with a history of heavy alcohol or other drug use, and one study stated that 23% of patients had a history of alcohol use in excess of four standard drinks per day.

    “Eight studies excluded patients with a history of heavy alcohol or other drug use and one study stated that 23% of patients had a history of alcohol use in excess of four standard drinks per day.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data

    All six studies reported that mean scores either remained the same following withdrawal (four studies) or significantly improved (two studies).

    “All six studies reported that mean scores either remained the same following withdrawal (four studies) or significantly improved (two studies).”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductreasoning

    The significant negative correlation between age and overall effect size should be interpreted with caution due to the inflation in type I error rate associated with the number of analyses conducted.

    “However, it should be interpreted with caution due to the inflation in type I error rate associated with the number of analyses conducted.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    There may be some permanent cognitive deficits, or deficits that take longer than 6 months to completely recover, after long-term benzodiazepine use.

    “and suggest that there may be some permanent deficits or deficits that take longer than 6 months to completely recover.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    A conservative approach was adopted for coding significance levels in the meta-analysis.

    “With regard to the coding of significance levels, a conservative approach was adopted.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    According to Cohen (1988), effect sizes of d = 0.20, 0.50, and 0.80 are considered small, medium, and large in magnitude, respectively.

    “According to Cohen (1988), effect sizes of d = 0.20, 0.50, and 0.80 are considered small, medium, and large in magnitude, respectively.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Cohen (1988)

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning

    Each study was allowed to contribute only one effect size to each cognitive category by averaging effect sizes from multiple tests, giving equal weight to each study per category.

    “Each study was only allowed to contribute one effect size to each cognitive category by averaging together the effect sizes from the same study if more than one type of test was used to measure a particular category.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductnone given

    Effect sizes were weighted on the basis of their sample size.

    “Effect sizes were weighted on the basis of their sample size.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Premisecitation

    Rosenthal has suggested that a conservative estimate for the tolerance level is 5k + 10 where k is the number of studies retrieved.

    “Rosenthal has suggested that a conservative estimate for this tolerance level is 5k + 10 where k is the number of studies retrieved.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Rosenthal

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationown data

    Since the fail-safe n falls short of the tolerance level, the possibility of a file drawer problem should be considered for both meta-analyses, and the resulting effect sizes interpreted with caution.

    “Since the fail-safe n falls short of the tolerance level, the possibility of a file drawer problem should be considered for both meta-analyses, and the resulting effect sizes interpreted with caution.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationcitation

    The finding that improvement in attention/concentration after withdrawal decreases with age may be related to deficits in performance on tasks of attention and concentration observed to accompany normal aging.

    “This finding may be related to deficits in performance on tasks of attention and concentration observed to accompany normal aging (Lezak, 1995).”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Lezak, 1995

  • May combine two claims.

    Premisecitation

    In a previous meta-analysis by the authors, moderate to large effect sizes were found across all cognitive areas studied, indicating that long-term benzodiazepine users are impaired across many cognitive areas.

    “In a previous meta-analysis that we have conducted looking at the long-term cognitive effects of benzodiazepine use (Barker, Greenwood, Jackson & Crowe, in press) we found moderate to large effect sizes across all cognitive areas studied indicating that long-term benzodiazepine users are impaired across many cognitive areas.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Barker, Greenwood, Jackson & Crowe, in press

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Interpretationreasoning

    It seems unlikely that a dementing process contributed to the cognitive deficits, given that the mean age in nine studies was between 39 and 47 years, and the remaining study screened for and excluded patients with dementia.

    “This seems unlikely, however, given that the mean age in nine of the studies was between 39 and 47 years, and the remaining study, which examined mainly older adults, did in fact screen for and exclude patients with dementia.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Interpretationreasoning

    It may be the case that some real differences exist that were undetectable given the small sample size.

    “It may be the case that some real differences do exist that were undetectable given the small sample size.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data

    There may be some permanent deficits or deficits that take periods longer than 6 months to completely recover.

    “and suggests that there may be some permanent deficits or deficits that take periods longer than 6 months to completely recover.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductown data

    Due to the small number of studies that met inclusion criteria and the limited information provided on relevant characteristics, an analysis of heterogeneity (Q) was considered inappropriate.

    “Due to the small number of studies that met inclusion criteria in this meta-analysis, and the limited information provided on relevant characteristics, an analysis of heterogeneity (Q) (Hedges & Olkin, 1985) was considered inappropriate.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites Hedges & Olkin, 1985

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Premisecitation

    Gorenstein et al.'s (1995) well-controlled comparison of neuropsychological test performance in anxious drug-free patients and normal controls found no significant difference in performance on a variety of neuropsychological tests except for a ball-bearing test measuring motor co-ordination.

    “Gorenstein et al.'s (1995) well-controlled comparison of neuropsychological test performance in anxious drug-free patients and normal controls found no significant difference in performance on a variety of neuropsychological tests except for a ball-bearing test measuring motor co-ordination.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Gorenstein et al. (1995) source

  • May combine two claims.

    Interpretationreasoning

    Integrating all available evidence indicates that long-term benzodiazepine users may be affected in a generalized rather than specific way, with some areas being more affected than others.

    “integrating all of the available evidence indicates that long-term benzodiazepine users may be affected in a generalized rather than specific way, with some areas being more affected than others.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductnone given

    The presence or absence of dementia in older subjects was not typically reported, and the possibility of a dementing process as a contributing factor to the cognitive deficits observed cannot be ruled out.

    “The presence or absence of dementia in older subjects was not typically reported and the possibility of a dementing process as a contributing factor to the cognitive deficits observed cannot be ruled out.”
    Quote in the paperNot a substantial claim
34 of 34

  • The paper relies on this without support.

    Little is known regarding the potential long-term effects of benzodiazepines.

    “Despite such widespread use, little is known regarding the potential long-term effects of benzodiazepines.”
1 of 1

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • In a meta-analysis, Barker et al. concluded that long-term benzodiazepine users show recovery of function in visuospatial skills, attention/concentration, general intelligence, psychomotor speed, and nonverbal memory after withdrawal.

    “In a metaanalysis, Barker et al. (170) concluded that longterm BZD users show recovery of function in visuospatial skills, attention ⁄ concentration, general intelligence, psychomotor speed and nonverbal memory assessments after withdrawal.”

    From Benzodiazepines in epilepsy: Pharmacology and pharmacokinetics (2008) · cites this paper as [170]

  • The study data did not demonstrate complete restoration of function within the first 6 months after discontinuation of benzodiazepine use.

    “The authors reported that the study data did not demonstrate complete restoration of function within the first 6 months following discontinuation of BZD use”

    From Benzodiazepines in epilepsy: Pharmacology and pharmacokinetics (2008) · cites this paper as [170]

  • Longer than 6 months may be needed for recovery from some deficits after benzodiazepine discontinuation.

    “they suggested that longer than 6 months may be needed for recovery from some deficits.”

    From Benzodiazepines in epilepsy: Pharmacology and pharmacokinetics (2008) · cites this paper as [170]

  • For long-term benzodiazepine users, the possible degree of recovery and the extent of residual impairment after withdrawal are unclear.

    “For individuals who use BZDs long term, the possible degree of recovery and the extent of residual impairment that may remain after the drugs are withdrawn are unclear (170).”

    From Benzodiazepines in epilepsy: Pharmacology and pharmacokinetics (2008) · cites this paper as [170]

  • The same authors found that significant cognitive impairment was more persistent in long-term users than in control groups or normative data.

    “The authors found, however, that significant cognitive impairment was more persistent in long-term users than in control groups or in normative data.”

    From Benzodiazepines in epilepsy: Pharmacology and pharmacokinetics (2008) · cites this paper as [170]

Statistics recomputed

Every reported test result, recomputed from its test statistic and degrees of freedom. A result shows as not adding up only once an independent check has confirmed it.

  • Consistentr(8) = -0.72reported p < .05 · recomputed p = .019

    “r (8) = -.72,p < .05”

1 of 1

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Two weighted overall effects are given for meta-analysis A

    Minor

    The discussion gives 0.40 as the weighted effect, while Table 3 and the results section give 0.41.

  • Meta-analysis B median has the wrong sign

    Minor

    The reported summary is for the weighted cognitive-category effects in Table 4. Its two middle values are −0.49 and −0.47, giving a median of −0.48, not 0.48.

    Reported 0.48 · recomputed -0.48

  • The age correlation is assigned to different outcomes

    Minor

    The results identify attention/concentration as the outcome of the sole significant age correlation; the discussion instead calls it an overall-effect correlation.

  • Table 3 is not ordered by decreasing weighted effect

    Minor

    The table claims descending order, but verbal memory has a weighted effect of 0.36 and appears after motor control/performance at 0.21.

4 of 4

Cites

Flagged first, then by citations

1 of the 34 references couldn’t be matched to a known paper.

Research works printed but not linked: 2

  1. [7]7. * , † Birzele, H. J. (1992). Benzodiazepine induced amnesia after long-term medication and during withdrawal. European Review of Applied Psychology, 42, 277-282.
  2. [21]21. Lader, M. H. (1982). Psychological effects of buspirone. Journal of Clinical Psychiatry, 43, 62-67.

Cited by

Sorted by citations

Related

Authors
Barker, M.
Affiliations
La Trobe University (AU)
Corresponding author
Barker, M
Venue
Archives of Clinical Neuropsychology
Publisher
Oxford University Press (OUP)
Published
2004-04
Volume
19
Issue
3
Pages
437-454
ISSN
0887-6177 (ISSN-L)
DOI
10.1016/s0887-6177(03)00096-9
OpenAlex
W2169064904
Cited by
307
References
34
Citations by year
2026: 8 · 2025: 17 · 2024: 12 · 2023: 13 · 2022: 12 · 2021: 15 · 2020: 21 · 2019: 19 · 2018: 18 · 2017: 25 · 2016: 22 · 2015: 14 · 2014: 12 · 2013: 16 · 2012: 13
FWCI
0.94
Type
journal article
Language
en
Open access
closed
Locations
  • Archives of Clinical Neuropsychology (publishedVersion)
  • PubMed (publishedVersion)
  • RMIT Research Repository (RMIT University Library) (submittedVersion)
Topics
  • Health Sciences › Medicine › Geriatrics and Gerontology › Pharmaceutical Practices and Patient Outcomes
  • Health Sciences › Medicine › Psychiatry and Mental health › Dementia and Cognitive Impairment Research
  • Health Sciences › Medicine › Epidemiology › Substance Abuse Treatment and Outcomes
Keywords
benzodiazepines, long-term benzodiazepine use, benzodiazepine withdrawal, cognitive impairment, cognitive recovery, residual cognitive deficits, benzodiazepine withdrawal syndrome
MeSH terms
Benzodiazepines, Clinical Trials as Topic, Cognition Disorders, Humans, Long-Term Care, Neuropsychological Tests, Psychometrics, Substance Withdrawal Syndrome

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-11.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-11.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review