Habraken H, Soenen K, Blondeel L, et al. Gradual withdrawal from benzodiazepines in residents of homes for the elderly: Experience and suggestions for future research. European Journal of Clinical Pharmacology 1997;51:355-58
PubMed link: https://www.ncbi.nlm.nih.gov/pubmed/9049574
Full text article: https://link.springer.com/article/10.1007%2Fs002280050213
| Methods | Study design: Randomized double-blind placebo-controlled study with a parallel-design
Number of groups: Two groups |
| Participants | Number of participants: 55 randomized, 36 completed
· Active deprescribing group: 27 randomized, 17 completed · Control group: 28 randomized, 19 completed Age: 84 years Sex: 45 females, 10 males Participants with dementia: No Inclusion criteria: · 65 years and over · Use of benzodiazepine for at least one year · Intake of a constant dose of benzodiazepine once-daily during the month before the start of withdrawal Exclusion criteria: · Serious dementia · Ability of answer simple questions · Serious medical disease · Serious psychological problems · Recent psycho-traumatic experience (e.g. hospitalization, death of a relative) Country: Belgium Setting: Residential aged care facilities – 10 facilities in Ghent and the surrounding area with a total of 993 residents (328 of whom used benzodiazepines) |
| Interventions | Medicine: Benzodiazepine
Withdrawal schedule: Withdrawn over 5 weeks · 25% reduction per week for week 1, 2 and 3 · 12.5% reduction week 4 and 5 Comparator: Placebo compared to continued treatment |
| Outcomes | Level of daily functioning
Adverse drug withdrawal effects |
| Dates | Dates: Not described
Follow-up duration: 12 months |
| Funding sources | Minister of the Government of Flanders for Employment and Social Affairs |
| Notes |
Risk of bias table
| Bias | Authors’ judgment | Support for judgment |
| Random sequence generation (selection bias) | Unclear risk | Method not described.
“Participants were randomly allocated to the experimental or to the control group, after stratification by home.” |
| Allocation concealment (selection bias) | Unclear risk | Not described. Insufficient detail about the allocation was given. |
| Blinding of participants and personnel (performance bias) | Low risk | The authors state that the study was double-blind, but do not describe method of blinding for personal or outcome assessors nor do they describe the placebo.
“In the experimental group, the daily dose of lorazepam had to be gradually replaced by placebo over a period of 5 weeks: a reduction of 25% per week in the first 3 weeks of withdrawal, and a reduction of 12.5% per week in the following 2 weeks of withdrawal. Experimental subjects continued taking placebo medication for another year after withdrawal. In the control group, lorazepam was administered during the whole study period. The study design was double-blind.” |
| Blinding of outcome assessment (detection bias) | Unclear risk | As above. |
| Incomplete outcome data (attrition bias) | High risk | “Approximately one-third of subjects in both study groups dropped out before the end of the study (December 1994). Two persons in each group dropped out due to hospitalization. general practitioners stopped the study medication in eight persons of the placebo group (four during or shortly after withdrawal) and three of the lorazepam group (one just before withdrawal, two during withdrawal). The interruption was due to complaints of insomnia in five placebo users and two lorazepam users. The patients who dropped out were not different from those continuing the intake of study medication regarding personal characteristics or baseline scores on the questionnaires. However, lorazepam intake at the start of the study appeared different between drop-outs and stayers. First, all drop-outs had been converted to lorazepam for study purposes, compared to 70% of the stayers. Second, in the placebo group drop-outs were taking a higher daily dose of lorazepam than stayers (mean drop-outs = 2.0 (SD — 0.74); mean stayers = 1.3 (SD = 0.66)).”
Huge number of drop-outs. Reasons for drop-outs likely related to intervention. |
| Selective reporting (reporting bias) | High risk | Results not reported for sleep quality (subjective appraisal – used Groningen Sleep Quality Scale) despite stated it was measured at weekly intervals during withdrawal phase, and again at 6 and 12 months.
Outcome criteria were not clearly stated. The following were mentioned in the method: |
| Other bias | High risk | The choice of lorazepam as the standardized benzodiazepine is an inherent methodological design flaw as it is a relatively short half-life benzodiazepine and more likely to cause withdrawal symptoms than diazepam for instance. |
Leave a Reply