Long-term persistence of withdrawal of temazepam, zopiclone, and zolpidem in older adults: a 3-year follow-up study

Long-term benzodiazepine agonist (BZDA) use can result in adverse outcomes such as increased risk of falls, fractures, cognitive decline, and mortality [1‐3]. Prolonged benzodiazepine use as a hypnotic is considered inappropriate, according to guidelines on pharmacotherapy in the aged [4]. However, a considerable proportion of subjects with insomnia using BZDA hypnotics on a nightly basis for years do still experience sleep disturbances despite hypnotic medication [5, 6]. In a primary-care population, prevalence of BZDA hypnotic drug prescriptions is much higher for multimorbid patients than for patients without multimorbidity [7]. The prevalence of chronic BZDA use is particularly high in nursing-home residents, it has ranged in European nursing homes from 28% [8] to more than 50% [9].

Meta-analyses of interventions for reducing inappropriate long-term BZDA use in older adults have shown supervised BZDA withdrawal augmented with psychotherapy to be the most effective intervention, but for pragmatic reasons, a patient-centred approach with individual planning and monitoring or medication reviews is the recommendation [10, 11]. Very little data on the long-term persistence of BZDA withdrawal results exist, as most of the follow-ups have been short, 0.5 to 3 months [11]. Few studies have reported follow-ups lasting 12 months or longer [11‐13]. Morin et al. did a 24-month outcome study among community-dwelling residents who had used as hypnotics various benzodiazepine medicaments but not the widely used Z-drugs: zopiclone or zolpidem [14]. Vicens et al. [15] compared, in a 3-year study, educational methods with routine care in regard to cessation of benzodiazepine use. However, most of their patients seem to have used benzodiazepines for indications other than as hypnotics, and a medical withdrawal intervention group was lacking. Furthermore, whether BZDA withdrawal can affect the use of other central nervous system (CNS) -affecting drugs remains unknown. Our outpatient study is the first in which persistence of BZDA hypnotic (mainly Z-drugs) discontinuation has been followed for up to 3 years in chronic BZDA users, after melatonin/placebo-supported initial withdrawal.

We performed in older outpatients a psychosocially supported BZDA withdrawal intervention as described earlier [16]. BZDA withdrawal rapidly improved muscle strength and balance, but failed to improve cognitive performance [17, 18]. The present secondary data analysis aimed to study in patients of our original cohort their long-term persistence of BZDA withdrawal and possible changes in their medications. In short, our aim was to describe the persistence of temazepam, zolpidem, and zopiclone (BZDA) withdrawal among older outpatients at 3 years from the beginning of withdrawal, as well as any changes in their use of other medications.

Details of this temazepam, zopiclone, and zolpidem withdrawal study, its participants, interventions, measurements, and withdrawal results up to 6 months have appeared elsewhere [16‐18]. In short, the original Satauni study was a randomized, double-blind, placebo-controlled, parallel-group study on the efficacy of daily melatonin (2 mg) in BZDA withdrawal during a one-month period and during a double-blind 5-month follow-up. At baseline, a physician provided individual psychological support and sleep-hygiene counselling, including discussions about regular sleep rhythm and factors influencing sleep. The psychological support for all participants was continued by a nurse who provided a supportive visit once a week and was available by phone during the one-month withdrawal (period). The participants had follow-up meetings with a nurse at months 2 and 6, and with a physician at month 6 after withdrawal initiation. After this, the participants returned to normal, routine care at their health centres.

The main inclusion criteria at baseline were primary insomnia according to the criteria of the Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) [19], age ≥ 55 years, and long-term (> 1 month) regular night-time use of temazepam, zopiclone, or zolpidem to treat primary insomnia. The key exclusion criteria were BZDA use other than those uses identified above, current use of antipsychotic or antiepileptic medication, active alcohol or drug abuse or a history of abuse, anxiety disorder or other psychiatric disorder, neurological disease, smoking more than ten cigarettes a day, or autoimmune disease [16].

For the present 3-year follow-up study, we sent a letter to all our 92 initial participants and invited them for a follow-up meeting and measurements. The study nurse (JS, MS) met the participants and interviewed them. If participants were unable to schedule an in-person meeting, the study nurse called each one on the telephone and requested return of the completed questionnaire.

The Satauni study protocol was approved by the Ethics Committee of Satakunta Hospital District (2§/7/2008) and by the National Agency for Medicines of Finland (218/2008) and registered to EudraCT (2008–0006795-30). Written informed consent was received from each participant before the study began.

To our knowledge, this is the lengthiest study to ever report BZDA withdrawal results after medical withdrawal intervention (melatonin or placebo) with its follow-up of 3 years. Our cohort had a high follow-up retention rate after a structured BZDA withdrawal intervention in older outpatients, individuals who had for a long time regularly used short-acting BZDAs as hypnotics. Most (85%) of our participants had used chronically short-acting “Z-drugs,” zopiclone or zolpidem, and only 15% had temazepam as their nightly hypnotic. The short-term (at 1 month) withdrawal results in this cohort were good [16]. However, BZDA use increased markedly over time after the withdrawal period.

Three years after the one-month BZDA withdrawal period, roughly one-third of the initial participants remained non-users, one-third used BZDAs irregularly, and one-third took BZDAs regularly as a hypnotic. The withdrawal rates and persistence of the withdrawal seemed to be similar irrespective of the hypnotic. Controlled-release melatonin during the one-month BZDA withdrawal period did not affect the long-term BZDA-withdrawal result. One explanation for persistence of BZDA withdrawal was the psychosocial support during the first month and at 2 and 6 months for all participants. The number of all medications (median four drugs per patient at baseline) did not increase within 3 years in non-users, whereas during that period—for irregular users and regular users—it increased. However, the number of participants using antidepressants, dopamine agonists, melatonin, and NSAIDs/paracetamol increased in all groups. Worryingly, some long-acting benzodiazepines and BZDA combination preparations with anticholinergic antidepressants were prescribed by the time of the 3-year follow-up.

Most patient characteristics at baseline were fairly similar in all groups despite the patients’ differing BZDA use at the 3-year follow-up point. Yet, higher BMI at baseline, i.e., before start of withdrawal, was associated at the 3-year follow-up with regular BZDA use. Higher BMI may be associated with sleep apnoea, but treatment more appropriate than BZDA would be weight loss and positive airway pressure therapy. As no clear markers as yet exist, it is, however, impossible to predict at baseline who will be a successful BZDA withdrawer and who will not.

Expected health between the groups at baseline was not uniform. As there were no major differences in majority background variables, we decided that BMI and expected health were not clinically meaningful background variables to be adjusted for. Furthermore, small sample size would have complicated the multivariate analyses. Of note, most—meaning 23 of the 26 final non-users (at 3 years)—had used BZDA hypnotics regularly for at least 5 years before entering this withdrawal study. Thus, even after their very long-time use, BZDA hypnotics, at the usual therapeutic doses, may not reduce the likelihood of withdrawal of a patient who is well motivated. On the other hand, a clear distinction arises, as pointed out, in withdrawal success between abusers of high doses of BZDAs and those using therapeutic doses as hypnotics [23].

All our patients were outpatients, and nine of them could not be contacted for the 3-year follow-up; at least two of them had died, and one moved away before the 3-year follow-up point. Here, percentages of withdrawal persistence were calculated conservatively by the intention-to-treat (ITT) principle, meaning comparison of numbers withdrawing to those 92 who had entered the withdrawal study. Thus, the true withdrawal persistence could be somewhat better than reported here.

According to recent withdrawal studies, spontaneous BZDA withdrawal rate without any interventions ranges from 5 to 26% [10, 11, 15, 24]. The study other than ours with a 3-year follow-up reporting the results of educational withdrawal intervention [15], with one-time counselling or with counselling combined with follow-up meetings lasting between 2 and 3 weeks, produced respective withdrawal rates of 41 and 39%. In their control group (routine care), 26% had withdrawn [15]. In our study, the BZDA withdrawal rate was 28% at 3 years after psychosocially supported BZDA withdrawal augmented with melatonin.

Persistence of BZDA withdrawal over years might have been better had psychosocial support continued after the one-month period. Additionally, there exist several BZDA withdrawal interventions with shorter follow-ups: In Canada, counselling at a pharmacy produced a 27% persistence rate of BZDA withdrawal and dose reductions for 11% of its participants at 6 months [24]. An American study [25] randomized long-term BZDA users into three groups, among which one group received cognitive behavioural therapy, one placebo therapy (biofeedback), and one received physician-given counselling for gradual BZDA withdrawal. In all groups, the BZDA use decreased 84% compared to baseline, and BZD withdrawal persistence remained at one-third compared to baseline in all groups for up to 1 year [25]. However, no direct comparisons between previous studies can be made due to different withdrawal methods and follow-ups times.

Why does BZDA withdrawal induce increased use of other CNS medications? First, hypnotic drugs have a strong placebo effect [26]. Second, it can be hypothesized that BZDA withdrawal reveals previously untreated conditions. The increase in symptomatic medications can be explained by such symptoms that had been eased by BZDA. Insomnia may precede comorbid depression. The undiagnosed depressive symptoms, depression, and muscle aches or progressive osteoarthritis, as well as sleep problems explain at least partially participants’ increased antidepressant use. By 3 years, melatonin seems to have replaced BZDAs as the hypnotic for some (27%) of the BZDA non-users. Interestingly, melatonin itself, in a recent meta-analysis, did not improve BZDA withdrawal [27]. The minor increase in low-dose pramipexole (dopamine agonist) use in BZDA non-users and in irregular users may be related to its use for restless legs which had contributed to sleep disturbances; in fact, BZDAs may have masked previously unrecognized symptoms of restless legs syndrome.

The main strengths of this study are its long follow-up time of up to 3 years and the very high follow-up retention rate of participants (90%), of whom most had for a lengthy period used zopiclone or zolpidem up until the withdrawal. One potential weakness is the relative small sample size that limits validity of conclusions and does not allow comparisons between individual hypnotics. Another potential weakness is that use of BZDAs and other medications at the 6-month and 3-year follow-up points was based on interview data verified from medical records but not by blood- or urine drug determinations. However, no significant discrepancy between patient-reported BZDA use and plasma levels emerged at baseline or at the one-month follow-up point in these participants [15]. Additionally, we have no data on the medications between the 6-month and 3-year follow-up points.

Psychosocially supported gradual BZDA withdrawal was effective in discontinuation of long-term hypnotic use, but withdrawal persistence decreased over time. At 3 years after withdrawal, nearly one-third of the previous chronic users were BZDA-free, one-third used it irregularly, and one-third continued nightly use. High BMI seems to predict poor withdrawal persistence, but melatonin given during the withdrawal month failed to improve persistence results.