Objectives: To evaluate the clinical and cost impact of providing, in routine general practice settings, a cognitive-behaviour therapy (CBT) package for insomnia to long-term hypnotic drug users with chronic sleep difficulties; and to identify factors associated with variations in clinical outcomes.
Design: A pragmatic cluster randomised controlled trial with two treatment arms (a CBT-treated 'sleep clinic' group, and a 'no additional treatment' control group), with post-treatment assessments starting at 3, 6 and 12 months.
Setting: Twenty-three general practices in Sheffield, UK.
Participants: In total, 209 patients (aged 31-92 years) with chronic sleep problems who had been receiving repeat hypnotic drug prescriptions for at least 1 month (mean = 13.4 years) were recruited into the trial.
Interventions: The intervention consisted of six 50-minute sessions as follows: introduction and sleep assessment, basic sleep hygiene, stimulus control and sleep restriction procedures, progressive relaxation, cognitive treatments, and review and discharge.
Main outcome measures: These included: global sleep quality [as measured by the Pittsburgh Sleep Quality Index (PSQI)], frequency of hypnotic drug use, mean dose of hypnotics consumed, health-related quality of life [as measured by the Short-Form 36 (SF-36)], NHS service costs and overall cost utility.
Results: At 3- and 6-month follow-ups, patients treated with CBT showed improved global PSQI scores as well as improvements in the SF-36 dimensions of vitality at 3 months and physical functioning and mental health at 6 months. CBT-treated patients also reported reductions in the frequency of hypnotic drug use compared with the control group, with many CBT-treated patients reporting zero drug use at the follow-up assessments. Clinical improvements were maintained within the CBT group at the 12-month follow-up, with PSQI scores and the frequency of hypnotic drug use continuing to show significant reductions relative to the control group. Multiple regression analyses of PSQI scores within the sleep clinic group alone indicated that the magnitude of pre- to post-treatment change in overall sleep quality was closely related to Hospital Anxiety and Depression Scale depression scores at 3-, 6-and 12-month follow-ups. In each model higher depression scores at baseline were associated with poorer treatment outcomes. No significant relationship was found between the patient's age and PSQI outcomes in any of these analyses. Within the sleep clinic group, reductions in drug use showed no significant association with the hypnotic product consumed. At the 3-month follow-up low-frequency drug use was reported by 22.9% (8/35) of temazepam users, 33.3% (5/15) of nitrazepam users and 38.9% (7/18) of zopiclone users. The total cost of service provision was GBP154.40 per patient (1999/2000 prices). The mean incremental cost per quality-adjusted life-year (QALY) at 6 months was GBP3418; this figure was insensitive to changes in costs. A simple model also showed that extending the evaluation period beyond 6 months may improve the cost-effectiveness of CBT. The incorporation of hidden costs associated with hypnotic drug treatment (e.g. accidents) also reduces the cost per QALY ratio, although to a much lesser degree.
Conclusions: In routine general practice settings, psychological treatment for insomnia can improve sleep quality, reduce hypnotic drug use, and improve health-related quality of life at a favourable cost among long-term hypnotic users with chronic sleep difficulties. These positive outcomes appear robust over time, persisting for at least 1 year among the more treatment-adherent patients. While these benefits may be reduced among those patients presenting with higher levels of psychological distress, the present study clearly indicates that older age per se presents no barrier to successful treatment outcomes. Further research should assess the long-term clinical and cost-effectiveness of psychological treatments for insomnia among non-hypnotic-using patients, and establish the minimum psychological treatment input required.