Brief Summary

Context. Sleep disorders complaints are common in elderly and hypnosedative drugs are widely prescribed in community-dwelling elderly. Furthermore, acute hospitalisation may induce sleep disorders and hypnosedative initiation occurs in 14 to 29% of elderly during a hospital stay. These hospital-induced sleep disorders should spontaneously disappear after discharge and, because of their potential impact on falls, hip fractures, psychiatric side effects and induced dependence, hypnosedative drugs should therefore be discontinued at discharge in these naïve-treated patients. Adhesion to this recommendation is poor. Recent and on-going research on this topic mostly concerns adverse effects although these are already substantially documented and evidence-based, while there is a poor interest on developing research on potential strategies susceptible to practically improve the current adhesion to recommendations.

Design and objectives. This project proposes multi-component intervention and is directed towards hospital prescribers, patients, and their general practitioner. It aims at discontinuing, at hospital discharge, the hypnosedative treatment that was initiated during hospitalization in naïve-treated elderly (age≥65) patients. The value of the intervention, as compared to usual care management, will be estimated in a multicentre (6 hospitals gathering 8 centres: 5 internal medicine departments, 3 cardiology departments) randomised, cross-over, two-period trial. Two hospitals will gather 2 centres (2x2 centres) and 4 hospitals will gather only 1 centre (4x1 centres). An equilibrated randomization will be applied to the 8 centres, making sure that, in hospitals gathering 2 centres, these 2 centres will apply the same strategy in a given period and that the 2 hospitals will apply alternative strategies. This randomisation is set up to avoid the risk of a potential contamination between the 2 strategies from one centre to the other within the same hospital. As a result, 3 hospitals (including 4 medical departments) will apply the intervention during the first 11-month period, while the 3 others (including 4 medical departments) will apply usual care during this period, as a result of an equilibrated hospitals randomization. During the second 11-month period, each hospital will apply the alternative management. The two periods will be separated by a 1-month wash-out period.

In all 240 patients (15 patients / centre) will be enrolled (120 in the intervention group and 120 in the usual care group).

Patient follow-up duration will be 12 months after discharge. Patient status in regards with quality of sleep (study primary objective), hypnosedative consumption and frequency of falls (secondary objectives) will be collected by telephone interviews 1, 3, 6 12 months after discharge.

Expected results. The results of the study should contribute to guide research and public decisions to practically decrease hypnosedative prescription and consumption, and associated adverse events.

Context

Hypnosedative prescriptions in hospital can be either the continuation of the drug already prescribed before admission or result from an initiation during hospitalization. In the hospitalised population, 36.7% to 62.7% of the patients indeed complain of significant insomnia. Furthermore, it has been clearly established that a number of factors associated with hospitalization such as the effects of illness and new medications, environmental factors including noise and disruption of usual sleeping schedules induce insomnia and a decrease in the quality of sleep.

For these reasons, up to 33% to 60% of hospitalised patients are treated with hypnosedatives.

Hospital initiation of hypnosedatives was reported in 14% to 29% of the patients who were not taking a hypnosedative before hospitalization. Unfortunately, only a very few studies have been made on the discontinuation of such drugs at hospital discharge, still they showed that the drug was maintained in 20 to 36% of the initiated patients.

In the elderly, hypnosedatives provide only a small improvement in the quality of sleep while falls and cognitive disorders are significantly increased.

Due to dependence to benzodiazepines, a withdrawal syndrome may occur at discontinuation after a long term treatment.

Therefore, the benefit/risk balance of hypnosedative is not in favour of their use in elderly. Moreover, sleep disturbances appearing during an acute hospital stay are most of the time transient and the continuation should be systematically debated. However, to our best knowledge, no data has yet reported the evaluation of an intervention aiming at discontinuing the prescription at discharge from hospital in formerly hypnosedative-naïve patients.

Design

The HYPAGE study is a multicentre, cluster-randomized, two-period cross-over study that will test an intervention (compared to usual care) for discontinuing hypnosedative prescription at the time of hospital discharge in patients aged 65 years and over.

Six hospitals gathering 8 centres (5 internal medicine departments, 3 cardiology departments) located in the Paris area will participate in the study. Two hospitals will gather 2 centres (2x2 centres) and 4 hospitals will gather only 1 centre (4x1 centres). An equilibrated randomization will be applied to the 8 centres, making sure that, in hospitals gathering 2 centres, these 2 centres will apply the same strategy in a given period and that the 2 hospitals will apply alternative strategies. This randomisation is set up to avoid the risk of a potential contamination between the 2 strategies from one centre to the other within the same hospital. As a result, 3 hospitals (including 4 medical departments) will apply the intervention during the first 11-month period, while the 3 others (including 4 medical departments) will apply usual care during this period, as a result of an equilibrated hospitals randomization. During the second 11-month period, each hospital will apply the alternative management. The two periods will be separated by a 1-month wash-out period.

Description of the intervention.

The intervention under study includes two components involving a local hospital pharmacist, the first component towards physicians, and the second towards patients.

The intervention towards physicians is composed of the following items:

1. A written and oral information will be delivered to the hospital physicians. Such information includes:

   • Place of hypnosedatives among drugs at risk in elderly
   • Adverse drug events induced by hypnosedatives: prevalence of falls in elderly, psychiatric adverse events.
   • Procedures of hypnosedative withdrawal (French recommendations, HAS 2007)
   • The oral presentation will be presented during a staff meeting to the physicians in every centre entering the study at the beginning of the intervention period. The oral presentation will be delivered by local pharmacist in all centres entering the trial, according to the standardized slide show (i.e. slides + accompanying talk) prepared by the scientific committee of the study.

2. In patient in whom a hypnosedative has been initiated, medication reconciliation at discharge will involve the local hospital pharmacist. Medication reconciliation refers to the process of reviewing the patient's complete medication regimen at the time of admission, transfer, and discharge and comparing it with the regimen being considered for the new setting of care. The pharmacist will compare the medication list during the hospital stay with the discharge prescription. If the hypnosedative is maintained at discharge, after discussion with the physician on the opportunity of the hypnosedative continuation, two alternative decisions may be chosen:

   • If the physician decides to maintain the hypnosedative drug, the reason will be registered
   • If the physician decides to stop the hypnosedative, a standardized text will be included in the usual letter transmitted to the patient's general practitioner after hospital discharge. It will indicate that the hypnosedative initiated in patient during hospitalization was stopped at discharge (or progressively decreased with the aim of discontinuation), and will briefly mention the adverse effects related to continuation of hypnosedative use, advocating that any reintroduction should be based on sound arguments. Therefore, the intervention includes a component oriented towards the continuity of care after discharge, this component being directed towards general practitioners.

Intervention towards patients

Intervention towards patients consists in the delivery of a written information. The written information and the accompanying talk of the pharmacist during this delivery both heighten awareness about the hypnosedative-related risks of adverse drug events: relation between hypnosedatives and falls, uselessness of hypnosedatives at home after hospital discharge.

Patient follow-up

Patients will be included the day of discharge and will be followed-up up to 12 months. A study pharmacist will be enrolled and will be in charge for collecting follow-up data of all patients, whatever their admission centre. The study pharmacist will call (phone calls) the patient or the surrogate, the general practitioner and the community pharmacist 1, 3, 6, and 12 months after hospital discharge. He will assess the current treatment and will collect data concerning:

• The quality of sleep (Insomnia Severity Index score)
• The current hypnosedative treatment

Frequency of falls will be assessed by a fall diary. This fall diary will be provided to the patient at discharge and will be filled by the patient. The diary will be sent to the study pharmacist on a monthly basis (envelopes and stamps will be provided at hospital discharge).

Objectives

The primary objective of the study is to assess if the intervention does not produce a deterioration of the quality of sleep quality (non-inferiority design) compared to usual care.

The primary criterion will be the quality of sleep in the patients 1, 3, 6, and 12 months after hospital discharge. The Insomnia Severity Index score should not differ by more than

The secondary objectives are the following:

• Comparing the frequency of falls in the intervention and usual care groups.
• Measuring the frequency of hypnosedative reintroduction 1, 3, 6 and 12 months after discharge when hypnosedative prescription is stopped at hospital discharge.
• Comparing the Insomnia Severity Index score in treated and untreated patients with an hypnosedative drug at 1, 3, 6 and 12 months.
• Comparing the frequency of falls in treated versus untreated patients with an hypnosedative drug at 1, 3, 6 and 12 months.

Sample size

There is no formula for non inferiority design in cluster cross-over trials, therefore we chose to calculate the sample size for a traditional parallel group design and have applied the design effect linked with the clustering in centres and over periods. The calculation is based on the following hypotheses: a mean ISI score in patients treated with hypnosedative medications of 8 and a standard deviation of 5, a non inferiority limit of 3, a power of 90% and a one-sided type I error of 0.025. The sample size for a non inferiority trial without accounting for intra-cluster and interperiod correlations, is 63 per arms, i.e. 126 patients in total. With a hypothesized intracluster correlation of 0.03 and an interperiod correlation of 0.01 (NB: in cross-over clustered trials, it is widely accepted that the intracluster correlation should be at least twice the interperiod correlation, and these values should be low), the design effect is 1.29 [36]. The resulting calculated sample size is 162, rounded to 180 in order to take into account the patients lost to follow up. Based on the participation of 6 centres (original number of centres), this results in 15 patients per period and per centre. We kept this number of 15 patients per period per centre after the addition of 2 centres (8 including centres) in order to achieve at least the same power. This leads to 240 patients in all. With regards to the average length of stay and the number of beds, we estimate that such a number of patients can be enrolled during one semester in each center (worst case calculation).

Statistics

Results on quantitative variables will be provided as means and standard deviations or median and interquartile, results on qualitative variables will be provided as numbers and percentages. A type I error is set at 0.05. To account for clustering and the cross-over matching, the comparison between groups will be made using Mantel-Haenszel and Wilcoxon signed rank tests. Confidence intervals will be calculated using resampling bootstrap methods.

Analysis set. The primary analysis will be conducted according to the intent-to-treat principle (ITT), at the individual level, considering every patient included irrespective of the patient's follow-up and will take into account the clustering nature of data. Due to the non-inferiority design, secondary sensitivity analyses in the per-protocol population will also be conducted.

Missing values in the ITT analysis: we will check whether the proportion of missing values does not differ from one arm to the other. For patients with a missing endpoint we will assume there was no modification of the quality of sleep after discharge (i.e. a variation of the ISI score of 0). More conservative sensitivity analyses will be conducted, with an imputed increase of the ISI score above 3 (non inferiority limit) in the intervention arm vs imputed increase of the ISI score of 0 in the control arm.

Methods for Main endpoint analysis. The analysis will be performed at the individual level. A comparison will be made of the mean ISI score for the two arms. We will use a linear mixed effect model to perform the comparison, in which the fixed effects are: the period, the arm, the cluster, and the random effects are: the order of strategies and individual residual error. The conclusion of non inferiority will be based by the confrontation of the lower limit of the confidence interval and the pre-specified non inferiority limit (3 points).

Methods for Secondary endpoint analyses. The analysis of secondary endpoints will be performed using the same methodological framework taking into account the clustering and the period effect. For binomial variables, alternating logistic regression models will be used.

Inclusion criteria

• Patients hospitalized in one of the investigation centres
• Individuals (male and female) aged 65 years and above
• Patients prescribed an hypnosedative drug during the hospitalization (after the first 48 hours)
• Patients from whom a non-opposition of participation in the research is obtained
• Patients with no history of hypnosedative at the admission (and during the first 48 hours of hospitalization)
• Individuals not suffering from cognitive impairment suggested by a score of less than 24 on the Mini-Mental State Examination
• Patients returning home at discharge
• Patients reachable by phone

Exclusion criteria

• Patients who gave their opposition to participate in the study
• Patient with physical disability
• Patients suffering from severe depression with suicide risk or other severe psychopathologic conditions
• Individuals suffering from cognitive impairment suggested by a score of less than 24 on the Mini-Mental State Examination
• Patients suffering from metastatic/terminal cancer or undergoing parenteral treatment (cytotoxic or targeted therapy) for cancer or in palliative care program
• Patients with severe alcohol or drug dependence
• Patients unable to answer a questionnaire in French
• Patients unable to undergo the intervention
• Patients not available by phone at discharge
• Patients not returning home at discharge