Improving Sleep and the Patient Experience in the Emergency Department

Study Design This will be a randomized controlled trial.

Setting Kingston General Hospital is the Regional Trauma and Tertiary Care Center in Kingston, Ontario. The Emergency Department sees approximately 56,000 patients annually. It is staffed by Emergency Medicine Physicians, and residents in both the FRCP and CCFP-EM Emergency Medicine programs, as well as off-service residents.

Population This study will include a convenience sample of recruitment on weekdays when we have research assistance available. Specifically included are emergency department patients aged ≥18 years, GCS 15, who at 2300h are deemed likely by study personnel to board in the Emergency Department until at least 0700h the following morning while they wait for consultant assessments, investigations, or transfers.

Excluded patients include those previously enrolled in this study or involved in sleep research within last month, those who are legally blind or have a history of hearing impairment, a documented history of cognitive impairment, a positive delirium screen at enrollment, and those with a presenting complaint of head trauma. Also excluded are prisoners and those who cannot comprehend English.

Outcomes The primary outcome will be sleep quality as measured by a modified Richards-Campbell Sleep Questionnaire (RCSQ), completed by the patient at the termination of intervention.

Secondary outcomes will include individual sleep domains on the Richards Campbell sleep questionnaire (depth, latency, awakenings, percentage time awake, and quality of sleep), patient satisfaction, perceived hours of sleep, blood pressure, heart rate, new-onset delirium, changes in the patient's sense of being well-rested and changes in the patient's overall sense of well-being.

Randomization Group allocation will be randomized by night, cluster randomization, in an effort to decrease contamination between patients enrolled during the same evening. All eligible patients will be screened and consented (see Level 1 Information Sheet and Consent), with primary information (see Enrollment Questionnaire) completed prior to randomization. A sequential opaque envelope, with a tin-foil wrapped allocation card, will then be opened. During treatment nights, patients will then be consented to the intervention (see Level 2 Information Sheet and Consent). Sleep aids will then be distributed to consenting patients (see Verbal Instructions at Enrollment). Treatment patients will be reminded to not disclose allocation until the last question of the Completion Questionnaire the following morning. Night randomization will be randomly permuted to boxes of 6-10. A randomization log will be kept and the date will be recorded in the log and on the case report form.

Patients in both groups will be informed of the randomized nature of the study, and the nature of both study groups at the completion of data collection.

Study procedures and consent Patients will be identified as potentially eligible by a trained research staff at 2300h on each day of recruitment. This will be accomplished through examination of electronic medical records in the emergency department and discussion with treating clinicians. Eligibility will be confirmed with Eligibility Confirmation prior to consent.

This study will employ a two-level consent procedure. This is necessary to avoid a measurement bias in patients ultimately randomized to the control group. Many of our outcomes are subjective, and patients randomized to the control group (no sleep mask or earplugs) might be prone to answer sleep quality and patient satisfaction questions more negatively if they gain knowledge that mask and sleep plugs had been available but withheld via the consent process.

Data sources and data collection Outcome data will be collected with the assistance of the Research Assistant prior to and at the termination of intervention (including modified-RCSQ and satisfaction VAS) and retrospectively (vital signs and overnight, sedating or stimulating medications used from the nursing records assessed via electronic medical records).

Efforts will be made to blind the research staff undertaking the morning measurements to study group. The nighttime treating nurse will remove the sleep mask and ear plugs prior to leaving at 0700hrs. Data collection will begin at 0800hrs to allow for a 1hr washout in the Emergency Department.

The delirium assessment will be made by the study staff at enrollment and termination of intervention (approximately 0700h) via standardized screening protocol (Delirium Triage Screen and Brief Confusion Assessment Method from Han, 2013).

"Nursing interaction" will be quantified at the end of the study period in the morning. We define one "Nursing interaction" as a visit to the patient bedside recorded in the electronic medical record (EDIS). Typical interventions will include medication delivery, vital sign assessment, or responding to a call bell. For example, two medications offered at once will count for a single intervention.

Should a patient leave the Emergency Department prior to collecting outcome data, a research staff member will mail the paper-based questionnaire to the patient. This will be followed by telephoning the patient once a week for 3 weeks or until their information is returned.

Emergency Department description: as in similar studies, serial decibel measurements will be made in each section of our emergency department on five nights. Each section will be represented by a mean dB reading and peak value. Similar measurements for light intensity will be made. This will provide comparisons between sections for analysis.

Statistical Methods

Analysis plan Baseline characteristics will be summarized by descriptive statistics (e.g. mean/SD for continuous variables, count/percent for categorical). The primary analysis will be an intension-to-treat analysis. The primary analysis will be unadjusted for patient or environmental characteristics. Mean Richards-Campbell Sleep Score will be compared between treatment groups using a mixed effects model. Patient and environmental characteristics will be compared between groups with descriptive statistics. These characteristics include age, gender, pain (as recorded on a visual analogue scale), emergency department location (section A, B, C, D, E), the presence of a companion in the room, lighting conditions (lights on/off), private room versus multi-patient area versus hallway/non patient area placement, day of week and the use of sedatives, home sleep quality, and chronic use of medications and sleep aids at home. A sensitivity analysis employing linear regression will be used to explore the association between treatment group and Richards-Campbell Sleep Score while controlling for these potential confounders. Secondary analysis will include a per-protocol analysis including only patients who actually used the ear plugs and sleep mask for >3 hours.

Power calculation Power calculations have been completed using mean RCSQ scores from previous work done in ICU patients (no data were available from patients in the emergency department). The mean score in this patient population is 57 +/- 28. For a two-sided alpha=0.05 and a power of 80% to detect a 20% difference in mean RCSQ between groups we will need 96 patients in each group. Because of the cluster analysis we will increase this to 150 patients in each group to account for the possible loss in power.

Recruitment and timeline This study will be registered on clinicaltrials.gov prior to participant enrolment.

From an informal review of the KGH ED patient census we anticipate there will be approximately 7 people waiting for morning imaging or CCAC referrals who could be assessed for eligibility. Making a conservative estimate of 4 eligible and consenting patients who complete the study per recruiting day, 75 recruiting days are required. Given the convenience sample with recruitment on weekdays only, the recruitment phase of the study will require 15 weeks.