Interim results of a quality improvement study evaluating subjective and objective sleep quality in hospitalized adults with acute leukemia

Abstract

Abstract Background: Impaired sleep in hospitalized patients is associated with a myriad of adverse health outcomes. However, how sleep disruption impacts cancer patients is poorly characterized. Patients with acute leukemia (AL) may be at particularly high risk for disrupted sleep due to frequent vital sign (VS) checks, administration of chemotherapy at night, transfusion of blood products with associated monitoring, and therapy-related toxicities. In this quality improvement (QI) pilot study, we used an established patient-reported outcome measure (PROM) and Fitbits to evaluate inpatient sleep quality in patients with acute leukemia, with the goal of establishing a baseline upon which to design future interventional studies focused on improving sleep in these patients. Methods: In this observational study, we included adults with acute leukemia or high-grade myeloid neoplasms (≥10% blasts in blood/marrow) if they were planned for or had received recent treatment with high-dose cytarabine-based chemotherapy, targeting 60 patients across three cohorts: A) active chemotherapy (n=20), B) febrile neutropenia (FN; n=20) and C) low-risk hospitalized (n=20). Cohort C includes patients who are medically stable and hospitalized only for monitoring during their pancytopenic period. Exclusion criteria include neurologic or mobility impairments, and for cohort C medical needs requiring frequent VS (<Q8H), supplemental O2, telemetry, or non-standard transfusion thresholds. Patients were on study during chemotherapy (cohort A) or 7 days/discharge (cohorts B and C). The primary outcome is to evaluate subjective sleep quality. Secondary outcomes include assessing feasibility of collecting surveys and use of a Fitbit device in hospitalized patients with AL to estimate sleep duration objectively. Other outcomes to be included in the final analysis include activity level, nighttime disruptions, quality of life, and association between sleep and glucose control, blood pressure, falls, severe infection, and ICU admission. Subjective sleep quality is assessed via the 6-item Richards-Campbell Sleep Questionnaire (RCSQ). Each question is scored from 0-100 using a visual analogue scale: a score of 0-25 correlates with very poor sleep, 26-50 with poor sleep, 51-75 with good sleep, and 76-100 with very good sleep. Objective sleep quality is assessed using a sleep-staging Fitbit (Inspire 3), which has been validated in a meta-analysis against polysomnography, which is the gold standard for sleep evaluation, but not feasible in the hospitalized AL population. Nighttime sleep is defined as the period from 6PM-8AM; daytime sleep from 8AM-6PM. Feasibility is defined as patient sleep surveys completed and Fitbit data successfully transmitted in ≥50% of patient days on study. Reported results are from a planned interim analysis at 20 patients enrolled. Results: Twenty patients (65% with AML, 30% with ALL, 5% with high-grade MDS) were enrolled from 5/2025-7/2025; 10 in Cohort A, 5 in Cohort B, and 5 in Cohort C. The median age was 52 years (range 25-77), and patients were on study for a median of 5 days (2-7). Sleep surveys were returned for 69% of study days. Fitbit data was available for 69% of nights. Based on Fitbit data, the median sleep duration was 6.9 hours per 24 hr period (IQR 5.0-8.5); 6.6 hrs (5.0-7.9) occurred at night. Despite this length of sleep at night, there was frequent daytime sleep (26.8% of days), lasting a median of 1.3 hrs (1.0-2.3). Subjective sleep quality was also rated poorly on the RCSQ total score (median 50, IQR 35-65). The lowest RCSQ subscores highlight light sleep (35, 22-66), poor sleep quality (50, 29-67), and increased awakenings (50, 34-67) as the primary drivers of poor sleep. Patients were able to fall asleep quickly (59, 39-79), fall back asleep if awakened (59, 37-71), and the environment was quiet (73, 50-88). There was moderate correlation between total objective sleep time at night and total RCSQ score (R=0.32, p=0.01) and subjective sleep quality (R=0.36, p=0.005). Conclusions: Our results suggest that detailed assessment of sleep quality is feasible in a larger study. There was also poor subjective sleep quality in hospitalized patients with AL, notably light sleep, increased awakenings, and poor sleep quality. Final results will provide baseline subjective and objective sleep quality to inform future interventional QI studies aimed at improving inpatient sleep quality and associated outcomes.