Prevention of Injury in Skilled Nursing Facilities Through Optimizing Medications

After hospitalization for fracture, approximately 60% of older adults receive post-acute care in Skilled Nursing Facilities (SNFs), but fewer than 20% of these individuals receive medication optimization, which is evidence-based care known to prevent additional injurious falls, fractures, and mortality. Therefore, there is an opportunity to substantially improve care and patient outcomes in this setting. The purpose of this study is to determine the comparative efficacy of a Deprescribing Care Model designed to deprescribe fall related medications (FRIDs), a Bone Heath Service Model designed to provide osteoporosis evaluation and management, or an Injury Prevention Service Model offering both services. All care models will be delivered remotely to patients in the SNF and after they transition home by a post-fracture nurse consultant supported by an interprofessional team. The primary outcome is injurious falls, with secondary outcomes measuring how the process of care is changed and capturing patient-reported outcomes valued by stakeholders.

Our Specific aims are to:

1. Conduct a randomized cluster crossover trial in fracture patients comparing the effectiveness of the Deprescribing Care Model, Bone Health Service Model, or Injury Prevention Service Model in 42 SNFs, with each SNF receiving 6 months of each intervention in random sequence. We will test for clinically important differences in patient-centered outcomes (living with medicines survey, PHQ-8 depression scale, anxiety, sleep, and pain scales) at 90 days and implementation measures osteoporosis prescriptions, drug burden) at the time of discharge. 2. Compare the effectiveness of the Deprescribing Care Model, Bone Health Service Model, or Injury Prevention Service Model on subsequent injurious fall and fractures for patients treated within the 42 randomized SNFs with those in 168 matched concurrent control SNFs. We will use data from the National Institute on Aging-funded Long-Term Care Data Cooperative which includes Minimum Data Set (MDS), Electronic Health Record (EHR), and Medicare (CMS) claims. To address decisional dilemmas, we will examine differences in efficacy by baseline estimated life expectancy and cognitive status. 3. Identify SNF characteristics associated with measures of effective model implementation. We hypothesize that organizational factors captured by SNF staff responses to surveys capturing organizational readiness for change will explain additional variance in implementation measures beyond factors known to be associated with care quality (staff turnover rates, for profit-status, region, case mix, and proportion of patients from racial and ethnic minority groups).

Study Design: In the Comparative Effectiveness Aim (Aim 1), we will conduct a pragmatic, cluster randomized crossover trial in 42 SNFs comparing the effectiveness of a Bone Health Service Model, a Deprescribing Care Model, or an Injury Prevention Service Model, all delivered remotely by a post-fracture nurse consultant supported by an interprofessional team. Every SNF will receive 6 months of each care model in random sequence. Follow-up for delivery of the care models will begin on the date that the first eligible patient from the facility is included in the study. All patients with osteoporotic (OP) fracture admitted to the SNF within the intervention time period will receive the full designated care model, even if their stay in the SNF extends into the next intervention period. One primary post-fracture nurse consultant will be assigned to each facility for the life of the study. To prevent contamination between models, the nurse consultant will double check the care model assigned before every patient/proxy encounter. Medication utilization (measured using EHR data) will be assessed at discharge and 90 days after SNF admission date, and Patient Reported Outcomes (PROs) (measured via survey) will be assessed at 90 days after SNF admission date, regardless of whether they have discharged home or remain in the SNF. We will therefore be able to compare the effectiveness of the models on important PROs, osteoporosis treatment rates, and FRID medication burden. Patients will be followed for injurious falls and fractures using administrative data and self-report during post-fracture nurse consultant calls until the end of the study period (up to 3 years, mean 2 years) censoring at death.

In the Overall Effectiveness Aim, we will compare rates of injurious falls and fractures for patients treated with the Deprescribing Care Model, Bone Health Service Model, or Injury Prevention Service Model to patients treated in 168 matched concurrent control SNFs, identified through the Long Term Care Data Cooperative. No new data will be collected for this aim.

In the Implementation Evaluation Aim, we will identify facility characteristics associated with high levels of intervention implementation. This aim will require the collection of surveys from facility leadership.

• Number of patients who initiate treatment for osteoporosis [Time frame: 1 day of discharge from SNF and one year]
• Medication Possession Ratio during follow-up, as measured by EHR and Medicare claims [Time frame: Up to 3 years]
• Number of patients who had deprescribing ordered [Time frame: 1 day of discharge from SNF and 90 days after SNF admission date]
• Modified Drug Burden Index during follow-up, as measured by EHR and Medicare claims [Time frame: Up to 3 years]
• Patient medication side effect burden, measured via Living with Medicines Questionnaire (LMQ3). [Time frame: 90 days after SNF admission date]
• Patient falls self-efficacy (i.e. fear of falling), measured via Falls Efficacy Scale [Time frame: 90 days after SNF admission date]
• Patient pain, measured via Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference Short Form [Time frame: 90 days after SNF admission date]
• Patient depression, measured via Patient Health Questionnaire survey (PHQ-8) [Time frame: 90 days after SNF admission date]
• Patient anxiety, measured via PROMIS anxiety short form survey [Time frame: 90 days after SNF admission date]
• Patient sleep, measured via PROMIS sleep disturbance short form [Time frame: 90 days after SNF admission date]