Discontinuation of Long-term Medications in Older People Entering Nursing Home Care

STUDY TITLE Medication Rationalization for Older People awaiting long-term nursing home care: a randomized controlled trial using the STOPPfrail criteria.

AIMS Our primary aim is to test the feasibility of applying explicit deprescribing criteria (STOPPfrail) to the medication regimens of a cohort of older, frail hospitalized patients for whom long-term care is considered necessary and compare this intervention with usual care.

Secondary aims are to estimate the effect of medication rationalization on mortality, acute hospital transfers, unscheduled medical reviews, falls and non vertebral fractures, prescription of neuroleptic antipsychotic medications and cost of medications.

BACKGROUND The majority of people transferring to nursing home care are frail and highly dependent. The average survival following admission to a nursing home in Ireland is 1.9 years.1 Despite frailty and shortened life expectancy, these patients are among the highest consumers of prescription medications. The SHELTER study reports the rate of polypharmacy (5-9 drugs) and excessive polypharmacy (≥10 drugs) in nursing home residents to be 48.7 and 24.3%, respectively.2 This is important because the number of drugs that a patient is taking is the most important predictor of adverse drug reactions (ADRs).3 ADRs are associated with ill health,4 hospitalization,5 and higher mortality.6

Transfer to a nursing home is major life event for older people and medication review in hospital prior to transfer is appropriate. Medication rationalization or deprescribing is the process of tapering and stopping medications. The goal is to prevent adverse drug reactions and to appropriately match medications to the goals of care and prognosis of the patient. Deprescribing trials have been promising to date. Garfinkel8 discontinued 332 different drugs in 119 nursing home residents (2.8 per person) and reported successful cessation of all target drugs in 82% of patients. The one-year mortality rate was 45% in the control group and 21% in the intervention group. Overall, deprescribing of medications resulted in a 53% reduction in mortality(p<0.001). Potter et al9 successfully deprescribed 207 medications in 47 residents (4.4 per patient) and while there was a trend towards improved survival in the intervention group, this did not reach statistical significance. Both trials use a similar deprescribing algorithm that requires the user to answer a series of questions about each medication before a deprescribing decision can be made. Expertise in geriatric pharmacotherapy as well as clinical judgement is required, and these factors, along with time constraints, may limit the use of these algorithms in everyday clinical practice.

STOPPfrail is a validated explicit list of 27 drugs and drug classes that are potentially inappropriate in older adults with high morbidity burden and poor 1 year survival prognosis.10 The criteria were developed by Delphi consensus survey of an expert panel comprising academic geriatricians, clinical pharmacologists, palliative care physicians, old age psychiatrists, general practitioners and clinical pharmacists. The criteria are explicit and concise and therefore easy-to-use and time efficient. Observational work by our group, led by Dr. Amanda Lavan, has shown that 90% of people awaiting transfer to a nursing home bed are prescribed at least one potentially inappropriate medication (PIM) according to the STOPPfrail criteria and the median number of PIMs per person is three. Of those people deemed suitable for long-term care in a nursing home, 64.3% were eligible for the STOPPfrail intervention.

In theory, the deprescribing of unnecessary, ineffective, potentially inappropriate medications should reduce medication burden as well as exposure to adverse drug events. Recent deprescribing trials have suggested a trend towards improved survival in patients who have their medications deprescribed. Deprescribing using the STOPPfrail criteria has not yet been evaluated in a randomized controlled trial.

Recruitment Hospitalized patients applying for the Nursing Home Support Scheme undergo a multidisciplinary team assessment and the report of this assessment is compiled in the Common Summary Assessment Report (CSAR) form. At the Local Placement Forum (LPF), which takes place fortnightly in Cork University Hospital (CUH), city-wide applications are reviewed and suitability for long-term care (LTC) is determined.

Potential participants will be identified at the Local Placement Forum (LPF) in CUH. Inpatients in CUH, the Mercy University Hospital and St. Finbarr's Hospital (Cork City Hospitals) who are transitioning to long-term care will be considered potentially eligible for recruitment. The CSAR form, which includes information about medical history as well as cognitive and functional status will be used to screen participants for eligibility.

Prior to approaching potential participants, a senior physician (consultant, registrar overseeing the patient's care in hospital or the patient's own GP), will be contacted and requested to answer the 'surprise question'(SQ). If the physician indicates he or she "would not be surprised if the participant died in the next year" (SQ+), permission will be requested to review the patient's chart.

Establishing informed consent (figure 2) The unblinded researcher will approach potential participants directly, explain the study aims and provide a participant information form (PIF). Any patient who declines to participate will not be enrolled in the study. Cognition will be assessed using the Mini-Mental Status Examination (MMSE) in individuals who agree to participate. Participants with a MMSE ≥24 will be considered competent to formally consent to inclusion in the study. If these participants agree, the unblinded researcher will also inform the nominated next-of-kin (NOK) of the participant's desire to participate in the trial and provide the NOK with a study information sheet. If a participant wishes to participate in the study, but is not competent to formally consent (MMSE <24), the unblinded researcher will request written agreement from the participant's nominated NOK.

Baseline data A detailed medical history of each participant will be documented. The unblinded researcher will record the generic name, indication, dosage, and frequency of all medications taken by the participant, including all prescribed and over-the-counter medications and any herbal or mineral supplements. The unblinded researcher will determine the indication for each medication by discussion with the participant, the participant's doctor and by reviewing the participant's medical records. The unblinded researcher will determine medication compliance as well as challenges associated with administrating individual medications by communicating with nursing staff and reviewing the medication kardex. Cognitive status (MMSE) and functional status (Barthel index) as well as weight and blood pressure will be recorded. Quality of life will be measured using the ICECAP-O (proxy version) and, for those with a diagnosis of dementia, the QUALIDEM.

Randomization:

Each participant will be randomized to the intervention arm (STOPPfrail tool) or to the control arm receiving routine care on a 1:1 ratio using block randomization.

Intervention group:

The unblinded researcher will advise intervention group participants' doctors that their patient has been randomised to the intervention group. The unblinded researcher (an experienced Specialist Registrar in geriatric medicine) will offer medication advice based on the STOPPfrail criteria. Where there is a risk of an adverse drug withdrawal event, it will be recommended that that particular medication be withdrawn slowly according to an evidence based protocol .Prior to recommending the dose reduction or cessation of a medication, the unblinded researcher will inform the participant, NOK and nursing staff of potential adverse withdrawal effects associated with that drug. The nurse manager and medical team will have the mobile phone number of the primary researcher and will be encouraged to make contact if there are any concerns about adverse withdrawal effects in the participant.

If the participant is discharged to a nursing home during the period of drug withdrawal, a written drug withdrawal plan will be sent to the nursing home director of nursing as well as to the medical officer overseeing the care of patients in that nursing home. The mobile phone number of the unblinded researcher will be included in this written plan and the unblinded researcher will be available to respond to queries or concerns.

Follow-up data In order to minimize the risk of ascertainment bias, a different researcher, who is blinded to the group allocation of participants will provide follow-up and obtain outcome measurements. Follow-up will take place at the time of discharge from hospital and at 3- months post-enrolment. All medications, falls and non-vertebral fractures, emergency transfers to hospital, unscheduled medical reviews and deaths will be recorded. Quality of life will be reassessed, and weight and sitting blood pressure will be measured.

Serious Adverse Events (SAEs) SAEs are defined as death, emergency hospital admission, fall, non-vertebral fracture, non-fatal vascular event (MI or CVA), and an unscheduled medical review. The nurse managers overseeing the care of the participants will be asked to advise the unblinded researcher as soon as practical if any of these events occur in study participants. All SAEs are to be reported to the principle investigator (PI) within 24 hours upon becoming aware of the event.

The principal investigator will evaluate whether the SAE is related to the deprescribing intervention. It is the PIs responsibility to report SAEs that are related or possibly related to the intervention within 7 days to the local ethics committee.

Sample Size Calculation

The investigators plan to enrol 160 participants (80 participants in each group). This allows for an anticipated attrition rate (deaths and dropouts) of 30%. This will allow a power to detect a difference of 2 in the mean number of medications between the groups (α = .05 and 1-β =0.8, population variance 14). Of note, Garfinkel and Potter achieved a mean reduction of 2.8 and 4.4 medications per subject respectively.