Deprescribing research in nursing home residents using routinely collected healthcare data: a conceptual framework

Demographics include age, sex, race, and ethnicity, which influence treatment selection and intensity by way of treatment guidelines and perceived or actual risk for disease or disease-related complications for preventive treatments. Demographics may also influence access to care and prescriber biases in prescribing of medications - i.e., pharmacoequity [39]. Socioeconomic status is another important determinant that may exhibit an influence on deprescribing by way of access to healthcare services and prescribers to optimize medications, or by introducing cost incentives to simplify medication regimens.

Index condition attributes encompass the symptoms or potential complications of the specific condition related to the medication being deprescribed. These may also include the current level of disease control as well as predisposition to medication-induced adverse events. Taken together, these characteristics often warrant individualized consideration of more intense or less intense treatment goals. For individuals with a greater number of related risk factors or those with poorly controlled disease, more intensive treatment goals and medication use may be warranted. However, those with otherwise adequate disease management and low risk for long-term complications may be more likely to be targeted for deprescribing. For example, for individuals with diabetes, deprescribing may be more strongly indicated in individuals with the lowest A1c values (e.g., < 6.0%), or those with overall lower risk of diabetes complications (e.g., no prior history of CVD or stroke) [40]. As another example, individuals with severe stage dementia may be perceived as least likely to benefit from continued treatment, and therefore may be more likely to be targeted for deprescribing. Prior treatment history may also affect the appropriateness of deprescribing. For example, bisphosphonates may be more appropriate for deprescribing in individuals who have already received a treatment course of more than 5 years, as the therapeutic benefits of these medications can persist for up to 2 years after discontinuation [41, 42]. Finally, it is important to consider predisposition to adverse events, which may also be influenced by their current level of disease control. For example, a prescriber may be more likely to deprescribe medications that may have higher risk for hypoglycemia or hypotension in individuals with a prior history of falls.

The specific attributes of the medication being re-evaluated may affect the feasibility or safety of deprescribing. This may include the benefit vs. harm evaluation, based on an individual’s prognosis and the time to benefit of the medication [43]. Medications in a treatment regimen that have the highest risk for medication-induced adverse effects may be more likely to be deprescribed than others. This risk is also affected by the specific pharmacokinetics and pharmacodynamics of a medication, which can change with aging. As an example, insulin and sulfonylureas, which carry a significantly higher risk for hypoglycemia may be more likely to be deprescribed than other diabetes medications [15]. Difficulty of administration may also influence the likelihood of deprescribing. For example, oral administration of bisphosphonates, which require the patient to remain upright for at least 30 min to avoid esophagitis, may prove difficult for individuals with dysphagia or severe dementia without appropriate oversight by nursing staff. Administration of injectable medications like insulin multiple times per day may also become challenging for individuals with behavioral issues or may be viewed as an unnecessary burden that opposes comfort care for patients approaching end of life.

Prognosis encompasses both general comorbidity burden as well as specific clinical signs and symptoms that may help to estimate an individual’s clinical trajectory or may impact their ability to take oral medications. Prognostic factors are highly influential in aligning medication use with life expectancy and goals of care. Preventive medications are more beneficial for individuals who have a projected life expectancy exceeding the time to benefit, which varies by medication class. For example, statins being used for primary prevention have a time to benefit of 2–5 years and bisphosphonates being used for fracture prevention have a time to benefit of 8–19 months [43, 44]. Thus, an individual with a projected life expectancy of less than 6 months likely would not experience benefit from continued treatment with either medication. A number of diagnoses and specific clinical characteristics can be used to infer or estimate prognosis. For example, the MDS Mortality Risk Index [35, 36] is a composite score based on clinical characteristics and diagnoses that has been shown to be highly predictive of 6-month mortality in NH residents. It includes clinical characteristics such as cancer diagnosis, renal failure, sudden weight loss, loss of appetite, shortness of breath, dehydration, activities of daily living, and changes in mental status. In addition to the presence of specific life-limiting conditions, measures of overall comorbidity burden, such as the Charlson Comorbidity Index [45] or the Elixhauser Index [46], which are based on the accumulation of chronic conditions, may also be useful for determining prognosis. Frailty is another important consideration, which modifies an individual’s vulnerability to adverse health outcomes and medication-related adverse events and may affect prescribers’ perceptions of the likelihood for continued benefit relative to harms. Frailty can be measured using several claims-based algorithms [47‐50] or by using proxy measures of physical function, like activities of daily living (ADLs). Individual characteristics may also be informative for inferring an individual’s goals of care and whether quantity or quality of life is more likely to be prioritized, thus identifying opportunities for deprescribing. For example, individuals with swallowing difficulty or severe dementia may have difficulty ingesting multiple medications per day and may experience improved quality of life. Hospice enrollment, an obvious marker of prognosis, can also be a clear indication for deprescribing and a proxy for likelihood of imminent death.

It is incredibly important to consider the entire medication regimen as part of the deprescribing process. Polypharmacy is one of the driving forces behind deprescribing, with the goal of reducing medication regimen complexity and medication burden. Medications used concurrently with the index medication have the potential to modify the risks and benefits associated with the index medication by way of drug-drug interactions, prescribing cascades, and synergistic therapeutic effects. For example, use of proton pump inhibitors may affect the risk of gastrointestinal bleeding for patients taking aspirin for primary or secondary prevention and may modify the assessment of risks associated with continued use [51]. Similarly, continuing diabetes medications may be more likely among individuals who are also prescribed psychotropic medications with metabolic adverse effects that may affect weight gain and A1c [52, 53].

The influence of family caregivers is likely a function of their level of engagement in care and decision-making. The relationship between residents and caregivers likely serves a key role [54]. Previous research among non-institutionalized older adults has shown that family caregivers’ own use of potentially inappropriate medications predicts potentially inappropriate medication use in care recipients with dementia [55, 56]. Thus, their own preferences and medication experiences may – to varying degrees – continue to influence decision-making about medications and deprescribing after NH admission. For example, spousal caregivers may be more involved in decision-making and have stronger preferences for continuing versus deprescribing medications, whereas siblings or adult children may be less involved. Engagement may also vary by geographic proximity of the caregiver.

Healthcare providers may differ in their predisposition towards recommending deprescribing based on their role or specialty training [57]. Some nursing facilities may have advanced practice providers (i.e., Physician Assistants and Nurse Practitioners) who see residents more regularly than a physician who spends a smaller portion of their clinical time on site. These advanced care providers may be more inclined or comfortable with deprescribing medications, given their familiarity with residents’ care needs. Specialty training in geriatrics or palliative care may also influence whether certain prescribers are inclined to deprescribe medications [25, 26].

Organizational structure, including the type of facility (e.g., continuing care retirement community, skilled nursing facility, VA community living center), ownership (i.e., federal, for-profit, non-profit), and academic affiliation may exert an influence on the culture around deprescribing and to what degree it may be prioritized by prescribers. For example, facilities with an academic affiliation are likely to have prescribers with specialized training in geriatrics and/or palliative care who recognize the importance of deprescribing. Other facility factors such as bed size and staffing hours or turnover may affect the feasibility of addressing deprescribing as part of routine assessment. Larger facilities with a greater resident to staff ratio and few advanced practice providers to assist with oversight may less bandwidth to address deprescribing. Finally, availability of resources and services such as hospice care or specialized dementia care units with trained staff may serve as markers for higher quality care and also facilitate deprescribing.

It is also important to consider the influence of other healthcare team members that may contribute to fragmented care delivery. In the NH setting, one provider typically oversees care management, but often this individual may not be the patient’s usual primary care provider. Thus, the decision to change medication regimens may be at the discretion of recommendations from other healthcare providers, and these decisions may ultimately offset each other in certain circumstances. Additionally, a patient’s clinical trajectory and the sequence of events or care transitions leading up to NH residence (i.e., emergency visits, hospitalizations) may introduce opportunities for further care fragmentation that may lead to both for appropriate and inappropriate medication changes.