Despite clinicians’ best intentions to prescribe conservatively, as patients age and medical conditions accumulate, drugs frequently “pile up.” The problems listed in Table 53–1 can be ameliorated by decreasing the number and complexity of a patient’s medication regimen. At present, no study has demonstrated that trimming a patient’s medication list reduces morbidity or mortality. However, studies have demonstrated that pruning a patient’s medication list reduces the chance of adverse medication effects, lowers pharmacy cost, and improves patient compliance to their remaining medications. A Cochrane review of multiple systematic approaches to reduce inappropriate medications, for example, found that even with a modest reduction in the number of medications, ADRs could be reduced by 35%. For these reasons alone, unprescribing should become as routine a practice as prescribing.
The first step in assessing for inappropriate medications is to determine which medications the patient is taking (including dose and frequency). The traditional way of accomplishing this task is the “brown bag medication review,” in which the patient and/or family members collect and bring in all medications, including pills and creams, vitamins and supplements, herbal, and nonprescription medications. As each medication is removed from the brown bag, the physician (or office staff) can assess: (a) what the patient is actually taking; (b) what he/she understands about each medication; (c) the medication’s effectiveness; and (d) any suspected side effects. Finally, the provider updates the office medication list with the patient’s medications. This process may be time-consuming, but much of it can be completed by nonphysician office staff.
Approaches to unprescribing draw on evidence, expert opinion, physician judgment, and patient/caregiver preference. A number of approaches have been described to assist in assessing the appropriateness of a medication, including STOPP (Screening Tool of Older Persons’ Potentially Inappropriate Prescriptions), MAI (Medication Appropriateness Index), and ARMOR (Assess, Review, Minimize, Optimize, Reassess). Figure 53–1 illustrates an approach to determine if a medication is a candidate for unprescribing. This approach is built on two published rubrics: the Good Palliative-Geriatric Practice algorithm and Holmes’ Model for Appropriate Prescribing for Patients Late in Life.
Filtering algorithm for unprescribing medications.
After conducting a “brown bag” review, step 2 requires the clinician to reconsider the evidence (or lack thereof) supporting each medication. Older patients are often excluded or underrepresented in trials and geriatric clinical guidelines are often extrapolated from studies of younger patients. Therefore, it is important to remember that what’s good for younger patients may not necessarily be good for older patients. Step 2 prompts clinicians to consider unprescribing when a patient is not “comparable” to the population originally examined in a research study. For example, a number of studies support the use of antihypertensive medications in older patients. Similarly, studies support the use of statins in some older populations. However, studies of digoxin for heart failure or atrial fibrillation are limited in older patients. Likewise, analyses of oral hypoglycemic agents or studies supporting specific target A1cs in older patients are scarce. The drugs that lack evidence may be candidates for unprescribing.
Step 3 requires the clinician to personalize the medication that “passed” Step 2. Holmes’ “Model for Appropriate Prescribing for Patients Late in Life” is useful. This model suggests that prescribing clinicians should apply 4 considerations to “filter” the medications that passed step 2. These considerations include: (a) the patients’ goals of care; (b) individualized treatment targets; (c) life expectancy; and (d) the drug’s “time until benefit.” Applying this model will result in a personalized set of medications.
To apply the Holmes model, the patient’s goals of care may range from curative to solely palliative; treatment targets may range from preventive to symptom management only. A drug’s “time until benefit” can vary from minutes (diuretics for congestive heart failure) to years (HMG CoA reductase inhibitors for primary prevention of coronary heart disease). For example, in an older adult with severe emphysema, a long-acting β-agonist inhaler is an evidenced-based, patient-appropriate medication because it improves symptoms in a short timeframe. Statins are also evidenced-based for this patient’s hyperlipidemia; however, the patient’s limited life expectancy would filter out this preventive medicine.
In step 4, the physician identifies any symptoms of an ADR. It is important to note that symptoms of an ADR may be subtle, such as fatigue or weakness. A good axiom to remember is that all new symptoms are caused by a medication until proven otherwise. The physician should consider unprescribing a medication when a new symptom occurs shortly after that medication was started. This approach will also avoid treating a new symptom by prescribing yet another medication.
In step 5, the physician must decide whether the potential for an adverse effect is likely. The “Beers” listing of potentially inappropriate medications is one helpful resource. Special scrutiny should also be applied to high-risk medications, including hypoglycemic agents, anticoagulation/antiplatelet medications, digoxin, narcotics, antianxiety/sleep aids, antidepressants, and medications with anticholinergic properties. If a drug is the likely cause of an adverse effect, it should be stopped. When unprescribing, clinicians must withdraw drugs carefully. When a medication requires titration up, for example, it will also require titration down. Titrating down is especially important for medications that promote tolerance, such as opiates, sedative-hypnotics, β blockers, clonidine, gabapentin, and selective serotonin reuptake inhibitors. Avoid abrupt, nondiscriminant “drug holidays,” and counsel patients to watch for withdrawal symptoms. (Exceptions to titrating down can be made in special circumstances where immediate discontinuation is required.) Finally, step 6 attempts to simply reduce the dose. Reducing dosage and/or frequency may also improve adherence. Further techniques to improve adherence are discussed in the next section.
Interventions beyond physician decision making can also improve prescribing practices. For example, attempts should be made to limit the number of providers who are prescribing medications for a given patient. Studies show that with each additional prescriber there is a 29% increase in ADRs. Likewise, using multiple pharmacies increases this risk. Therefore, clinicians should try to keep a patient’s medications with 1 prescriber and 1 pharmacy. Similarly, electronic prescribing (e-prescribing) also holds the potential to decrease the number of inappropriate medications in all patients by instantly reporting drug–drug and drug–disease interactions. E-prescribing programs generate an alert that is sometimes followed by a suggestion of a more appropriate medication. One analysis examined several trials that tested e-prescribing in the ambulatory, hospital, and nursing home settings. The majority of trials showed some decrease in inappropriate prescriptions. However, this decrease was variable, ranging from a decrease of 24% to less than 1%. In addition, excessive (and often irrelevant) alerts were not helpful, leading to “alert fatigue” for physicians. Although e-prescribing holds much promise, the magnitude of the effect on polypharmacy remains uncertain.