Managing Medications in Clinically Complex Elders: Title and subTitle Break“There's Got to Be a Happy Medium”

Mr L is an 84-year-old man with dementia and a medical history of atrial fibrillation, diabetes mellitus, hypertension, hyperlipidemia, chronic kidney disease, gastritis, and gastroesophageal reflux disease. His past surgeries include a transurethral bladder resection for bladder cancer with subsequent urinary incontinence and a lumbar decompression for spinal stenosis in 2008.

Mr L lives with his wife, Mrs L, who also cares for him. He is a retired writer and editor and a lifelong tennis player. On first presentation, his initial concerns were forgetfulness, difficulty walking, and falling. His wife reported that he was “doing almost nothing,” maintaining a sedentary lifestyle at home, and following her around. He needed considerable help with bathing and dressing, some assistance with toileting and transferring, and was dependent in most instrumental activities of daily living including shopping, housekeeping, and preparing meals. Mrs L hired a home health aide for several hours a day to help alleviate the substantial burden of caregiving.

At his initial visit, Mr L's blood pressure was approximately 135/60 mm Hg, his heart rate (beats/min) was in the 50s, and his estimated creatinine clearance was 42 mL/min/1.73 m². He scored 13 of 29 points on a Folstein Mini-Mental State Examination (MMSE) performed shortly before the visit, consistent with Dr S's clinical impression of moderately severe cognitive impairment. His oral medications were glyburide, 2.5 mg; digoxin, 0.125 mg; warfarin (varying dose); etodolac, 400 mg; docusate sodium, 100 mg; a multivitamin and iron, each taken daily; memantine, 10 mg; metoprolol, 25 mg; and gabapentin, 300 mg; twice daily; essential fatty acids, 3 times daily; and on an as-needed basis, acetaminophen, 650 mg every 6 hours; and lactulose, for a total of 13 medications at 16 scheduled doses per day.

Mr L's hemoglobin A_1c was 5.9% so Dr S discontinued glyburide. Dr S referred him to receive physical therapy and also to social services to discuss options regarding caregiving, social engagement, and long-term care plans. His warfarin dose was managed by nurse practitioners in a nearby hospital's anticoagulation clinic, and his international normalized ratios were maintained in the desired range between 2.0 and 3.0.

Mr L's drug regimen included ongoing use of etodolac and gabapentin after his 2008 lumbar laminectomy, despite no longer reporting pain. Dr S sequentially tapered off both medications, watching for increased reports of pain. Mr L reported no return of pain, his walking improved, and he had no further falls. Dr S also tapered the digoxin, starting by reducing the dose by half for 1 week. His heart rate remained between 50 and 70 beats/min so digoxin was discontinued. He began going to yoga with his wife and then to the gym twice weekly.

Mr L's initial laboratory results had shown a normal hemoglobin level of 13 g/dL, and Dr S discontinued iron; his hemoglobin level subsequently remained stable. Seven months after Mr L's first visit, Dr S asked Mrs L whether memantine was helping her husband's memory (he previously had not tolerated donepezil). She was unsure, and together they decided to try tapering Mr L's use of it. Thereafter, he had greater difficulty with nouns and names so Dr S referred him to speech therapy for cognitive exercises and resumed the memantine at its full dose. He initially showed some improvement, but within 6 months cognitive decline was again apparent.

Mrs L continues to pay for her husband's medication under his Medicare Part D plan. She reports that his activities of daily living have been stable. Socially, he is improved.

Mrs L and Dr S were interviewed by a Care of the Aging Patient editor in December 2009.

Mrs L: Just looking at [some of his medication] you realized that you could keep taking it, but you don't really have to. . . . It's better to pull it out.

Use of multiple medications is a common source of concern for patients and clinicians. Nearly 20% of community-dwelling adults aged 65 years and older take 10 or more medications, a figure that can easily be reached by following practice guidelines for a small number of coexisting conditions.^{1 - 2} Multiple medication use is associated with greater use of inappropriate medications and with nonadherence, and imposes substantial cost burdens on older patients even when they have prescription drug insurance.^{3 - 5} In addition, the frequency of adverse drug events increases in proportion to the number of medications used, including drug-specific phenomena as well as nonspecific syndromes including weight loss, falls, and decline in functional and cognitive status.^{6 - 10} Such adverse drug events affect an estimated 5% to 35% of older patients living in the community per year, and are responsible for approximately 10% of hospital admissions in older adults.^{11 - 15}

Despite legitimate concerns regarding multiple medication use, believing that Mr L is taking too many medicines does not help the clinician know which ones to stop. Moreover, such labels can distract from addressing underuse of potentially beneficial medications, which is as prevalent in older adults taking many drugs as in those taking relatively few.^{16 - 17} The task for the clinician is not to determine whether too many or too few medications are being taken, but to determine if the patient is taking the right medications—tailored to the patient's individual circumstances, including his or her constellation of comorbidities, goals of care, preferences, and ability to adhere to medications.

We conducted several systematic literature reviews. Our main review evaluated the effect of interventions to improve on suboptimal prescribing across the medication regimen (ie, without focus on a single drug class or disease) for elders in ambulatory settings who were taking multiple medications. Searching PubMed and International Pharmaceutical Abstracts from 1975 through March 2010, the search used a combination of the terms polypharmacy, multiple medications, polymedicine, suboptimal prescribing, medication misuse, inappropriate prescribing, elderly, geriatric, and aged, and was restricted to randomized clinical trials published in the English language, involving patients aged 65 years and older, and reporting both process measures assessing prescribing and clinical outcome measures. We also reviewed studies of the effects of discontinuing specific types of medications taken by Mr L. Details of the search strategies are available in an eAppendix).

Before optimizing Mr L's medication regimen, Dr S's first task was to assess what drugs Mr L thought he should be taking, what he actually was taking, and the benefits and harms he was experiencing from his drugs.

A good medication review is essential because discrepancies are common between patients' understanding of what they should be taking, what they actually are taking, and what physicians record on their medication lists.^{18 - 19} There is little direct evidence to support one specific method of medication review over another in ambulatory settings.²⁰ However, a “brown bag” review in which patients are asked to bring in all of their medicines (including all prescription and over-the-counter medicines, vitamins, supplements, and herbal preparations) can provide a useful snapshot of the patient's current medication use. The clinician can review each medication and inquire about how the patient takes it (eg, by asking “tell me how you take this medication”).

Quiz Ref IDBrown bag reviews often present an opportune time to review the effectiveness of medications (eg, control of pain, constipation, or depressed mood) as well as their adverse effects. Patients often do not report drug-related symptoms to their physicians, in part due to limited physician efforts to solicit this information.^{21 - 22} In one major study, such communication gaps were responsible for 37% of remediable adverse drug events.²³ The question “In the past XX months, have you noticed any side effects, unwanted reactions, or other problems with medications you have taken?” has been validated as an effective way to inquire about adverse drug events.¹⁵ Directed questions about common or high-risk symptoms may also be necessary—for example, inquiring about postural symptoms in a patient taking antihypertensive medications.

Mrs L: I opened his 7-day pill container on a Monday and it was wet. It turned out that he had been taking them out and moving them around and had spilled water in there somehow.

Mr L had several red flags for adherence problems including dementia, a complex medication regimen, and previous adverse drug events.^{5 ,24} Approximately one-half of older patients have problems with adherence to taking at least 1 medication, being evenly split between occasional, frequent, and near-universal omissions of drug doses, although patients nonadherent to taking one of their medications are commonly adherent with others.^{24 - 25}

Patients are often reluctant to admit to nonadherence so a multifaceted approach to evaluating adherence is necessary.^{24 ,26} Quiz Ref IDDuring medication review, clues about adherence can be deduced from observing medication organization, pill counts, and refill history (using information on the refill date and quantity dispensed printed on the label). Asking patients and their caregivers about their understanding of why they take each medication can also be useful. Although studies have found that age itself does not predict adherence and that older adults understand the purpose of as much as 88% of their medications, lack of understanding increases risk of nonadherence and provides a ready target for intervention.^{5 ,27} More generally, nonadherence can be elicited by nonjudgmental questions such as “I know it must be difficult to take all your medications regularly. How often do you miss taking them?”²⁴ If nonadherence is identified, the patient should be asked why, with prompting as necessary for common reasons such as those listed in Table 1.^{5 ,24} Interventions to improve medication use and adherence are most likely to succeed when they address the reasons underlying these problems.

For many physicians, ideal medication reviews and adherence assessments are an improbable reality given the time pressures of office-based practice.²⁸ In this setting, focusing on the highest-risk and highest-benefit drugs can yield good return on a limited time investment. Better yet is sharing these responsibilities with other health care professionals. Contacting community pharmacists regarding concerns about patients can help engage their expertise in identifying and crafting solutions to problems with adherence or medication regimens. Where health systems permit, nurses and clinic-based pharmacists should share medication management responsibilities as articulated in the patient-centered medical home model of care.²⁹ Some medication management programs are available through pharmacy benefit management plans serving Medicare Part D patients (Resources available below). Eligibility criteria and scope of these programs are often limited, although more widespread benefits are mandated for implementation by 2013.³⁰

Mrs L: The family is all guilt-ridden and they tell themselves that they have to keep dear old dad alive . . . . My stake is that he himself has a decent day-to-day life as much as he can.

Like many older patients in the final chapter of their lives, Mr L and his caregivers are facing choices about using medications that might increase his longevity but negatively affect his quality of life.³¹ When getting to know Mr L, one of Dr S's first tasks was to learn what he and his family were trying to achieve through medication use including extension of longevity, reduction in symptoms, and/or minimization of pill burden, medication adverse effects, and costs.³² Many patients would like to achieve all these goals, but often they come into conflict. The physician's role is thus to understand and clarify the relative prioritization of these values, which usually emerges from multiple conversations about specific medication decisions and general goals of care discussions.

Understanding the life expectancy of patients through application of prognostic tools and clinical judgment can help inform goal-driven decisions about prescribing (Resources available below).³³ A short life expectancy affords patients limited opportunity to be helped by medications that require several years to achieve a clinical benefit, such as drugs to improve glycemic control in diabetes.^{31 - 32 ,34} In addition, for patients with advanced dementia, poor prognosis, or both, consensus panels do not recommend (and in some cases advocate against) medications such as statins, bisphosphonates, and cholinesterase inhibitors, although these positions are not universally endorsed.^{35 - 37}

Given Mr L's complex medication regimen and multiple comorbidities, he seems to have been a good candidate for structured medication review and management. The evidence base to guide such approaches is limited. Among 6 studies of medication management that met inclusion criteria in our literature review (see “Methods” section), 3 tested the effect of a clinical pharmacist working with a general practice or general medicine clinic, 2 examined a comprehensive interdisciplinary medication review in a geriatrics clinic, and 1 examined the effect of expert clinician recommendations through computer-based feedback (Table 2).^{38 - 44} Overall, these programs improved markers of pharmaceutical care quality such as reducing medication burden, correcting underuse of medications, and improving a multicomponent score of medication appropriateness. Less evidence is available about the effect of these interventions on clinical outcomes. In the largest study of its type using a multidisciplinary intervention, Schmader et al⁴³ reduced the rate of serious adverse drug events from 0.6 to 0.4 events per 1000 person-days (P = .02). A similar, but nonsignificant (P = .19), degree of reduction in all adverse drug reactions was observed in a study of veterans aged 65 years and older by Hanlon et al,³⁸ with adverse events in 30% of patients receiving medication management vs 40% in patients receiving usual care. There is little conclusive evidence about the effect of comprehensive medication management on other clinical outcomes, including quality of life, health services utilization, and major clinical events; in general, these studies were underpowered for these outcomes.

Of note, most studies on improving medication prescribing for elders with multiple medication use evaluated an external intervention such as pharmacist review or referral to a geriatric evaluation and management clinic. Few studies have evaluated clinicians' own attempts to integrate medication management principles into their practice.⁴⁶ However, limited data suggest that physicians who are provided structured assessment tools for medication review are able to identify and correct medication problems in a large percent of their patients, although time limitations impede widespread implementation of such reviews.^{45 - 47} For example, one study of a guided approach to optimize prescribing found that the proportion of correct medication decisions in a series of clinical vignettes increased from 35% without the method to 48% with it, with a corresponding decline in the number of potentially harmful decisions from a mean of 3.3 to 2.4 per case.⁴⁸

Although few data are available about the effect of structured medication management on patient health and well-being, such approaches are endorsed by experts, in part due to clear evidence of beneficial effects on markers of prescribing quality.⁴⁹ A simple and effective approach to systematically identify prescribing problems is to match each of the patient's conditions with medications that he or she is taking (Table 3). Areas of mismatch can highlight drugs that are being overused (ie, used with no indication), underused (ie, conditions that may benefit from drug therapy that is not currently being offered), and misused (ie, drugs given for an appropriate indication that could be improved by changing the dose, frequency, or substituting another drug with a better profile of benefits, harms, and costs).⁵⁰

Of note, the proper match between clinical conditions and medications is defined not only by guideline recommendations and best practices, but by how medication treatment for a given condition will help the patient attain the goals of care. Thus, the optimized medication regimen for a patient desiring a palliative approach that minimizes medication burden may look quite different than the regimen for a similar patient with the same conditions whose overriding goal is maximizing longevity.

Dr S: A lot of the pain complaints that he used to have had disappeared after he had a lumbar surgery in 2008. [His wife] didn't know if he still needed the pain medication, but was too worried to stop them.

Without knowing anything else about Mr L, the fact that he was taking 13 medications when he first met Dr S suggests a high probability that 1 or more of his medications could or should be stopped.^{16 ,51} Studies of community-based older patients have documented an average of 1 unnecessary drug per patient, including drugs with no identifiable indication or that provide little benefit for the indication for which they are prescribed.^{52 - 53} Perpetuation of unnecessary medications is particularly acute in older adults with multiple prescribers or transitions of care (eg, recent hospital visits).^{54 - 57} In the hospital setting, a large study found that 44% of hospitalized frail older patients were discharged with at least 1 unnecessary medication; common culprits include proton pump inhibitors, central nervous system medications, and vitamin and mineral supplements.^{55 ,58 - 59}

In addition, drugs given for a useful clinical purpose are often misprescribed. For example, highly anticholinergic antihistamines, tricyclic antidepressants, and other high-risk drugs described in drugs-to-avoid lists for older patients are used by approximately 20% to 30% of adults older than 65 years, whereas in many cases, drugs with better safety and/or efficacy would be a more appropriate choice for the target condition.^{60 - 64} Other common problems with misprescribing include use of inappropriately high or low doses, drug-drug and drug-disease interactions, incorrect directions, and choice of expensive drugs when less expensive alternatives would provide similar benefit at lower cost.⁵⁰

As shown in Table 3, matching Mr L's medications with his conditions shows several drugs for which he lacks a clear current indication, including etodolac, gabapentin, acetaminophen, multivitamins, and iron. These should be among the first drugs considered for discontinuation. Next are drugs for which Mr L has a current indication but that may, given his circumstances, provide limited or no benefit. For Mr L, such drugs include memantine for dementia, glyburide for diabetes, and digoxin for rate control of atrial fibrillation. Finally, certain drugs may have benefits but an unfavorable risk profile and should be substituted for others with a more favorable ratio of benefits to harms.

Troublesome symptoms obviously caused by a drug provide a clear signal to consider discontinuation. However, the adverse effects of many drugs are nonspecific and can mimic underlying disease processes, such as Mr L's generalized functional decline. Often, the only way to know whether or not a symptom is an adverse effect is to temporarily stop the drug(s) and see whether the symptoms improve.⁶⁵ Although these are individualized clinical decisions, it can be useful to remember the adage that “any symptom in an older patient should be considered a drug side effect until proven otherwise.”⁶⁶

With limited exceptions, very few studies address the benefits and harms of discontinuing specific types of medications.^{65 ,67} In the case of Mr L, we could not identify any controlled studies that evaluated outcomes of withdrawing digoxin for rate control in atrial fibrillation, discontinuing hypoglycemic medications in diabetes, or withdrawing memantine in dementia (although we identified 1 randomized trial and 2 poorly controlled trials about withdrawal of cholinesterase inhibitors, which suggested worsening of cognition after stopping the drug).^{68 - 70}

In the absence of high-quality trial data on discontinuing medications, decisions about the cessation of certain drugs should be guided by the epidemiology of prescribing problems and by common sense. In assessing harms, particular attention should be paid to drugs that carry a high risk of serious adverse effects, including warfarin, hypoglycemic medications, and digoxin (Table 4), which account for one-third of all emergency department visits in older patients due to adverse drug events.⁷⁶ In the case of Mr L, this provides extra reason to critically evaluate Mr L's diabetes regimen and digoxin. Mr L likely did not need medications for his diabetes, on the basis of guidelines and evidence suggesting that tight glycemic control in the setting of advanced age or multiple comorbidities can result in greater harms than benefits.^{34 ,71} Even if Mr L did require medication for glycemic control, glyburide would be a poor choice because this agent is relatively contraindicated for patients with creatinine clearance less than 50 mL/min/1.73 m² and carries a higher risk of severe hypoglycemia than other sulfonylureas.^{77 - 79} Nonetheless, the presence of a high-risk drug should not automatically mandate a medication change without further exploration of context.⁵² Quiz Ref IDFor example, tricyclic antidepressants are often problematic in elders due to a high frequency of anticholinergic adverse effects. However, if an older patient is already taking a tricyclic antidepressant for a valid indication and reports excellent symptom control, has no anticholinergic symptoms, and is reluctant to switch medications, it may be reasonable to continue the medication while educating the patient to be vigilant for potential future adverse effects.

Although use of ineffective or harmful medications is common in older adults, the same patients often are not prescribed potentially beneficial medications, for example warfarin for atrial fibrillation, antidepressants for major depression, pain medications, and laxatives.^{16 ,50 ,80} Mr L has some conditions that might benefit from additional drug therapy beyond what he is receiving (Table 3). Quiz Ref IDIn patients in their final years of life, preference usually should be given to ensuring that troublesome symptoms such as pain and depressed mood are adequately treated. However, some forms of primary prevention can be appropriate if consistent with goals of care. For example, vitamin D deficiency is common in older patients and has been implicated in falls and fracture risk (along with an emerging variety of other conditions), and repletion can reduce risk of these outcomes.^{81 - 82} Thus, vitamin D supplementation (≥ 800 IU/d) should be considered for Mr L, particularly if his serum 1,25-dihydroxyvitamin D level is low.⁸² For many conditions, the relative paucity of drug trials that include old-old adults or those with extensive comorbidities limits the evidence basis for treating patients such as Mr L. However, in many cases, it appears likely that the relative risk reduction observed in middle-aged and young-old adults is not radically different in the old-old.⁸³

Dr S: I very rarely stop things cold—especially something such as a pain medicine, which could very well be helping the patient; that might be the reason he's not complaining of pain.

When starting drugs in older adults, geriatricians often begin drugs one at a time and follow the dosing mantra of “start low and go slow.” Limited evidence is available about the best ways to stop medications in older people, although in clinical practice many follow a similarly sequential, step-wise approach to discontinuing drugs.⁶⁵ In certain circumstances, an all-at-once approach may be warranted when dangerous signs or symptoms are thought likely to be due to drugs but the exact culprit cannot be identified, or when tendencies toward clinical inertia in a patient or practice environment suggest that future opportunities for medication modification will be limited.

Medications can typically be effectively withdrawn once the decision has been made to do so, although unwanted reactions in the period after withdrawal are common.^{84 - 85} In one of the only broad-based studies of the topic in ambulatory older patients, 26% of drug discontinuations were accompanied by worsening of the underlying disease (eg, recurrence of angina or high blood pressure) and 4% were accompanied by physiologic withdrawal reactions (mostly to β-blockers and benzodiazepines).⁸⁶ For many drugs, risk of adverse withdrawal events can be minimized by slow, careful tapering of drug dose. This is particularly true for drugs to which the body adapts over time, for example through up- or down-regulation of end-organ receptors, which produce a physiologic withdrawal reaction if the drug is withdrawn abruptly.⁶⁵ Although the scientific basis for how to withdraw specific drugs is scant, a rule of thumb is that drugs can usually be tapered down at the same rate at which they are titrated up at the initiation of drug therapy. Common drugs that require tapering include opioids, β-blockers, clonidine, gabapentin, selective serotonin reuptake inhibitors, serotonin-norepinephrine reuptake inhibitors, and tricyclic antidepressants.⁶⁵ Regardless of the speed of the taper, patients should be monitored for adverse withdrawal events, including educating and activating patients to recognize and report concerning symptoms.^{13 ,29} Communication with other health care professionals involved in prescribing for the patient is critical when stopping related drugs. For example, Mr L's warfarin dosing might have been affected by discontinuing other drugs and the anticoagulation clinic should have been alerted to the change.

Sometimes drugs are stopped on a trial basis to determine if potential adverse drug effects resolve or symptoms of the underlying disease worsen. Such assessments can be complicated by fluctuations of symptoms and biomarkers in an individual patient; for example, it may be difficult to ascertain whether improvement in symptoms after withdrawing a drug was the result of stopping the drug or natural fluctuations in the disease course. In this case, a formal rechallenge with the drug (ie, as part of an n-of-1 trial) may help to establish causality.⁸⁷

Mrs L: We looked and saw how confused he was so I told him I was going to take over all of his medicines.

The benefits of changing the medication regimen are contingent on the patient adhering to the revised plan of care. Improving adherence requires diagnosing barriers to proper medication use and devising strategies to overcome those barriers (Table 1).

Randomized controlled trials of strategies to improve adherence to chronic medications have yielded mixed results, and often have studied multifaceted interventions in a manner that makes it difficult to unpack the contribution of each component to improving adherence.^{88 - 90} However, several lessons emerge from the data. First, education through oral counseling or written instruction is important, but often insufficient unto itself. Most randomized trials of intensive educational interventions have yielded minimal to moderate impacts on adherence and little effect on clinical outcomes.^{88 - 89} Nonetheless, common sense suggests it is useful to briefly discuss and write out instructions for taking a medication that is being newly prescribed or modified. A teach-back approach, in which the patient or caregiver is asked to describe the purpose of the drug, instructions for its use, and adverse effects to be aware of can help to ensure comprehension.

In contrast to a focus on education, a potent intervention to improve adherence is simplifying medication dosing schedules. Observational studies have found that adherence drops steeply with increasing number of doses per day, with average adherence falling from roughly 80% in patients taking once-daily regimens to 50% in those taking 4-times-per-day regimens.⁹¹ Randomized controlled trials have found large differences in adherence in patients randomized to medications requiring different numbers of doses per day, although effects on downstream clinical outcomes were mixed.⁸⁸ Thus, whenever possible, clinicians should minimize dosing frequency by prescribing longer-acting medications and dosing different drugs at the same time. In addition, pill burden can be reduced by using medications that can treat 2 or 3 conditions simultaneously (for example, β-blockers in a patient with hypertension, heart failure, and atrial fibrillation with rapid ventricular response). Attempts to reduce dosing frequency may be particularly potent for patients with cognitive difficulties, but are also helpful for cognitively intact patients or caregivers (such as Mr L's wife), who can also frequently forget to take or administer medicines and may resist the pill burden and lifestyle impacts that come with multiple dosings.²⁴

Quiz Ref IDOther approaches can help address common barriers to adherence, including behavioral interventions (eg, cues, medication organizers, packaging), involvement of family and friends (eg, support, monitoring, and administering medications, as was done for Mr L), and by having patients demonstrate ability to self-medicate in a controlled environment (eg, in the hospital or long-term care facility) before discharge to home without support.^{88 - 90} In addition, addressing medication costs, for example, by prescribing lower-cost generic alternatives instead of brand-name drugs, can reduce cost-related nonadherence as well as negative effects on other aspects of the patient's financial well-being.⁹² Many patients will need a combination of approaches, and pharmacists can be helpful partners in devising and following strategies to improve adherence.

Ongoing monitoring for the toxicity and effectiveness of drug therapy is critical to providing quality care and improving outcomes, but current practices often fall short.^{93 - 94} Approximately one-third to two-thirds of patients taking angiotensin-converting enzyme inhibitors, digoxin, carbamazepine, and other drugs that require laboratory-based safety monitoring fail to receive minimum standards for monitoring.^{49 ,95 - 96} If suboptimal monitoring or frequent deviations from target levels have been present, barriers to monitoring and safe drug dosing should be assessed. If such barriers cannot readily be remediated, the clinician should consider discontinuing the drug.

Finally, systematic review of a patient's medication list (eg, using the brown bag framework suggested in this article) is a form of monitoring that should be done periodically. Although the frequency of such reviews should be tailored to patient circumstances, good starting points include recommendations by the National Committee for Quality Assurance and the Assessing Care of Vulnerable Elders (ACOVE) project, which consider medication review at least once per year to be an important measure of care quality in older adults.^{97 - 98} Declines in function and the onset or worsening of geriatric syndromes such as cognitive decline or falls may represent adverse drug effects or signal a change in goals of care and should also precipitate medication review.

Prescribing for older patients is an extraordinarily complex endeavor. However, as illustrated by Dr S, Mr L, and his caregiver Mrs L, a thoughtful, systematic approach to addressing the medication regimen can bring order to complexity and make a meaningful difference in patient outcomes. The success of Dr S's care of Mr L was not in knowing the “right” answer for her patient from the beginning, but rather from using a careful, stepwise process that merged key principles of pharmacologic care with the clinical reality, social situation, and goals of care of the patient.