Antihypertensive Prescribing Cascades as High-Priority Targets for Deprescribing

Older adults frequently take many medications, with two-fifths taking 5 or more. While most older adults are exposed to polypharmacy as part of evidence-based treatment of 1 or more chronic conditions, with a rising tide of medications comes higher risk of adverse drug events. One strategy to prevent harms associated with polypharmacy is to identify medications that are unnecessary or inappropriate and proactively deprescribe them. The American Geriatric Society Beers Criteria¹ and similar lists have provided a starting point for identifying medications that are most often inappropriate in older adults. However, the appropriateness of many other medications is highly dependent on the clinical context of the individual patient. One potentially valuable strategy for deprescribing efforts is to identify common prescribing cascades, instances in which a second (potentially avoidable) medication is administered in response to an adverse effect or drug reaction caused by another medication. In this issue of JAMA Internal Medicine, Savage and colleagues² describe an example of a prescribing cascade: the prescription of loop diuretics following calcium channel blocker initiation.²

The authors used administrative health databases from Ontario, Canada, to conduct a population-based retrospective cohort study of older adults with hypertension, but not heart failure, who started taking calcium channel blockers between 2011 and 2016. They hypothesized that because calcium channel blockers commonly cause peripheral edema, their initiation may lead clinicians to subsequently prescribe diuretics to treat this adverse effect. As edema associated with calcium channel blockers is due to fluid redistribution rather than fluid overload, this prescribing cascade is not only unnecessary but creates a risk for severe adverse events, as loop diuretic use may cause overdiuresis in patients with euvolemia. The study found that 1.4% of patients prescribed calcium channel blockers were subsequently prescribed a loop diuretic within 90 days and 3.5% were prescribed a loop diuretic within 1 year. In contrast, 0.7% of patients prescribed other antihypertensives were prescribed a loop diuretic at 90 days and 1.8% at 1 year, yielding an absolute risk difference of 0.7% at 90 days and 1.7% at 1 year. As the underlying population studied was all patients prescribed a calcium channel blocker rather than all those prescribed a calcium channel blocker who then developed edema, these small values obscure the high frequency of this prescribing cascade. Assuming that 10% of patients prescribed a calcium channel blocker develop peripheral edema (literature range, 2% to 25%), potentially 7% to 14% of people who develop edema while taking a calcium channel blocker may then receive a loop diuretic. Thus, Savage and colleagues² add real-world evidence of a common prescribing cascade with 3 important implications.

First, their study adds to other examples of prescribing cascades, which include the use of antihypertensives following initiation of nonsteroidal anti-inflammatories, gout treatment following thiazide diuretics, and anticholinergic drugs following cholinesterase inhibitors.^3,4 The frequency of antihypertensive medications among the causes and consequences of prescribing cascades is notable. Given the high prevalence of hypertension in older adults, 6 of the 10 most commonly prescribed medications in older adults are antihypertensives,⁵ and use is likely to increase given the lowering of blood pressure treatment thresholds and targets in recent guidelines. Each antihypertensive class has predictable adverse effects that while, typically mild, may lead to predictable prescribing cascades, including: (1) calcium channel blocker causing edema leading to loop diuretic prescription, (2) angiotensin-converting enzyme inhibitors causing dry cough leading leading to antitussive (eg, codeine syrup) prescription, (3) β-blocker causing sexual dysfunction leading to phosphodiesterase-5 inhibitor prescription, (4) diuretic causing urinary frequency leading to antimuscarinic prescription, and (5) all antihypertensives causing dizziness leading to antihistamine or benzodiazepine prescription.

Given clinicians’ familiarity with these common drug classes, why do prescribing cascades like the one described by Savage and colleagues² occur? Drug-induced symptoms frequently go underreported and when reported are often mis-attributed as the manifestation of new disease, in part because physicians do not routinely ask about medication adverse effects.⁴ The risk of this diagnostic error may be amplified by episodic care and a lack of clinician continuity. When a patient follows-up with a different clinician than the one who initially prescribed the calcium channel blocker, this new clinician may be unaware of the chronology of prescribing and less likely to consider the possibility of lower extremity edema being an adverse drug event. Even when symptoms are recognized as possibly associated with a medication, some clinicians may apply a shotgun approach of pursuing multiple strategies to diagnose and resolve a new problem at once rather than systematically excluding medication-associated symptoms before moving on to new treatments or tests. A shotgun approach to new lower extremity edema might include ordering a new medication to treat the symptom (loop diuretic), stopping the potentially offending medication (calcium channel blocker), and ordering diagnostic testing to rule out alternative causes (eg, echocardiogram, lower extremity dopplers, and urine albumin). This approach may be encouraged by patients, as receiving a medication to treat a symptom, even if clinically inappropriate, may improve patient satisfaction by facilitating the feeling that something is actively being done.⁶ Similarly, diagnostic testing to rule out an unlikely disease and provide reassurance is a more active approach than awaiting the resolution of an adverse effect.

Second, as Savage and colleagues² note, the initial prescribing cascade can set off many other negative consequences, including adverse drug events, potentially avoidable diagnostic testing, and hospitalizations. While their study did not examine downstream outcomes, characterizing the aftershocks which follow common prescribing cascades is an area ripe for future population-based research. In the specific case of calcium channel blockers and lower extremity edema, the prescription of loop diuretics may lead to electrolyte disturbances and hypovolemia, which in turn may lead to emergency department or hospital stays. The additional ordering of an echocardiogram to evaluate for heart failure as a cause of lower extremity edema may lead result in a large bill for the patient and substantial anxiety about a potentially serious diagnosis. Unnecessary testing may identify incidentalomas that trigger further diagnostic testing cascades in a cyclical feedback loop.⁷

Third, it is important to consider that antihypertensives may be a trigger and consequence of prescribing cascades. Many commonly used medications may raise blood pressure levels, including nonsteroidal anti-inflammatories, corticosteroids, estrogens, testosterones, certain antidepressants, and common over-the-counter cold remedies and supplements. Thus, patients with polypharmacy are at heightened risk of being exposed to series of prescribing cascades if their current use of medications is not carefully discussed before the decision to add a new antihypertensive.

Identifying prescribing cascades and their consequences is an important step to stem the tide of polypharmacy and inform deprescribing efforts. Ultimately curtailing polypharmacy will require clinicians to routinely assess for adverse drug events following all medication initiations so that when adverse drug events do occur we can minimize their morbidity and avoid further contributing to prescribing cascades.