Common Drug Side Effects and Drug-Drug Interactions in Elderly Adults in Primary Care

Susan E. Merel, MD; Douglas S. Paauw, MD


J Am Geriatr Soc. 2017;65(7):1578-1585. 

In This Article

Important Side Effects of Drugs Commonly Prescribed in Primary Care


Statins are the most commonly prescribed drug in elderly adults, with as many as half of community-dwelling elderly taking these agents.[3] Statin myopathy occurs in 10% to 15% of individuals, and the clinical presentation varies in severity from mild myalgia, fatigue, and weakness to life-threatening rhabdomyolysis. Women, frail individuals, those with a low body mass index, hypothyroidism, or polypharmacy; and those who abuse alcohol are at higher risk of statin myopathy.[4] Vitamin D deficiency also increases the risk of myalgia in individuals taking statins, with proposed mechanisms including high beta-oxidation and preferential shunting of CYP 3A4 for vitamin D hydroxylation, resulting in lower availability of CYP 3A4 for statin metabolism and resulting higher serum statin levels.[5] Small studies suggest that individuals taking statins who have a vitamin D level less than approximately 30 ng/ml may be more likely to have myalgia and that treating individuals with 50,000 to 100,000 IU of ergocalciferol weekly may be associated with resolution of myalgia.[6,7] Coenzyme Q10 may also have a role in statin-related myopathy, but a recent metaanalysis did not find any significant effect of coenzyme Q10 supplementation on muscle pain or creatine kinase (CK) levels in individuals taking statins.[8] The authors' overall approach to individuals taking statins presenting with myalgia is as follows. CK, thyroid-stimulating hormone, and vitamin D levels are checked, and vitamin D deficiency and thyroid disease are treated if present. The statin is then stopped; if symptoms disappear and the individual did not have a significantly high CK level or clinically important rhabdomyolysis, the statin is restarted at a lower dose, or a different statin is started. Because simvastatin has greater muscle toxicity and drug interactions, it should be avoided. If symptoms are recurrent, high-dose fluvastatin (80 mg) or low-dose rosuvastatin daily, every other day, or twice a week is tried. If symptoms persist, ezetimibe is tried.[4]

Within the past few years, data have been published about the association between statin use and both cognitive impairment and diabetes mellitus. Multiple studies have suggested a slightly greater risk of incident diabetes mellitus with statin use, but the vascular benefits of statins far outweigh the slightly greater risk; for every 255 individuals treated with statins over 4 years, one will be diagnosed with diabetes mellitus, but 5.4 vascular events will be prevented.[9] Data regarding the association between statin use and cognitive decline or incident dementia has been mixed, with most studies showing a small protective effect, some negative studies, and some studies showing no association.[10] The U.S. Food and Drug Administration (FDA) added safety label changes to statins in 2012 warning of the potential for reversible cognitive side effects and hyperglycemia, adding that the cardiovascular benefits outweigh these risks.[11] The authors do not routinely consider the possible associations between statin use and cognitive impairment and diabetes mellitus in the risk-benefit analysis when prescribing statins to older adults. Although there is strong evidence supporting statin use for secondary prevention in older adults with heart disease, trials of statins in primary prevention have included few participants aged 75 and older; there is a need for more evidence of the benefits of statins in primary prevention in individuals aged 75 and older.[12] Statin use for primary prevention should be reconsidered in many individuals after age 75, especially those who are frail and have a limited life expectancy and polypharmacy. This is especially important when individuals have bothersome statin side effects.

Proton Pump Inhibitors

Proton pump inhibitors (PPIs) are widely used in this population, with a weighted prevalence estimate of 18.5% of community-dwelling older adults using them in a 2010–11 cohort.[3] The 2015 Beers Criteria recommend against the use of PPIs for longer than 8 weeks except in high-risk individuals, but many clinicians may not be following this recommendation.[1,13] This warning was based on the risks of Clostridium difficile infection (CDI) and bone loss and fractures. An association between PPI use and CDI has been found in several metaanalyses (odds ratio (OR) = 1.74, P < .001).[14] Studies have also shown greater risk of recurrent CDI and enteric infections with organisms such as Salmonella and Campylobacter.[14,15] A 2012 FDA safety warning advises clinicians to consider CDI in individuals taking PPIs who develop diarrhea that does not improve, to advise individuals taking PPIs to seek medical care for symptoms suggestive of CDI, and to prescribe PPIs at the lowest possible dose and shortest duration.[16]

There are also well-established associations between PPI use and altered bone metabolism, with resultant effects on bone density and fracture risk. PPIs reduce bone mineral density primarily by reducing calcium absorption and inhibiting osteoclast activity. A 2012 prospective cohort study showed a 35% higher risk of hip fracture in women regularly using PPIs for at least 2 years; the association was stronger in current and former smokers, with a more than 50% greater risk.[17] In addition to calcium, individuals taking PPIs are less able to absorb iron, magnesium, vitamin B12, and some medications, including azole antifungals and thyroid hormone, with the strongest effect on absorption of magnesium and calcium. Periodic monitoring of magnesium levels in individuals taking long-term PPIs is advised, especially those taking diuretics or digoxin.[18]

Recent evidence suggests associations between PPI use and dementia and acute and chronic kidney disease. The association between PPIs and acute interstitial nephritis (AIN) is well established.[20] A recent study of individuals aged 65 and older initiating PPI therapy found a greater hazard of acute kidney injury than in matched controls (hazard ratio (HR) = 2.52, 95% confidence interval (CI) = 2.27–2.79); the authors postulated that many of those cases were AIN.[21] Newer data suggests an association between PPI use and chronic kidney disease. A large prospective cohort study found a greater risk of incident CKD in PPI users than in nonusers (adjusted HR (aHR) = 1.50, 95% CI = 1.14–1.96).[22] The mechanism of the possible association with CKD is unclear. Two prospective German cohort studies found greater risk of incident dementia in patients taking PPIs (HR = 1.44, 95% CI = 1.36–1.52, P < .001 in the most recent study).[23,24] This association is biologically plausible because PPIs cross the blood–brain barrier, decrease the absorption of vitamin B12, and increase beta-amyloid levels in the brain in mouse models, although this association may not be causal; PPI use may be a marker of greater healthcare use and polypharmacy, both of which are common in individuals with dementia in primary care. Although neither of these associations proves causality, they add to the overall evidence that PPIs should be used at the lowest dose for the shortest possible duration. The authors use PPIs at the lowest dose for the shortest possible duration and pay particular attention to individuals who have recently been discharged from the hospital because PPIs are often added during a hospitalization without a clear plan for discontinuation. AIN is considered in the differential diagnosis for individuals taking PPIs who have acute kidney injury, but otherwise the current evidence is not strong enough to stop PPIs in individuals with CKD or consider dementia risk when prescribing these agents.

Oral Antimicrobials in Common use in Primary Care

Side effects of two antimicrobials commonly prescribed to elderly adults in the primary care setting will be briefly reviewed: trimethoprim-sulfamethoxazole (TMP/SMX) and fluoroquinolones. TMP/SMX is widely used in primary care, and use has increased because of its relative efficacy against community-acquired methicillin-resistant staphylococcus aureus. Clinicians should be aware of the risk of hyperkalemia with trimethoprim, which is more common in elderly adults, those with renal insufficiency, and those taking an angiotensin-converting enzyme inhibitor (ACE-I) or angiotensin receptor blocker (ARB). Trimethoprim acts like the potassium-sparing diuretic amiloride and reduces urinary potassium excretion by 40%. A nested case–control study of individuals with an average age of 82 found that the risk of hyperkalemia requiring hospitalization was almost seven times as high in individuals taking TMP-SMX and an ACE-I or ARB as in those those taking amoxicillin.[25] A second study in older adults taking TMP-SMX and an ACE-I or ARB found a greater risk of sudden death (adjusted OR (aOR) = 1.54, 95% CI = 1.29–1.84), corresponding to approximately three sudden deaths within 14 days per 1,000 TMP-SMX prescriptions.[26] TMP/SMX should be avoided when possible in older adults who are taking an inhibitor of the renin-angiotensin system, although quinolones and azithromycin can also put individuals at risk of sudden death, with the mechanism for those agents being QT prolongation.

Fluoroquinolones are also commonly used in primary care and are associated with a number of important side effects, including peripheral neuropathy, central nervous system side effects, and collagen-related adverse effects. The risk of peripheral neuropathy with these agents has been known for decades, and a FDA Drug Safety Communication in 2013 emphasized that this side effect can start as soon as 24 hours after starting the agent and sometimes last for months or years or can be permanent.[27] Central nervous system effects are also well described; the most common include anxiety, restlessness, euphoria, dizziness, and insomnia, with an incidence of up to 2%; seizures and psychiatric events are also reported but are rarer.[28] Data have also emerged within the past few years supporting the known association between fluoroquinolones and collagen-associated adverse events. In a large population-based longitudinal cohort study of older adults, current fluoroquinolone use was associated with greater risk of tendon rupture (adjusted HR (aHR) = 2.40, 95% CI = 2.02–2.49) and aortic aneurysm (aHR = 2.24, 95% CI = 2.24–2.57); both were greater than risks associated with amoxicillin.[29] There may be a slightly greater risk of retinal detachment as well, but the data are not conclusive. Of the collagen-associated adverse events, rupture of the Achilles tendon is probably the most common; this risk is greatest in older adults and in those also taking corticosteroids. In July 2016, the FDA issued a warning emphasizing that the collagen- and nervous system-associated side effects can be disabling and permanent. The FDA recommended against quinolones being used for uncomplicated urinary tract infections, acute bacterial sinusitis, or acute bacterial exacerbations of chronic obstructive pulmonary disease unless there are no other options; these agents should be used only for serious bacterial infections when the benefits clearly outweigh the risks.[30] With the risk of tendinopathy with concurrent corticosteroid use, quinolones should be avoided in individuals taking steroids.


The risks of hypnotics, including benzodiazepine receptor agonists such as zolpidem, in elderly adults are well known, and the Beers criteria recommendation on this class of agents was strengthened in 2015 to recommend avoiding them even for short-term use. These agents increase the risk of delirium, falls, fractures, and motor vehicle crashes and have only a minimal effect on sleep latency and duration.[1] New studies have raised concern about risks of dementia and mortality in individuals taking zolpidem, but the evidence is not conclusive. There is relatively strong evidence from multiple studies over the past 30 years suggesting excess mortality with hypnotics, but most of these studies included benzodiazepine receptor agonists and benzodiazepines.[31] Two recent retrospective cohort studies examined the association between benzodiazepine receptor antagonists and mortality, with conflicting results; one study showed greater risk of mortality, and the other showed a potentially dose-responsive protective effect.[32,33] A retrospective case–control study of older adults in Taiwan found an association between zolpidem use and two ICD-9 codes for dementia (aOR = 1.33, 95% CI = 1.24–1.41), but the association between zolpidem use and neurologist-diagnosed Alzheimer's disease only held for doses between 170 and 819 mg/yr and not for higher or lower doses.[34] The authors do not consider the potential association between zolpidem and dementia to be relevant to clinical decision-making. Nevertheless, given strong evidence of harm, benzodiazepine receptor agonists and benzodiazepines should be avoided in elderly adults. No pharmacological agents are recommended for treatment of insomnia in elderly adults; nonpharmacological treatments such as cognitive behavioral therapy are recommended instead.[35]

Nonsteroidal Anti-inflammatory Drugs

Nonsteroidal anti-inflammatory drugs (NSAIDS) are listed as PIMs in the Beers Criteria because of risk of gastrointestinal bleeding, peptic ulcer disease, heart failure exacerbation, and acute kidney injury, especially in individuals aged 75 and older; those with heart failure or kidney disease; and those taking corticosteroids, anticoagulants, or antiplatelet agents. Evidence published since 2011 suggests that NSAIDs can increase the risk of heart attack or stroke as early as the first week of NSAID use and in individuals with or without a history of cardiovascular or cerebrovascular disease. A nationwide cohort study in Denmark of individuals taking NSAIDS after their first myocardial infarction (MI) found a significant association between NSAID use and recurrent MI after as little as 1 week of treatment (HR = 1.45, 95% CI = 1.29–1.62).[36] A large metaanalysis of 280 trials comparing NSAIDS (including traditional NSAIDS and selective cyclooxygenase (COX)-2 inhibitors) with placebo or other NSAIDS found that major vascular events were approximately one-third more likely with selective COX-2 inhibitors or diclofenac; ibuprofen also significantly increased major coronary events. Naproxen was not associated with an increase in major vascular events or vascular death. The increased risk of events was independent of baseline characteristics, including vascular risk. All NSAIDs roughly doubled the risk of heart failure and increased the risk of upper gastrointestinal complications (selective COX-2 inhibitors had the lowest risk (relative risk (RR) = 1.81, 95% CI = 1.17–2.81); naproxen had the highest risk (RR = 4.33, 95% CI = 2.71–6.56)).[37] Based on these and other recent studies, the FDA strengthened its warning regarding the cardiovascular risks of NSAIDS in July 2015, emphasizing the risks even with shorter-term use, that the risks are greater even in those without known cardiovascular disease, and that the evidence is not sufficient to determine whether cardiovascular risks are lower with particular NSAIDS.[38] The authors avoid prescribing NSAIDS in elderly adults and have a risk–benefit conversation with individuals who are taking over-the-counter NSAIDS, pointing out the serious side effects of these agents.

Selective Serotonin Reuptake Inhibitors

Although selective serotonin reuptake inhibitors (SSRIs) are listed in the Beers Criteria as potentially inappropriate because of their associations with risk of falls and hyponatremia, they are still widely used and considered first-line pharmacologic therapy for depression in elderly adults. SSRIs are associated with hyponatremia due to the syndrome of inappropriate secretion of anti-diuretic hormone; estimates of incidence of this side effect vary widely (0.5–30%), and risk factors include older age, female sex, low body weight, concurrent use of diuretics, and baseline hyponatremia.[39] Selective norepinephrine reuptake inhibitors are sometimes recommended as an alternative in individuals at risk of falls but are also associated with hyponatremia and fragility fractures, although risk of hyponatremia may be higher with SSRIs. One study in older adults with depression suggested that those taking the antidepressants escitalopram or fluoxetine had greater decreases in serum sodium than those treated with other SSRIs or SNRIs.[40] SSRIs and SNRIs have been associated with risk of fragility fractures in a cohort of individuals aged 50 and older (HR = 1.88, 95% CI = 1.48–2.39).[41] Sodium levels should be monitored when starting therapy or changing doses of SSRIs, especially in older women and those with a history of hyponatremia or falls or those taking other agents that can contribute to hyponatremia, such as diuretics. The authors use SSRIs with appropriate monitoring in individuals with depression who fit criteria for pharmacotherapy because there are not safer antidepressants. Risks and benefits are reassessed if these individuals fall, but the benefit for depression often outweighs the risk because falls are multifactorial.

Dipeptidyl Peptidase 4 Inhibitors

Dipeptidyl peptidase 4 (DPP4) inhibitors are newer agents for the treatment of type 2 diabetes mellitus and act primarily by decreasing glucagon-like peptide-1 levels. Although they are not first-line agents because of expense and modest efficacy, they may be attractive to some older adults with type 2 diabetes mellitus because they are once-daily oral agents, do not cause hypoglycemia, and are weight neutral. Two case–control studies suggest an association between these agents and severe joint pain, with joint pain resolving with withdrawal of the agent in most cases.[42,43] The FDA issued a warning in 2015 about the risk of severe and disabling joint pain and recommended that clinicians stop these agents in individuals complaining of joint pain.[44]