Drug-drug interactions and polypharmacy

Drug-drug interactions (DDIs) occur when one drug alters the effect of another, potentially causing harm. Drug-disease interactions occur when a drug worsens an existing condition. Both are particularly relevant in multimorbidity—patients with multiple long-term conditions who often receive multiple medications, a situation termed polypharmacy.

Polypharmacy lacks a universally accepted definition. Studies have used cut-offs ranging from ≥3 to ≥10 medications, and in thirty-one studies no definition was given at all. A more clinically useful approach focuses on whether each medication is actually indicated, rather than simply counting drugs. This matters because potentially serious DDIs are common among drugs recommended by guidelines for different conditions. ^[4]

The British National Formulary (BNF) is the primary source UK clinicians use to check drug-drug and drug-disease interactions. The BNF defines a potentially serious interaction as one where "concomitant administration of the drugs involved should be avoided (or only undertaken with caution and appropriate monitoring)". Importantly, this designation reflects the seriousness of potential harm if the interaction occurs—not the likelihood of it happening. ^[3]

A study of three NICE guidelines—type 2 diabetes, depression, and heart failure—examined interactions between guideline-recommended drugs and those recommended for 11 other common conditions. The expert panel systematically searched the BNF for drug-disease warnings, defining important warnings as contraindications or clear cautions to avoid in all or most people with the condition. They also identified potentially serious DDIs between drugs recommended by each index guideline and drugs from any of the 12 guidelines. ^[3]

The results were striking. For type 2 diabetes, 133 potentially serious DDI pairs were found; for depression, 89; and for heart failure, 111. Few of these interactions were highlighted in the guidelines themselves. Drug-disease interactions were relatively uncommon, except in patients with chronic kidney disease. ^[3]

Polypharmacy is most commonly defined as the concomitant use of five or more different medications per day by a single person. However, there is currently no official international consensus on the exact cut-off value used to define this state. Historically, studies have utilised a wide range of thresholds, including three or more, six or more, or even ten or more medications. Because of this lack of standardisation, A definition focusing on whether the medication is clinically indicated may be more appropriate

The clinical relevance of polypharmacy lies in the increased risk of adverse drug reactions (ADRs) and drug-drug interactions (DDIs). These risks are particularly pronounced when combining prescription medications with over-the-counter (OTC) drugs. In patients with chronic kidney disease (CKD), the prevalence of DDIs is estimated to be between 56.9% and 89.1%. This high prevalence is often a result of the complex therapeutic regimens required to manage multimorbidity. ^[7]

Beyond simple binary definitions of interaction, modern computational approaches now seek to identify specific "polypharmacy side effects"—clinical manifestations that arise specifically from the interaction of a drug pair and cannot be attributed to either drug alone. Traditional models often treated interactions as scalar values representing overall strength, whereas newer multirelational models, such as Decagon, attempt to predict the specific type of side effect associated with a drug pair by analysing protein-protein and drug-protein interaction networks. ^[9]

DDIs arise through several mechanisms. Pharmacokinetic interactions alter drug absorption, distribution, metabolism, or excretion—for example, one drug inhibiting or inducing the cytochrome P450 enzymes that metabolise another. Pharmacodynamic interactions occur when drugs have additive, synergistic, or antagonistic effects at the same receptor or physiological system, such as two drugs that both prolong the QT interval or both increase bleeding risk.

The types of harm from co-prescription varied by guideline. For diabetes drugs, harms were most commonly cardiovascular and "other"; for depression, bleeding and "other"; and for heart failure, bleeding and cardiovascular. ^[3]

High-risk combinations identified in the study include:

• Potassium-sparing diuretics with ACE inhibitors—explicitly mentioned in the type 2 diabetes guideline ^[3]
• Potassium-sparing diuretics with angiotensin receptor blockers—also mentioned in the diabetes guideline
• SSRIs with NSAIDs or aspirin—highlighted in the depression guideline due to increased bleeding risk ^[3]

These three combinations were the only ones explicitly discussed in the index guidelines. The heart failure guideline contained no explicit discussion of any of its 111 identified DDIs. ^[3]

Not all recommended drugs interact. Nine of the drugs recommended for diabetes had no DDIs at all. This highlights that careful drug selection can reduce interaction burden. ^[3]

The clinical implications are substantial. A patient with type 2 diabetes, depression, and heart failure—a common combination—could receive drugs from three guidelines with hundreds of potential interactions between them. The study identified 133 DDIs for diabetes drugs, 89 for depression drugs, and 111 for heart failure drugs when cross-referenced against all 12 guidelines. A patient on multiple medications could easily encounter several of these.

Given the extensive number of potentially serious interactions, innovative interactive approaches to guideline production and dissemination are needed, allowing clinicians and patients with multimorbidity to make informed drug selections. Guideline developers should more explicitly account for interactions in people with multimorbidity and use epidemiological evidence to identify when interactions are likely to be common and serious enough to warrant specific mention. ^[3]

For medical students, the key takeaway is practical: when prescribing for a patient with multiple conditions, systematically check interactions using the BNF, pay particular attention to first-line drugs, and remember that chronic kidney disease markedly increases drug-disease interaction risk. The absence of an interaction warning in a guideline does not mean the interaction does not exist.

Older adults carry the highest drug-interaction burden. Adverse drug events as a cause for seeking ambulatory care nearly doubled in the United States between 1995 and 2005, with increasing age and increasing polypharmacy the predominant characteristics of affected patients. Ageing and multimorbidity increase the number of drugs, and so the potential for interactions. The American Geriatrics Society identifies the consideration of drug-disease and drug-drug interactions as a key element of optimal care for older adults with multimorbidity. ^[3]

Older people living with frailty are often exposed to polypharmacy and potential harm from medications, and targeted deprescribing represents an important component of optimizing medication. Sedatives, antipsychotics and anticholinergics are frequent targets. Only six studies (two were RCTs) with overall good quality reported deprescribing outcomes in older people with reliably identified frailty. Deprescribing reduced the number of medications and PIMs with potential cost saving, and mortality and hospitalisation rates did not increase. ^[5]

Mortality evidence is mixed: one review reported that deprescribing reduced mortality in non-randomized studies but no changes were observed in RCTs, while other reviews suggested a reduction in all-cause mortality in nursing home residents. Functional status has been graded on a 5 point scale (1 = independent, 2 = frail, 3 = mild disability, 4 = disability, 5 = severe disability). ^[5]

Chronic kidney disease dominates drug-disease interactions across major clinical guidelines. In a review of 12 national guidelines, 27 of 32 potentially serious drug-disease interactions for type 2 diabetes recommendations involved chronic kidney disease, representing 84% of the total. For depression and heart failure guidelines, all six and ten drug-disease interactions respectively were also with chronic kidney disease. Yet the guidelines themselves rarely address this: the type 2 diabetes and heart failure guidelines each discussed just one of these identified interactions, and for type 2 diabetes the only specific recommendation was to avoid thiazolidinediones in people with comorbid heart failure. This gap matters because adverse drug events cause an estimated 6.5% of unplanned hospital admissions in the UK, accounting for 4% of hospital bed capacity, with bleeding and renal injury predominating when admissions end in death. ^[3]

Renal impairment alters drug handling through several mechanisms. The kidney clears water-soluble drugs and their active metabolites by glomerular filtration and tubular secretion. As estimated glomerular filtration rate (eGFR) falls, drug and metabolite accumulation occurs, prolonging half-life and increasing toxicity risk. For drugs with narrow therapeutic windows, this accumulation is clinically dangerous. Levofloxacin, a fluoroquinolone cleared renally, requires dose adjustment in kidney failure, and therapeutic drug monitoring (TDM) is helpful to ensure adequate exposure after such adjustments. Ethambutol, used for tuberculosis, shows a biphasic decline in plasma concentration with a terminal half-life of about 10–12 hours in persons with good renal function; in renal impairment this half-life extends considerably, and accumulation can cause optic neuritis. ^[6]

Hepatic impairment affects drug handling differently. The liver metabolises lipophilic drugs via phase I (oxidation, e.g. cytochrome P450 enzymes) and phase II (conjugation) reactions, and also excretes drugs into bile. In cirrhosis, reduced hepatocyte mass and portosystemic shunting decrease first-pass metabolism, increasing oral bioavailability of high-extraction drugs. Protein binding is also affected: many drugs bind to albumin, and hypoalbuminaemia in liver disease raises the free (unbound) fraction. Only the free drug can diffuse through biological membranes to exert its pharmacological effect. Therefore, for highly protein-bound drugs, total concentration measurements can mislead; the free concentration should ideally guide dosing, though in practice MIC determinations and free drug concentrations are not measured because of logistical or financial limitations. ^[6]

Drug interactions compound these effects. Moxifloxacin exposure is approximately 25% decreased when used concomitantly with rifampicin, owing to increased moxifloxacin clearance. Rifampicin induces hepatic enzymes, accelerating metabolism of co-administered drugs. In renal or hepatic impairment, such interactions are amplified because the margin between efficacy and toxicity narrows. TDM integrates pharmacokinetic and pharmacodynamic knowledge to personalise therapy, keeping concentrations within the targeted therapeutic range: below this range treatment is less effective, whereas high concentrations may result in toxicity. When selecting a population pharmacokinetic model for TDM, it is important to consider whether the population in which the model was developed is representative of the population where TDM is conducted. ^[6]

Special populations require particular caution. In older adults, age-related decline in renal function, polypharmacy, and reduced homeostatic reserve increase vulnerability. In pregnancy, increased glomerular filtration and altered protein binding change drug concentrations unpredictably. In children, immature hepatic enzymes and renal function necessitate weight-based dosing with careful monitoring. In low-resource settings, where TDM is often unavailable, clinicians must rely on clinical judgment and standard dose adjustments, making awareness of drug-disease interactions even more critical. Guidelines should systematically assess interactions and make explicit decisions about whether they require discussion, similar to the requirement for treatment benefits to be systematically assessed. The key issue is that interactions and risks should be systematically assessed, with details of expected harm informing clinicians about alternative drug choices or discussions with individual patients. ^[3]

Dose adjustment in polypharmacy is not merely a matter of reducing strength; it is a systematic response to altered pharmacokinetics driven by organ function and concurrent drug therapy. In chronic kidney disease, the British National Formulary (BNF) warnings often explicitly recommend dose adjustment, a requirement that is uniquely emphasized for this condition compared to other comorbidities. For instance, metformin requires caution in renal impairment; clinicians must review the dose if the estimated glomerular filtration rate (eGFR) is less than 45 mL/min/1.73 m2, and the drug should be avoided entirely if the eGFR falls below 30 mL/min/1.73 m2. This threshold-based approach prevents the accumulation of the drug, which is critical when managing type 2 diabetes in patients with reduced renal clearance. ^[3]

Beyond renal function, drug-drug interactions significantly alter the effective dose a patient receives. Rifampicin presents a complex case where repeated daily administration leads to auto-induction, decreasing the drug’s own exposure and half-life due to increased clearance. This non-linear pharmacokinetic profile means that increasing the dose results in a more than proportional increase in plasma exposure. Consequently, standard dosing may be insufficient. Current evidence suggests that a dose of 35 mg/kg daily is safe and tolerable for three months, producing a ten-fold higher area under the curve (AUC0–24 h) compared to the lower end of the dose-response curve. Furthermore, a meta-analysis indicates that at least 25 mg/kg dosing is required to achieve pharmacokinetic/pharmacodynamic (PK/PD) targets. ^[6]

Other agents require dose modifications to balance efficacy against toxicity. Linezolid, which has nearly 100% oral bioavailability, is associated with myelosuppression, lactic acidosis, and neurotoxicity during long-term use at the standard adult dose of 600 mg twice daily. To mitigate these cumulative toxicities, a dose of 600 mg, or occasionally 300 mg, once daily is often recommended for patients with tuberculosis. Similarly, meropenem, used off-label for tuberculosis, shows a significant drop in microbiologic activity when the dose is reduced from 6 g per day to 3 g per day. Most of an intravenous meropenem dose is excreted unchanged in the urine, highlighting the importance of renal function in dosing. ^[6]

These adjustments illustrate that dose modification is a dynamic process. In heart failure, interactions causing severe hypotension or increased digoxin concentrations are cited as potentially serious. Therefore, dose adjustments must account for both the primary disease and the pharmacokinetic impact of co-administered drugs, ensuring that therapeutic levels are maintained without crossing into toxicity.

Drug-drug interactions and polypharmacy rarely occur in isolation; they cluster in patients with multimorbidity, frailty, and age-related physiological change. Understanding the diseases most closely linked to these problems helps you anticipate which patients are at risk and why.

Multimorbidity is the core driver. Clinical guidelines are largely written as though patients have a single condition, and the cumulative impact of treatment recommendations from multiple clinical guidelines is not generally considered. In people with several conditions, simply applying recommendations from multiple single-disease guidelines can result in complex multiple drug regimens (polypharmacy) with the potential for implicitly harmful combinations of drugs. The more conditions a patient accumulates, the more guidelines apply, and the more drugs accumulate—each new prescription adding another potential interaction surface. ^[3]

Frailty is a distinct and independent risk state. Frailty is identifiable in middle-aged as well as older adults, and is associated with mortality independently of the extent of multimorbidity as well as age, sex, sociodemographic, and lifestyle factors. This means a frail patient with one condition may be at greater risk of harm from polypharmacy than a robust patient with several. Frailty is also associated with a range of long-term conditions, creating a vicious cycle: more disease → more drugs → more frailty-related vulnerability to adverse effects. The management of frailty is an important clinical priority for all health professionals, not just those caring for older people. ^[1]

Specific disease clusters illustrate the problem. A 72-year-old man with a new diagnosis of type 2 diabetes who has pre-existing heart failure, atrial fibrillation, and chronic kidney disease exemplifies the challenge—8.9% of people aged 70–79 with type 2 diabetes have heart failure, 9.4% have atrial fibrillation, and 10.4% have chronic kidney disease. Here treatment choices are more constrained in the face of multiple drug-drug and drug-disease interactions, and effective diabetes treatment might require more global treatment optimisation. Each disease brings its own guideline-mandated medications, and the interactions between them—for example, between antidiabetic agents, anticoagulants, and drugs excreted renally—multiply rapidly. ^[3]

Lifestyle disease and ageing trajectories. Many people have the opportunity to adopt a lifestyle that will either give protection from these diseases or an alternate lifestyle that will make them more susceptible to lifestyle diseases. Physical activity and exercise optimise physiology when engaged in with adequate doses and modalities, and evidence from randomised controlled trials demonstrates favourable effects of specific PA/exercise modalities on age-related physiological changes, disease prevention, and treatment of older adults with chronic disease and disability. This matters for polypharmacy because lifestyle modification can reduce the number of drugs needed—but the choice of lifestyle is not equally available to all, as socioeconomic conditions do not always allow for it. ^[2]

Why older adults are disproportionately affected. The occurrence of drug-drug interactions and adverse drug reactions among older adults is a prevalent problem in various countries. Older adults usually do better use of medicines when their care is managed by a multidisciplinary team, consisting of a physician (geriatrician), clinical pharmacist and nurse. The involvement of a dentist in this team seems to be relevant, as demonstrated in a study. This team-based approach recognises that the diseases of polypharmacy are not just medical but also functional and social. ^[4]

In summary, the diseases of this topic are not single entities but constellations: multimorbidity, frailty, and lifestyle-related chronic disease, often in older adults. Recognising these patterns early—and involving a multidisciplinary team—is the first step toward safer prescribing.

Investigations in polypharmacy focus on identifying adverse drug reactions (ADRs) and drug-drug interactions (DDIs) that often manifest as laboratory abnormalities or clinical deterioration. While specific diagnostic tests are not always definitive for interactions, monitoring key parameters allows clinicians to detect toxicity early. For instance, in patients with type 2 diabetes, cardiovascular harm such as significant hypotension or bradycardia is a common outcome of serious interactions, alongside renal or serum potassium associated harms. Similarly, for depression, the most commonly identified harm is risk of bleeding, particularly with selective serotonin reuptake inhibitors recommended as first line, followed by lithium toxicity.

Laboratory monitoring is critical when managing patients with chronic kidney disease, where drug-disease interactions are more common than in other conditions. The prevalence of comorbidity with chronic kidney disease is about 4% in patients with depression, 14% in patients with type 2 diabetes, and 23% in patients with heart failure. These figures suggest that guideline developers should consider chronic kidney disease with heart failure and possibly with type 2 diabetes, but possibly not with depression. In frail older people, harms might outweigh benefits for some medications, such as intensive blood glucose control in type 2 diabetes, or where the known time to benefit exceeds projected life expectancy, such as with statins. ^[3]

Computer-based screening and technologies for monitoring DDIs and recognizing ADRs are increasingly used to identify clinically significant interactions. The software must have high sensitivity and specificity, and high positive and negative predictive value to be effective. Utilizing computerized databases for reviewing medication prescriptions helps practitioners recognize potential interactions that might otherwise go unnoticed.

Specific drug classes require targeted monitoring. For example, patients using clopidogrel with proton pump inhibitors, tamoxifen, co-prescription of macrolide antibiotics and calcium-channel blockers, trimethoprim/sulfamethoxazole, antidepressants, warfarin, or benzodiazepines are at increased risk. Additionally, aminoglycoside use is limited by significant toxicity after prolonged use and should only be used if drug susceptibility testing results confirm susceptibility and audiometry monitoring can be ensured. A 10% probability of ototoxicity occurred with a threshold cumulative AUC > 87 g h/L. The probability of ototoxicity increased sharply starting after 6 months of amikacin therapy to near maximum at 9 months. ^[6]

When treating patients with multiple conditions, the first challenge is that guideline-recommended therapy is extensive. Across three common chronic disease guidelines, there were 23 drugs recommended in the type 2 diabetes guideline (four first line), 13 drugs in the depression guideline (one first line), and 11 in the heart failure guideline (two first line). This means that a patient with diabetes, depression, and heart failure could be offered nearly 50 different drug options, with only seven designated as first-line for all or nearly all patients. The sheer number creates the substrate for drug-drug interactions before any individual prescription is written. ^[3]

The treatment approach must therefore be selective and interaction-aware. In heart failure, for example, amlodipine should be considered for the treatment of comorbid hypertension and/or angina in patients with heart failure, but verapamil, diltiazem, or short acting dihydropyridine agents should be avoided. This distinction matters clinically: amlodipine is a dihydropyridine calcium channel blocker with a long half-life and a safer haemodynamic profile in heart failure, whereas verapamil and diltiazem have negative inotropic effects and short-acting dihydropyridines cause reflex sympathetic activation. When choosing drugs for coexisting conditions, prescribers must check whether the proposed agent is in the "avoid" or "consider" category for each of the patient's diseases. ^[3]

For tuberculosis, treatment optimisation is a separate but instructive model. For most anti-TB drugs, the AUC0–τ (often AUC0–24 h)/MIC is the most predictive PK/PD index, although a specific AUC0–24 h/MIC target value has not yet been established for all anti-TB drugs. This matters because individual MIC values might not always be reliable and often are not available early in treatment, and MIC values might be significantly different when using different methods (e.g., solid or liquid media). In practice, this means that even when a drug-drug interaction alters drug exposure, interpreting whether that change matters requires knowing both the AUC and the MIC—and the MIC may be method-dependent. ^[6]

For multidrug-resistant TB, the evidence base for therapeutic drug monitoring is uneven. Only for the fluoroquinolones, linezolid, and aminoglycosides does a substantial body of evidence support TDM; the newer drugs such as bedaquiline, delamanid, and pretomanid lack this abundance of data, although publications on acquired resistance, drug-drug interactions, and significant variability of bioavailability based on concomitant food intake already provide a rationale that TDM may be of use for the newer drugs as well. A well-designed prospective study comparing TDM in M/XDR-TB is still lacking. ^[6]

The bi-directional relationship between polypharmacy and frailty has been reported: drugs and frailty might interact through a network of connections, including physiological changes, multiple pathologies and chronic diseases, life expectancy and functional or cognitive status. Frailty may influence drug pharmacokinetics and pharmacodynamics, toxicity, and therapeutic efficacy, and these factors may in turn be involved in the development of frailty. Treatment in this context should focus on improvement in overall functionality, complementary to the diagnosis and treatment of specific diseases. ^[5]

Polypharmacy does not affect all populations equally. The burden falls disproportionately on older adults, those with multimorbidity, and increasingly on middle-aged individuals with complex health needs. Frailty—a state of reduced physiological reserve—is strongly associated with older age and multimorbidity (two or more long-term conditions). However, frailty is not confined to the elderly; it also occurs in middle-aged people, where it carries a significant mortality risk even after adjusting for the number of long-term conditions, sociodemographics, and lifestyle factors. This challenges the traditional view that frailty is solely a geriatric concern. ^[1]

The global scale of the problem is substantial. An integrative review analysing 47 full-text studies involving 14,624,492 older adults (aged 60 years and over) found that polypharmacy is a multifactorial process, with predictors and inappropriate prescribing consistently associated with negative health outcomes, including increased frequency and types of adverse drug reactions (ADRs) and drug-drug interactions (DDIs) across different drug classes^[1]. Of these studies, 51.1% concerned ADRs, 29.8% concerned DDIs, and 19.1% investigated both^[1]. Drug-drug interactions and adverse drug reactions among older adults remain a significant issue worldwide^[1]. ^[4]

A concrete example illustrates the risk. Selective serotonin reuptake inhibitors (SSRIs) are recommended as the first-line antidepressant by NICE guidance, but they have potentially serious interactions with aspirin, clopidogrel, and over-the-counter ibuprofen. Alternative antidepressants such as mirtazapine, duloxetine, and reboxetine do not share this interaction, although venlafaxine does. This matters because patients rarely declare every over-the-counter purchase to their prescriber. ^[3]

For vulnerable populations, exercise offers a protective strategy. Multicomponent training programmes—combining strength, power, gait, balance, and functional training—should include gradual increases in volume, intensity, and complexity, and can be prescribed even to acutely ill, hospitalised older patients or institutionalised older adults. Conversely, targeting glycaemic control without addressing central obesity and sedentary lifestyle may hasten disease complications and add to the burden of polypharmacy in insulin-resistant individuals. Weight loss by diet alone, without resistance exercise, leads to losses of lean tissue that could exacerbate age-related sarcopenia and osteopenia. ^[2]

Deprescribing in older people living with frailty has limited research, but existing studies suggest it can be safe, feasible, well tolerated, and lead to important benefits. Health-care services must be reconfigured to meet the needs of increasingly complex patient populations, shifting focus from single conditions to the requirements of these patients. ^{[5] [1]}

Drug–drug interactions (DDIs) are not abstract pharmacology; they are clinical predictions. The first pearl is to think in terms of prevalence × severity. A potentially fatal interaction that affects almost no one is less important in daily practice than a modest interaction that affects many. For example, statins (first line for type 2 diabetes) and ciclosporin (second line for rheumatoid arthritis) carry a risk of myopathy and rhabdomyolysis. Yet only 1.4% of people with type 2 diabetes also have rheumatoid arthritis, and ciclosporin is used only as second line in that subgroup. This interaction is real but rare—so rare that guideline groups may never explicitly address it. Contrast this with the co-prescription of selective serotonin reuptake inhibitors (SSRIs, first line for depression) and tramadol (second line for pain). Over 27% of people with depression also have painful conditions, and tramadol is widely used in the UK. Serotonin syndrome is potentially fatal, though the risk appears low and is poorly quantified. The pearl: when you see an SSRI and tramadol together, you are looking at a common, potentially deadly combination; when you see a statin and ciclosporin, you are looking at a rare one. Both deserve respect, but the former will cross your desk far more often. ^[3]

The second pearl concerns enzyme induction kinetics. Rifampicin is a potent inducer of phase I and II metabolic enzymes and drug transporters, especially cytochrome P450 3A4. Induction begins after a few doses and is complete after one week; after stopping rifampicin, baseline enzyme activity returns in about two weeks. This matters for any drug you add or remove. If a patient on rifampicin starts a CYP3A4 substrate (e.g. a statin, an azole, a calcium-channel blocker), the substrate’s exposure will fall—often dramatically—and therapeutic effect may be lost. If you stop rifampicin, the enzyme activity does not normalise immediately; for roughly two weeks, the patient remains in a hyper-metabolic state, and doses that were safe during induction may now become toxic. The pearl: always ask about rifampicin, and always remember the two-week tail. ^[6]

The third pearl is about dose-dependent toxicity in the setting of renal impairment. Ethambutol is used at 15–20 mg/kg daily, maximum 25 mg/kg daily. Higher doses and exposures, especially with decreased renal function, are associated with a higher incidence of retrobulbar (optic) neuritis. This is a classic DDI-adjacent problem: not a drug–drug interaction, but a drug–organ interaction that mimics one. If a patient on ethambutol develops blurred vision or colour vision loss, check renal function and dose, not just the drug list. Similarly, delamanid prolongs the cardiac QT interval, an effect linked to its metabolite DM-6705, which has a half-life of 121–425 h. Any other QT-prolonging drug (e.g. a macrolide, an antipsychotic, methadone) will add to this risk. The pearl: when a drug has a long-lived active metabolite, the interaction window extends far beyond the parent drug’s half-life. ^[6]

The first pitfall is over-indexing on rare interactions. A guideline development group may not even consider a DDI if the affected population is tiny. The statin–ciclosporin example is instructive: it is serious (myopathy, rhabdomyolysis) but affects only a sliver of patients. If you spend your clinical energy memorising every rare interaction, you will miss the common ones. The trap is to treat all DDIs as equally likely. They are not. Prevalence data should drive your vigilance.

The second pitfall is forgetting the induction tail. Rifampicin’s enzyme induction is not instantaneous, and it does not vanish when the drug is stopped. A common error: a patient stops rifampicin, and the prescriber immediately restores the previous dose of a CYP3A4 substrate, assuming the interaction is over. For about two weeks, enzyme activity is still elevated. The reverse error is also common: starting rifampicin and expecting the interaction to be fully expressed on day one. It takes a few doses to begin and a week to complete. The pitfall is thinking in binary terms—interaction on or off—when induction is a gradual process.

The third pitfall is ignoring the dose–toxicity axis in renal impairment. Ethambutol’s optic neuritis risk rises with higher doses and exposures, particularly when renal function is reduced. The pitfall is to assume that a drug is safe because it is “first-line” or “standard.” Standard doses are for standard kidneys. If a patient has chronic kidney disease, the same dose becomes a toxic dose. This is not a DDI in the strict sense, but it behaves like one: the drug’s effect is amplified by a patient factor, and the result is harm. Always check renal function before prescribing renally cleared drugs, and re-check when a new drug is added that might affect renal function (e.g. an ACE inhibitor plus a diuretic plus an NSAID).

The fourth pitfall is underestimating food and formulation effects. Delamanid’s bioavailability is enhanced two- to four-fold by food, and its exposures are less than dose proportional. If a patient takes delamanid fasting on some days and with food on others, the exposure swings wildly, and the QT-prolonging effect becomes unpredictable. Similarly, PAS absorption is improved by acidic food or yoghurt. The pitfall is to treat a drug as a fixed entity when its absorption is highly variable. Ask not only “what” but “how” and “when.” ^[6]

The fifth pitfall is confusing bacteriostatic with bactericidal in combination therapy. PAS is generally regarded as bacteriostatic, though one study described a bactericidal effect at high (20 g) once-daily dosing. Ethambutol’s early bactericidal activity is negligible, and it has very little sterilizing activity; it is added to prevent rifampin resistance in case of unrecognised isoniazid resistance. The pitfall is to assume that every drug in a regimen contributes equally to killing. Some drugs are there to protect the regimen, not to kill. Removing them to simplify a regimen can precipitate resistance. ^[6]

In practice, DDI awareness is applied in three settings: guideline development, deprescribing, and individual prescribing decisions.

In guideline development, the key question is whether a DDI is common enough to warrant explicit recommendation. The statin–ciclosporin interaction is a textbook example of a rare interaction that guideline groups may reasonably ignore. In contrast, the SSRI–tramadol combination is common because of the overlap between depression and painful conditions (over 27%) and the widespread use of tramadol. A guideline group should address this one. The practical lesson: when you read a guideline, ask whether the population it describes matches the population you see. Real-world data from large electronic primary care datasets can define the population, its comorbidity, and its current prescribing. This is where DDI knowledge becomes population medicine.

In deprescribing, explicit and implicit criteria guide which drugs to stop. Two studies used explicit criteria (lists of drug names): the STOPP criteria and the STOPPFrail tool. Two used implicit criteria (evaluative questions or a process), including the Garfinkel algorithm for concurrent deprescribing of multiple medications and guidelines for targeted deprescribing of anticholinergic and sedative medicines. Two used a combination: one used STOPP and Beers criteria alongside pharmacist’s judgment, and another used an algorithm based on STOPP, Beers, Choose Wisely, and Choose Wisely Canada. In practice, this means you do not deprescribe blindly. You use a structured tool to identify high-risk drugs, then apply clinical judgment. The Garfinkel algorithm, for example, allows concurrent deprescribing of multiple medications rather than one at a time. This is particularly relevant in polypharmacy, where the interaction is not between two drugs but among five or ten. ^[5]

In individual prescribing, the practical workflow is: identify the patient’s full drug list, check for known interactions, consider the prevalence of the combination, and adjust for patient factors (renal function, age, food intake). The flowchart below outlines this process.

In practice, the SSRI–tramadol interaction is managed by choosing an alternative analgesic (e.g. an NSAID if safe, or a non-serotonergic opioid) or by using the lowest effective tramadol dose with explicit patient education about serotonin syndrome symptoms. The statin–ciclosporin interaction is managed by avoiding the combination if possible, or by using a statin that is less dependent on CYP3A4 and monitoring creatine kinase. The rifampicin interaction is managed by adjusting the substrate dose during induction and for two weeks after discontinuation. The ethambutol interaction is managed by dose adjustment for renal function and baseline and follow-up visual acuity testing. The delamanid interaction is managed by checking the QT interval at baseline and during treatment, and by avoiding other QT-prolonging drugs.

Finally, remember that some drugs are used precisely because they are bacteriostatic or have negligible early bactericidal activity. Ethambutol is added to prevent rifampin resistance, not to kill bacteria. PAS retains a role in extensively drug-resistant TB (XDR-TB) when other options are limited. In these cases, the “interaction” is protective: the drug prevents resistance to a more potent partner. Do not remove these drugs casually.