Nature Reviews Nephrology | Volume 20 | June 2024 | 386–401 386 nature reviews nephrology Review article https://doi.org/10.1038/s41581-024-00823-3 Check for updates Drug stewardship in chronic kidney disease to achieve effective and safe medication use Rasheeda K. Hall1 , Rümeyza Kazancıoğlu 2 , Teerawat Thanachayanont3 , Germaine Wong4 , Dharshana Sabanayagam4 , Marisa Battistella5 , Sofia B. Ahmed 6 , Lesley A. Inker7 , Erin F. Barreto8 , Edouard L. Fu9 , Catherine M. Clase10 & Juan J. Carrero 11 Abstract People living with chronic kidney disease (CKD) often experience multimorbidity and require polypharmacy. Kidney dysfunction can also alter the pharmacokinetics and pharmacodynamics of medications, which can modify their risks and benefts; the extent of these changes is not well understood for all situations or medications. The principle of drug stewardship is aimed at maximizing medication safety and efectiveness in a population of patients through a variety of processes including medication reconciliation, medication selection, dose adjustment, monitoring for efectiveness and safety, and discontinuation (deprescribing) when no longer necessary. This Review is aimed at serving as a resource for achieving optimal drug stewardship for patients with CKD. We describe special considerations for medication use during pregnancy and lactation, during acute illness and in patients with cancer, as well as guidance for the responsible use of over-the-counter drugs, herbal remedies, supplements and sick-day rules. We also highlight inequities in medication access worldwide and suggest policies to improve access to quality and essential medications for all persons with CKD. Further strategies to promote drug stewardship include patient education and engagement, the use of digital health tools, shared decision-making and collaboration within interdisciplinary teams. Throughout, we position the person with CKD at the centre of all drug stewardship eforts. Sections Introduction Burden of medication-related problems Approaches to inform clinical decisions Drug dosing for people with CKD Drug stewardship for safe medication choice Medication review and reconciliation Safe deprescribing Strategies to promote drug stewardship in CKD Conclusions A full list of affiliations appears at the end of the paper. e-mail:
[email protected] Nature Reviews Nephrology | Volume 20 | June 2024 | 386–401 387 Review article but recognize that many of the principles discussed are universal. This Review was developed alongside the Kidney Disease: Improving Global Outcomes 2024 Clinical Practice Guideline for the Evaluation and Management of CKD8 , and readers are therefore encouraged to refer to those guidelines for a practical summary of medication management recommendations for patients with CKD. Burden of medication-related problems Medication-related problems — defined as problems that involve medications that can interfere with patient outcomes — are common in people with CKD1 . Inappropriate medication dosing is reported in nearly 70% of studies in hospital settings and 34% of studies in long-term care settings9 . The risk of an adverse drug event — defined as an unde- sirable effect of a drug — is 3 to 10 times higher for a person with CKD than for an age-matched person without CKD10. Medication-related problems are traditionally classified into nine categories (Supplemen- tary Table 1)11,12. These problems are consistently associated with poor outcomes, including reduced quality of life, increased hospitalization rates, increased mortality and increased health care costs1 . Risk factors for medication-related problems in CKD include comorbid conditions, complex treatment regimens and polyphar- macy, frequent medication changes, the use of medications with a narrow therapeutic window, drug interactions, the involvement of complex care teams and concerns about treatment burden, cost and non-adherence1 . These risks likely increase as GFR declines as a conse- quence of changes in drug pharmacokinetics (that is, the absorption, distribution, metabolism and excretion of the drug). Re-evaluation of the drug and dose appropriateness in this setting is essential to patient safety. Approaches to inform clinical decisions Poor representation of people with CKD in clinical trials has often led to indirect evidence of the benefits and harms of medications in this group of patients2 . When clinical trials have included patients with CKD, the participants are not always representative of the population in whom the medication will be used13; moreover, the size or duration of the trial may not be sufficient to explore the effects of the medica- tion on the risk of progression to kidney failure. Some studies on which our understanding of drug dosing are based are old, originating from eras when polypharmacy was less common, or focus on inpatients, for whom different medications and monitoring protocols are now used14. Fortunately, recognition by medical agencies15 of changes in GFR decline and albuminuria as surrogate end points of kidney failure has facilitated the design of trials to assess the benefit and/or risk of medications on the progression of CKD in a more practical time frame than previously possible. Acute nephrotoxicity events are rarely well characterized in clini- cal trials16,17 despite the fact that 14–26% of acute kidney injury (AKI) events in adults are attributable to medications18–20. Evidence of these and other adverse events often originates from pharmacovigilance systems (such as FDA MedWatch), which rely heavily on the voluntary reporting of cases by patients and health care professionals. How- ever, fewer than 1% of adverse drug reactions21 are detected through spontaneous reporting in pharmacovigilance systems, which may be influenced by external factors such as media interest or safety alerts (a phenomenon known as notoriety bias). As such, pharmacovigilance systems are not sufficient to capture the true incidence of adverse drug reactions and susceptibility factors. Individual patient chart review by pharmacists is a more comprehensive method by which to Key points • Medication reconciliation and review is an essential first step in patient-centred drug stewardship. • Doses should be adjusted according to a patient’s glomerular filtration rate (GFR). Most pharmacokinetic information available is derived from creatinine-based equations, and for many medications, diferent creatinine-based equations to estimate GFR are interchangeable; however, use of equations that utilize both creatinine and cystatin C, or direct measurement of GFR, is advocated when the therapeutic range is narrow and precision is needed. • Drug choice should consider relative and absolute contraindications by GFR, and the increased nephrotoxicity of nephrotoxic medications to people with chronic kidney disease. • Acute illness and fluctuations in GFR should prompt frequent reassessment of GFR and medications; however, the implementation of routine sick-day rules has a substantial cost or opportunity cost, is not supported by evidence and may cause more harm than good. • For people with chronic kidney disease who might become pregnant, education around potential teratogens is important, but people should not be denied medication on the basis that they might become pregnant. • Culturally appropriate, multi-faceted educational and empowerment activities may include written instructional materials and digital technologies. • Health care workers and patients should collaborate to promote more equitable access to evidence-based medications within their countries and globally. Introduction People with chronic kidney disease (CKD) are prescribed, on average, 12 medications per day in addition to non-prescription therapies, creating a complex polypharmacy1 . Evidence of the harms and effec- tiveness of these medications is often indirect, requiring clinical acu- men and interpretation. The exclusion of people with advanced CKD from clinical trials2 creates uncertainty about the generalizability of results. The central role of the kidneys in the elimination of various medications and their metabolites means that many medications require dose adjustments or discontinuation in people with a low glomerular filtration rate (GFR)3 . CKD increases the nephrotoxicity of some medications, and nephrotoxicity is critical to avoid in people with low GFR. Patients with CKD can also be susceptible to the adverse effects of medications, which might further contribute to suboptimal pharmacological management4,5 . Drug stewardship typically describes the responsible use of antimicrobials6 ; however, here we apply this term to describe the effective, safe and sustainable use of medications for people with CKD7 . We discuss key concepts and definitions for effective drug stew- ardship in adults with CKD with a focus on all medications, not just those used in CKD management. We do not cover drug stewardship for children or for patients undergoing kidney replacement therapy,
Nature Reviews Nephrology | Volume 20 | June 2024 | 386–401 388 Review article identify adverse drug reactions; however, this approach is prohibitively expensive and time-consuming for research and/or administrative use. Additional data to guide prescribing strategies are derived from post-marketing surveillance22. Studies of routine care provide, in many cases, the best source of information to guide practice where trials are not available or sufficient, and are a critical source of information on rare harms. For example, diabetic ketoacidosis or Fournier gangrene are serious but rare adverse events of sodium-glucose cotransporter-2 (SGLT2) inhibitors that were identified through observational reports from routine care23,24. Large datasets, innovative systems and innova- tive approaches to evaluating causal inference from observational data have improved the validity of conclusions drawn from such reports25 and represent a critical addendum to classic randomized controlled trials (RCTs; Supplementary Table 2). As a case example, the STOP-ACEi trial26 screened over 14,000 individuals in order to recruit 411 patients with advanced CKD who were then randomized to continue or to stop their angiotensin-converting-enzyme inhibi- tor (ACEi) treatment for 3 years. The trial was designed to confirm or refute long-standing myths of medication-induced harm that origi- nated from biased observational research. The researchers observed no evidence of harm (in that both treatment strategies had simi- lar risks of death or kidney replacement therapy) but a numerically higher proportion of major adverse cardiovascular events (MACE) among those who discontinued treatment. Given the cardioprotective effects of ACEi treatment, the trial was interpreted to support contin- ued renin–angiotensin system inhibitor (RASi) use in these patients. However, 2 years earlier an observational study using the target trial emulation framework from the Swedish Renal Registry27 and involv- ing over 10,000 patients reached similar conclusions, including a clear association between treatment discontinuation and the risk of MACE. Thus, data from trials and routine care have the potential to complement each other and assist health care professionals in making evidence-based decisions. Drug dosing for people with CKD The GFR is considered the best surrogate of renal clearance and is there- fore pivotal in assessing the eligibility and/or dosing of drugs that are cleared by the kidneys. Failure to account for GFR increases the risk of treatment failure and adverse drug events10,28. Biomarkers to guide drug dosing Standard approaches to the dosing and monitoring of medications in patients with CKD are based on serum creatinine concentration. Creatinine is freely filtered at the glomerulus. However, 10–40% of creatinine clearance is derived from proximal tubular secretion. There- fore, measurements of creatinine clearance modestly overestimate true GFR; this difference becomes more pronounced as GFR declines29. Select medications (for example, trimethoprim) interfere with the tubular secretion of creatinine and alter estimated GFR (eGFR) but not measured GFR. Serum creatinine is the terminal by-product of skeletal muscle catabolism. Non-renal factors including diet, sex, body compo- sition (particularly skeletal muscle mass), conditions such as cirrhosis, malnutrition or cachexia30,31 and factors that modify extrarenal elimina- tion can affect serum creatinine concentrations32. In these situations, the creatinine-based eGFR (eGFRcr) may overestimate the true GFR. The recognized limitations of creatinine have led to the identi- fication of alternative filtration markers as adjuncts or alternatives to estimate GFR for drug dosing and monitoring. Cystatin C is the most widely available of these alternatives; however, serum cystatin C concentrations can be affected by adiposity, smoking, hypothyroid- ism, hyperthyroidism, glucocorticoid excess, chronic inflammation and malignancy, independent of changes in the GFR31,33–43. The eGFR from cystatin C in combination with creatinine (eGFRcr-cys) is recom- mended as a supportive test for eGFRcr when eGFRcr is thought to be inaccurate and clinical decisions are impacted by the level of the GFR8 . Consequences of mismatches between eGFR equations Recommendations for medication use and dosing in CKD vary across pharmacopoeia and medication agencies44. Only in the past few years have regulatory agencies required pharmacokinetic studies for approval of a renally cleared drug. However, available pharma- cokinetic studies are generally small, short and conducted in patients with minimal comorbidities45. Many pharmacokinetic studies — both past and present — estimated the GFR using the Cockcroft–Gault (CG) equation for estimating creatinine clearance46. However, that equa- tion was developed prior to the standardization of creatinine assays in a small population of white, male patients from a single centre. More accurate equations were subsequently developed to assess the GFR in people of different sex, age groups, ancestry, ethnicity and weight. A large validation study published in 2022 confirmed that the CG equation is less precise, more biased and less accurate than more contemporary eGFR equations such as the Modification of Diet in Renal Disease, CKD Epidemiology Collaboration (CKD-EPI), Lund Malmo Revised, and European Kidney Function Consortium creatinine-based equations47. Major regulatory agencies now recognize that “any contemporary, widely accepted, and clinically applicable eGFR equation is consid- ered reasonable to assess kidney function in pharmacokinetic stud- ies”10,45. For drugs approved using the CG equation, it is likely that inconsistencies in the development of dosing recommendations will only be resolved with the emergence of new, observational data from post-marketing studies. For example, metformin was originally con- traindicated in people with serum creatinine >1.5 mg/dl (133 μmol/l) because of an increased risk of drug accumulation, lactic acidosis and diabetes-related death48,49. However, dosing by serum creatinine level was inconsistent with best practices, as in the current era, dosing is based on the eGFR. Observational studies subsequently defined dosing thresholds for metformin based on eGFR, identifying an eGFR threshold of <30 ml/min per 1.73m2 as higher risk50,51. Practical challenges also exist in the translation of approved label- ling to real-world practice. Drug dosage labels rarely include informa- tion on which eGFR equation was used in the pharmacokinetic studies used to derive dosing instructions. Although many medications were historically approved based on the CG equation, laboratory report- ing of eGFR typically uses other equations, such as the Modification of Diet in Renal Disease52 or CKD-EPI53,54 equations. Electronic health record and clinical decision support tool calculations enable clinicians to apply local eGFR preferences to dose recommendations without consideration of the original measurement method. For equations based on serum creatinine level (such as the CG, CKD-EPI, Lund-Malmo- Revised and European-Kidney-Function-Consortium equations) the impact of this difference is likely modest, although it is worth noting that the units of CG are usually ml/min and those of the other equa- tions ml/min per 1.73 m2 (ref. 55). Calculation of eGFR using alternative biomarkers such as cystatin C (cystatin-C-based eGFR (eGFRcr-cys) or eGFRcys) and applying it to thresholds based on creatinine-only calcu- lations may result in different drug-dose recommendations. Such an approach might be appropriate if eGFRcr is not an accurate reflection
Nature Reviews Nephrology | Volume 20 | June 2024 | 386–401 389 Review article of GFR. However, better understanding of these differences and their implications is needed, especially as cystatin-C-based measurements become more widely used in clinical practice. eGFR selection for drug-dose adjustments in people with CKD Guidelines recommend that GFR assessment for drug dosing should be individualized and consider the risks and benefits of the therapy in question and the health status of the patient8 . Our suggested approach to drug dosing is to use initial and supportive testing to develop a final assessment of the most likely true GFR (Fig. 1). The initial test for evalu- ating drug dosing in most circumstances will be creatinine-based eGFR (eGFRcr). The eGFRcr is widely available, routinely evaluated as part of a basic metabolic panel, is low cost and has sufficient precision for most medication-dose adjustments. If eGFRcr is expected to be inac- curate, measurement of cystatin C could be considered to calculate eGFRcys alone or in combination with creatinine (eGFRcr-cys), which is the most accurate estimate in most populations studied thus far. An eGFRcys often provides discordant GFR estimates to eGFRcr. An analysis of data from 158,601 adults in Stockholm, Sweden56, found that 1 in 4 patients with same-day testing of creatinine and cystatin C differed in their eGFR estimates by 30% or more, leading to uncertainty about which eGFR estimate to apply. A qualitative study identified a variety of factors — including creatinine, cystatin C, urine output and severity of illness — that influenced the drug-dose decision mak- ing of acute care providers; however, this study also found that many clinicians chose ‘the lowest eGFR number’ to be conservative57. When discrepant eGFR estimates exist, use of a lower value would increase the likelihood of dose reductions. This prioritization of safety over effectiveness may not be clinically appropriate. For example, it may be appropriate in a patient with severe infection, if using an antibiotic with a wide therapeutic window, to use a higher dose. In general, if concern about the appropriateness of eGFRcr has led to measurement of cystatin C, we suggest remeasuring creatinine at the same time and using a combination equation (eGFRcr-cys). Use of both biomarkers in the eGFR equation to some extent manages or mitigates the non-renal determinants of either alone. A 2023 evaluation of the accuracy of eGFR equations versus iohexol clearance identified that concordant eGFRcr and eGFRcys values (within 15 ml/min per 1.73 m2 or 20–30% of each other) resulted in similar accuracy of eGFRcr, eGFRcys and eGFRcr-cys58, and consistent dose recommendations across the equations. However, when eGFRcr and eGFRcys were dis- cordant, eGFRcr-cys was generally more accurate than either eGFRcr or eGFRcys58. A separate study in which eGFRcr-cys was used to individual- ize drug doses led to improved pharmacokinetic and pharmacodynamic target attainment of vancomycin therapy compared with standard approaches using eGFRcr59. If eGFRcr-cys is expected to be inaccurate, or if more accurate assessment of GFR is needed for medication dosing in people with CKD, then a measured glomerular filtration rate (mGFR) using plasma or urinary clearance of exogenous filtration markers should be considered. Ultimately, local resources and turnaround time, and the need for additional visits for blood draws and the time commitment of mGFR impact the ability to apply these non-creatinine alternatives to drug dosing in people with CKD. Normalized and non-normalized eGFR equations Drug-dosing adjustments should be based on the GFR in units of ml/min as the clearance of a medication is related to the absolute clearance level of an individual, not that of another person. The native units for eGFRcr, eGFRcys and eGFRcr-cys for many contemporary equations are normalized to a nominal standard body surface area (BSA) of 1.73 m2 and expressed as ml/min per 1.73 m2 . To re-express the result in non- normalized, absolute terms (ml/min) clinicians should use the follow- ing equation: non-normalized eGFR = eGFR in ml/min/1.73 m2 × [BSA (m2 )/1.73 m2 ]. The greatest impact of this nuance will be in patients with a BSA that is markedly different from 1.73 m2 (ref. 60). Sex and gender Women are less likely than men to be recognized as having CKD61 and thus be at risk of potentially inappropriate prescribing62. Sex differ- ences in drug safety and efficacy in people with CKD are understudied, and very little is known about the optimal assessment of the GFR in people who are transgender63. Compared with men, women are also more likely to report adverse drug reactions to commonly used CKD medications64. It has been postulated that the excess risk of adverse drug reactions in women is explained by overdosing, given that most drugs are prescribed to women and men at the same dose, whereas pharmacokinetic values are often higher in women65. Sex differences in body weight and composition, as well as physiological differences can indeed affect drug metabolism and response; of note, sex hormones seem to mediate at least some of the observed sex-based differences in pharmacokinetics across the life cycle66,67. Interestingly, studies suggest Initial test, eGFRcr No, use eGFRcr Yes, measure cystatin C Consider potential sources of error in eGFRcr and need for more accurate assessment Evaluate discordance between eGFRcr and eGFRcys No, use eGFRb Yes, measure GFR Consider potential sources of error in eGFRcr-cys and need for even more accurate assessment No, use eGFRcr Yes, use eGFRcr-cysa Evaluation of GFR for clinical application 15 30 45 60 90 120 ml/min/1.73 m2 Fig. 1 | Suggested approach to GFR evaluation for drug dosing. This algorithm describes our suggested approach to the evaluation of the glomerular filtration rate (GFR) and how it can be applied to drug dosing, although we acknowledge that facilities for cystatin-C testing might not be widely available. a Consider cystatin-C–based calculations of eGFR (eGFRcys) rather than calculations based on cystatin C in combination with creatinine (eGFRcr-cys) in otherwise healthy populations with decreased creatinine generation due to reduced muscle mass, or decreased creatinine secretion or extra-renal elimination owing to use of specific medications. b Use creatinine-based calculations of eGFR (eGFRcr) or eGFRcr-cys depending on discordance between eGFRcr and eGFRcys.
English