Hyperpolypharmacy in patients with chronic kidney disease and its impact on clinical outcomes

Abstract

Hyperpolypharmacy (≥ 10 daily medications) is frequent in patients with chronic kidney disease (CKD), but its impact remains poorly characterized. This study, based on 3,011 non-dialyzed, non-transplant CKD outpatients from the CKD-REIN cohort (eGFR < 60 mL/min/1.73 m²) aimed to describe drug burden and assess associations between hyperpolypharmacy and adverse outcomes. Drug prescription, kidney function, adverse drug reactions (ADRs), hospitalizations, kidney replacement therapy and deaths before KRT were prospectively recorded over five years. Median age was 69 years and mean eGFR was 34 mL/min/1.73 m². At baseline, 80% of the cohort had polypharmacy (≥ 5 daily medications), and 33% had hyperpolypharmacy. These rates remained stable over time. Diabetes, dyslipidemia, and a history of cardiovascular and respiratory diseases were the main contributors to hyperpolypharmacy status. Hyperpolypharmacy was associated with greater likelihoods of an ADR (hazard ratio (HR) [95% confidence interval (CI)] 1.21 [1.04–1.40]), hospitalization (HR [95%CI] 1.34 [1.18–1.51]) and death before KRT (HR [95%CI] 1.46 [1.17–1.82]). Among patients with eGFR ≥ 30 mL/min/1.73m², hyperpolypharmacy also raised the risk of KRT initiation (HR [95%CI] 1.46 [1.00–2.13]), but not in those with eGFR < 30 (HR [95%CI] 0.94 [0.78–1.14]). These results identify hyperpolypharmacy as a significant concern in CKD and underscore the importance of regular medication reviews to reduce adverse outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-25632-x.

Introduction

Nephrologists deal with some of the most complex patients, due largely to the high prevalences of morbidity and mortality in people with renal conditions. Indeed, patients with chronic kidney disease (CKD) often present several comorbidities (such as hypertension, diabetes, cardiovascular disease and metabolic disorders) that contribute to an elevated medication burden. In the context of CKD, optimal medication therapy management is essential but can be challenging; the decline in the patient’s glomerular filtration rate (GFR) significantly alters the pharmacokinetics and pharmacodynamics of drugs eliminated by both renal and non-renal mechanisms^1–3.

Polypharmacy (defined typically as the daily intake of five or more medications) is highly prevalent among patients with CKD^1,3–7. In a recent meta-analysis, the estimated prevalence of polypharmacy was 80% among patients with CKD overall and 70% among those with stage 3–5 CKD. Moreover, the prevalence of hyperpolypharmacy (defined typically as the daily intake of 10 or more drugs per day) is 46% among patients with CKD overall and 24% among those with stage 3–5 CKD⁴. Factors known to influence polypharmacy in CKD include age, comorbidities, disease severity, and healthcare resource use^1,7–9. However, factors specifically associated with hyperpolypharmacy have been insufficiently investigated.

Although polypharmacy is intended to optimize treatment, it is associated with adverse outcomes^2,4. Polypharmacy is not specific to patients with CKD; it is also common among older adults, where an estimated prevalence of 37% has been reported¹⁰. In this population, polypharmacy has been linked to elevated morbidity, hospitalization, adverse drug reactions (ADRs), and elevated mortality^11,12. Moreover, patients with CKD tend to be older^1,4,5 and are exposed to a greater risk of ADRs than older patients without CKD². However, to the best of our knowledge, adverse outcomes such as ADRs, hospitalization, kidney replacement therapy and death have not been simultaneously evaluated in a CKD population.

Given that (i) most patients with CKD experience polypharmacy and (ii) reducing the number of daily medications to fewer than five may not be feasible in this population, the present study focused on hyperpolypharmacy. Indeed, patients with CKD are typically prescribed six to twelve different medications per day¹ and the estimated median [95% confidence interval (CI)] number of daily prescriptions is 8 [7–9]⁴. Understanding the specific consequences of hyperpolypharmacy is essential because it might put patients at even greater risk of adverse outcomes and might provide a clearer cut-off for mitigating these risks.

The objectives of the present study of non-dialyzed, non-transplant patients with an eGFR < 60 mL/min/1.73 m² being managed by a nephrologist in France were to (i) describe the therapeutic classes prescribed and the prevalence of hyperpolypharmacy during a 5-year period of active follow-up, (ii) identify factors associated with hyperpolypharmacy, and (iii) analyze the impact of hyperpolypharmacy on clinical outcomes like ADRs, hospitalization, kidney failure, and death before KRT.

Methods

Study design and participants

The Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) is a representative, prospective cohort study carried out at 40 nephrology outpatient facilities in France. The inclusion criteria were age 18 years or over, management by a nephrologist, a diagnosis of CKD with an eGFR < 60 mL/min/1.73 m², and no history of chronic dialysis or kidney transplant. From July 2013 to April 2016, 3,033 patients were included and actively followed up for 5 years or until the initiation kidney replacement therapy (KRT), whichever occurred first. Details of the study protocol have been published elsewhere¹³.

All methods were carried out in accordance with relevant guidelines and regulations. The study protocol was approved by the institutional review board at the French National Institute of Health and Medical Research (INSERM; reference: IRB00003888) and registered at ClinicalTrials.gov (NCT03381950). Written informed consent was obtained from all participants prior to inclusion in the study.

In the present analysis, we omitted patients with missing data on drug prescriptions at baseline (n = 8) or on follow-up (n = 14). Hence, the final analysis encompassed 3011 patients.

Study data

Using interviews, medical records, and patient self-questionnaires, trained clinical research assistants (CRAs) collected data at baseline and then at 1-year intervals. The variables included patient characteristics (age, sex, years of full-time education, body mass index (BMI), and medication adherence on the Girerd scale¹⁴), comorbidities (primary kidney disease, a history of cardiovascular or respiratory disease, dyslipidemia, diabetes, acute kidney injury (AKI), gastric bleeding, cirrhosis, cancer, history of depression, past and/or present cancer, the Mini Mental Status Examination (MMSE) score, arthrosis, osteoporosis, and gout) and the frequency of medical consultations with a family physician, nephrologist and/or cardiologist (Supplementary Table 1). Routine laboratory tests were performed in hospital central laboratories and private medical laboratories, as part of the patients’ usual care. We used the 2009 Chronic Kidney Disease Epidemiology Collaboration equation to estimate the GFR¹⁵.

Patients were requested to present all their prescriptions (regardless of the prescribing physician) for drugs taken in the three months prior to the baseline interview with a CRA and at each yearly of follow-up visit (i.e. the number of medications was determined at each annual time point based on drugs taken during the preceding three months, allowing for a consistent assessment of medication use over time). The CRAs used a study-specific electronic case report form with an integrated Anatomical Therapeutic and Chemical (ATC) thesaurus to record standardized drug class codes. For each prescription, the prescription period, trade name, ATC classification, unit dose, daily dosage, and pharmaceutical formulation were documented. Over-the-counter (OTC) medications were not recorded.

Depending on the number of prescription drugs taken daily, patients were categorized into two groups: no hyperpolypharmacy (< 10 drugs/d) and hyperpolypharmacy (≥ 10 drugs/d). Secondary analyses were performed by categorizing the patients into two other groups: no polypharmacy (< 5 drugs/d) and polypharmacy (≥ 5 drugs/d).

Study outcomes

Several outcomes were prospectively recorded throughout the follow-up period.

Kidney replacement therapy

KRT was defined as the initiation of maintenance dialysis or preemptive kidney transplantation. Information about KRT was obtained from medical records, patient interviews, and/or linkage to the French national Renal Epidemiology and Information Network registry (REIN registry).

Adverse drug reactions

An ADR is defined as “an appreciably harmful or unpleasant reaction, resulting from an intervention related to the use of a medicinal product, which predicts hazard from future administration and warrants prevention or specific treatment, or alteration of the dosage regimen, or withdrawal of the product”¹⁶. An ADR was considered to be serious when the patient’s outcome was death or a life-threatening situation, hospitalization (initial or prolonged), disability, permanent damage, or another important medical event¹⁷.

Briefly, and as described previously, we used a study-specific electronic case report form to collect data on ADRs that occurred during the five-year follow-up period¹⁸. ADRs were identified through patient interviews and/or the inspection of medical records and hospital reports. Each drug prescribed to the patient at the time of each ADR was documented and reviewed by two pharmacists, who classified the ADR and determined whether it was likely to be related to any of the patient’s drugs. Serious ADRs were evaluated further for imputability (using Bégaud’s method¹⁹ and the Naranjo algorithm²⁰) and severity by a committee of pharmacologists.

Hospitalization

Hospital stays were documented prospectively by CRAs using medical records, hospital reports, or participant interviews. Based on the hospital admission and discharge dates, the CRAs noted the length of the hospital stay. All types and lengths of hospital stay were included in the analyses, except for scheduled hospitalizations. A short hospital stay was defined as a stay of less than one night.

Death before KRT

Deaths were ascertained from death certificates, hospital records, reports by family members, and linkage to the national vital status registry. Only deaths occurring before the initiation of KRT were included in the analysis, as the start of KRT marked the end of active follow-up.

Patients were censored at the occurrence of a competing event (death and/or KRT depending on the studied outcome), the completion of their five-year follow-up, or upon loss to follow-up.

Statistical analyses

Baseline characteristics and prescribed drugs were described for all participants and for patients with and without hyperpolypharmacy. The number of drugs taken daily was evaluated as a function of the baseline eGFR (< 30 or ≥ 30 mL/min/1.73 m²). Variables were expressed as the mean ± standard deviation (SD), the median [interquartile range (IQR)], or the frequency (percentage). We used a Sankey plot to describe the change in the number of drugs taken daily over the follow-up period according 3 categories: no polypharmacy: < 5, polypharmacy: 5–9; hyperpolypharmacy: ≥ 10. Baseline drug prescriptions were described by hyperpolypharmacy status and compared using standardized mean differences (SMD).

A logistic regression model was used to assess factors associated with hyperpolypharmacy at baseline. The covariates included in the model were preselected in a review of the literature. Variables with a p-value < 0.20 in the crude model were included in the multivariate analyses. Age and sex was systematically included as a covariate in the adjusted models. The variables considered for inclusion in the model included sociodemographic factors (age, sex, educational level, and treatment compliance), clinical and laboratory variables (the urinary albumin-to-creatinine ratio (ACR), anemia, C-reactive protein (CRP), and eGFR), the patient’s medical history (obesity, cardiovascular and respiratory diseases, arthrosis, depression, possible neurocognitive disorders, diabetes, dyslipidemia, and systolic blood pressure (SBP)) and the frequency of consultations with a physician (family physicians, nephrologists, and/or cardiologists) in the 12 months prior to study inclusion. The odds ratio (OR) and the corresponding 95% confidence interval (CI) were calculated for each of the included variables.

Incidence rates (IRs) and the corresponding 95% CIs were calculated for the first ADR, the first hospital admission, KRT, or death before KRT. A cause-specific Cox proportional hazards model was used to examine the association between hyperpolypharmacy at baseline and the incidences of each clinical outcome. Baseline covariates included in the model were preselected in a review of the literature, and variables with p < 0.20 in the crude model were included in the multivariate analyses. The selected covariates depended on the study outcome (Supplementary Table 2). For ADRs and hospitalization, the data in the Cox proportional hazards model were censored at the date of death, KRT, or last follow-up. For KRT, the data were censored at the date of death or last follow-up. Lastly, for death before KRT, the data were censored at the date of KRT or last follow-up. In each model, patients who presented the outcome of interest on the day of inclusion in the study were excluded (Fig. 1). We tested for interactions (i) between hyperpolypharmacy and sex, (ii) between hyperpolypharmacy and age, and (iii) between hyperpolypharmacy and eGFR. When a significant interaction was detected, the models were stratified accordingly. The proportional hazards assumption was assessed by examining the Schoenfeld residuals.

Fig. 1.

Study flowchart.

We performed two sensitivity analysis considering serious ADRs and short hospital stays as outcomes. To address potential early-event bias, we conducted a sensitivity analysis excluding events within the first 6 months.

Lastly, to handle missing data, we used multiple imputation by chained equations with 20 iterations and (i) 40 datasets for the model with an ADR as the outcome, (ii) 45 datasets for associated factors, kidney failure, or hospitalization as the outcome, and (iii) 60 datasets for death before KRT as the outcome. The imputation models included all the covariates used in the adjusted analyses. A separate imputation model was built for each analysis. The missing data patterns were consistent with a missing-at-random mechanism.

The threshold for statistical significance was set to p < 0.05. Statistical analyses were performed using R software (version 4.3.0)²¹.

Results

The results of this cohort study are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines²².

Baseline characteristics

At baseline, the median [IQR] age was 69 [60–76], and the majority of the participants (66%) were males. The mean ± SD eGFR was 34.1 ± 13.1 mL/min/1.73 m² (Table 1). When compared with patients taking fewer than 10 drugs per day, patients taking 10 or more drugs per day (i.e. those with hyperpolypharmacy) were older and had a lower eGFR and higher prevalences of poor compliance, anemia, BMI, cardiovascular and respiratory diseases, diabetes, and dyslipidemia (Table 1).

Table 1.

Baseline characteristics of the patients in the CKD-REIN cohort, stratified by hyperpolypharmacy status.

		Hyperpolypharmacy		Missing data (n = 3011)
	Whole cohort (n = 3011)	< 10 drugs daily (n = 2006)	≥ 10 drugs daily (n = 1005)
Sociodemographic factors
Age (years)	69 [60–76]	67 [58–75]	71 [65–77]	-
Male	66%	66%	64%	-
Educational level				1.2%
< 9 years	15%	12%	20%
9–11 years	49%	47%	53%
> 11 years	36%	41%	27%
Poor compliance	62%	56%	74%	0.9%
Clinical and laboratory variables
eGFR (mL/min/1.73 m2)	34.1 ± 13.1	35.4 ± 13.3	31.6 ± 12.4	-
Urinary albumin-to-creatinine ratio				11.5%
A1 (< 30 mg/g)	28%	31%	23%
A2 (30–300 mg/g)	35%	35%	34%
A3 (> 300 mg/g)	37%	34%	44%
Anemia	42%	35%	55%	5.1%
Albuminuria (g/L)	40.8 [38.1–43.1]	41.2 [38.9–43.5]	39.9 [37.3–42.3]	8.6%
C-reactive protein (mg/L)	2.6 [1.2–6.0]	2.3 [1.0–5.2]	3.3 [1.5–7.9]	15.8%
Systolic blood pressure (mmHg)	140 [129–153]	138 [127–151]	142 [130–157]	0.6%
Medical history
Body mass index (kg/m2)	27.9 [24.6–31.8]	27.0 [24.1 – 30.4]	30.2 [26.3 – 34.9]	2.4%
Obesity (≥ 30 kg/m2)	35%	27%	52%
Cardiovascular disease	53%	43%	73%	1.4%
Dyslipidemia	74%	66%	89%	0.2%
Diabetes	43%	31%	67%	0.2%
Respiratory disease	24%	16%	39%	2.2%
MMSE score (out of 30)	28 [25–29]	28 [26–29]	27 [24–29]	0.9%
< 24	13%	10%	19%
Arthrosis	20%	16%	26%	2.1%
Medical consultations*
Family physician				14.4%
No consultations	2%	2%	2%
1 or 2 consultations	13%	15%	9%
3 consultations or more	85%	83%	89%
Nephrologist				13.8%
No consultations	2%	2%	1%
1 or 2 consultations	63%	66%	57%
3 consultations or more	35%	31%	42%
Cardiologist				17.1%
No consultations	32%	40%	15%
1 or 2 consultations	57%	52%	67%
3 consultations or more	11%	7%	18%
Number of drugs per day	8 [5–10]	6 [4–8]	12 [10–13]

eGFR, estimated glomerular filtration rate; MMSE, Mini Mental State Examination.

* during the year before inclusion in the CKD-REIN study.

The data are reported as the median [interquartile range], the mean ± SD, or a percentage.

Hyperpolypharmacy

The median [IQR] number of daily prescribed medications at baseline was 8 [5–10], 80% of the patients took 5 or more drugs per day, and 33% took 10 or more drugs per day. Most therapeutic classes were more commonly prescribed to patients with hyperpolypharmacy than to other patients. Notably, 58% of patients with hyperpolypharmacy were taking drugs for gastro-intestinal acid-related disorders, and 46% were taking antigout preparations (allopurinol, febuxostat and colchicine) (Table 2). Over the 5-year follow-up period, most patients remained in their baseline polypharmacy category (no polypharmacy/polypharmacy/hyperpolypharmacy; Fig. 2). Some of the patients moved from one category to another; this change was generally towards higher medication counts and illustrated the progressive increase in the treatment burden in CKD. Between baseline and year 5, the proportion of patients with hyperpolypharmacy increased from 33 to 42%, and the proportion of patients with fewer than 5 drugs per day decreased from 20 to 15% (Fig. 2).

Table 2.

Baseline drug prescriptions for patients in the CKD-REIN cohort, stratified by hyperpolypharmacy status.

	Hyperpolypharmacy		SMD
	< 10 N = 2006	≥ 10 N = 1005
Drugs for acid-related disorders	460 (23%)	579 (58%)	0.76
Drugs used in diabetes	458 (23%)	626 (62%)	0.87
Insulins and analogues	221 (11%)	422 (42%)	0.75
Oral antidiabetic drugs	325 (16%)	387 (39%)	0.52
Glucagon-like peptide-1 analogues	17 (0.8%)	28 (2.8%)	0.15
Mineral supplements	225 (11%)	253 (25%)	0.37
Vitamin D	859 (43%)	587 (58%)	0.32
Antianemic	201 (10%)	269 (27%)	0.44
Iron preparations	126 (6.3%)	177 (18%)	0.34
Erythropoiesis-stimulating Agents	92 (4.6%)	139 (14%)	0.32
Antithrombotic agents	801 (40%)	779 (78%)	0.83
Platelet aggregation inhibitors (excluding heparin)	635 (32%)	600 (60%)	0.59
Anticoagulants	193 (9.6%)	244 (24%)	0.40
Cardiac therapy	170 (8.5%)	259 (26%)	0.47
Cardiac glycosides	6 (0.3%)	23 (2.3%)	0.18
Antiarrhythmics. class I and III	111 (5.5%)	133 (13%)	0.27
Cardiac stimulants (excluding cardiac glycosides)	4 (0.2%)	4 (0.4%)	0.04
Vasodilators used in cardiac diseases	36 (1.8%)	98 (9.8%)	0.35
Other cardiac preparations	21 (1.0%)	30 (3.0%)	0.14
Antihypertensives	1826 (91%)	988 (98%)	0.33
Diuretics	878 (44%)	765 (76%)	0.70
Beta-blocking agents	659 (33%)	605 (60%)	0.57
Calcium channel blockers	825 (41%)	625 (62%)	0.43
Agents acting on the renin-angiotensin system	1480 (74%)	806 (80%)	0.15
Other antihypertensives	216 (11%)	299 (30%)	0.49
Lipid-modifying agents	1081 (54%)	821 (82%)	0.62
Urologicals	176 (8.8%)	168 (17%)	0.24
Corticosteroids for systematic use	107 (5.3%)	111 (11%)	0.21
Thyroid therapy	171 (8.5%)	191 (19%)	0.31
Calcium balance drugs (cinacalcet)	4 (0.2%)	6 (0.6%)	0.06
Antibacterials for systematic use	81 (4.0%)	101 (10%)	0.24
Immunosuppressants	86 (4.3%)	81 (8.1%)	0.16
Antiinflammatory and antirheumatic products	24 (1.2%)	19 (1.9%)	0.06
Antiinflammatory and antirheumatic products. non-steroids	24 (1.2%)	19 (1.9%)	0.06
Antigout preparations	558 (28%)	465 (46%)	0.39
Osteoporosis drugs	30 (1.5%)	22 (2.2%)	0.05
Analgesics	305 (15%)	412 (41%)	0.60
Opioids	63 (3.1%)	136 (14%)	0.38
Other analgesics and antipyretics	252 (13%)	333 (33%)	0.51
Antimigraine preparations	5 (0.2%)	3 (0.3%)	0.01
Antiepileptics	80 (4.0%)	107 (11%)	0.26
Psycholeptics	202 (10%)	296 (30%)	0.50
Antipsychotics	80 (4.0%)	107 (11%)	0.26
Anxiolytics	127 (6.3%)	182 (18%)	0.37
Hypnotics and sedatives	89 (4.4%)	150 (15%)	0.36
Psychoanaleptics	119 (5.9%)	120 (12%)	0.21
Drugs for obstructive airways diseases	74 (3.7%)	152 (15%)	0.40
Antihistaminic	41 (2.0%)	79 (7.9%)	0.27
All other therapeutic products
Potassium binding resins	177 (8.8%)	208 (21%)	0.34
Phosphate binders	45 (2.2%)	66 (6.6%)	0.21
Unclassified ATC
Sodium bicarbonate	49 (2.4%)	26 (2.6%)	0.01

A difference between groups was deemed to exist when Standardized Mean Difference (SMD) > 0.1

Fig. 2.

Changes in the number of daily prescription medications over the 5-year follow-up period.

Note: The Sankey plot describes the number of prescription medications taken daily during the 5-year follow-up period. The bars depict the distribution of patients receiving less than 5 drugs, 5 to 9 drugs, or 10 and more drugs per day, for each year of follow-up. The connections between the bars represent the flow of patients transitioning from one status to another.

Factors associated with polypharmacy at baseline

Age, sex, poor compliance, obesity, anemia, low eGFR, cardiovascular and respiratory diseases, arthrosis, possible neurocognitive disorders, diabetes, and dyslipidemia were associated with a greater likelihood of hyperpolypharmacy (Table 3). Seeing a nephrologist or cardiologist more than three times in the year before study inclusion was also associated with a greater likelihood of hyperpolypharmacy (Table 3).

Table 3.

Factors associated with hyperpolypharmacy (≥ 10 daily drugs) at baseline.

	Unadjusted model		Adjusted model
	OR [95%CI]	p-value	OR [95%CI]	p-value
Sociodemographic factors
Age (years)	1.03 [1.03–1.04]	< 0.001	1.01 [1.00–1.02]	0.003
Sex
Male	Ref		Ref
Female	1.11 [0.95–1.31]	0.18	1.52 [1.24–1.87]	< 0.001
Educational level
< 9 years	Ref		Ref
9–11 years	0.68 [0.55–0.85]	0.001	0.95 [0.72–1.25]	0.71
≥ 12 years	0.40 [0.31–0.50]	< 0.001	0.85 [0.62–1.15]	0.28
Compliance
Good	Ref		Ref
Bad	2.20 [1.86–2.59]	< 0.001	1.94 [1.60–2.36]	< 0.001
Clinical and laboratory variables
Body mass index
< 30 kg/m2	Ref		Ref
≥ 30 kg/m2	2.83 [2.41–3.32]	< 0.001	1.45 [1.19–1.76]	< 0.001
Urinary albumin-to-creatinine ratio
A1 (< 30 mg/g)	Ref		Ref
A2 (30–300 mg/g)	1.30 [1.05–1.61]	0.02	1.06 [0.82–1.37]	0.65
A3 (> 300 mg/g)	1.80 [1.47–2.20]	< 0.001	1.16 [0.89–1.52]	0.27
Anemia	2.29 [1.96–2.68]	< 0.001	1.89 [1.55–2.30]	< 0.001
C-reactive protein (mg/L)	1.02 [1.01–1.02]	< 0.001	1.00 [0.99–1.01]	0.42
eGFR: a decrease of 5 mL/min/1.73 m2	1.12 [1.09–1.16]	< 0.001	1.06 [1.02–1.10]	0.01
Systolic blood pressure (mmHg)	1.01 [1.01–1.01]	< 0.001	1.00 [1.00–1.01]	0.82
Medical history
Cardiovascular disease	3.48 [2.95–4.11]	< 0.001	2.16 [1.76–2.65]	< 0.001
Respiratory disease	3.28 [2.75–3.91]	< 0.001	2.17 [1.77–2.68]	< 0.001
Arthrosis	1.75 [1.45–2.11]	< 0.001	1.37 [1.09–1.73]	0.01
Depression	1.43 [1.08–1.89]	0.01	1.23 [0.88–1.73]	0.23
Possible neurocognitive disorders	2.01 [1.67–2.41]	< 0.001	1.33 [1.05–1.68]	0.02
Diabetes	4.52 [3.84–5.32]	< 0.001	2.71 [2.24–3.29]	< 0.001
Dyslipidemia	4.41 [3.53–5.50]	< 0.001	2.62 [2.03–3.38]	< 0.001
Frequency of physician consultations (per year)
Family physician
< 3	Ref		Ref
≥ 3	1.73 [1.35–2.22]	< 0.001	1.11 [0.82–1.49]	0.51
Nephrologist
< 3	Ref		Ref
≥ 3	1.57 [1.33–1.86]	< 0.001	1.34 [1.07–1.67]	0.01
Cardiologist
< 3	Ref		Ref
≥ 3	2.71 [2.12–3.48]	< 0.001	1.69 [1.25–2.29]	0.001

OR, odds ratio; CI, confidence interval.

The presence of a possible neurocognitive disorder was defined as a Mini Mental State Examination score of less than 24 out of 30. In the present analysis, the reference for comparison with the hyperpolypharmacy group was the group of patients taking less than 10 drugs daily.

Impact of hyperpolypharmacy on clinical outcomes

Adverse drug reactions, hospitalization and death before KRT

During a median follow-up of 4.7 years, 964 patients experienced ADR, including 386 in the hyperpolypharmacy group (Table 4). A total of 1,418 patients had a first hospitalization, with 601 in the hyperpolypharmacy group (median follow-up: 4.9 years). In addition, 545 patients died before KRT, of whom 265 were in the hyperpolypharmacy group (median follow-up: 4.9 years).

Table 4.

Number of events and the incidence rate for clinical outcomes and hyperpolypharmacy (≥ 10 drugs per day).

	ADR N = 3004	Hospitalization N = 3006	Death before KRT N = 3011	Kidney replacement therapy N = 3008
				eGFR < 30 mL/min/1.73 m2	eGFR ≥ 30 mL/min/1.73 m2
Median [IQR] follow-up	4.7 [4.7–4.8]	4.9 [4.8–4.9]	4.9 [4.9–4.9]	4.9 [4.9–5.0]	5.0 [5.0–5.0]
Incidence rate (per 100 person-years)	10.1 [9.5–10.7]	16.1 [15.3–17.0]	4.6 [4.2–5.0]	13.9 [12.8–15.0]	2.3 [2.0–2.7]
Number of events	964	1418	545	589	176
Number of events in the hyperpolypharmacy group	386	601	265	232	64
Unadjusted HR [95%CI] for hyperpolypharmacy	1.61 [1.41–1.83]	1.92 [1.73–2.13]	2.31 [1.95–2.73]	1.12 [0.95–1.33]	1.77 [1.30–2.42]
Adjusted HR* [95%CI] for hyperpolypharmacy	1.21 [1.04–1.40]	1.34 [1.18–1.51]	1.46 [1.17–1.82]	0.94 [0.78–1.14]	1.46 [1.00–2.13]

ADR, adverse drug reaction; eGFR, estimated glomerular filtration rate; IQR, interquartile range; HR, hazard ratio; CI, confidence interval.

* The ADR model was adjusted for age, sex, compliance, educational level, (eGFR (per 5 mL/min/1.73m² decrease), urinary albumin-to-creatinine ratio (ACR) with log transformation, anemia, albumin, body mass index (BMI), cardiovascular disease, systolic blood pressure (SBP), diabetes, acute kidney injury (AKI), and consultation with a nephrologist.

The hospitalization model was adjusted for age, sex, educational level, compliance, living alone, eGFR (per 5 mL/min/1.73m² decrease), log ACR, anemia, albumin, BMI, cardiovascular disease, SBP, diabetes, AKI, dyslipidemia, respiratory disease, gastric bleeding, cirrhosis, depression, possible neurocognitive disorders, arthrosis, osteoporosis, and gout.

The death before KRT model was adjusted for age, sex, educational level, compliance, eGFR (per 5 mL/min/1.73m²), log ACR, anemia, albumin, calcium, phosphate, C-reactive protein (CRP), BMI, cardiovascular disease, diabetes, AKI, polycystic kidney disease, respiratory disease, gastric bleeding, cirrhosis, cancer, possible neurocognitive disorders, arthrosis, drugs for acid-related disorders, antithrombotic agents, antihypertensive (diuretics, beta blocking agent, calcium channel blockers, agents acting on the renin-angiotensin system), and lipid-modifying agents.

The KRT model was adjusted for age, sex, eGFR (per 5 mL/min/1.73m²), log ACR, calcium, phosphate, bicarbonate, albumin, polycystic kidney disease, SBP and diabetes.

After adjustments for sociodemographic factors, laboratory variables, comorbidities, and the frequency of nephrologist consultations (Supplementary Table 2), hyperpolypharmacy was associated with higher risks of ADR occurrence (hazard ratio (HR) [95%CI] 1.21 [1.04–1.40]), hospitalization (HR [95%CI] 1.34 [1.18–1.51]) and death before KRT (HR [95%CI] 1.46 [1.17–1.82]) (Table 4, Fig. 3, Supplementary Table 4–6). Similar trends were observed for polypharmacy (Supplementary Fig. 1, Supplementary Table 3).

Fig. 3.

Adjusted hazard ratios for ADR, hospitalization, death before KRT and hyperpolypharmacy (≥ 10 drugs/day). HR, hazard ratio; CI, confidence interval. P-value: NS, non-significant (≥ 0.05); < 0.05*, ≤ 0.01** ; ≤ 0.001.

Note: The ADR model was adjusted for age, sex, compliance, educational level, (eGFR (per 5 mL/min/1.73m² decrease), urinary albumin-to-creatinine ratio (ACR) with log transformation, anemia, albumin, body mass index (BMI), cardiovascular disease, systolic blood pressure (SBP), diabetes, acute kidney injury (AKI), and consultation with a nephrologist. The hospitalization model was adjusted for age, sex, educational level, compliance, living alone, eGFR (per 5 mL/min/1.73m² decrease), log ACR, anemia, albumin, BMI, cardiovascular disease, SBP, diabetes, AKI, dyslipidemia, respiratory disease, gastric bleeding, cirrhosis, depression, possible neurocognitive disorders, arthrosis, osteoporosis, and gout. The death before KRT model was adjusted for age, sex, educational level, compliance, eGFR (per 5 mL/min/1.73m²), log ACR, anemia, albumin, calcium, phosphate, C-reactive protein (CRP), BMI, cardiovascular disease, diabetes, AKI, polycystic kidney disease, respiratory disease, gastric bleeding, cirrhosis, cancer, possible neurocognitive disorders, arthrosis, drugs for acid-related disorders, antithrombotic agents, antihypertensive (diuretics, beta blocking agent, calcium channel blockers, agents acting on the renin-angiotensin system), and lipid-modifying agents.

Kidney replacement therapy

In the analysis of KRT outcome, a statistically significant interaction between hyperpolypharmacy and eGFR was observed (p for interaction = 0.001). Accordingly, subgroup analyses were performed on patients with an eGFR ≥ 30 and those with an eGFR < 30 mL/min/1.73 m². During the median follow-up period of 4.9 years for patients with an eGFR < 30 mL/min/1.73m² and 5.0 years for patients with an eGFR ≥ 30 mL/min/1.73 m², respectively 589 and 176 patients started KRT (Table 4). Among patients with an eGFR ≥ 30 mL/min/1.73 m² at baseline, the HR was 1.46 [95% CI: 1.00–2.13]—suggesting a trend towards a higher risk. In contrast, the risk of kidney failure associated with hyperpolypharmacy was not statistically significant among patients with an eGFR < 30 mL/min/1.73 m² at baseline (HR [95%CI] 0.94 [0.78–1.14]) (Fig. 4, Supplementary Table 7). Again, similar trends were observed for polypharmacy (Supplementary Fig. 2).

Fig. 4.

Adjusted hazard ratios for kidney replacement therapy and hyperpolypharmacy (≥ 10 drugs/day), as a function of the eGFR at baseline. HR, hazard ratio; CI, confidence interval; eGFR, estimated glomerular filtration rate. P-value: NS, non-signifiant (≥ 0.05); < 0.05*, ≤ 0.01** ; ≤ 0.001.

Note: A statistically significant interaction between hyperpolypharmacy and eGFR was found (p for interaction = 0.001). The KRT model was adjusted for age, sex, eGFR, log ACR, albumin, calcium, phosphate, bicarbonate, polycystic kidney disease, diabetes, and systolic blood pressure.

Sensitivity analysis

When considering the 964 patients with a first documented ADR, 358 of the reactions were classified as serious. Hyperpolypharmacy was also associated with a greater likelihood of a first serious ADR (HR [95%CI] 1.35 [1.06–1.71]) (Supplementary Fig. 3).

When considering the 1418 patients with a first hospitalization, 574 of the stays were short (< 1 night). In the analysis of the short hospital stay outcome, a statistically significant interaction between hyperpolypharmacy and eGFR was observed (p for interaction = 0.01). Accordingly, we analyzed two subgroups of patients: those with an eGFR ≥ 30 mL/min/1.73 m² and those with an eGFR < 30 mL/min/1.73 m². Hyperpolypharmacy was again associated with a significantly increased risk of short hospital stay in patients with an eGFR ≥ 30 (HR [95%CI] 1.43 [1.09–1.87]), but not in those with eGFR < 30 mL/min/1.73 m² (HR [95%CI] 0.98 [0.72–1.34]) (Supplementary Fig. 3).

Results remained consistent in the sensitivity analysis excluding events within the first 6 months (Supplementary Table 8).

Discussion

We found that polypharmacy and hyperpolypharmacy were highly prevalent in a cohort of non-dialyzed, non-transplant CKD patients with an eGFR < 60 mL/min/1.73 m². Diabetes, dyslipidemia, and a history of cardiovascular and respiratory diseases were the main contributors to hyperpolypharmacy status. Furthermore, hyperpolypharmacy was associated with a greater likelihood of ADRs, hospitalization, kidney replacement therapy, and death before KRT.

Polypharmacy is common in patients with CKD^1,4,5,8. In the present CKD-REIN study, 80% of the cohort were taking 5 or more drugs per day, and 33% were taking 10 or more drugs per day. Our findings are in line with previous studies. In the Renal Risk in Derby study ⁵ of 1,741 people with CKD stage 3, the prevalences of polypharmacy and hyperpolypharmacy were respectively 59% and 11%. Similarly, in a German study of a cohort of 5,217 patients receiving nephrology care, the prevalences of polypharmacy and hyperpolypharmacy were respectively 80% and 20%^8. Lastly, in Oosting et al.’s meta-analysis,⁴ the estimated prevalences of polypharmacy and hyperpolypharmacy in patients with moderate-to-severe CKD were respectively 70% and 24%. Furthermore, we found that the proportions of patients with polypharmacy and hyperpolypharmacy remained stable during the follow-up period. To our knowledge, only one study in CKD has examined the evolution of polypharmacy, reporting a slight decrease in its prevalence over time (from 80 to 76%)⁸.

Given that most patients with CKD require multiple medications to manage hypertension, diabetes, cardiovascular disease, and other complications, reducing the number of prescription drugs to below the polypharmacy threshold of five may not be a realistic goal. We therefore focused our analyses on hyperpolypharmacy (i.e. the prescription of 10 or more daily drugs). In the present study, comorbidities such as anemia, cardiovascular and respiratory disease, diabetes, dyslipidemia, and arthrosis were associated with a greater likelihood of hyperpolypharmacy. Older age and lower eGFR were both associated with a higher medication burden, likely reflecting the greater number of comorbidities observed in these populations. Similar observations were made in the German CKD cohort, where age, BMI, diabetes, cardiovascular disease, hypertension, and dyslipidemia were linked to polypharmacy, however, hyperpolypharmacy was not reported in this cohort⁸.

There is a growing body of evidence of sex-based differences in kidney health and disease;¹ conditions such as anemia, thyroid disease, anxiety, and arthrosis are more prevalent among women than among men^23–26. In the present study, the likelihood of hyperpolypharmacy was greater in women than in men.

Patients who consulted a nephrologist or cardiologist more than three times in the year preceding their inclusion in the present study had a higher probability of taking 10 or more drugs per day. This may reflect a more severe disease and/or more comorbidities, which require frequent consultations with a specialist physician. Given the greater number of comorbidities, patients with CKD are more likely to be managed by several physicians^1,5,10. This situation might result in patients receiving a greater number of prescriptions drugs.

Our data revealed a significant association between hyperpolypharmacy and several clinical outcomes, including ADRs, hospitalization, KRT initiation, and death before KRT. Indeed, 32% of the patients in the CKD-REIN cohort experienced a first ADR during the 5-year follow-up period. We found that hyperpolypharmacy was associated with a greater likelihood of ADRs and serious ADRs. Although these associations are not unexpected (given that hyperpolypharmacy is linked to a greater risk of drug interactions and medication errors), the findings of Oosting et al.’s meta-analysis highlighted that only two cohorts, including the CKD-REIN cohort, have evaluated this topic in cohorts of patients with CKD⁴. Our observations are in line with Corsonnello et al.’s report of a link between polypharmacy and the occurrence of ADRs in 363 inpatients with diabetes and CKD²⁷.

In the present study, we found that hyperpolypharmacy was associated with a 1.3-fold greater risk of hospitalization. These findings are in line with previous studies. In a large Korean study²⁸ including 1.3% of patients with CKD, polypharmacy was associated with a higher risk of hospitalization (HR [95%] 1.16 [1.6–1.7]). In Oosting et al.’s meta-analysis,⁴ polypharmacy was associated with a greater risk of hospitalization (pooled risk ratio [95%CI] 1.12 [1.02–1.23]).

Hyperpolypharmacy was also linked to CKD progression (defined as KRT initiation) in our cohort but was only statistically significant among individuals with an eGFR above 30 mL/min/1.73 m² at baseline. In patients with eGFR < 30 mL/min/1.73 m², the lack of significant association may reflect the influence of factors beyond hyperpolypharmacy, with greater impact in this high-risk group for KRT. In the Fukushima CKD cohort²⁹ polypharmacy and hyperpolypharmacy were significantly associated with a higher risk of kidney failure requiring KRT (HR [95%CI] 2.3 [1.0–5.2] and 2.8 [1.2–6.7], respectively), when compared with patients taking fewer than 5 drugs per day. Oosting et al.’s meta-analysis⁴ further showed an association between CKD progression and polypharmacy (HR [95%CI] 1.4 [1.0–1.9]).

The medication burden is a well-established risk factor for mortality in older adults²⁸. Our results were in line with the literature data, and we observed that hyperpolypharmacy was associated with a 1.5-fold greater risk of death before KRT. As discussed above, hyperpolypharmacy may reflect more serious disease and the management of multiple comorbidities and may result in more frequent medications errors and drug-drug interactions¹⁸.

Deprescribing has been defined as “discontinuing drugs that are either potentially harmful or no longer required”^30. Indeed, deprescribing has been shown to reduce the likelihood of falls, hospitalization, and mortality in older adults³¹. In the present analysis, we deeply described medications types in a CKD population, with a view to identifying those that could be deprescribed. Although deprescription in CKD is a critical aspect of managing patients, it is not always possible or appropriate for each medication. Rather than simply reducing the number of drugs, a thorough medication review is essential for identifying both inappropriate prescriptions and potential gaps in therapy. Whereas certain medications (such as statins and renin-angiotensin system inhibitors) are often under-prescribed, other drugs (such as proton pump inhibitors, uric-acid-lowering agents in asymptomatic hyperuricemia, benzodiazepines, and anticholinergics) may pose risks in patients with CKD and should be considered for deprescription. These medications can contribute to kidney damage, cognitive decline, or other adverse outcomes, especially as kidney function worsens^32–34. Indeed, just over a third of the patients in our study cohort and 58% of those with hyperpolypharmacy used a drug for a gastro-intestinal acid-related disorder at baseline. These drugs are frequently prescribed inappropriately and have been associated with higher risks of acute kidney injury, disease progression, and mortality in patients with CKD³³. Furthermore, only 45% of the patients included in the CKD-REIN study taking uric-acid-lowering agents at baseline had a documented history of gout in their medical records³². Beyond medication reviews and (when feasible) deprescribing, strategies such as enhanced patient education and optimized management of uncontrolled risk factors should be considered, particularly when regimen simplification is not possible. Such an approach helps ensure that patients receive the most appropriate treatment while minimizing risks.

The present study had several strengths. Firstly, it is based on a large, prospective cohort of non-dialyzed, non-transplant patients with CKD with an eGFR < 60 mL/min/1.73 m² managed by nephrologists in routine care across France. Secondly, we provided a detailed, longitudinal description of the medication burden over time in this population, using repeated and standardized assessments. Lastly, we evaluated the association between medication burden and multiple clinical outcomes, including ADRs and hospitalizations. These two outcomes are rarely reported in studies on polypharmacy⁴, despite their clinical relevance in patients with CKD.

Our study had some limitations. Firstly, we analyzed prescription medications only and not OTC medications. In fact, some OTC medications are not recommended in patients with CKD¹ and so the number of drugs taken daily was probably underestimated. Secondly, the number of ADRs was possibly underestimated (particularly for non-hospitalized patients), due to memory bias or underreporting. Although patients are likely to report ADRs with significant negative impacts on their health, some hospital stays and drug effects may go unnoticed. However, the prospective process of documenting ADRs probably helped to limit this potential underestimation. Thirdly, our study focused on patients with CKD being managed by a nephrologist, and so the results cannot be generalized to all patients with an eGFR < 60 mL/min/1.73 m². It is possible that nephrologists manage the treatment regimens of patients with CKD with great caution, given the complexity of these cases. Lastly, our study’s observational design prevented us from distinguishing between the effects of specific medications and the effects of the underlying comorbidities that lead to polypharmacy. Hence, associations with hyperpolypharmacy should not be interpreted as causal effects of the number of medications per se.

In conclusion, polypharmacy is a major concern in CKD, and hyperpolypharmacy in particular was associated with worse clinical outcomes. Beyond medication reviews and (when feasible) deprescription, strategies like patient education and better control of risk factors are crucial for reducing the frequency of adverse clinical outcomes in patients with CKD.

Author contributions

The study was designed by SL and SML. Statistical analyses were performed by AM, SL, and SML. AM, SL, and SML drafted the initial version of the manuscript. AM, MM, NAP, CJ, ML, ZAM, SL, and SML critically revised the manuscript and provided important intellectual input. NAP coordinated the CKD-REIN cohort. All authors reviewed and approved the final version of the manuscript. Each author contributed important intellectual content during manuscript drafting or revision and agrees to be personally accountable for the individual’s own contributions and to ensure that questions pertaining to the accuracy or integrity of any portion of the work—even one in which the author was not directly involved—are appropriately investigated and resolved, including with documentation in the literature if appropriate.

Funding

CKD-REIN is funded by the French Agence Nationale de la Recherche through the 2010 “Cohortes-Investissements d’Avenir” program (ANR-IA-COH-2012/3731) and by the 2010 national Programme Hospitalier de Recherche Clinique. CKD-REIN is also supported through a public–private partnership with GlaxoSmithKline (GSK) since 2012, Boehringer Ingelheim France since 2022, Novo Nordisk since 2024, Fresenius Medical Care from 2012 to 2024, Vifor France from 2018 to 2023, Sanofi-Genzyme from 2012 to 2015, Baxter and Merck Sharp & Dohme-Chibret (MSD France) from 2012 to 2017, Amgen from 2012 to 2020, Lilly France from 2013 to 2018, Otsuka Pharmaceutical from 2015 to 2020, and AstraZeneca from 2018 to 2021. Inserm Transfert set up and has managed this partnership since 2011. A specific project was funded by the National Agency for the Safety of Medicines and Health Products (ANSM) through the EPI-PHARE group of scientific interest. It should be noted that the authors of this article were solely responsible for interpreting the data; ANSM was not involved. The funding sources had no roles in study design, conduct, reporting or the decision to submit for publication.

Data availability

The data that support the findings of this study are available upon reasonable request by contacting the CKD-REIN study coordination staff at ckdrein@inserm.fr.

Declarations

Competing interests

A.M., S.L., S.M.L., M.M., M.L. have nothing to declare. N.A.P. declare financial support from pharmaceutical companies involved in the CKD-REIN study’s public–private partnership: Fresenius Medical Care, GlaxoSmithKline (GSK), Vifor France, and Boehringer Ingelheim; all the grants were made to Paris Saclay University. Z.A.M. reports having received grants for CKD-REIN and other research projects from Amgen, Baxter, Fresenius Medical Care, GlaxoSmithKline, Merck Sharp & Dohme-Chibret, Sanofi- Genzyme, Lilly, Otsuka, AstraZeneca, Vifor and the French government, as well as fees and grants to charities from AstraZeneca, Boehringer Ingelheim, and GlaxoSmithKline. However, none of the authors has any direct competing financial and/or non-financial interests to disclose in relation to the research presented in this manuscript.