Views and ﬁnally Assessed for Causality, Preventability, and Immediate Therapeutic Management Dr

Article

Adverse Drug Reactions in Patients with CKD

Sole`ne M. Laville ,1 Vale´rie Gras-Champel,2 Julien Moragny,2 Marie Metzger,1 Christian Jacquelinet,1,3 Christian Combe ,4,5 Denis Fouque ,6 Maurice Laville,6 Luc Frimat,7,8 Bruce M. Robinson,9 Be´ne´dicte Stengel,1 Ziad A. Massy,1,10 and Sophie Liabeuf ,2,11 on behalf of the Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) Study Group*

Abstract Background and objectives Little is known about the burden of adverse drug reactions in CKD. We estimated the incidence of overall and serious adverse drug reactions and assessed the probability of causation, preventability, Due to the number of contributing authors, and factors associated with adverse drug reactions in patients seen by nephrologists. the affiliations are listed at the end of Design, setting, participants, & measurements The Chronic Kidney Disease-Renal Epidemiology and Information this article. Network cohort included 3033 outpatients (65% men) with CKD and eGFR,60 ml/min per 1.73 m2,withfollow- up for 2 years. Adverse drug reactions were identified from hospitalization reports, medical records, and Correspondence: participant interviews and finally assessed for causality, preventability, and immediate therapeutic management Dr. Beńe´dicte Stengel, E´quipe E´pide´miologie by experts in pharmacology. Clinique, Centre for Epidemiology and Results Median (interquartile range) age was 69 (60–76) years old; 55% had eGFR$30 ml/min per 1.73 m2, and 45% Population Health had eGFR,30 ml/min per 1.73 m2. Participants were prescribed a median (range) of eight (five to ten) drugs. Over (CESP) — Inserm U1018, 16 Avenue 2 years, 536 patients had 751 adverse drug reactions, 150 (in 125 participants) classified as serious, for rates of fi fi P. Vaillant Couturier, 14.4 (95% con dence interval, 12.6 to 16.5) and 2.7 (95% con dence interval, 1.7 to 4.3) per 100 person-years, Villejuif Cedex, respectively. Among the serious adverse drug reactions, 32% were considered preventable or potentially France. Email: preventable; 16 caused death, directly or indirectly. Renin-angiotensin system inhibitors (15%), antithrombotic benedicte.stengel@ agents (14%), and diuretics (10%) were the drugs to which the most adverse drug reactions were imputed, but inserm.fr antithrombotic agents caused 34% of serious adverse drug reactions. The drug was discontinued in 71% of cases, at least temporarily. Adjusted hazard ratios for serious adverse drug reaction were significantly higher in patients with eGFR,30 versus $30 ml/min per 1.73 m2 (1.8; 95% confidence interval, 1.3 to 2.6), in those prescribed more than tenversusless than fivemedications(2.4;95%confidence interval,1.1 to 5.2),orinthosewithpoor versus good adherence (1.6; 95% confidence interval, 1.4 to 2.4).

Conclusions Adverse drug reactions are common and sometimes serious in patients with CKD. Many serious adverse drug reactions may be preventable. Some speciﬁc pharmacologic classes, particularly antithrombotic agents, are at risk of serious adverse drug reactions.

Clinical Trial registry name and registration number Chronic Kidney Disease-Renal Epidemiology and In- formation Network (CKD-REIN), NCT03381950 CJASN 15: 1090–1102, 2020. doi: https://doi.org/10.2215/CJN.01030120

Introduction few have examined ADRs in outpatient settings (4). Despite the highly regulated process of drug market- Although clinicians recognize ADRs as a major prob- ing authorization, no medicine is completely safe. lem in patients with CKD, few studies have investi- Adverse drug reactions (ADRs) are relatively com- gated the incidence of and factors associated with mon; they cause 2%–7% of overall hospitalizations ADRs in this population. (1–4). Two French studies (2,5) have reported that Drug-related nephrotoxicity is frequent and well 3.2%–3.6% of hospital admissions are related to ADRs. documented (15). However, the kidney also plays an The concept of ADR has evolved over time, and today, important role in the clearance of many drugs and it includes any harmful and unwanted reaction to a toxic metabolites that can cause ADR due to their drug occurring at doses normally used in humans or accumulation as kidney function declines. Despite resulting from drug misuse or error or from accidental impaired pharmacokinetics and pharmacodynamics or willful overdose (6). Most available studies have (16–18), patients with CKD use multiple medicines focused on ADR-related hospitalizations (with the and are often exposed to some that are inappropri- ADR the cause of admission [2,3,5,7–10], occurring ately prescribed (19). Until now, studies in patients during hospitalization [11,12], or both [1,13,14]); very with CKD have been on the basis of self-reported

ADRs (20), have concerned specific drugs during clinical insurance, but none of these differences affect the recording trials (21,22), or have been restricted to ADRs during and processing of prescriptions. Accordingly, drug pre- hospitalization (23) or to specific types of ADRs (24). None scriptions were continuously recorded from 3 months of the studies reported therapeutic management after the preceding inclusion through the end of follow-up. We ADR. No comprehensive evaluation exists of the in- used the international Anatomic Therapeutic and Chemical cidence, probability of causation, and preventability of thesaurus (28) to code treatments and recorded their start ADRs in both inpatients and outpatients belonging to and discontinuation dates (with causes, if any). Kidney this population. failure events, defined as dialysis start or preemptive The primary objective of this study was to estimate the transplantation, and deaths were reported by the partici- incidence rates of overall and serious ADRs according to pants or their families, or they were identified from medical eGFR in patients with moderate or advanced CKD treated records or record linkage with the national kidney fail- by a nephrologist. Secondary objectives aimed at assessing ure registry (29). their causation, preventability, associated factors, and immediate therapeutic management. Identification and Validation of Adverse Drug Reactions An ADR is defined as “an appreciably harmful or Materials and Methods unpleasant reaction, resulting from an intervention related Study Design and Participants to the use of a medicinal product, which predicts hazard fi The Chronic Kidney Disease-Renal Epidemiology and from future administration and warrants prevention or speci c Information Network (CKD-REIN) is a prospective cohort treatment, or alteration of the dosage regimen, or withdrawal ” study conducted in 40 nationally representative nephrol- of the product (6). An ADR is considered serious when the ogy outpatient facilities in France. Eligible patients were at patient outcome is death (or a life-threatening situation), least 18 years of age, had a confirmed diagnosis of hospitalization (initial or prolonged), disability or permanent moderate or advanced CKD, had an eGFR,60 ml/min damage, or another important medical event (30). per 1.73 m2, were not on dialysis, and had not been We collected ADRs over a 2-year follow-up via an fi transplanted. From July 2013 to March 2016, CKD-REIN electronic form designed speci cally to include information included 3033 patients. Details of the study protocol and critical for this study. We used several sources to identify flow chart have been published elsewhere (25). The in- ADRs: (1) medical records (examined by CRAs), (2) stitutional review board of the French National Institute of participant interviews by CRAs, and (3) hospital reports fi Health and Medical Research (reference: IRB00003888) (Figure 1). Hospitalizations were identi ed from (1)elec- approved the protocol, and the study was registered at tronic medical records, (2) nephrology records, and (3) ClinicalTrials.gov (NCT03381950). participant interviews. For each hospitalization, we obtained a report to confirm the period and the cause. Causes of hospitalization were coded by a physician according to Information the tenth revision of the International Classification of Data were collected at baseline and then annually by Diseases. Every drug prescribed to the patient at the time of trained clinical research associates (CRAs) from participant each ADR was recorded. Each identified ADR was re- interviews and medical records from the nephrology viewed by two pharmacists (S.L. and S.M.L.) who evalu- centers that included them. All of them contain patient ated the potential causation of reported drug-related ADRs histories, hospitalizations reports, imaging, and laboratory and coded the types of effect according to the medical data from every ward of the hospital/clinic, but they are dictionary for regulatory activities (MedDRA Dictionary), not standardized and may differ between these centers. the severity of the ADR (nonserious or serious), the drug Data collected included sociodemographic characteristics suspected of responsibility for the ADR, its dosage, and and a history of hypertension, diabetes, cardiovascular immediate medication management: discontinuation of the disease, dyslipidemia, or AKI, as defined previously (25). product, dose adaptation, or no change. If the ADR was Medication adherence was assessed with the Girerd score considered serious, a larger committee of expert pharma- (26), which was calculated by asking six questions that cologists (J.M., S.L., S.M.L., and V.G.-C.) from the Amiens explore the primary determinants of adherence to chronic pharmacovigilance center further evaluated the potential medication (its timing, remembering to take it, and re- causal relation of each drug (prescribed at the time of the membering to renew the prescription). Serum creatinine, ADR) and the preventability of the ADR. albumin, and hemoglobin were measured, as was urinary albumin or protein. We used the Chronic Kidney Disease Epidemiology Collaboration equation to estimate GFR (27). Assessment of the Causes and Preventability of Serious Participants were asked to bring to their inclusion appoint- Adverse Drug Reactions ment all of their current drug prescriptions for the previous We applied the Bégaud imputability method (31) 3 months (regardless of the prescribing physician) and all (Supplemental Table 1A), which is the official procedure prescriptions for the year to each annual follow-up ap- used in French pharmacovigilance centers to report serious pointment. In France, all prescriptions are reimbursed ADRs to the French drug authorities. This algorithmic similarly, except for medications determined by the Min- method attributes an intrinsic score on the basis of istry of Health to have only moderate or insufficient chronological and semiological criteria. The cause and medical benefits. Some differences in reimbursement also effect relationship is assessed independently for each drug exist for patients with chronic expensive diseases and for taken by the patient before the occurrence of the event and patients who have not purchased supplementary health is not influenced by the extent of imputability to other 1092 CJASN

Collection of adverse drug reactions in the CKD-REIN cohort

Patient Hospitalization Medical records interviews reports

751 adverse drug reactions assessed and validated during the two years following inclusion

Review by pharmacists

Adverse drug reactions considered as serious (n=150)

Reviewed by a committee of expert pharmacologists in a pharmacovigilance center

Nonserious adverse drug reactions (n=601) Serious adverse drug reactions (n=150)

Figure 1. | Description of the process for identifying and validating adverse drug reactions in the Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) cohort. drugs. This method allowed us to identify the drug most (percentages). Fisher exact, t, or chi-squared tests were used responsible for serious ADRs (i.e., that with the highest to compare categorical variables. A sensitivity analysis intrinsic score). During the ADR validation process, we was conducted by using the last eGFR preceding the ADR globally evaluated all of the risk factors for ADRs, in- to describe it. cluding a review of all drugs prescribed at the time of the Crude incidence rates and 95% confidence intervals (95% ADR and evaluation of potential pharmacokinetic/phar- CIs) of ADRs and serious ADRs per 100 person-years were macodynamic interactions. estimated by Poisson regression for the overall population In addition, we used the Naranjo ten-question algorithm and by subgroups according to baseline eGFR; they were (Supplemental Table 1B) to confirm the causal relation of corrected for overdispersion by using the quasilikeli- each reported serious ADR by determining the probability hood approach. that an ADR is actually due to a drug rather than to any We used cause-specific Cox proportional hazard models other factor (32). This analysis considers only ADRs to investigate patient clinical characteristics associated with categorized as definite, probable, or possible with both the ADR risk. Data were censored at the end of the 2-year the Bégaud and Naranjo methods. follow-up, the last patient visit, death, or kidney failure, The preventability of ADRs was assessed with a seven- whichever came first (competing events). Variables from item ADR preventability scale (33) (Supplemental Table 1, preselected by a literature review, were analyzed Table 1C) that classified ADRs in four categories: “pre- in a crude model. Variables with a P value .0.10 in the ventable,”“potentially preventable,”“not assessable,” and crude model were excluded from the multivariable anal- “not preventable.” When items related to adherence to yses. Age and sex were forced into the final model: that is, recommendations and the patient’s need for the prescrip- included because they were considered necessary for the tion were uncertain, the expert committee rated prevent- model’svalidity. ability as “not assessable.” Because of the possibility of multiple ADRs per patient Both outpatient and inpatient medical records were used during the follow-up, we performed a sensitivity analysis to assess causality and preventability. using the Prentice, Williams, and Peterson gap-time recurrent event time-to-event analysis, with sandwich var- iance estimators, to determine the factors associated with Statistical Analyses ADRs (34,35). Baseline characteristics were described for all partici- To deal with the missing data (Supplemental Table 2), pants and by subgroup according to baseline eGFR (,30 or multiple imputations were performed (fully conditional $30 ml/min per 1.73 m2). ADRs were also described specification method [36], ten iterations, ten datasets), according to baseline eGFR. Results were expressed as including all patient characteristics from Table 1, the total means 6 SDs, medians (interquartile ranges), or numbers number of ADRs per patient, and the number of serious CJASN 15: 1090–1102, August, 2020 Adverse Drug Reactions in CKD, Laville et al. 1093

Table 1. Baseline characteristics of participants in the Chronic Kidney Disease-Renal Epidemiology and Information Network

Baseline eGFR, ml/min per 1.73 m2 Participants with Missing Characteristics All, n53033 Data,a %, n53033 $30, n51670 ,30, n51363

Age, yr 69 [60–76] 68 [59–75] 70 [61–78] 0 ,60 716 (24%) 421 (25%) 295 (22%) 60–75 1400 (46%) 803 (48%) 597 (44%) $75 917 (30%) 446 (27%) 471 (35%) Men 1982 (65%) 1121 (67%) 861 (63%) 0 High school diploma or higher 1094 (36%) 648 (39%) 446 (33%) 1.7 BMI 28 [25–32] 28 [25–32] 28 [25–32] 2.1 $30 kg/m2 1075 (35%) 573 (34%) 502 (37%) Serum albumin 4.0 [3.8–4.3] 4.1 [3.8–4.3] 4.0 [3.7–4.3] 18.9 ,3.5 g/dl 293 (10%) 138 (8%) 155 (11%) UACR, mg/g 11.2 ,30 846 (28%) 625 (37%) 221 (16%) 30–300 955 (31%) 540 (32%) 415 (31%) .300 1232 (41%) 505 (31%) 727 (53%) Anemiab 1236 (41%) 490 (29%) 746 (55%) 0.9 Smoking status 0.8 Smoker 360 (12%) 197 (12%) 163 (12%) Nonsmoker 1252 (41%) 701 (42%) 551 (40%) Ex-smoker 1421 (47%) 773(46%) 648 (48%) Diabetes 1304 (43%) 702 (42%) 602 (44%) 0.2 AKI history 710 (23%) 350 (21%) 360 (26%) 8.0 Cardiovascular history 1611 (53%) 854 (51%) 757 (56%) 1.4 Hypertension 2751 (91%) 1493 (89%) 1258 (92%) 0.2 Dyslipidemia 2229 (73%) 1216 (73%) 1011 (74%) 0.5 No. of drugs 8[5–10] 7 [5–10] 8 [6–11] 0.6 ,5 593 (19%) 412 (25%) 181 (13%) 5–10 1694 (56%) 915 (54%) 779 (57%) .10 747 (25%) 344 (21%) 403 (30%) Poor adherence to medications 1888 (62%) 1010 (60%) 878 (64%) 1.0

Median (interquartile range) or n (%). BMI, body mass index; UACR, urine albumin-creatinine ratio. aMissing data were imputed as specified in Materials and Methods. bAnemia is defined by the 1968 World Health Organization (51) definition: ,12 g/dl for women and ,13 g/dl for men.

ADRs per patient. Data patterns suggest that the assump- Tables 2 and 3). Five percent of participants had more than tion that data were missing at random was plausible. Cox one ADR. Sixty percent of ADRs were reported only in the model regression coefficients were estimated separately in each imputed dataset and combined according to p < 0.001 25 Rubin rules. Statistical analyses were performed with SAS software (version 9.4; SAS Institute, Cary, NC) and R software 20 (version 3.5.0; Foundation for Statistical Computing, Vienna, Austria). 15 p = 0.06 10 Results Baseline Characteristics Participants were predominantly men; 43% had diabetes, 5 53% had cardiovascular disease, and 91% had hypertension Incidence (per 100 person-years) (Table 1). Those with eGFR,30 ml/min per 1.73 m2 at 0 baseline were older and had less education than those with Overall ADRs Serious ADRs higher eGFR. They also had anemia and a history of AKI Baseline estimated glomerular filtration rate (eGFR) more often, and they were prescribed more medications. eGFR > or = 30 ml/min/1.73m2 2 Incidence and Description of Adverse Drug Reactions eGFR < 30 ml/min/1.73m Over the 2-year follow-up, 751 ADRs were reported in 536 (18%) of 3033 participants; 150 ADRs in 125 partici- fi Figure 2. | Higher incidence rates of adverse drug reactions (ADRs) pants (4%) were classi ed as serious (i.e., 14.4; 95% CI, 12.6 and serious ADRs in patients with eGFR<30 ml/min per 1.73 m2. to 16.5 and 2.7; 95% CI, 1.7 to 4.3 per 100 person-years, Incidence rates are represented with their 95% confidence interval respectively). Both were nearly twice as high in participants whiskers. P values test the difference between incidence rates with eGFR,30 versus $30 ml/min per 1.73 m2 (Figure 2, according to baseline eGFR. 04CJASN 1094

Table 2. Associations of participant characteristics with adverse drug reactions

Incidence Ratea Unadjusted Model Adjusted Model N (%) with Adverse Drug Characteristics [95% [95% [95% Reactions, n5536 Incidence Hazard P Hazard P Confidence Confidence Confidence Rate Ratio Value Ratio Value Interval] Interval] Interval]

Age, yr 0.10 0.10 ,60 113 (16) 13.2 [9.9 to 17.6] Reference Reference 60–75 271 (19) 15.2 [12.5 to 18.5] 1.25 [1.00 to 1.56] 1.06 [0.84 to 1.34] $75 152 (17) 14.0 [10.9 to 18.0] 1.08 [0.85 to 1.38] 0.85 [0.65 to 1.11] Sex 0.08 0.04 Men 332 (17) 13.2 [11.1 to 15.7] Reference Reference Women 204 (19) 16.6 [13.4 to 20.6] 1.17 [0.98 to 1.39] 1.21 [1.01 to 1.45] Educational level 0.09 0.99 High 179 (16) 13.0 [10.2 to 16.5] Reference Reference Low 357 (18) 15.2 [12.8 to 17.9] 1.17 [0.97 to 1.40] 1.00 [0.83 to 1.21] eGFR, ml/min per 1.73 m2 ,0.001 ,0.001 $30 244 (15) 10.8 [8.9 to 13.2] Reference Reference ,30 292 (21) 19.4 [16.2 to 23.2] 1.74 [1.47 to 2.07] 1.56 [1.30 to 1.87] Serum albumin, g/dl 0.06 0.31 $3.5 476 (17) 13.9 [12.0 to 16.0] Reference Reference ,3.5 60 (20) 19.7 [12.9 to 29.9] 1.38 [1.00 to 1.90] 1.19 [0.85 to 1.65] UACR, mg/g 0.06 0.90 ,30 139 (16) 13.1 [10.0 to 17.1] Reference Reference 30–300 165 (17) 12.9 [10.0 to 16.7] 1.09 [0.85 to 1.39] 0.95 [0.75 to 1.22] .300 232 (19) 16.6 [13.5 to 20.4] 1.29 [1.03 to 1.62] 1.00 [0.79 to 1.27] Anemiab ,0.001 0.06 Without anemia 287 (16) 12.3 [10.3 to 14.8] Reference Reference With anemia 249 (20) 17.7 [14.6 to 21.5] 1.45 [1.23 to 1.72] 1.19 [0.99 to 1.42] Diabetes 0.01 0.78 Without diabetes 274 (16) 12.1 [10.0 to 14.7] Reference Reference With diabetes 262 (20) 17.4 [14.4 to 21.0] 1.31 [1.10 to 1.55] 1.03 [0.85 to 1.24] AKI history ,0.001 0.01 Without AKI history 379 (16) 12.9 [11.0 to 15.2] Reference Reference With AKI history 157 (22) 19.4 [15.0 to 25.1] 1.48 [1.21 to 1.81] 1.32 [1.08 to 1.61] Cardiovascular history ,0.001 0.03 Without cardiovascular 215 (15) 11.3 [9.1 to 14.0] Reference Reference history With cardiovascular history 321 (20) 17.2 [14.6 to 20.4] 1.42 [1.19 to 1.69] 1.24 [1.02 to 1.50] Hypertension 0.06 0.71 Without hypertension 38 (13) 9.4 [5.4 to 16.1] Reference Reference With hypertension 498 (18) 14.9 [13.0 to 17.1] 1.37 [0.99 to 1.91] 1.07 [0.76 to 1.51] CJASN 15: 1090–1102, August, 2020 Adverse Drug Reactions in CKD, Laville et al. 1095

medical record or hospitalization report, 20% were reported in both the medical record and participant interview, P 0.01 0.01

Value and 20% were only reported in the participant interview. Among the serious ADRs, 145 were associated with hospitalization: 65% as its cause and 35% as its consequence (Figure 3). Sixteen deaths resulted from an ADR, directly or ﬁ

dence indirectly, as did ve life-threatening events. Five partici- ﬁ [95% pants with serious ADRs (three medically important and two Interval]

Con resulting in permanent disability) were not hospitalized.

Adjusted Model Renal and urinary disorders were the most frequent type Reference Reference of ADR, particularly AKI, followed by gastrointestinal (mostly diarrhea), musculoskeletal, and connective tissue 1.501.71 [1.12 to 2.02] [1.21 to 2.41] 1.36 [1.12 to 1.64]

Ratio disorders (Table 4, Supplemental Table 3). Renal disorders Hazard and hemorrhages or bleeding accounted for two thirds of the 150 serious ADRs (Figure 3). Of the 16 deaths linked to an ADR, 11 were related to hemorrhages, which also P 0.001 0.001 accounted for 40% of the serious ADRs in participants with Value , , baseline eGFR ,30 ml/min per 1.73 m2 and 19% in those with eGFR$30 ml/min per 1.73 m2 (P50.009) (Figure 3). Using the last eGFR before the ADR did not signiﬁcantly change the proportions reported in Figure 3 and Table 4 dence ﬁ

[95% (Supplemental Tables 4 and 5).

Interval] Renin-angiotensin system (RAS) inhibitors, antithrom- Con botic agents, and diuretics were the medications most Unadjusted Model Reference Reference frequently responsible for both nonserious and serious ADRs (Figure 4, Supplemental Table 6). Antithrombotic agents were responsible for 34% of the 150 serious ADRs: Ratio Hazard 34 were due to vitamin K antagonists, nine were due to 13 g/dl for men.

, heparin, six were due to platelet aggregation inhibitors, and two were due to direct factor Xa inhibitors. Three pharmacodynamic drug interactions were identiﬁed: two dence a

ﬁ between vitamin K antagonist and heparins and one [95%

Interval] between vitamin K antagonist and an antibiotic. Con

Factors Associated with Adverse Drug Reactions and Serious 12 g/dl for women and

Incidence Rate Adverse Drug Reactions ,

7.5 [5.0 to 11.2] Participants with a baseline eGFR ,30 ml/min per 13.921.7 [11.6 to 16.7] [17.3 to 27.3] 1.86 2.59 [1.41 to 2.44] [1.93 to 3.46] 16.7 [14.3 to 19.6] 1.52 [1.26 to 1.83] 10.7 [8.3 to 13.7] Rate 1.73 m2 had a risk 1.6 times higher of an ADR than those Incidence 2 nition: with an eGFR$30 ml/min per 1.73 m after adjustment for fi other associated variables (Table 2). The risk of ADR also significantly increased with the participant’s baseline number of prescribed drugs, history of cardiovascular disease, 536 history of AKI, and poor treatment adherence. Women 5 n were at higher risk of an ADR than men, but not of a serious ADR. There was no significant association with age. 62 (10) 301 (18) 173 (9) 378 (20) 158 (14) Hazard ratios for serious ADRs were significantly higher in , 2

Reactions, participants with eGFR 30 ml/min per 1.73 m compared with

(%) with Adverse Drug eGFR$30 ml/min per 1.73 m2 as well as in those prescribed N more than ten compared with less than five medications and in participants with poor adherence(Table3).Ageandsexwere not associated with a higher risk of serious ADR. A sensitivity analysis confirmed that the factors men- tioned above were associated with ADRs (Supplemental Table 7A). It also showed that anemia was significantly associated with the risk of serious ADRs but that poor

ned by the 1968 World Health Organization (51) de adherence no longer was (Supplemental Table 7B). ﬁ

Characteristics Immediate Management of Adverse Drug Reactions and 10 5 10

– Preventability , 5 . Poor Good values used the Wald chi-squared test for global variable effect. UACR, urine albumin-creatinine ratio. Anemia is de Incidence rates are expressed per 100 person-years. After an ADR, the drug considered responsible was Table 2. (Continued) a b Baseline no. of drugs per patient Adherence to medication P discontinued in 71% of cases (e.g., in 78% of statin-linked 06CJASN 1096

Table 3. Associations of participant characteristics with serious adverse drug reactions

Incidence Ratea Unadjusted Model Adjusted Model N (%) with Adverse Drug Characteristics [95% [95% [95% Reactions, n5125 Incidence Hazard P Hazard P Confidence Confidence Confidence Rate Ratio Value Ratio Value Interval] Interval] Interval]

Age, yr 0.15 0.32 ,60 26 (4) 2.4 [0.9 to 6.9] Reference Reference 60–75 52 (4) 2.4 [1.1 to 5.1] 1.02 [0.64 to 1.63] 0.71 [0.44 to 1.17] $75 47 (5) 3.4 [1.6 to 7.5] 1.45 [0.90 to 2.34] 0.91 [0.54 to 1.51] Sex 0.91 0.74 Men 82 (4) 2.7 [1.5 to 4.8] Reference Reference Women 43 (4) 2.8 [1.3 to 6.0] 0.98 [0.68 to 1.42] 1.06 [0.73 to 1.55] eGFR, ml/min per 1.73 m2 ,0.001 0.01 $30 49 (3) 1.8 [0.9 to 3.4] Reference Reference ,30 76 (6) 4.0 [2.4 to 6.8] 2.17 [1.51 to 3.11] 1.82 [1.25 to 2.63] BMI, kg/m2 0.05 0.46 ,30 70 (4) 2.4 [1.3 to 4.5] Reference Reference $30 55 (5) 3.3 [1.6 to 6.8] 1.42 [0.99 to 2.04] 1.16 [0.79 to 1.70] Anemiab 0.001 0.15 Without anemia 60 (3) 2.0 [1.0 to 4.1] Reference Reference With anemia 65 (5) 3.9 [2.0 to 7.5] 1.77 [1.25 to 2.52] 1.31 [0.91 to 1.89] Diabetes 0.03 0.89 Without diabetes 60 (3) 2.3 [1.2 to 4.3] Reference Reference With diabetes 65 (5) 3.3 [1.8 to 6.1] 1.47 [1.03 to 2.09] 0.97 [0.65 to 1.44] AKI history 0.01 0.12 Without AKI history 84 (4) 2.3 [1.3 to 4.2] Reference Reference With AKI history 41 (6) 4.0 [1.8 to 9.3] 1.67 [1.12 to 2.49] 1.37 [0.92 to 2.05] Cardiovascular history ,0.001 0.01 Without 35 (2) 1.5 [0.7 to 3.4] Reference Reference cardiovascular history With 90 (6) 3.9 [2.4 to 6.3] 2.45 [1.65 to 3.63] 1.93 [1.25 to 2.98] cardiovascular history Baseline no. of drugs ,0.001 0.05 per patients ,5 10 (2) 1.1 [0.2 to 4.6] Reference Reference 5–10 62 (4) 2.4 [1.3 to 4.3] 2.29 [1.17 to 4.46] 1.53 [0.76 to 3.10] .10 53 (3) 5.0 [2.7 to 9.4] 4.70 [2.39 to 9.23] 2.31 [1.07 to 5.01] Adherence to medication 0.01 0.03 Good 30 (3) 1.8 [0.8 to 4.3] Reference Reference Poor 95 (5) 3.3 [2.0 to 5.4] 1.95 [1.29 to 2.94] 1.59 [1.05 to 2.42]

P values used the Wald chi-squared test for global variable effect. BMI, body mass index. aIncidence rates are expressed per 100 person-years. bAnemia is deﬁned by the 1968 World Health Organization (51) deﬁnition: ,12 g/dl for women and ,13 g/dl for men. CJASN 15: 1090–1102, August, 2020 Adverse Drug Reactions in CKD, Laville et al. 1097

Renal and urinary disorders (n=52) 33% 37%

Hemorrhages and bleeding (n=46) 19% 40%

Gastrointestinal disorders (n=12) 17% 3%

Blood and lymphatic system disorders (n=7) 7% 3%

Type of ADRs Cardiac disorders (n=4) 4% 2%

Other disorders (n=24) 20% 14%

0 10 20 30 40 0 10 20 30 40 Serious ADRs in patients with Serious ADRs in patients with eGFR > or = 30 mL/min/1.73m2 eGFR < 30 mL/min/1.73m2 (n=54) (n=91) Was the adverse drug reaction the cause or a consequence of hospitalization?

ADR was the cause of hospitalization ADR was a consequence of hospitalization

Figure 3. | Distribution of serious ADRs causing or resulting from hospitalization according to baseline eGFR (n5145). Renal disorders and bleeding are the most frequent serious ADRs. Results are expressed as percentages. The denominators used are the total number of ADRs in patients with eGFR,30 or $30 ml/min per 1.73 m2.

ADRs), at least temporarily, and the dose was adjusted in The Olivier ADR preventability scale (33) allowed us to 14% of cases; no change was made in the prescription in classify the 150 serious ADRs as preventable in 13% of 11% (4% missing data). When the ADR was serious, 83% of cases (n519) and potentially preventable in 19% the drugs blamed were discontinued at least temporarily (n528). A quarter of the preventable ADRs were right after the event. associated with participant self-medication. Overall,

Table 4. Description of adverse drug reactions according to baseline eGFR

All Adverse Drug Adverse Drug Reactions in Adverse Drug Reactions in Type of Adverse Drug Reaction Reactions, n5751 Patients with eGFR$30, n5331 Patients with eGFR,30, n5420

Renal and urinary disorders 150 (20%) 62 (19%) 88 (21%) AKI 102 41 61 Increased serum creatinine 40 20 20 Other type of renal and 81 7 urinary disorders Gastrointestinal disorders 119 (16%) 61 (18%) 58 (14%) Diarrhea 57 35 22 Gastrointestinal conditions 24 8 16 Other type of 38 18 20 gastrointestinal disorders Musculoskeletal and connective 68 (9%) 34 (10%) 34 (8%) tissue disorders Contractures 35 21 14 Muscle pain 22 10 12 Other type of musculoskeletal and 11 3 8 connective tissue disorders Hemorrhages and bleeding 67 (9%) 18 (5%) 49 (12%) Hemorrhages 34 8 26 Hematoma 19 7 12 Other type of hemorrhages 14 3 11 and bleeding General disorders and 58 (8%) 26 (8%) 32 (8%) administration site conditions Peripheral edema 30 11 19 Drug intolerance 8 3 5 Other type of general disorders and 20 12 8 administration site conditions Other type of adverse drug reactions 289 (38%) 130 (39%) 159 (38%)

eGFR is expressed in milliliters per minute per 1.73 m2. Results are expressed as n (%). The denominator used in column 2 is the total number of adverse drug reactions, and the denominators used in columns 3 and 4 are those of the adverse drug reactions in patients with eGFR,30 versus $30, respectively. P50.03 tests the difference in the distribution of adverse drug reaction type according to patient baseline eGFR. 1098 CJASN

Agents acting on the RAS (n=115) 18% 13% Antithrombotic agents (n=107) 8% 19% Diuretics (n=77) 11% 10% Lipid modifying agents (n=50) 8% 6% Calcium channel blockers (n=45) 5% 6% Antibacterials for systemic use (n=36) 5% 5% Drugs used in diabetes (n=31) 6% 2% Analgesics (n=29) 4% 4% Immunosuppressants (n=27) 5% 3% Antigout preparations (n=26) 3% 4% Antineoplastic agents (n=22) 4% 2%

Pharmacological classes Contrast media (n=21) 2% 3% Antianemic preparations (n=20) 3% 3% Beta blocking agents (n=15) 2% 2% Antihypertensives (n=14) 1% 3% Other drug classes (n=116) 15% 16%

0105152001051520 ADRs in patients with ADRs in patients with eGFR > or = 30 mL/min/1.73m2 eGFR < 30 mL/min/1.73m2 (n=331) (n=420)

Nonserious ADRs Serious ADRs

Figure 4. | The most common pharmacological classes responsible for ADRs and serious ADRs are renin-angiotensin system inhibitors, antithrombotic agents, and diuretics. Adverse drug reactions caused by agents acting on the renin-angiotensin system (RAS), antithrombotic agents, or diuretics accounted for 39% of ADRs. Serious ADRs caused by these three classes accounted for 58% of serious ADRs. Results are expressed as percentages. The denominators used are the total number of ADRs in patients with eGFR,30 or $30 ml/min per 1.73 m2.

37% of ADRs were inevitable, and 31% were not and without hospitalization; they were identified from an assessable (Supplemental Table 6). extensive review of medical records, hospitalization reports, and participant interviews. We compared rates according to eGFR and showed that ADR incidence in- Discussion creased when eGFR was lower than 30 ml/min per 1.73 m2. This study presents a global descriptive view of the Overall, renal, urinary, gastrointestinal, musculoskeletal, magnitude and diversity of ADRs in a well phenotyped and connective tissue disorders were the most commonly CKD population. The central message here is that ADRs are reported ADRs in our study, but renal disorders and common, often serious, and potentially preventable in bleeding largely predominated among the serious ADRs. patients with CKD and that these patients are vulnerable Several studies have reported similar results in hospital- and their treatment is complex. It shows that three drug ized patients with unknown CKD status (5,8–10). The classes among those most prescribed in this population are known high susceptibility of patients with CKD for AKI responsible for almost 40% of ADRs, including RAS explains the high frequency of renal and urinary ADRs. inhibitors, antithrombotic agents, and diuretics. The study Cardiovascular medicines stand out among the most especially points out the severity of ADRs caused by common suspected drugs in several studies of hospitalized antithrombotic agents, to which one third of the serious patients with unknown CKD status (2–5,9,10). Similarly, events were imputed. In addition, we identify some care we found that RAS inhibitors, antithrombotic agents, and and patient characteristics that increase ADR risk; these diuretics were the pharmacologic classes to which ADRs include eGFR (,30 ml/min per 1.73 m2), a higher number were most commonly imputed in our study. The high of prescribed drugs, and poor adherence to medications. prevalence of CKD-related cardiovascular complications Importantly, a significant proportion of these ADRs may be explains the high use of cardiovascular drugs in patients preventable. with CKD (37), despite their high risk of ADR due to the Our findings are difficult to compare with those of other combined effect of low eGFR, hemorrhagic risk, and studies, which differ from ours in several ways. Most of electrolyte disturbance. The principal drugs suspected of them assessed ADR incidence at hospital admission causing ADRs were usually not directly nephrotoxic, and (1–3,5,7–10,13) or during hospitalization (1,11–13,23), and most ADRs resulted from reduced renal clearance. This they used highly heterogeneous study settings, definitions conclusion has important clinical implications, notably of ADRs, and methods for ascertaining ADRs (spontaneous regarding drug prescriptions for patients with CKD and report, intensive chart review, or both). In this study, we the need to focus on regularly reassessing use or dose according evaluated ADR incidence in nephrology outpatients, with to eGFR (especially when it drops below 30 ml/min per CJASN 15: 1090–1102, August, 2020 Adverse Drug Reactions in CKD, Laville et al. 1099

1.73 m2), as well as on potential nephrotoxic agents. related to their lower number of inappropriate prescriptions However, the clinical benefits of some of these drugs (19). This may indicate more careful prescriptions by phy- have been demonstrated by a high level of evidence. For sicians for elderly patients. Declining kidney function seems instance, a moderate increase (of 20%–30%) in creatinine can to be an important risk factor for patients with CKD and be expected with RAS inhibitors; nephrologists may find eGFR,30 ml/min per 1.73 m2 compared with $30 ml/min this an acceptable trade-off in view of these drugs’ pro- per 1.73 m2, in line with the results reported by Sharif-Askari tective nature in the long term and their ability to slow CKD et al. (23) in hospitalized patients with CKD. Corsonello et al. progression. This moderate increase is well below the (44) found a similar association between ADR risk and Kidney Disease Improving Global Outcomes definition declining kidney function in elderly hospitalized patients. used for drug-related AKI events in our study, which is This higher rate of ADR at eGFR,30 ml/min per 1.73 m2 on the basis of a rise in creatinine of at least 50% (25). seems related mainly to bleeding events due to antithrom- Patients seen by nephrologists require the most complex botic agents, mostly vitamin K antagonists. This highlights care because of their multiple comorbidities and the the importance of reassessing prescriptions regularly as complications associated with decreased kidney function kidney function declines, especially for drugs such as vitamin (38). These result in the use of multiple medications as K antagonists. Although these medications are metabolized shown here: CKD-REIN participants were prescribed a by the liver, excreted in an inactive form in stool and urine median of eight different drugs daily. Moreover, the (45) and, thus, not renally eliminated, CKD can impair their number of prescribed drugs increases as CKD progresses disposition (18). Specifically, it can cause the accumulation of (19,39). High rates of ADRs in patients with CKD add to the uremic toxins in blood and organs with deleterious effects complexity of their care. As we showed, treatments often (46): for example, indoxyl sulfate, which impairs plate- need to be stopped, at least temporarily, or dosages need to let activity (47). be adjusted because of ADRs, which may affect therapeutic Major strengths of this study include its large sample size management and ultimately reduce the likelihood of slow- of patients with confirmed CKD diagnoses recruited from a ing CKD progression and decreasing its complications. representative sample of nephrology outpatient facilities. Lipid-lowering agents are a good example of treatments The high sensitivity and specificity of our process for that are strongly recommended in patients with CKD identifying and grading ADRs are also unique. Indeed, a because of their high cardiovascular risk but are often major limitation in ADR research is the lack of reliable data discontinued because of contractures, cramps, or myalgia about the true burden of these reactions related to their (stopped in 78% of statin-linked ADRs). Schneider et al. (40) reporting process; the traditional ascertainment method, noted the underprescription of statins in patients with through the spontaneous reporting system in pharmaco- CKD, which may be partly explained by ADRs; a similar vigilance, results in their vast under-reporting. A previous underprescription rate exists in the CKD-REIN cohort (41). study showed that health professionals spontaneously The challenge for physicians is to assess the benefit-risk report only 6% of serious ADRs (48). Our study used ratio between treating a new complication and adding a different sources to capture these events, especially serious new drug. Increased awareness by the medical community ADRs, through hospitalization reports, medical records, of this difficulty and of the necessity to reassess this benefit- and participant interviews. Furthermore, all cases were risk ratio regularly, especially when eGFR decreases, is reviewed by pharmacists, and serious cases were evaluated essential. Pharmacists, too, must play a role given that most by a committee of experts. Finally, the use of standard ADRs occur in outpatients. Finally, the importance of international coding systems for both drug classes and patient education in terms of drug use must be enhanced ADRs, the availability of start and end dates for each because one quarter of preventable serious ADRs were due treatment, and the reasons for treatment discontinuation to participant misuse. contribute to the quality and novelty of our findings. The major factors identified in this study associated with Our study also has limitations. Despite our sensitive ADRs overall and with serious ADRs are the number of method for identifying ADRs, their number may still be prescribed drugs, cardiovascular disease, poor adherence underestimated, mainly for those not hospitalized during to treatment, and eGFR. Reports have regularly shown that follow-up; these may not have been reported by physicians the number of drugs is a risk factor in hospitalized patients in medical records or by participants either to physicians or with unknown CKD status (8,9,14,23). Other than the during interviews due to memory bias or because an ADR increased risk of ADR, polypharmacy is also associated is so well known that it tends to be poorly reported in with deleterious health outcomes in elderly patients (42). hospitalization reports and medical records. However, Poor adherence to medication is associated with higher participants were probably most likely to report ADRs ADR risk, possibly reflecting patients’ misuse of drugs. with the most negative effects on their health and well- However, the healthy user effect cannot be ruled out (43). being. We may not have captured all hospitalizations and Two other studies have also described cardiovascular may also have missed drug effects that have never been disease as a risk factor for ADR (7,23). Although numerous reported as a potential ADR. Other sources of ADRs might studies show women to be at higher risk of serious ADR in exist but were not available to us. Several decision algorithms populations with an unknown CKD status (2,9,10,14), this for causality assessment in ADRs exist (49). However, none of was not the case in our CKD population. Age was not them have been accepted as a gold standard, and comparing associated with the risk of either ADR or serious ADR, them could lead to discrepancies in results (50). These consistent with previous reports (8,14,23). The risk of ADR algorithms do not replace medical diagnosis. tended to be slightly lower in older patients in the multivari- Another limitation of this study is the heterogeneous able analyses, although not significantly; this finding may be nature of the outcome (ADR) used in the multivariable 1100 CJASN

analyses. Although these analyses allowed us to Disclosures identify some specific and potentially modifiable Z.A. Massy reports receiving grants for the CKD-REIN and other risk factors, our basic aim was to describe the general research projects from Amgen, Baxter, Fresenius Medical Care, characteristics of patients at high risk for ADR. We GlaxoSmithKline, Merck Sharp and Dohme-Chibret, Sanofi- will further explore risk factors associated with ADRs Genzyme, Lilly, Otsuka, and the French government, as well as more specifically for a number of drug classes fees and grants to charities from Amgen, Astellas, Daichii, and and outcomes. Sanofi-Genzyme. These sources of funding are not necessarily re- Finally, our study is generalizable to patients with CKD lated to the content of this manuscript. B. Stengel reports receiving seen by nephrologists but not to all patients with an grants for the CKD-REIN from Amgen, Baxter, Fresenius Medical eGFR,60 ml/min per 1.73 m2 in the general population. Care, GlaxoSmithKline, Merck Sharp and Dohme-Chibret, Sanofi- However, because nephrologists probably handle drugs Genzyme, Lilly, Otsuka, and Vifor Fresenius, as well as speaker more carefully, the ADR incidence would likely have been honoraria at the French Society of Diabetology from Lilly and even higher in the broader population of patients not seen at the French-speaking Society of Nephrology, Dialysis and by nephrologists. Transplantation from MSD. All remaining authors have nothing The burden of ADRs is high in patients with moderate to disclose. to advanced CKD, and incidence was higher when CKD was severe. Our results highlight the major risk of Funding specific pharmacological classes, particularly antith- The CKD-REIN is funded by the Agence Nationale de la “ ’ ” rombotic agents, which must be used cautiously in Recherche through the 2010 Cohortes-Investissements d Avenir patients with CKD, especially at low eGFRs. Greater program (ANR-IA-COH-2012/3731) and by the 2010 National awareness by the medical community of the importance Programme Hospitalier de Recherche Clinique. The CKD-REIN is of eGFR level in prescribing medications, increased also supported through a public-private partnership with Amgen, involvement of pharmacists in systematically verifying Fresenius Medical Care, and GlaxoSmithKline since 2012; Lilly eGFR for patient prescriptions, and enhanced patient France since 2013; Otsuka Pharmaceutical since 2015; Baxter and fi education are key elements for preventing ADRs in this Merck Sharp and Dohme-Chibret from 2012 to 2017; Sano - population at high risk. The effect of ADRs on health Genzyme from 2012 to 2015; and Vifor Fresenius and AstraZeneca resources and patients’ quality of life requires fur- since 2018. French National Institute of Health and Medical Research ther evaluation. Transfert set up and has managed this partnership since 2011. A specific project on drug optimization in patients with CKD has been funded by the French National Agency for Medicines and Health Acknowledgments Products Safety. The authors thank the CKD-REIN study coordination staff for Supplemental Material their efforts in setting up the CKD-REIN cohort: M.M., Elodie This article contains the following supplemental material online at Speyer, Céline Lange, Sophie Renault, Reine Ketchemin, Natalia http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN. Alencar de Pinho, and all of the CRAs. We thank Jo Ann Cahn for 01030120/-/DCSupplemental. editing the English version. Supplemental Table 1. Algorithms to assess causation and pre- Dr. Solène Marie Laville, Prof . Sophie Liabeuf, Prof. Ziad Massy, ventability. Dr. Marie Metzger, and Dr. Bénédicte Stengel designed this project; Supplemental Table 2. Baseline characteristics of participants in Dr. Valérie Gras-Champel, Dr. Solène Marie Laville, Prof. Sophie the CKD-Renal Epidemiology and Information Network (be- Liabeuf, and Dr. Julien Moragny evaluated the serious ADRs; Dr. fore imputation). Solène Marie Laville, Prof. Sophie Liabeuf, and Dr. Marie Metzger Supplemental Table 3. Details of types of adverse drug reactions. analyzed the data; Dr. Valérie Gras-Champel, Dr. Solène Marie Supplemental Table 4. Descriptions of adverse drug reactions Laville, Prof. Sophie Liabeuf, Prof. Ziad Massy, Dr. Marie Metzger, according to the last eGFR reported before the reaction. Dr. Julien Moragny, and Dr. Bénédicte Stengel contributed to the Supplemental Table 5. Distribution of serious adverse drug re- interpretation of the results; Dr. Solène Marie Laville, Prof. Sophie actions causing or resulting from hospitalization according to the Liabeuf, Prof. Ziad Massy, and Dr. Bénédicte Stengel wrote the first last eGFR reported before the reaction (n5145). draft of the article; Prof. Christian Combe, Prof. Denis Fouque, Prof. Supplemental Table 6. Details of imputed drugs responsible for Luc Frimat, Dr. Valérie Gras-Champel, Dr. Christian Jacquelinet, adverse drug reactions and of type of adverse drug reaction for the Prof. Maurice Laville, Dr. Solène Marie Laville, Prof. Sophie Liabeuf, five pharmacological classes most frequently imputed. Prof. Ziad Massy, Dr. Marie Metzger, Dr. Julien Moragny, Dr. Supplemental Table 7. Sensitivity analyses by the Prentice, Wil- Robinson, and Dr. Bénédicte Stengel provided critical feedback; liams, and Peterson gap-time recurrent event time-to-event analysis. Prof. Christian Combe, Prof. Denis Fouque, Prof. Luc Frimat, Dr. Valérie Gras-Champel, Dr. Christian Jacquelinet, Prof. Maurice Laville, Dr. Solène Marie Laville, Prof. Sophie Liabeuf, Prof. Ziad References Massy, Dr. Marie Metzger, Dr. Julien Moragny, Dr. Robinson, and 1. Lazarou J, Pomeranz BH, Corey PN: Incidence of adverse drug Dr. Bénédicte Stengel helped shape the research, analysis, and final reactions in hospitalized patients: A meta-analysis of prospective draft of the manuscript; Prof. Christian Combe, Prof. Denis Fouque, studies. JAMA 279: 1200–1205, 1998 2. Pouyanne P, Haramburu F, Imbs JL, Be´gaud B: Admissions to Prof. Luc Frimat, Dr. Valérie Gras-Champel, Dr. Christian Jacque- hospital caused by adverse drug reactions: Cross sectional in- linet, Prof. Maurice Laville, Dr. Solène Marie Laville, Prof. Sophie cidence study. French Pharmacovigilance Centres. BMJ 320: Liabeuf, Prof. Ziad Massy, Dr. Marie Metzger, Dr. Julien Moragny, 1036, 2000 Dr. Robinson, and Dr. Bénédicte Stengel approved the version to be 3. Kongkaew C, Noyce PR, Ashcroft DM: Hospital admissions as- published; and Dr. Bénédicte Stengel is the principal investigator. sociated with adverse drug reactions: A systematic review of CJASN 15: 1090–1102, August, 2020 Adverse Drug Reactions in CKD, Laville et al. 1101

prospective observational studies. Ann Pharmacother 42: D, Gro¨nhagen-Riska C, Dasgupta T, Lewis D, Herrington W, 1017–1025, 2008 Mafham M, Majoni W, Wallendszus K, Grimm R, Pedersen T, 4. Tache´ SV,So¨nnichsen A, Ashcroft DM: Prevalence of adverse drug Tobert J, Armitage J, Baxter A, Bray C, Chen Y, Chen Z, Hill M, events in ambulatory care: A systematic review. Ann Pharmac- Knott C, Parish S, Simpson D, Sleight P,Young A, Collins R; SHARP other 45: 977–989, 2011 Investigators: The effects of lowering LDL cholesterol with sim- 5. Beńard-Laribie`re A, Miremont-Salame´ G, Pe´rault-PochatM-C, vastatin plus ezetimibe in patients with chronic kidney disease Noize P,Haramburu F; EMIR Study Group on behalf of the French (Study of Heart and Renal Protection): A randomised placebo- network of pharmacovigilance centres: Incidence of hospital controlled trial. Lancet 377: 2181–2192, 2011 admissions due to adverse drug reactions in France: The EMIR 22. Liabeuf S, RyckelynckJ-P, El Esper N, Urena~ P, Combe C, Dussol B, study. Fundam Clin Pharmacol 29: 106–111, 2015 Fouque D, Vanhille P, Frimat L, Thervet E, Mentaverri R, Prie´ D, 6. Edwards IR, Aronson JK: Adverse drug reactions: Definitions, Choukroun G; FRENCH Study collaborators: Randomized clinical diagnosis, and management. Lancet 356: 1255–1259, 2000 trial of sevelamer Carbonate on serum Klotho and fibroblast growth 7. Chan SL, Ang X, Sani LL, Ng HY,Winther MD, Liu JJ, Brunham LR, factor 23 in CKD. Clin J Am Soc Nephrol 12: 1930–1940, 2017 Chan A: Prevalence and characteristics of adverse drug reactions 23. Sharif-Askari FS, Syed Sulaiman SA, Saheb Sharif-Askari N, Al at admission to hospital: A prospective observational study. Br Sayed Hussain A: Development of an adverse drug reaction risk J Clin Pharmacol 82: 1636–1646, 2016 assessment score among hospitalized patients with chronic kid- 8. Alexopoulou A, Dourakis SP, Mantzoukis D, Pitsariotis T, Kandyli ney disease. PLoS One 9: e95991, 2014 A, Deutsch M, Archimandritis AJ: Adverse drug reactions as a 24. Pecoits-Filho R, Fliser D, Tu C, Zee J, Bieber B, Wong MMY,Port F, cause of hospital admissions: A 6-month experience in a single Combe C, Lopes AA, Reichel H, Narita I, Stengel B, Robinson BM, center in Greece. Eur J Intern Med 19: 505–510, 2008 Massy Z; CKDopps Investigators: Prescription of renin-angio- 9. Onder G, Pedone C, Landi F, Cesari M, Della Vedova C, Bernabei tensin-aldosterone system inhibitors (RAASi) and its determinants R, Gambassi G: Adverse drug reactions as cause of hospital ad- in patients with advanced CKD under nephrologist care. J Clin missions: Results from the Italian group of Pharmacoepidemiol- Hypertens (Greenwich) 21: 991–1001, 2019 ogy in the elderly (GIFA). J Am Geriatr Soc 50: 1962–1968, 2002 25. Stengel B, Metzger M, Combe C, Jacquelinet C, Briançon S, Ayav 10. Pirmohamed M, James S, Meakin S, Green C, Scott AK, Walley TJ, C, Fouque D, Laville M, Frimat L, Pascal C, HerpeY-E´, Morel P, Farrar K, Park BK, Breckenridge AM: Adverse drug reactions as DeleuzeJ-F, Schanstra J, Lange C, Legrand K, Speyer E, Liabeuf S, cause of admission to hospital: Prospective analysis of 18 820 Robinson BM, Massy ZA: Risk profile, quality of life and care of patients. BMJ 329: 15–19, 2004 patients with moderate and advanced CKD: The French CKD- 11. Corsonello A, Pedone C, Corica F, Mussi C, Carbonin P, Antonelli REIN Cohort Study. Nephrol Dial Transplant 34: 277–286, 2019 Incalzi R; Gruppo Italiano di Farmacovigilanza nell’Anziano 26. Girerd X, Radauceanu A, Achard JM, Fourcade J, Tournier B, (GIFA) Investigators: Concealed renal insufficiency and adverse Brillet G, Silhol F, Hanon O: [Evaluation of patient compliance drug reactions in elderly hospitalized patients. Arch Intern Med among hypertensive patients treated by specialists]. Arch Mal 165: 790–795, 2005 Coeur Vaiss 94: 839–842, 2001 12. Danial M, Hassali MA, Ong LM, Khan AH: Survivability of hos- 27. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF 3rd, pitalized chronic kidney disease (CKD) patients with moderate to Feldman HI, Kusek JW, Eggers P, Van Lente F, Greene T, Coresh J; severe estimated glomerular filtration rate (eGFR) after experi- CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration): encing adverse drug reactions (ADRs) in a public healthcare A new equation to estimate glomerular filtration rate [published center: A retrospective 3 year study. BMC Pharmacol Toxicol 19: correction appears in Ann Intern Med 155: 408, 2011]. Ann Intern 52, 2018 Med 150: 604–612, 2009 13. Bouvy JC, De Bruin ML, Koopmanschap MA: Epidemiology of 28. World Health Organization Collaborating Centre for Drug Sta- adverse drug reactions in Europe: A reviewof recent observational tistics Methodology: Guidelines for ATC classification and DDD studies. Drug Saf 38: 437–453, 2015 assignment 2016, 20th Ed, 2016. Available at: www.whocc.no/ 14. Fattinger K, Roos M, Verge`res P, Holenstein C, Kind B, Masche U, atc_ddd_index/. Accessed September 10, 2019 Stocker DN, Braunschweig S, Kullak-Ublick GA, Galeazzi RL, 29. Couchoud C, Stengel B, Landais P, AldigierJ-C, de Cornelissen F, Follath F, Gasser T, Meier PJ: Epidemiology of drug exposure and Dabot C, Maheut H, Joyeux V, Kessler M, Labeeuw M, Isnard H, adverse drug reactions in two swiss departments of internal Jacquelinet C: The renal epidemiology and information network medicine. Br J Clin Pharmacol 49: 158–167, 2000 (REIN): A new registry for end-stage renal disease in France. 15. Pierson-Marchandise M, Gras V, Moragny J, Micallef J, Gaboriau Nephrol Dial Transplant 21: 411–418, 2006 L, Picard S, Choukroun G, Masmoudi K, Liabeuf S; French Na- 30. World Health Organization Quality Assurance and Safety of tional Network of Pharmacovigilance Centres: The drugs that Medicines Team: Safety of medicines: A guide to detecting and mostly frequently induce acute kidney injury: A case - noncase reporting adverse drug reactions: Why health professionals need study of a pharmacovigilance database. Br J Clin Pharmacol 83: to take action, 2002. Available at: https://apps.who.int/iris/ 1341–1349, 2017 bitstream/handle/10665/67378/WHO_EDM_QSM_2002.2.pdf? 16. LeveyAS,CoreshJ: Chronickidney disease.Lancet379:165–180, sequence51&isAllowed5y. Accessed September 10, 2019 2012 31. Be´gaud B, Evreux JC, Jouglard J, Lagier G: [Imputation of the 17. Hassan Y, Al-Ramahi R, Abd Aziz N, Ghazali R: Drug use and unexpected or toxic effects of drugs. Actualization of the method dosing in chronic kidney disease. Ann Acad Med Singapore 38: used in France]. Therapie 40: 111–118, 1985 1095–1103, 2009 32. Naranjo CA, Busto U, Sellers EM, Sandor P,Ruiz I, Roberts EA, Janecek 18. Nolin TD: A synopsis of clinical pharmacokinetic alterations in E, Domecq C, Greenblatt DJ: A method for estimating the probability advanced CKD. Semin Dial 28: 325–329, 2015 of adverse drug reactions. Clin Pharmacol Ther 30: 239–245, 1981 19. Laville SM, Metzger M, Stengel B, Jacquelinet C, Combe C, 33. Olivier P,Caron J, Haramburu F, ImbsJ-L, Jonville-Be´raA-P,Lagier Fouque D, Laville M, Frimat L, Ayav C, Speyer E, Robinson BM, G, Sgro C, Vial T,MontastrucJ-L, Lapeyr-Mestre M: [Validation of a Massy ZA, Liabeuf S; Chronic Kidney Disease-Renal Epidemi- measurement scale: Example of a French Adverse Drug Reactions ology and Information Network (CKD-REIN) Study Collaborators: Preventability Scale]. Therapie 60: 39–45, 2005 Evaluation of the adequacy of drug prescriptions in patients with 34. Prentice RL, Williams BJ, Peterson AV: On the regression analysis chronic kidney disease: Results from the CKD-REIN cohort. Br of multivariate failure time data. Biometrika 68: 373–379, 1981 J Clin Pharmacol 84: 2811–2823, 2018 35. Yang W,Jepson C, Xie D, Roy JA, Shou H, Hsu JY,Anderson AH, Landis 20. Ginsberg JS, Zhan M, Diamantidis CJ, Woods C, Chen J, Fink JC: JR, He J, Feldman HI; Chronic Renal Insufficiency Cohort (CRIC) Study Patient-reported and actionable safety events in CKD. JAmSoc Investigators: Statistical methods for recurrent event analysis in cohort Nephrol 25: 1564–1573, 2014 studies of CKD. Clin J Am Soc Nephrol 12: 2066–2073, 2017 21. Baigent C, Landray MJ, Reith C, Emberson J, Wheeler DC, Tomson 36. vanBuurenS:Multipleimputationofdiscreteandcontinuous databyfully C, Wanner C, Krane V, Cass A, Craig J, Neal B, Jiang L, Hooi LS, conditional specification. Stat Methods Med Res 16: 219–242, 2007 Levin A, Agodoa L, Gaziano M, Kasiske B, Walker R, Massy ZA, 37. Whittaker CF, Miklich MA, Patel RS, Fink JC: Medication safety Feldt-Rasmussen B, Krairittichai U, Ophascharoensuk V, principles and practice in CKD. Clin J Am Soc Nephrol 13: Fellstro¨m B, Holdaas H, Tesar V,Wiecek A, Grobbee D, de Zeeuw 1738–1746, 2018 1102 CJASN

38. Tonelli M, Wiebe N, Manns BJ, Klarenbach SW,James MT,Ravani P, regulator of tissue factor stability and an antithrombotic target in Pannu N, Himmelfarb J, Hemmelgarn BR: Comparison of the uremia. J Am Soc Nephrol 27: 189–201, 2016 complexity of patients seen by different medical subspecialists in a 48. Hazell L, Shakir SAW: Under-reporting of adverse drug reactions: universal health care system. JAMA Netw Open 1: e184852, 2018 A systematic review. Drug Saf 29: 385–396, 2006 39. Schmidt IM, Hu¨bner S, Nadal J, Titze S, Schmid M, Ba¨rthlein B, 49. Agbabiaka TB, SavovicJ, Ernst E: Methods forcausalityassessment Schlieper G, Dienemann T, Schultheiss UT, Meiselbach H, of adverse drug reactions: A systematic review. Drug Saf 31: Ko¨ttgen A, Flo¨ge J, Busch M, Kreutz R, Kielstein JT, EckardtK-U: 21–37, 2008 Patterns of medication use and the burden of polypharmacy in 50. Hutchinson TA, Flegel KM, HoPingKong H, Bloom WS, Kramer patients with chronic kidney disease: The German chronic kidney MS, Trummer EG: Reasons for disagreement in the standardized disease study. Clin Kidney J 12: 663–672, 2019 assessment of suspected adverse drug reactions. Clin Pharmacol 40. Schneider MP,Hu¨bner S, Titze SI, Schmid M, Nadal J, Schlieper G, Ther 34: 421–426, 1983 Busch M, Baid-Agrawal S, Krane V, Wanner C, Kronenberg F, 51. World Health Organization: Nutritional Anemias. Report of a WHO Scientific Group, Geneva, Switzerland, World Health EckardtK-U: Implementation of the KDIGO guideline on lipid Organization, 1968 management requires a substantial increase in statin prescription rates. Kidney Int 88: 1411–1418, 2015 Received: January 24, 2019 Accepted: May 13, 2020 41. Massy ZA, Ferrie`res J, Bruckert E, Lange C, Liabeuf S, Velkovski- Rouyer M, Stengel B; CKD-REIN Collaborators: Achievement of low-density lipoprotein cholesterol targets inCKD. KidneyInt Rep *The CKD-REIN Study Group steering committee and coordinators 4: 1546–1554, 2019 are as follows: Natalia Alencar De Pinho, Carole Ayav, Serge 42. FriedTR,O’Leary J,TowleV,GoldsteinMK, Trentalange M, Martin Briançon, Dorothée Cannet, C.C., D.F., L.F., Yves-Edouard Herpe, DK: Health outcomes associated with polypharmacy in C.J., M.L., Z.A.M., Christophe Pascal, B.M.R., B.S., Céline Lange, community-dwelling older adults: A systematic review. JAm Karine Legrand, S.L., M.M., and Elodie Speyer. The CKD-REIN Geriatr Soc 62: 2261–2272, 2014 investigators/collaborators are as follows: Thierry Hannedouche, 43. Brookhart MA, Patrick AR, Dormuth C, Avorn J, Shrank W, Bruno Moulin, Sébastien Mailliez, Gaétan Lebrun, Eric Magnant, Cadarette SM, Solomon DH: Adherence to lipid-lowering therapy Gabriel Choukroun, Benjamin Deroure, Adeline Lacraz, Guy and the use of preventive health services: An investigation of the Lambrey, Jean Philippe Bourdenx, Marie Essig, Thierry Lobbedez, healthy user effect. Am J Epidemiol 166: 348–354, 2007 Raymond Azar, Hacène Sekhri, Mustafa Smati, Mohamed Jamali, 44. Corsonello A,Pedone C,LattanzioF,OnderG, Antonelli Incalzi R; Alexandre Klein, Michel Delahousse, C.C., Séverine Martin, Isabelle Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA): As- Landru, Eric Thervet, Z.A.M., Philippe Lang, Xavier Belenfant, sociation between glomerular filtration rate and adverse drug Pablo Urena, Carlos Vela, L.F., Dominique Chauveau, Viktor reactions in elderly hospitalized patients: The role of the esti- Panescu, Christian Noel, François Glowacki, Maxime Hoffmann, mating equation. Drugs Aging 28: 379–390, 2011 45. Fawzy AM, Lip GYH: Pharmacokinetics and pharmacodynamics Maryvonne Hourmant, Dominique Besnier, Angelo Testa, François of oral anticoagulants used in atrial fibrillation. Expert Opin Drug Kuentz, Philippe Zaoui, Charles Chazot, Laurent Juillard, Stéphane Metab Toxicol 15: 381–398, 2019 Burtey, Adrien Keller, Nassim Kamar, D.F., and M.L. 46. Liabeuf S, Neirynck N, Druëke TB, Vanholder R, Massy ZA: Clinical studies and chronic kidney disease: What did we learn Published online ahead of print. Publication date available at recently? Semin Nephrol 34: 164–179, 2014 www.cjasn.org. 47. Shivanna S, Kolandaivelu K, Shashar M, Belghasim M, Al-Rabadi L, BalcellsM, ZhangA,WeinbergJ,FrancisJ,PollastriMP,Edelman ER, See related editorial, “Adverse Drug Effects in Patients with CKD: Sherr DH, Chitalia VC: The aryl Hydrocarbon receptor is a critical Primum Non Nocere,” on pages 1075–1077.

AFFILIATIONS

1Paris-Saclay University, Versailles Saint-Quentin-en-Yvelines University, National Institute of Health and Medical Research, Center for research in Epidemiology and Population Health (CESP), Clinical Epidemiology Team, Villejuif, France 2Department of Clinical Pharmacology, Amiens University Hospital, Amiens, France 3Renal Epidemiology and Information Network Registry, Biomedicine Agency, Saint Denis, France 4Department of Nephrology Transplantation Dialysis, Centre Hospitalier Universitaire de Bordeaux, Bordeaux, France 5Inserm Unit 1026, University of Bordeaux Segalen, Bordeaux, France 6Nephrology Department, Centre Hospitalier Lyon Sud, Universite´ de Lyon, Carmen, Pierre-Be´nite, France 7Nephrology Department, Centre Hospitalier Re´gional Universitaire de Nancy, Vandoeuvre-le`s-Nancy, France 8Lorraine University, APEMAC, Vandoeuvre-le`s-Nancy, France 9Arbor Research Collaborative for Health, Ann Arbor, Michigan 10Division of Nephrology, Ambroise Pare´ University Hospital, Assistance publique – Hoˆpitaux de Paris, Boulogne-Billancourt/Paris, France 11MP3CV Laboratory, EA7517, University of Picardie Jules Verne, Amiens, France Adverse Drug Reactions in Patients with Chronic Kidney Disease

S.M. Laville, V. Gras-Champel, J. Moragny, M. Metzger, C. Jacquelinet, C. Combe, D. Fouque, M. Laville, L. Frimat, B.M. Robinson, B. Stengel, Z.A Massy, S. Liabeuf, on behalf of the Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) Study Group

Clinical Journal of the American Society of Nephrology

Corresponding author: Bénédicte Stengel Director, Inserm U1018, Team 5 UPS-UVSQ, CESP, Centre for Epidemiology and Population Health EpReC Team, Renal and Cardiovascular Epidemiology 16, avenue P. Vaillant Couturier F-94 807 Villejuif cedex France Phone: + 33-145-595-039 E-mail: [email protected]

1 Supplemental Material Table of Contents

Supplementary Material 1: Algorithms to assess causation and preventability

Supplementary Material 2: Baseline characteristics of participants in the Chronic

Kidney Disease-Renal Epidemiology and Information Network (before imputation)

Supplementary Material 3: Details of types of adverse drug reactions

Supplementary Material 4: Descriptions of adverse drug reactions according to the last estimated glomerular filtration rate reported before the reaction

Supplementary Material 5: Distribution of serious adverse drug reactions causing or resulting from hospitalization according to the last estimated glomerular filtration rate reported before the reaction (n=145).

Supplementary Material 6: Details of imputed drugs responsible for adverse drug reactions

Supplementary Material 7: Sensitivity analyses by the Prentice, Williams, and

Peterson (PWP) gap-time recurrent event time-to-event analysis

2 Supplementary Material 1: Algorithms to assess causation and preventability

A. Bégaud imputability method1

Table a - Decision table for the chronological criteria (C)

Event onset CHALLENGE: Very suggestive Compatible Incompatible Rechallenge (R) DECHALLENGE (DISCONTINUE) R (+) R (0) R (-) R (+) R (0) R (-) Suggestive: C C C C C C C Regression appears linked to drug discontinuation 3 3 1 3 2 1 0 Inconclusive: Regression seems spontaneous or induced by nonspecific treatment known to be effective, or course C3 C2 C1 C3 C1 C1 C0 unknown or follow up too brief or lesions irreversible (or drug not discontinued) Not suggestive: No regression of reversible event (or complete C1 C1 C1 C1 C1 C1 C0 regression without drug withdrawal) R (+): positive, the event recurs; R (0): no rechallenge or lack of information; R (-): negative, the event does not recur. C3: suggestive chronology; C2: possible chronology; C1: dubious chronology; C0: incompatible chronology.

Table b - Decision table for the semiological criteria (S)

Suggestive of the drug SEMIOLOGY (clinical or extraclinical): studied (and/or very Other cases favorable factor) RELIABLE AND SPECIFIC LABORATORY TEST (L) ALTERNATE NON-DRUG RELATED EXPLANATION L (+) L (0) L (-) L (+) L (0) L (-)

None (after an appropriate search) S3 S3 S1 S3 S2 S1 Possible or present S3 S2 S1 S3 S1 S1 L (+): positive laboratory test; L (0): no such test for the event-drug pair under consideration; L (-): negative laboratory test (if it is sensitive enough). S3: suggestive semiology; S2: possible semiology; S1: dubious semiology.

Table c - Intrinsic imputability decision table This is obtained from the score of the chronological (C - Table a.) and semiological (S – Table b.) imputabilities.

Semiology Chronology S1 S2 S3 C0 I0 I0 I0 C1 I1 I1 I2 C2 I1 I2 I3 C3 I3 I3 I4 The drug-effect relation can be: I4: very likely; I3: likely; I2: possible; I1: dubious; I0: unlikely (appears excluded).

3 B. Naranjo’s adverse drug reaction probability scale2

To assess the adverse drug reaction, please answer the following questionnaire and give the pertinent score Yes No Do not know Score 1. Are there previous conclusive reports on this reaction? +1 0 0 2. Did the adverse event appear after the suspected drug was +2 -1 0 administered? 3. Did the adverse reaction improve when the drug was +1 0 0 discontinued or a specific antagonist was administered? 4. Did the adverse reaction reappear when the drug was +2 -1 0 readministered? 5. Are there alternative causes (other than the drug) that could -1 +2 0 on their own have caused the reaction? 6. Did the reaction reappear when a placebo was given? -1 +1 0 7. Was the drug detected in the blood (or other fluids) in +1 0 0 concentrations known to be toxic? 8. Was the reaction more severe when the dose was increased, +1 0 0 or less severe when the dose was decreased? 9. Did the patient have a similar reaction to the same or similar +1 0 0 drugs in any previous exposure? 10. Was the adverse event confirmed by any objective evidence? +1 0 0 Total score

Score intervals Label ≥ 9 Definite reaction 5 to 8 Probable reaction 1 to 4 Possible reaction 0 Doubtful reaction

4 C. Olivier adverse drug reaction (ADR) preventability scale3

THE DRUG Score A- Adherence with recommendations a. Neither adherence with recommendations nor lack of precaution played any role in ADR occurrence +3 b. Not assessable 0 c. The prescriber or the patient ignored relevant recommendations -5

THE PATIENT B- Other risk factors identified in the patients a. Present, easy to detect -3 b. Present, difficult to detect -1 c. Absent +2 d. Not assessable 0

C- Adaptation of the prescription to the patient's living conditions or environment a. Correct +1 b. Not assessable 0 c. Inadequate -1

THE PRESCRIPTION D- This medication (prescribed or not) is probably essential for the patient a. Yes +2 b. Not assessable 0 c. No -4 TOTAL

Score intervals Label -13 to -8 Preventable ADR -7 to -3 Potentially preventable ADR -2 to +2 Not assessable +3 to +8 Not preventable

1. Bégaud B, Evreux JC, Jouglard J, Lagier G: Imputation of the unexpected or toxic effects of drugs. Actualization of the method used in France. Therapie 40: 111–118, 1985 2. Naranjo CA, Busto U, Sellers EM, Sandor P, Ruiz I, Roberts EA, Janecek E, Domecq C, Greenblatt DJ: A method for estimating the probability of adverse drug reactions. Clinical Pharmacology & Therapeutics [Internet] 30: 239–245, 1981 Available from: http://ascpt.onlinelibrary.wiley.com/doi/abs/10.1038/clpt.1981.154 [cited 2018 Sep 26] 3. Olivier P, Caron J, Haramburu F, Imbs J-L, Jonville-Béra A-P, Lagier G, Sgro C, Vial T, Montastruc J- L, Lapeyre-Mestre M: Validation d’une échelle de mesure : exemple de l’échelle française d’évitabilité des effets indésirables médicamenteux. Thérapie [Internet] 60: 39–45, 2005 Available from: http://www.sciencedirect.com/science/article/pii/S0040595716302669 [cited 2018 Sep 26]

5 Supplementary Material 2: Baseline characteristics of participants in the Chronic Kidney Disease-Renal Epidemiology and Information Network (before imputation)

Baseline eGFR in mL/min per 1.73m² All ≥30 <30 p-value (n=3033) (n=1670) (n=1363) Age (years) 69 [60 - 76] 68 [59 - 75] 70 [61 - 78] 0.001 <60 years 24% 25% 22% 60 to 75 years 46% 48% 44% ≥ 75 years 30% 27% 35% Men 65% 67% 63% 0.02 High school diploma or higher 36% 38% 32% 0.001 Missing 2% 2% 1% BMI 28 [25 - 32] 28 [25 - 32] 28 [25 - 32] 0.30 ≥ 30 kg/m² 35% 33% 36% Missing 2% 2% 2% Serum Albumin 4.0 [3.8 – 4.3] 4.1 [3.8 – 4.3] 4.0 [3.7 – 4.3] <0.001 <3.5 g/dL 8% 6% 9% Missing 19% 21% 17% UACR <0.001 < 30 mg/g 5% 33% 14% 30 – 300 mg/g 28% 29% 27% > 300 mg/g 36% 27% 48% Missing 11% 11% 11% Anemia* 40% 29% 55% <0.001 Missing 1% 1% 1% Smoking status 0.21 Smoker 12% 12% 12% Nonsmoker 41% 42% 40% Ex-smoker 46% 46% 47% Missing 1% 0.5% 1% Diabetes 43% 42% 44% 0.43 Missing 0.2% 0.2% 0.3% AKI history 22% 20% 24% 0.001 Missing 8% 7% 9% Cardiovascular history 53% 51% 55% 0.06 Missing 1% 1% 1% Hypertension 91% 89% 92% 0.02 Missing 0.2% 0.3% 0.2% Dyslipidemia 73% 73% 74% 0.35 Missing 0.5% 1% 0.3% Number of drugs 8 [5 - 10] 7 [5 - 10] 8 [6 - 11] <0.001 < 5 drugs 19% 24% 13% 5 to 10 drugs 56% 54% 57% > 10 drugs 24% 21% 30% Missing 1% 1% 0.2% Poor adherence to medications 1129 (37%) 649 (39%) 480 (35%) 0.02 Missing 31 (1%) 22 (1%) 9 (1%)

AKI: Acute kidney injury, BMI: Body mass index, eGFR: estimated glomerular filtration rate using CKD-EPI equation, UACR: Urine albumin to creatinine ratio Median (Interquartile range, IQR) or n (%) *Anemia is defined by the 1968 WHO definition [World Health Organization. Nutritional anaemias. Report of a WHO scientific group. Geneva, World Health Organization; 1968]: <12 g/dL for women and <13 g/dL for men.

6 Supplementary Material 3: Details of types of adverse drug reactions

Frequency Type of ADR (n=751) Renal and urinary disorders 150 (20%) Acute kidney injury 102 Blood creatinine increased 40 Renal impairment 4 Chronic kidney disease 1 Renal tubular necrosis 1 Nephroangiosclerosis 1 Urinary retention 1 Gastrointestinal disorders 119 (16%) Diarrhea 57 Gastrointestinal disorder 24 Constipation 10 Nausea 10 Vomiting 7 Abdominal pain 2 Gingival hypertrophy 2 Discolored feces 1 Colitis 1 Abdominal pain upper 1 Lip edema 1 Pancreatitis 1 Pancreatitis acute 1 Aphthous ulcer 1 Musculoskeletal and connective tissue disorders 68 (9%) Muscle spasms 35 Myalgia 22 Arthralgia 4 Tendinous discomfort 2 Tendon pain 1 Pain in extremity 1 Muscular weakness 1 Rheumatoid arthritis 1 Periarthritis 1 Bleeding 67 (9%) Epistaxis 11 Muscle hemorrhage 6 Hematuria 5 Hemorrhagic shock 4 Gastrointestinal hemorrhage 4 Hematoma 3 Subcutaneous hematoma 3 Peritoneal hemorrhage 3 Ecchymosis 2 Hemarthrosis 2 Cerebral hematoma 2 Subdural haematoma 2 Hemoptysis 2 Rectal hemorrhage 2 Gingival bleeding 2 Implant site hematoma 1 Injection site hematoma 1 Abdominal wall hematoma 1 Pulmonary alveolar hemorrhage 1

7 Incision site hemorrhage 1 Arteriovenous fistula site hemorrhage 1 Conjunctival hemorrhage 1 Eye hemorrhage 1 Adrenal hemorrhage 1 Brain stem hemorrhage 1 Wound hemorrhage 1 Upper gastrointestinal hemorrhage 1 Melena 1 Hemorrhagic esophagitis 1 General disorders and administration site conditions 58 (8%) Edema, peripheral 30 Drug intolerance 8 Malaise 6 Fatigue 5 Asthenia 4 Injection site pain 1 General physical health deterioration 1 Vessel puncture site hematoma 1 Drug ineffective 1 Edema 1 Metabolism and nutrition disorders 44 (6%) Hypoglycemia 11 Hyperkalemia 10 Gout 6 Hypercalcemia 5 Decreased appetite 2 Hypokalemia 2 Polydipsia 2 Metabolic acidosis 1 Weight loss poor 1 Diabetes mellitus inadequate control 1 Dehydration 1 Hypernatremia 1 Hypocalcemia 1 Vascular disorders 40 (5%) Hypotension 21 Orthostatic hypotension 13 Hot flush 4 Flushing 1 Hypertension 1 Nervous system disorders 36 (5%) Somnolence 9 Headache 5 Dizziness 5 Paraesthesia 3 Neuropathy peripheral 2 Cholinergic syndrome 2 Depressed level of consciousness 1 Encephalopathy 1 Toxic encephalopathy 1 Hyperreflexia 1 Burning sensation 1 Neurological symptom 1 Anticholinergic syndrome 1 Tremor 1 Nervous system disorder 1 Status epilepticus 1

8 Injury, poisoning and procedural complications 35 (5%) Overdose 33 International normalized ratio increased 1 Prothrombin time shortened 1 Skin and subcutaneous tissue disorders 34 (5%) Rash 6 Pruritus 4 Dermatitis allergic 3 Urticaria 3 Toxic skin eruption 3 Angioedema 2 Eczema 2 Skin reaction 2 Skin atrophy 1 Dermatitis bullous 1 Swelling face 1 Hyperhidrosis 1 Pemphigoid 1 Pruritus allergic 1 Photosensitivity reaction 1 Red man syndrome 1 Skin ulcer 1 Respiratory, thoracic and mediastinal disorders 21 (3%) Cough 14 Throat irritation 2 Interstitial lung disease 2 Dyspnea 1 Nasal discomfort 1 Lung disorder 1 Blood and lymphatic system disorders 14 (2%) Thrombocytopenia 4 Anemia 3 Eosinophilia 2 Agranulocytosis 1 Febrile bone marrow aplasia 1 Bicytopenia 1 Febrile neutropenia 1 Polycythemia 1 Cardiac disorders 10 (1%) Bradycardia 9 Tachycardia 1 Ear and labyrinth disorders 10 (1%) Vertigo 9 Tinnitus 1 Endocrine disorders 10 (1%) Hyperthyroidism 4 Hypothyroidism 4 Adrenal insufficiency 1 Blood thyroid stimulating hormone decreased 1 Psychiatric disorders 9 (1%) Confusional state 4 Insomnia 3 Abnormal behaviour 1 Loss of libido 1 Investigations 7 (1%) Creatinine renal clearance decreased 1 Blood creatine phosphokinase increased 1 Gamma-glutamyltransferase increased 1

9 Electrocardiogram QT prolonged 1 Lipase increased 1 Hormone level abnormal 1 Urine output increased 1 Immune system disorders 5 (1%) Hypersensitivity 2 Anaphylactic reaction 2 Anaphylactic shock 1 Hepatobiliary disorders 5 (1%) Hepatocellular injury 3 Cholestasis 1 Drug-induced liver injury 1 Infections and infestations 4 (1%) Chorioretinitis 1 Upper respiratory tract infection 1 Oral fungal infection 1 Rash pustular 1 Eye disorders 3 (1%) Eye allergy 1 Visual acuity decreased 1 Retinal toxicity 1 Reproductive system and breast disorders 2 (0.3%) Priapism 1 Retrograde ejaculation 1

Results are expressed as n (%).

10 Supplementary Material 4: Descriptions of adverse drug reactions according to the last estimated glomerular filtration rate reported before the reaction.

Last eGFR before ADR in mL/min/1.73m² ADRs in patients with ADRs in patients with All ADRs Type of ADRs eGFR ≥30 eGFR <30 (n=751) (n=314) (n= 437) Renal and urinary disorders 150 (20%) 51 (16%) 99 (23%) Acute kidney injury 102 32 70 Increased serum creatinine 40 16 24 Other type of renal and urinary disorders 8 3 5 Gastrointestinal disorders 119 (16%) 61 (19%) 58 (13%) Diarrhea 57 35 22 Gastrointestinal conditions 24 8 16 Other type of gastrointestinal disorders 38 18 20 Musculoskeletal and connective tissue disorders 68 (9%) 34 (11%) 34 (8%) Contractures 35 20 15 Muscle pain 22 10 12 Other type of musculoskeletal and connective tissue 11 4 7 disorders Hemorrhages and bleeding 67 (9%) 20 (6%) 47 (11%) Hemorrhages 34 9 25 Hematoma 19 8 11 Other type of hemorrhages and bleeding 14 3 11 General disorders and administration site conditions 58 (8%) 24 (8%) 34 (8%) Peripheral edema 30 12 18 Drug intolerance 8 3 5 Other type of general disorders and administration site 20 9 11 conditions Other type of ADRs 289 (38%) 124 (40%) 165 (38%) ADRs: adverse drug reactions, eGFR: estimated glomerular filtration rate expressed in mL/min/1.73m² Results are expressed as n (%). The denominator used in column 1 is the total number of ADRs, and in columns 2 and 3, those of the ADRs in patients with eGFR < vs ≥ 30 respectively. P-value=0.015 tests the difference in the distribution of ADR type according to patient eGFR at ADR occurrence. The median time between the GFR estimate and the ADR was 36 (interquartile range, 11-97) days.

11 Supplementary Material 5: Distribution of serious adverse drug reactions causing or resulting from hospitalization according to the last estimated glomerular filtration rate reported before the reaction (n=145).

Renal and urinary disorders were mostly acute kidney injuries or increased serum creatinine. Diarrhea was the most common gastrointestinal disorder. Blood and lymphatic system disorders consist of anemia, febrile neutropenia, thrombocytopenia, agranulocytosis, febrile aplasia, and bicytopenia. Cardiac disorders were bradycardia. Other disorders are described in the supplementary material 3. The median time between the GFR estimate and the ADR was 36 (interquartile range, 11-97) days. Results are expressed as %. The denominators used are the total number of ADRs in patients with eGFR < or ≥ 30 mL/min/1.73m2.

12 Supplementary Material 6: Details of imputed drugs responsible for adverse drug reactions and of type of adverse drug reaction for the 5 pharmacological classes most frequently imputed.

Preventability for serious ADR Frequency Serious ATC Pharmacological classes Not (n=751) (n=150) Preventable* Inevitable assessable C09 Agents acting on the renin-angiotensin system 115 (15.3%) 18 7 2 9 Acute kidney injury 25 15 6 2 7 Blood creatinine increased 19 0 Hypotension 12 0 Cough 12 0 Diarrhea 9 0 Other 38 3 2 0 1 B01 Antithrombotic agents 107 (14.2%) 51 13 19 19 Overdose/INR increased/ 31 2 2 0 0 Prothrombin time shortened Epistaxis 11 6 0 4 2 Muscle hemorrhage 6 5 2 1 2 Hematuria 4 3 0 1 2 Gastrointestinal hemorrhage 4 4 1 0 3 Other 53 31 8 13 10 C03 Diuretics 77 (10.3%) 18 6 6 6 Acute kidney injury 40 18 6 6 6 Blood creatinine increased 11 0 Gout 5 0 Hyperkalemia 4 0 Hypotension 3 0 Other 14 0 C10 Lipid modifying agents 50 (6.7%) 0 0 0 0 Muscle spasms 27 0 Myalgia 16 0 Other 7 0 C08 Calcium channel blockers 45 (6.0%) 0 0 0 0 Edema peripheral 26 0 Orthostatic hypotension 4 0 Other 15 0 J01 Antibacterials for systemic use 36 (4.8%) 11 4 6 1 A10 Drugs used in diabetes 31 (4.1%) 2 2 0 0 N02 Analgesics 29 (3.9%) 6 2 3 1 L04 Immunosuppressants 27 (3.6%) 6 1 1 4 M04 Antigout preparations 26 (3.5%) 3 2 1 0 L01 Antineoplastic agents 22 (2.9%) 5 0 5 0 V08 Contrast media 21 (2.8%) 9 1 6 2 B03 Antianemic preparations 20 (2.7%) 0 0 0 0 C07 Beta blocking agents 15 (2.0%) 4 1 2 1 C02 Antihypertensives 14 (1.9%) 1 1 0 0 A12 Mineral supplements 12 (1.6%) 1 1 0 0 C01 Cardiac therapy 11 (1.5%) 1 0 1 0 H03 Thyroid therapy 8 (1.1%) 2 1 1 0 N03 Antiepileptics 8 (1.1%) 0 0 0 0 V03 All other therapeutic products 8 (1.1%) 0 0 0 0 H02 Corticosteroids for systemic use 7 (0.9%) 1 1 0 0 M01 Antiinflammatory and antirheumatic products 7 (0.9%) 1 1 0 0 J04 Antimycobacterials 6 (0.8%) 0 0 0 0 M05 Drugs for treatment of bone diseases 5 (0.7%) 2 0 1 1 G04 Urologicals 4 (0.5%) 1 0 0 1 N06 Psychoanaleptics 4 (0.5%) 1 1 0 0

13 N07 Other nervous system drugs 4 (0.5%) 0 0 0 0 H01 Pituitary and hypothalamic hormones and analogues 3 (0.4%) 1 0 1 0 H05 Calcium homeostasis 2 (0.3%) 0 0 0 0 J05 Antivirals for systemic use 2 (0.3%) 0 0 0 0 J06 Immune sera and immunoglobulins 2 (0.3%) 0 0 0 0 L02 Endocrine therapy 2 (0.3%) 0 0 0 0 N05 Psycholeptics 2 (0.3%) 0 0 0 0 P01 Antiprotozoals 2 (0.3%) 0 0 0 0 R03 Drugs for obstructive airway diseases 2 (0.3%) 0 0 0 0 NO CODE No code 2 (0.3%) 0 0 0 0 A01 Stomatological preparations 1 (0.1%) 0 0 0 0 A02 Drugs for acid related disorders 1 (0.1%) 1 1 0 0 A03 Drugs for functional gastrointestinal disorders 1 (0.1%) 1 1 0 0 Antidiarrheals, intestinal antiinflammatory/antiinfective A07 1 (0.1%) 0 0 0 0 agents A11 Vitamins 1 (0.1%) 0 0 0 0 D06 Antibiotics and chemotherapeutics for dermatological use 1 (0.1%) 0 0 0 0 D07 Corticosteroids, dermatological preparations 1 (0.1%) 0 0 0 0 G03 Sex hormones and modulators of the genital system 1 (0.1%) 0 0 0 0 M02 Topical products for joint and muscular pain 1 (0.1%) 1 0 0 1 M03 Muscle relaxants 1 (0.1%) 1 0 1 0 N01 Anesthetics 1 (0.1%) 1 0 0 1 N04 Anti-Parkinson drugs 1 (0.1%) 0 0 0 0 R06 Antihistamines for systemic use 1 (0.1%) 0 0 0 0 ADR: adverse drug reaction, ATC: Anatomical Therapeutic and Chemical Results are expressed as n (%). * Either definitely preventable or potentially preventable

14 Supplementary Material 7: Sensitivity analyses by the Prentice, Williams, and Peterson (PWP) gap-time recurrent event time-to-event analysis

A. Hazard ratios for adverse drug reactions (overall) according to patient characteristics

Unadjusted model Adjusted model HR [95% CI] P-value HR [95% CI] P-value Age (years) 0.494 0.15 <60 Reference Reference 60 to 75 1.11 [0.92; 1.35] 0.95 [0.77; 1.15] ≥ 75 1.04 [0.83; 1.29] 0.81 [0.65; 1.02] Sex 0.05 0.02 Men Reference Reference Women 1.16 [0.99; 1.35] 1.20 [1.02; 1.40] eGFR <0.001 <0.001 ≥ 30 mL/min per 1.73m² Reference Reference <30 mL/min per 1.73m² 1.59 [1.36; 1.85] 1.47 [1.26; 1.72] Serum Albumin 0.05 0.29 ≥3.5 g/dL Reference Reference <3.5 g/dL 1.29 [1.01; 1.67] 1.15 [0.89; 1.49] UACR 0.04 0.59 < 30 mg/g Reference Reference 30 – 300 mg/g 1.01 [0.81; 1.24] 0.90 [0.73; 1.12] > 300 mg/g 1.23 [1.01; 1.50] 0.99 [0.81; 1.21] Anemia* <0.001 0.14 Without anemia Reference Reference With anemia 1.34 [1.15; 1.55] 1.12 [0.96; 1.31] Diabetes <0.001 0.35 Without diabetes Reference Reference With diabetes 1.32 [1.14; 1.52] 1.08 [0.92; 1.27] AKI history <0.001 0.01 Without AKI history Reference Reference With AKI history 1.35 [1.14; 1.61] 1.25 [1.05; 1.49] Cardiovascular history <0.001 0.01 Without cardiovascular history Reference Reference With cardiovascular history 1.39 [1.20; 1.61] 1.25 [1.06; 1.48] Hypertension 0.01 0.30 Without hypertension Reference Reference

15 With hypertension 1.43 [1.08; 1.90] 1.18 [0.86; 1.61] Baseline number of drugs/ patients <0.001 0.01 < 5 Reference Reference 5 to10 1.65 [1.26; 2.15] 1.35 [1.04; 1.76] > 10 2.32 [1.75; 3.07] 1.60 [1.19; 2.17] Adherence to medication <0.001 0.01 Good Reference Reference Poor 1.40 [1.19; 1.65] 1.26 [1.06; 1.48]

AKI: Acute kidney injury, CI: confidence interval, eGFR: estimated Glomerular filtration rate, HR: hazard ratio, UACR: Urine albumin to creatinine ratio *Anemia is defined by the 1968 WHO definition [World Health Organization. Nutritional anaemias. Report of a WHO scientific group. Geneva, World Health Organization; 1968]: <12 g/dL for women and <13 g/dL for men.

16 B. Hazard ratios for serious adverse drug reactions according to patient characteristics

Unadjusted model Adjusted model HR [95% CI] P-value HR [95% CI] P-value Age (years) 0.33 0.33 <60 Reference Reference 60 to 75 0.98 [0.63; 1.52] 0.71 [0.45; 1.11] ≥ 75 1.27 [0.81; 1.98] 0.78 [0.49; 1.25] Sex 0.92 0.63 Men Reference Reference Women 1.02 [0.72; 1.43] 1.09 [0.77; 1.54] eGFR <0.001 0.01 ≥ 30 mL/min per 1.73m² Reference Reference <30 mL/min per 1.73m² 2.08 [1.47; 2.94] 1.73 [1.22; 2.44] Anemia* <0.001 0.04 Without anemia Reference Reference With anemia 1.83 [1.33; 2.53] 1.42 [1.01; 1.99] Diabetes 0.04 0.88 Without diabetes Reference Reference With diabetes 1.40 [1.01; 1.94] 1.03 [0.73; 1.45] AKI history 0.01 0.12 Without AKI history Reference Reference With AKI history 1.60 [1.11; 2.30] 1.34 [0.92; 1.95] Cardiovascular history <0.001 0.001 Without cardiovascular history Reference Reference With cardiovascular history 2.30 [1.56; 3.38] 1.94 [1.29; 2.90] Baseline number of drugs/ patients <0.001 0.06 < 5 Reference Reference 5 to10 2.07 [1.06; 4.06] 1.47 [0.77; 2.81] > 10 3.90 [1.97; 7.73] 2.11 [1.04; 4.25] Adherence to medication 0.02 0.12 Good Reference Reference Poor 1.64 [1.09; 2.46] 1.35 [0.92; 1.97]

AKI: Acute kidney injury, CI: Confidence interval, eGFR: estimated Glomerular filtration rate, HR: hazard ratio *Anemia is defined by the 1968 WHO definition [World Health Organization. Nutritional anaemias. Report of a WHO scientific group. Geneva, World Health Organization; 1968]: <12 g/dL for women and <13 g/dL for men.