Pedrós et al. BMC Nephrology (2019) 20:193 https://doi.org/10.1186/s12882-019-1389-8 STUDY PROTOCOL Open Access Lactic acidosis associated with metformin in patients with moderate to severe chronic kidney disease: study protocol for a multicenter population-based case-control study using health databases Consuelo Pedrós1,2, Mónica Ávila3* , Ainhoa Gómez-Lumbreras4,5, Marcela Manríquez2, Rosa Morros4,6,7 and ALIMAR-C2 Study Group Abstract Background: The use of metformin in patients with type 2 diabetes mellitus has been associated with lactic acidosis. However, the information available in patients with moderate-severe chronic kidney disease is scarce. Methods: The ALIMAR-C2 study is a case-control study to assess the association between metformin and lactic acidosis in patients with type 2 diabetes mellitus and moderate-severe chronic kidney disease. The study will be performed with computerized registered electronic health records from eight Spanish hospitals linked to their corresponding primary care health areas from 2010 to 2016, comprising approximately 22.1 million person-years of follow-up. Logistic regression will be used to assess the crude and adjusted risk of lactic acidosis associated with metformin use overall and stratifying by use and dose categories, and chronic kidney disease stage. The overall case fatality rate of lactic acidosis, as well as the case fatality rate stratified by chronic kidney disease stage, will be calculated. Discussion: The ALIMAR-C2 study will provide useful information about the risk of lactic acidosis in type 2 diabetes mellitus patients with renal impairment using metformin. Keywords: Metformin, Type 2 diabetes mellitus, Chronic kidney failure, Lactic acidosis, Case-control studies, Electronic health records Background contraindicated in patients with an estimated glomerular Metformin is the first-line treatment for type 2 diabetes filtration rate (eGFR) below 30 mL/min. mellitus (DM2) when diet and exercise do not result in Nevertheless, the association between metformin and adequate glycaemia control, especially in overweight pa- LA has been a controversial issue due to conflicting re- tients [1]. It was the first antidiabetic agent that showed sults from different studies. The recommendation for to reduce diabetic complications and overall mortality in patients with eGFR between 30 and 60 mL/min has been overweight and obese DM2 patients [2]. a matter of debate [4]. Currently, some studies analyzing Its use has been related to the occurrence of lactic the risk of LA failed to show an increase of risk with acidosis (LA), a rare but severe adverse effect, especially metformin use [5–7]. in patients with renal disease [3]. For this reason, it is Information regarding patients with impaired renal function is scarce. In this line, some observational stud- * Correspondence: mavila@bellvitgehospital.cat ies have shown an increase of incidence of LA in pa- 3 Clinical Research and Clinical Trials Unit, Bellvitge Biomedical Research tients exposed to metformin in parallel to the degree of Institute (IDIBELL), L’Hospitalet de Llobregat, Barcelona, Spain impairment of renal function [8], as well as an increase Full list of author information is available at the end of the article © The Author(s). 2019 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Pedrós et al. BMC Nephrology (2019) 20:193 Page 2 of 8 of LA risk in patients with eGFR < 60 mL/min mainly la Santa Creu i Sant Pau) and their corresponding public due to higher risk in patients with eGFR < 45 mL/min institutions for primary care (Institut Universitari d’In- [9]. More recently, a study commissioned by the Euro- vestigació en Atenció Primària [IDIAP] Jordi Gol in pean Medicines Agency (EMA) to assess the use and Catalonia, and Gerencia Asistencial de Atención Pri- safety of metformin in real clinical practice in patients maria in Madrid). Initially, other eight hospitals were in- with or without renal failure showed a greater risk of LA vited to participate in the study but they were excluded in metformin users than in other glucose-lowering agent after a feasibility assessment. Table 1 describes the popu- users. Additionally, the incidence rates of LA increased lation covered, the study period, and the with decreasing baseline eGFR [10]. Afterward, the EMA population-time of follow-up in each participant centre. carried out a referral procedure in order to review the Overall, the study includes approximately 22.1 million evidence justifying the contraindication of metformin person-years of follow-up. use in chronic kidney disease (CKD). In October 2016, that safety review concluded that metformin could be Data sources used in patients with moderately reduced kidney func- The Catalonian hospitals have data warehouses (DWs) tion (eGFR 30–59 mL/min) [11]. with administrative, clinical and laboratory data col- Before the beginning of European referral procedure, lected during clinical practice. Systems Applications and the ALIMAR-C2 study (“Riesgo de Acidosis Láctica aso- Products in Data Processing Business Objects (SAP BO) ciada al uso de MetforminA en pacientes diabéticos tipo is used for the data mining of the DWs. Hospital Clinic 2 con enfermedad Renal crónica moderada-severa: estu- of Barcelona, Hospital de la Santa Creu i Sant Pau, and dio de Casos y Controles”) was designed to provide fur- University Hospital Vall d’Hebron do not have some la- ther data on risk of LA associated with metformin use in boratory results fully integrated with their corresponding diabetic patients with CKD. DWs. The aim of this publication is to present the study DW of Bellvitge University Hospital is the only one protocol in detail. that also integrates the information corresponding to clinical and primary care data. For the other Catalan Methods/design hospitals, the data source for primary care in Catalonia Aims is SIDIAP (Information System for Research in Primary The primary objective of the ALIMAR-C2 study is to as- Care) which contains anonymized clinical information of sess the association between the use of metformin and all the primary care centres of the Institut Català de la LA in patients with DM2 and moderate to severe CKD. Salut (ICS) [12]. It covers more than 5.8 million patients Secondary objectives include: (1) to evaluate the asso- (approximately 80% of the Catalan population, which ciation according to daily dose of metformin and stage represents more than 10% of the Spanish population). of CKD, to analyse the effect of comorbidities and con- The information comes from Estació clínica d’atenció comitant medications, and to estimate the case fatality primària (ECAP™; electronic records in primary health- rate and the incidence of admission to critical care units care), and it includes sociodemographic characteristics, of LA; (2) to assess the association of LA with the use of health conditions registered as International Classifica- other non-insulin antidiabetic drugs (NIADDs) and insu- tion of Diseases, 10th revision (ICD-10) codes, clinical lin; and (3) to analyse the existence of a detection bias parameters, toxic habits, laboratory data, and general that affects the diagnosis of LA depending on the expos- practitioners’ prescriptions identified through anatomical ure to metformin. therapeutic chemical (ATC) codes. The information on the hospital environment corre- Study design sponding to the centres located in Madrid will be ob- This is a population-based case-control study using hos- tained from different sources. In the case of Ramón y pital healthcare databases linked to primary health care Cajal University Hospital, they will be obtained from the databases. Conjunto Mínimo Básico de Datos (CMBD; Basic mini- mum set of data) and the laboratory data from the Setting Openlab system. For Fundación Jiménez Díaz University The study will be conducted through collaboration of re- Hospital, data will be obtained from the hospital infor- searchers at eight hospitals from Madrid (Ramón y Cajal mation system HIS, which integrates clinical and labora- University Hospital, Fundación Jiménez Díaz University tory data. Hospital Clínico San Carlos will get the Hospital, and Hospital Clínico San Carlos) and Catalonia clinical data from the system HIS Clinica and the labora- (Bellvitge University Hospital [coordinating centre], tory data through the EoLIS System™. Hospital Germans Trias i Pujol, Hospital Clinic of Barce- For the Madrid hospitals, the data source for primary lona, University Hospital Vall d’Hebron, and Hospital de care will be obtained from the database that contains the Table 1 Overall description of data sources and study period Database/Hospital Catalonia Madrid Bellvitge Hospital Hospital University Hospital de la SIDIAP Hospital Ramón y Fundación Gerencia de University Germans Clinic of Hospital Vall Santa Creu i Clínico San Cajal Jiménez Díaz Atención Hospital Trias i Pujol Barcelona d’Hebron Sant Pau Carlos University University Primaria Hospital Hospital Pedrós et al. BMC Nephrology Population covered (inhabitants) 373.000 810.000 540.000 450.000 403.000 N/A 366.000 574.000 424.000 N/A Database type Hospital Data + + + + + N/A + + + N/A (Electronic available medical register) Clinical DW Hospital Hospital Hospital DW Hospital DW N/A Hospital CMBD Hospital DW N/A data DW DW DW (2019) 20:193 Laboratory DW Hospital ServoLab Modulab, Openlab N/A Laboratory Openlab Hospital DW N/A data DW clinical database records Primary Data + N/A N/A N/A N/A + N/A N/A N/A + care available Clinical DW N/A N/A N/A N/A ECAP N/A N/A N/A Ap-Madrid data Laboratory DW N/A N/A N/A N/A ECAP N/A N/A N/A Ap-Madrid data Starting year of electronic data 2010 2008 2003 2009 2011 2006 2009 2009 2009 2002 collection Study period April 2011– January January January January January 2010– January January January 2010– January 2010– December 2016 2014– 2010– 2010– 2012– December 2016 2010– 2010– December 2016 December 2016 December December December December December December 2016 2016 2016 2016 2015 2015 Disease and procedure coding system ICD 9 ICD 9 ICD 9 ICD 9 ICD 9 ICD 10 ICD 9 ICD 9 ICD 9 ICPC-2 Hospital admission and discharge data Yes Yes Yes Yes Yes No Yes Yes Yes No available Drug dictionary code WHO-ATC, N/A N/A N/A N/A WHO-ATC, N/A N/A N/A WHO-ATC, National Drug National Drug National Drug Codes Codes Codes Drug information Prescriptions, N/A N/A N/A N/A Prescriptions, N/A N/A N/A Prescriptions, data on data on data on strength and strength and strength and length length length DW Data warehouse, CMBD, Conjunto mínimo básico de datos (Basic minimum set of data), ECAP, Estació clínica d’atenció primària (electronic health records in primary healthcare), Ap-Madrid Electronic Medical Record of Primary Care of Madrid, ICD 9 International Classification Diseases, 9th revision, ICD 10 International Classification Diseases, 10th revision, ICPC-2 International Classification Diseases, 2th revision, N/A not applicable Page 3 of 8 Pedrós et al. BMC Nephrology (2019) 20:193 Page 4 of 8 information included in the Electronic Medical Record controls need to be 18-year-old or older, have DM2 of Primary Care (AP-Madrid). This is a unique central- diagnosis before the index date, and CKD stage as de- ized electronic medical record containing clinical data fined for cases during the 2-year period before the index from all primary care centers of Servicio Madrileño de date (excluding the last 2 weeks). Additionally, they need Salud (SERMAS). It covers approximately 6.6 million to have any information recorded on the primary health- people. It includes clinical data similar to that one from care database within the 2-year period prior to the index Catalonia except for diagnoses coding, which runs with date. Exclusion criteria for controls include: (1) diagnosis the International Classification of Primary Care, 2nd re- of type 1 diabetes mellitus, human immunodeficiency vision (ICPC-2). virus disease or solid organ transplantation before the Detailed information about data sources is described index date, (2) diagnosis of malignant neoplasm (except in Table 1. skin cancer other than melanoma; including pheochro- mocytoma) within the 5-year period prior to index date, Participants and (3) patient not resident in the area of the study. A Cases will be patients admitted to hospital with LA, that patient included in the study as a case will not be able to is defined by pH < 7.35 and plasmatic lactic acid concen- be included as a control. tration > 5 mM/L within the first 24 and 72 h after ad- mission, respectively. These short observation periods Variables are targeted to exclude LA that develops during The following variables will be obtained from hospital hospitalization in patients admitted due to other reasons. databases: The day of admission will be the index date. Inclusion criteria include: (1) 18-year-old or older, (2) hospital or  hospital admission data: admission date, discharge primary healthcare diagnosis of DM2 previous to the diagnoses, in-hospital death, admission to critical index date (Additional file 1: Table S1), (3) moderate to care unit; severe CKD (stage 3a, 3b or 4 of the Kidney Disease Im-  demographic data: age and sex; proving Global Outcomes [KDIGO] classification) [13]  laboratory test data (values and dates): plasmatic during the 2-year period before the index date (exclud- lactic acid concentration (all values during the first ing the last 2 weeks; see below), taking into account data 72 h from hospital admission), pH (all values during from the primary healthcare database, and (4) to have the first 24 h from hospital admission), haemoglobin any information recorded on the primary healthcare (all values during the 30-day period before the index database within the 1-year period prior to the index date and the first 24 h from admission). date. Patients will be excluded as cases if they have any of the following diagnosis: (1) diabetic ketoacidosis dur- The following variables will be obtained from primary ing the current in-hospital stay; (2) hospital or primary healthcare databases: healthcare diagnosis of type 1 diabetes mellitus, human immunodeficiency virus disease or solid organ trans-  laboratory test data (values and dates): serum plantation before the index date; (3) hospital or primary creatinine (all values between 2 years and 2 weeks healthcare diagnosis of malignant neoplasm (except skin before the index date), haemoglobin (all values cancer other than melanoma; including pheochromocy- during the 30-day period before the index date). toma) within the 5-year period prior to index date (Add-  prescriptions drug data during the 1-year period be- itional file 1: Table S1). In Catalonia, patients not fore the index date: prescriptions of metformin, registered in the hospital referral area will also be other non-insulin antidiabetic drugs (NIADDs), in- excluded. sulin, diuretics, renin-angiotensin system (RAS) in- Controls will be randomly selected from the popula- hibitors, non-steroidal anti-inflammatory drugs tion assigned to the primary health care area of the hos- (NSAIDs) and dates of prescription (initial and final) pital cases and matched to them in a ratio 10:1 on age (Additional file 2: Table S2). The prescribed posology (within 2 years for cases aged 60 to 85-year-old, extend- and the National Drug Code (NDC) are retrieved able to 3 years if necessary; cases older than 85 or youn- for metformin prescriptions; each NDC corresponds ger than 60 years are matched to controls older than 85 to specific strength and quantity of the drug. or younger than 60, respectively), gender, CKD stage, and year (taking the date of admissions of the cases as a Additionally, the following diagnoses and their corre- reference). The same index date of each case will be sponding dates will be obtained from both hospital and assigned to their controls. All the information needed to primary healthcare databases: DM2, type 1 diabetes melli- assess controls for eligibility criteria will be obtained tus, diabetic ketoacidosis, diabetic target organ damage, from the primary healthcare databases. To be eligible, human immunodeficiency virus infection, organ Pedrós et al. BMC Nephrology (2019) 20:193 Page 5 of 8 transplantation, malignant neoplasm, alcohol use, acute Exposure to diuretics, RAS inhibitors, and NSAIDs alcohol intoxication, other intoxications (cyanide, metha- will be defined as prescription during the 30-day period nol, ethilenglicol, diethilenglicol, propilenglicol), cocaine prior to the index date. Additional file 3 contains the use, liver disease, acute myocardial infarction, heart fail- STROBE checklist completed for this study protocol. ure, peripheral arterial disease, dyslipidemia, hypertension, cerebrovascular disease, dementia, hemiplegia, connective Sample size calculation tissue disease, acute respiratory failure, chronic respiratory Assuming a prevalence of metformin exposure of 40% in disease, chronic pulmonary obstructive disease, surgery, DM2 patients with moderate-severe CKD, it was calcu- acute kidney failure, seizures, dehydration, diarrhoea, lated that 39 cases with 10 matched controls will be ne- vomiting, gastroenteritis, gastroduodenal ulcer, shock, cessary to have a power of 90% for detecting a risk of sepsis, thiamine deficit, and tests that require the use of LA associated with the use of metformin with an odds iodine-based contrasts (Additional file 1: Table S1). ratio ≥ 3 and a two-sided significance level of 0.05. Charlson comorbidity index will be estimated from the average of the accumulated score based on the presence Statistical analysis of determined comorbidities [14]. Baseline characteristics will be described for cases and Renal function will be assessed taking into account controls. For qualitative variables, absolute and relative values of serum creatinine concentration recorded in frequencies will be provided. For quantitative variables, primary healthcare database obtained between 2 years the main statistic parameters will be calculated (mean, and 2 weeks before the index date. The eGFR will be cal- standard deviation, median, interquartile range, mini- culated using the CKD-EPI formula that takes into ac- mum and maximum). count the gender, the age, the plasmatic concentration Unadjusted and adjusted risk of LA associated with of creatinine, and the race [15]. For this study, it will be metformin will be estimated through a stratified logistic assumed that no African American patients are included. regression. Odds ratio and 95% confidence intervals will A CKD stage was assigned to each GFR estimation, fol- be provided. The following covariables will be included lowing the KDIGO classification (stage 1: eGFR ≥90 mL/ in the model: age, gender, alcohol use, cocaine use, in- min/1,73 m2; stage 2: eGFR 60–89 mL/min/1,73 m2; toxications, severe anaemia, Charlson comorbidity index, stage 3a: eGFR 45–59 mL/min/1,73 m2; stage 3b: eGFR complications of diabetes mellitus, liver disease, acute 30–44 mL/min/1,73 m2; stage 4: eGFR 15–29 mL/min/ myocardial infarction, heart failure, surgery, anaesthesia, 1,73 m2; stage 5: < 15 mL/min/1,73 m2) [16]. In case of seizures, dehydration, vomiting, diarrhoea, gastroenter- several eGFR estimates resulting in different CKD stages itis, sepsis, shock, thiamine deficit, acute respiratory fail- for an individual patient across the 2-year period, the ure, chronic obstructive pulmonary disease, acute renal CKD stage closer to the index date will be assigned to failure, tests that require the use of iodine-based con- this patient. trasts, exposure to oral hypoglycaemic agents other than metformin, to insulin, to diuretics, to RAS inhibitors, and to NSAIDs. Variable selection will be performed by Exposure definition a stepwise procedure. Exposure to metformin, other NIADDs and insulin will Patients with complete data for these covariables will be defined as prescriptions during the 365-day period be taken into account for the main analysis. However, a prior to the index date and classified as current use (pre- sensitivity analysis will be performed using a Markov scription during the 30-day period prior to the index chain Monte Carlo method for multiple imputations to date) or past use (prescription before the 30-days prior missing values. to the index date). Subgroup analyses will be performed according to Starting and final prescription dates will define the daily dose (< 1 g, 1-2 g, and > 2 g), use levels (current and length of the exposure. Consecutive prescriptions within past use), disease stage (3a, 3b, and 4) and territory 30 days will be considered the same exposure period. (Catalonia and Madrid). Consecutive prescriptions with a 30-day or longer gap As secondary analyses, the risk of LA associated with will be considered as two different exposure periods. other hypoglycaemic drugs and insulin will also be esti- The prescribed daily dose of metformin will be calcu- mated. Additionally, the overall case fatality rate of LA lated taking into account the posology recorded by the as well as the case fatality rate stratified by CKD stage prescriber and the strength of the drug prescribed will be calculated from the number of deaths among pointed out by its National Drug Code. The calculated cases and the total number of cases. daily dose is categorized in < 1 g, 1-2 g, and > 2 g. The possibility of bias detection will be studied by ana- Other antidiabetic drugs will be classified into lyzing the frequency of determination of plasmatic lac- pharmacological subgroups. tate levels in patients with metabolic acidosis according Pedrós et al. BMC Nephrology (2019) 20:193 Page 6 of 8 to the status of metformin exposure. This analysis will The possibility of observational bias should also be be performed with data from two of the participating considered. If the diagnosis of LA or the determination hospitals in a sample of episodes of urgent hospital ad- of plasmatic lactate levels in a patient with acidosis is mission with pH < 7.35 during the first 24 h. not actually performed symmetrically in patients ex- All statistical analyses will be performed with R statis- posed to metformin and in those not exposed, the preva- tical package version 3.4.0 or higher. lence of metformin exposure in cases of LA and therefore the strength of association could be overesti- mated [10]. This phenomenon is planned to be analyzed Discussion in our study as a secondary objective. The ALIMAR-C2 study is a multicenter On the other hand, the major strengths of our study population-based case-control study using hospital and are the multicenter population-based design and our primary care health databases in order to assess the as- case definition, which is based on objective laboratory sociation between the use of metformin and LA in pa- parameters at arrival to the hospital and unrelated to tients with DM2 and moderate to severe CKD. registered diagnoses. Observational studies on the same exposure-outcome In conclusion, LA has been considered an infrequent association using different databases could be inconsist- though serious adverse effect of metformin, with CKD ent because of variations in methodological and record pointed out as the most important risk factor. With this factors intrinsically related to the databases as well as study, we try to provide additional evidence on the differences in the health care system [17]. This study benefit-risk balance of metformin in DM2 patients with aims to analyze information collected from different moderate to severe CKD. electronic healthcare databases with administrative and clinical data retrieved from different assistance levels Additional files (hospitals and primary care) from two Spanish regions. Study design and methodology are some of the factors Additional file 1: Table S1. Diagnosis codes according International contributing to the diversity and discrepancy of study re- Classification of Diseases version 9 (ICD-9) and 10 (ICD-10), and International Classification of Primary Care (ICPC-2) used to identify sults, even when using the same database [18]. When es- diagnoses. (DOCX 26 kb) timating the risk from different databases the way the Additional file 2: Table S2. Anatomical Therapeutic Chemical information has been collected and then extracted could Classification (ATC) codes of the drugs of interest. (DOCX 19 kb) also make the results inconsistent. The homogeneous Additional file 3: STROBE Checklist. STROBE Checklist of Study Protocol. methodology that will be used when collecting data and (DOCX 40 kb) the use of a uniform definition for drug exposure, out- come and confounders sharing the same protocol along- Abbreviations side the different data sets could bring more consistency CKD: Chronic kidney disease; CMBD: Basic minimum set of data; CREC: Clinical Research Ethics Committee; DM2: Type 2 diabetes mellitus; to our results. All data will be analysed as if it came eGFR: expected glomerular filtration rate; EMR: Electronic medical record; from one database. ENCePP: European Network of Centres for Pharmacoepidemiology and On the other hand, when analyzing the exposures ac- Pharmacovigilance; ER: Emergency Room; KDIGO: Kidney Disease Improving Global Outcomes; LA: Lactic acidosis; NIADD: Non-insulin antidiabetic drug; cording to the prescriptions of medicines and not to the NSAID: Non-steroidal anti-inflammatory drug; PHC: Primary health care; dispensation or to their actual intake, the exposure could RAS: Renin-angiotensin system; SAP BO: Systems, Applications and Products be overestimated, but this would happen in the same in Data Processing Business Objects way in cases and in controls [19]. Therefore, potential Acknowledgements overestimation is not expected to have a net effect on The authors want to acknowledge for their contribution to the following the results. investigators who participated in a preliminary stage of study conception Another important limitation, such as the existence of and critically reviewed the study proposal: Josep Maria Arnau (Bellvitge University Hospital), María Giner (Institut Universitari d’Investigació en Atenció confounding factors, will be minimized by matching Primària [IDIAP] Jordi Gol), Jordi Cortés (Institut Universitari d’Investigació en cases and controls for certain variables and performing a Atenció Primària [IDIAP] Jordi Gol), Hoi Tong (Hospital Universitario La Paz), proper statistical analysis. Antonio Carcas (Hospital Universitario La Paz), Mª Ángeles Lobo (Hospital Universitario Virgen del Rocío), Jaume Torelló (Hospital Universitario Virgen Regarding systematic errors that could bias a del Rocío), Judith Sanabria (Hospital Universitario Virgen de la Victoria), Mikel case-control study, the possible selection bias due to the Castaño (Hospital Universitario Cruces), Lorena Codesido (Hospital difficulty of an adequate selection of controls is mini- Universitario Donostia), Mª Teresa Rodrigo (Hospital Universitario Donostia), Ana Aldea (Hospital Universitario de Canarias), David Gualteros (Clinical mized since the controls are a random sample of the Research and Clinical Trials Unit. Bellvitge Biomedical Research Institute population where the cases come from, i.e. population [IDIBELL]), María Angeles Quijada (Clinical Research and Clinical Trials Unit. with DM2 and moderate-severe CKD. Therefore, they Bellvitge Biomedical Research Institute [IDIBELL]), Cristina Mustata (Unitat d’Estudis del Medicament. Institut Universitari d’Investigació en Atenció are expected to have the same probability as the cases of Primària [IDIAP] Jordi Gol). Additionally, we thank CERCA Programme/ having been exposed to metformin. Generalitat de Catalunya for institutional support. Pedrós et al. BMC Nephrology (2019) 20:193 Page 7 of 8 Members of the ALIMAR-C2 Study Group: Consuelo Pedrós (Clinical Pharma- of Barcelona] CREC: HCB/2019/0528; [University Hospital Vall d’Hebron] CREC: cology Department. Bellvitge University Hospital. Bellvitge Biomedical Re- EPA (AG)55/2016(4967); [Hospital de la Santa Creu i Sant Pau] CREC: 16/ search Institute [IDIBELL]), Mónica Ávila (Clinical Research and Clinical Trials 156(OBS); [Ramón y Cajal University Hospital] CREC: ALIMAR-C2; [Fundación Unit. Bellvitge Biomedical Research Institute [IDIBELL]), Marcela Manríquez Jiménez Díaz University Hospital] CREC: EOH 2016–37; [Hospital Clínico San (Clinical Research and Clinical Trials Unit. Bellvitge Biomedical Research Insti- Carlos] CREC: 16/326-E; [Institut Universitari d’Investigació en Atenció Primària tute [IDIBELL]), Sebastián Videla (Clinical Pharmacology Department. Bellvitge (IDIAP) Jordi Gol] CREC: P16/105, and [Gerencia Asistencial de Atención Pri- University Hospital. Bellvitge Biomedical Research Institute [IDIBELL]), Rosa maria] CREC:43/16. All CRECs accepted a waiver for participant consent, tak- Morros (Unitat d’Estudis del Medicament. Institut Universitari d’Investigació ing into account the characteristics of the study. en Atenció Primària [IDIAP] Jordi Gol). Departament de Farmacologia, Tera- The ALIMAR-C2 study was registered onto the European Union electronic pèutica i Toxicologia. Universitat Autònoma de Barcelona), Ainhoa Gómez- Register of Post-authorisation Studies (EU PAS Register Number: EUPAS13969, Lumbreras (Unitat d’Estudis del Medicament. Institut Universitari d’Investiga- June 2016) and received the European Network of Centres for Pharmacoepi- ció en Atenció Primària [IDIAP] Jordi Gol), Oriol Prat-Vallverdú (Unitat d’Estu- demiology and Pharmacovigilance (ENCePP) seal (http://www.encepp.eu/ dis del Medicament. Institut Universitari d’Investigació en Atenció Primària encepp/viewResource.htm?id=14215). [IDIAP] Jordi Gol), Inmaculada Fuentes (Servicio de Farmacología Clínica, Hos- Regarding the data contained in the databases and according to Spanish pital Universitari Vall d’Hebron. Vall d’Hebron Institut de Recerca), Angélica legislation about confidentiality and data protection (Ley Orgánica 3/2018, Valderrama (Vall d’Hebron Institut de Recerca, Barcelona), Cristina Aguilera de 5 de diciembre, de Protección de Datos Personales y garantía de los (Servicio de Farmacología Clínica, Hospital Universitari Vall d’Hebron), Ana derechos digitales), data included in this study consist of the identification of María Barriocanal (Clinical Research and Clinical Trials Unit-UPIC. Fundació incident cases of LA in hospitals and some specific information from primary Institut d’Investigació en Ciències de la Salut Germans Trias i Pujol), Joaquín care. In order to link this information, the same identity code is needed for Sáez-Peñataro (Servicio de Farmacología Clínica, Área del Medicamento, Hos- the patients. A codification system will be performed to conduct this linking pital Clínic de Barcelona; Universitat de Barcelona), Rosa Antonijoan (Servicio process maintaining the anonymization of the patients. de Farmacología Clínica, Hospital de la Santa Creu i Sant Pau), Claudia E. Del- gado (Clinical Research and Clinical Trials Unit, Sant Pau Biomedical Research Consent for publication Institute), Lucía Llanos (Clinical Research and Clinical Trials Unit, Fundación Not applicable. Jiménez Díaz, Health Research Institute), Leonor Laredo (Servicio de Farmaco- logía Clínica, Hospital Clínico San Carlos, Instituto de Investigación Sanitaria Competing interests del Hospital Clínico San Carlos IdISSC), Mónica Aguilar (Clinical Pharmacology The authors declare that they have no competing interests. Unit. Ramón y Cajal Hospital. IRyCIS), Teresa Sanz (Unidad de Apoyo a la Investigación. Gerencia Asistencial de Atención Primaria de Madrid), Montserrat Hernández (Gerencia Adjunta de Procesos Asistenciales. Asisten- Publisher’s Note cial de Atención Primaria de Madrid), José C. Estévez (Unidad de Apoyo Téc- Springer Nature remains neutral with regard to jurisdictional claims in nico. Gerencia Asistencial de Atención Primaria de Madrid), Sergio Ruiz published maps and institutional affiliations. (Gerencia Adjunta de Procesos Asistenciales. Gerencia Asistencial de Atención Primaria de Madrid), Lluís Murgui (Subdirectorate for Information Systems. Author details 1 Bellvitge University Hospital), Xavier Corbella (Internal Medicine Department. Clinical Pharmacology Department, Bellvitge University Hospital, L’Hospitalet Bellvitge University Hospital. Bellvitge Biomedical Research Institute [IDIBELL]), de Llobregat, Barcelona, Spain. 2Bellvitge Biomedical Research Institute Xavier Fulladosa (Nephrology Department. Bellvitge University Hospital. Bell- (IDIBELL), L’Hospitalet de Llobregat, Barcelona, Spain. 3Clinical Research and vitge Biomedical Research Institute [IDIBELL]), Manuel Pérez-Maraver (Endo- Clinical Trials Unit, Bellvitge Biomedical Research Institute (IDIBELL), crinology Department. Bellvitge University Hospital. Bellvitge Biomedical L’Hospitalet de Llobregat, Barcelona, Spain. 4Medicines Research Unit, Research Institute [IDIBELL]), Virginia Alonso (Intensive Medicine Department. Foundation University Institute for Primary Health Care Research Jordi Gol i Bellvitge University Hospital. Bellvitge Biomedical Research Institute [IDIBELL]), Gurina (IDIAPJGol), Barcelona, Spain. 5Department of Medical Sciences, Manel Mata-Cases (CAP La Mina. Institut Català de la Salut. USR Barcelona, Girona University, Girona, Spain. 6Departament de Farmacologia, Terapèutica Institut Universitari d’Investigació en Atenció Primària [IDIAP] Jordi Gol. i Toxicologia, Universitat Autònoma de Barcelona, Barcelona, Spain. 7Catalan GEDAPS). Health Institute, Barcelona, Spain. Received: 26 February 2019 Accepted: 16 May 2019 Funding The project received a research grant from the Carlos III Institute of Health, Ministry of Economy and Competitiveness (Spain), in 2015 (reference PI15/ References 00764) under the Health Strategy Action 2015, within the Technical, Scientific 1. National Institute for Health and Clinical Excellence. Type 2 diabetes in and Innovation Research National Plan 2013–2016 (Co-funded by European adults: management. https://www.nice.org.uk/guidance/ng28/resources/ Regional Development Fund. ERDF, a way to build Europe). The funders had type-2-diabetes-in-adults-management-pdf-1837338615493. Accessed 21 no role in study design, data collection, and analysis, decision to publish, or Nov 2018. preparation of the manuscript. We would like to confirm that our study 2. UK Prospective Diabetes Study (UKPDS) Group. Effect of intensive blood- protocol has undergone peer-review and awarded by Carlos III Institute of glucose control with metformin on complications in overweight patients Health. with type 2 diabetes (UKPDS 34). Lancet. 1998;352:854–65. 3. Lalau J-D. Lactic acidosis induced by metformin: incidence, management Availability of data and materials and prevention. Drug Saf. 2010;33:727–40. Not applicable for this publication. 4. del Pozo-Fernández C, Pardo-Ruiz C, Sánchez-Botella C, Blanes-Castañer V, López-Menchero R, Gisbert-Sellés C, et al. Discrepancies among consensus Authors’ contributions documents, guidelines, clinical practice and the legal framework for the CP is the coordinator and principal investigator and was actively involved in treatment of type 2 diabetes mellitus patients. Nefrologia. 2012;32:367–73. the conception and design of the study. MA, MM, RM, and AG are members 5. Bodmer M, Meier C, Krähenbühl S, et al. Metformin, sulfonylureas, or other of the coordinating group and were also involved in the conception and antidiabetes drugs and the risk of lactic acidosis or hypoglicemia. Diabetes design of the study. CP, MA, RM, and AG drafted the manuscript. All authors Care. 2008;31:2086–91. made contributions to the design of the study, critically reviewed the 6. Chang CH, Sakaguchi M, Dolin P. Epidemiology of lactic acidosis in type 2 manuscript and approved the final version. diabetes patients with metformin in Japan. Pharmacoepidemiol Drug Saf. 2016;25:1196–203. Ethics approval and consent to participate 7. Aharaz A, Pottegard A, Henriksen DP, Hallas J, Beck-Nielsen H, Lassen AT. The study protocol was approved by the CRECs of all participant centres Risk of lactic acidosis in type 2 diabetes patients using metformin: a case (study code in the coordinating centre [Bellvitge University Hospital] CREC: control study. PLoS One. 2018;13:e0196122. https://doi.org/10.1371/journal. EPA032/16; [Hospital Germans Trias i Pujol] CREC: PI-16-120; [Hospital Clinic pone.0196122 eCollection 2018. Pedrós et al. BMC Nephrology (2019) 20:193 Page 8 of 8 8. Richy FF, Sabidó-Espin M, Guedes S, Corvino FA, Gottwald-Hostalek U. Incidence of lactic acidosis in patients with type 2 diabetes with and without renal impairment treated with metformin: a retrospective cohort study. Diabetes Care. 2014;37:2291–5. 9. Eppenga WL, Lalmohamed A, Geerts AF, Derijks HJ, Wensing M, Egberts A, et al. Risk of lactic acidosis or elevated lactate concentrations in metformin users with renal impairment: a population-based cohort study. Diabetes Care. 2014;37:2218–24. 10. Li L, Jick S, Gopalakrishnan C, Heide-Jørgensen U, Nørrelund H, Sørensen HT, et al. Metformin use and risk of lactic acidosis in people with diabetes with and without renal impairment: a cohort study in Denmark and the UK. Diabet Med. 2017;34:485–9. 11. European Medicines Agency. Use of metformin to treat diabetes now expanded to patients with moderately reduced kidney function recommendations for patients with kidney impairment updated in product information. EMA/868987/2016. https://www.ema.europa.eu/documents/ referral/metformin-article-31-referral-use-metformin-treat-diabetes-now- expanded-patients-moderately-reduced_en.pdf. Accessed 21 Nov 2018. 12. Information System for Research in Primary Care. 2017. http://www.sidiap. org. Accessed 21 Nov 2018. 13. Levey AS, Eckardt KU, Tsukamoto Y, Levin A, Coresh J, Rossert J, De Zeeuw D, Hostetter TH, Lameire N, Eknoyan G. Definition and classification of chronic kidney disease: A position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int. 2005;67:2089–10. 14. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–83. 15. Levey AS, Stevens LA, Schmid CH, Zhang YL, Castro AF, Feldman HI, et al. A new equation to estimate glomerular filtration rate. Ann Intern Med. 2009; 150:604–12. 16. Levin A, Stevens PE. Summary of KDIGO 2012 CKD guideline: behind the scenes, need for guidance, and a framework for moving forward. Kidney Int. 2014;85:49–61. 17. Souverein PC, Abbing-Karahagopian V, Martin E, Huerta C, de Abajo F, Leufkens HGM, et al. Understanding inconsistency in the results from observational pharmacoepidemiological studies: the case of antidepressant use and risk of hip/femur fractures. Pharmacoepidemiol Drug Saf. 2016;25: 88–102. 18. Abbing-Karahagopian V, Kurz X, de Vries F, van Staa TP, Alvarez Y, Hesse U, et al. Bridging differences in outcomes of pharmacoepidemiological studies: design and first results of the PROTECT project. Curr Clin Pharmacol. 2014;9: 130–8. 19. Schneeweiss S, Avorn J. A review of uses of health care utilization databases for epidemiologic research on therapeutics. J Clin Epidemiol. 2005;58:323– 37.