Special report Pan American Journal of Public Health Developing a cooperative multicenter study in Latin America: Lessons learned from the Latin American Study of Nutrition and Health Project Mauro Fisberg,1 Irina Kovalskys,2 Georgina Gómez Salas,3 Rossina Gabriella Pareja Torres,4 Martha Cecilia Yépez García,5 Lilia Yadira Cortés Sanabria,6 Marianella Herrera-Cuenca,7 Attilio Rigotti,8 Viviana Guajardo,2 Ioná Zalcman Zimberg,9 Agatha Nogueira Previdelli,10 Luis A. Moreno,11 Michael Pratt,12 Berthold Koletzko,13 Katherine L. Tucker,14 and the ELANS Study Group15 Suggested citation Fisberg M, Kovalskys I, Gómez Salas G, Pareja Torres RG, Yépez García MC, Cortés Sanabria LY, et al. Developing a cooperative multicenter study in Latin America: Lessons learned from the Latin American Study of Nutrition and Health Project. Rev Panam Salud Publica. 2017;41:e111. doi: 10.26633/RPSP.2017.111 ABSTRACT This report examines the challenges of conducting a multicenter, cross-sectional study of coun- tries with diverse cultures, and shares the lessons learned. The Latin American Study of Nutrition and Health (ELANS) was used as a feasibility study involving the most populous cities of eight countries in Latin America (Argentina, Brazil, Chile, Colombia, Costa Rica, Ecuador, Peru, and Venezuela) in 2014–2015, about 40% of the population of the Americas. The target sample included 9 000 individuals, 15–65 years of age, and was stratified by geo- graphic location (only urban areas), gender, age, and socioeconomic status. Six principal challenges were identified: team structuring and site selections; developing a single protocol; obtaining ethic approvals; completing simultaneous fieldwork; ensuring data quality; and extracting data and maintaining consistency across databases. Lessons learned show that har- monization, pilot study, uniformity of procedures, high data quality control, and communication and collaboration across sites are imperative. Barriers included organizational complexity, recruit- ment of collaborators and research staff, institutional cooperation, development of infrastructure, and identification of resources. Consensus on uniform measures and outcomes and data collection methodology, as well as a plan for data management and analysis, communication, publication, and dissemination of study results should be in place prior to beginning fieldwork. While challeng- ing, such studies offer great potential for building a scientific base for studies on nutrition, physical activity, and other health topics, while facilitating comparisons among countries. Keywords Multicenter study; nutritional surveillance; nutrition surveys; Latin America. 1 Instituto Pensi, Fundação Jose Luiz Egydio 7 Centro de Estudios del Desarrollo, Universidad Aragón, Centro de Investigación Biomédica en Setubal, Hospital Infantil Sabara, São Paulo, Central de Venezuela/Fundación Bengoa, Caracas, Red Fisiopatología de la Obesidad y Nutrición, Brazil. Send correspondence to Mauro Fisberg, Venezuela. University of Zaragoza, Zaragoza. mauro.fisberg@gmail.com 8 Departamento de Nutrición, Diabetes y Meta­ 12 Nutrition and Health Sciences Program, Hubert 2 Committee on Nutrition and Wellbeing, Inter­ bolismo, Centro de Nutrición Molecular y Department of Global Health, Rollins School of Public national Life Science Institute, Buenos Aires, Enfermedades Crónicas, Escuela de Medicina, Health, Emory University, United States of America. Argentina. Pontificia Universidad Católica, Santiago, Chile. 13 Division of Metabolic and Nutritional Medicine, 3 Departamento de Bioquímica, Escuela de Medic ina, 9 Departamento de Psicobiologia, Universidade Dr. Von Hauner Children’s Hospital, University Universidad de Costa Rica, San José, Costa Rica. Federal de São Paulo, São Paulo, Brazil. of Munich Medical Center, Munich, Germany. 4 Instituto de Investigación Nutricional, Lima, Peru. 10 Faculdade de Ciências Biológicas e da Saúde, 14 Department of Clinical Laboratory and Nutri­ 5 Colegio de Ciencias de la Salud, Universidad San Universidade São Judas Tadeu, São Paulo, Brazil. tional Sciences, University of Massachusetts, Francisco de Quito, Quito, Ecuador. 11 Growth, Exercise, Nutrition and Development Lowell, Massachusetts, United States. 6 Departamento de Nutrición y Bioquímica, Pontificia Research Group, Instituto Agroalimentario de 15 The ELANS Study Group. For a complete list of Universidad Javeriana, Bogotá, Colombia. Aragón, Instituto de Investigación Sanitaria members, see the Acknowledgements. This is an open access article distributed under the terms of the Creative Commons Attribution­NonCommercial­NoDerivs 3.0 IGO License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited. No modifications or commercial use of this article are permitted. In any reproduction of this article there should not be any suggestion that PAHO or this article endorse any specific organization or products. The use of the PAHO logo is not permitted. This notice should be preserved along with the article’s original URL. Rev Panam Salud Publica 41, 2017 1 Special report Fisberg et al. • Lessons from a multicenter study in Latin America Growing interest in community­based cross­sectional, multicenter study, that Key factors in the ELANS project im­ public health and policy interventions was conducted simultaneously in the ur­ plementation were the selection of team that reduce obesity and improve nutri­ ban populations of the most populous members and the inclusion of multidisci­ tion has prompted multicenter studies to cities of eight countries in Latin America. plinary professional, technical, and sci­ understand population­based behaviors This paper highlights the principle les­ entific advisors with complementary and outcomes. In contrast to multi­site sons learned from the ELANS, including expertise in epidemiology, energy bal­ studies whose main purpose is to obtain the key decisions, challenges and barri­ ance, physical activity, and statistics. a larger sample, multicenter studies are ers, and logistical strengths and limita­ PIs were selected based on their ability designed jointly by Principal Investiga­ tions encountered during the design to conduct epidemiological studies (in­ tors (PIs) at all sites. This means that all process, data collection, and data entry. vestigator fieldwork experience, research are involved in planning the study proto­ background, and academic support col and procedures, are scientifically re­ SURVEY METHODOLOGY structure for research) and diversity of sponsible for the study results, and geographic location in Latin America. participate actively in manuscripts and ELANS is a household­based, multi­ All of the academic/research teams other dissemination activities. These national, cross­sectional survey that aims were composed of researchers, under­ studies have been increasingly valued to describe the nutritional status of peo­ graduates, and postgraduate health sci­ due to large sample sizes, ability to ex­ ple in Latin America and to investigate ence students based in universities or plore differences across sites, and the in­ food and nutrient intake, as well as phys­ in association with universities, with creased generalization of results (1). ical activity levels among representative the exception of one based at a research Multicenter collaborations also allow re­ samples of urban populations. The total institute (Table 2). The organizational cruitment of more diverse populations, population of the eight study countries— structure had three branches: academic within a much shorter time frame. While Argentina, Brazil, Chile, Colombia, Costa team, technical support team (physical challenging, such studies offer great po­ Rica, Ecuador, Peru, and Venezuela— activity and food intake assessment, an­ tential for building a scientific base for represents about 40% of the population thropometrics), and operational team the study of obesity and for planning of the Americas. The target sample in­ (Figure 1). The organizational structure health policies and intervention pro­ cluded 9 000 individuals, 15 – 65 years of of the ELANS was defined during the grams. However, such studies require age, and was stratified by geographic lo­ first design meetings among the re­ heightened attention to detail, simultane­ cation (only urban areas), gender, age, searchers, including the scope of the ity, comprehensive planning, and collab­ and socioeconomic status. The rationale study, the complexity of communica­ oration with colleagues (2). and design of the study are reported in tions, and what agreements would have Several large, multicenter, observational more detail elsewhere (7). In brief, the to be reached. The study management studies have been conducted to investigate ELANS protocol was designed to collect and quality control strategies are de­ the nutritional and physical activity status data at the individual level using ques­ scribed elsewhere (7). of various populations (3 – 6). The major­ tionnaires (sociodemographic, dietary It is important to highlight that project ity of these studies have been performed intake, and physical activity) and objec­ coordination was centralized in an inter­ separately by each country and unified tive measurements (accelerometry and national team composed of two chair­ later. In most cases, differing methodolo­ anthropometry). persons, a co­chair, and two international gies were used to assess food consumption project managers. This centralized ap­ and physical activity, or samples of each CHALLENGE 1: STRUCTURING A proach ensured that any barriers that country were not representative of specific TEAM AND SELECTING SITES arose would be resolved rapidly, that populations. In Latin America, few studies communication among all teams would have been conducted that represent the re­ Selected ELANS countries differed in be effective, and that research uniformity ality of each country and region. demographic and some socioeconomic would be maintained. By bringing to­ The Latin American Study of Nutrition indicators, reflecting the differences gether expert researchers in the field and Health (ELANS), however, is a present in Latin America (Table 1). of physical activity and statistics, the TABLE 1. Site characteristics of the eight countries participating in the Latin American Study of Nutrition and Health, 2013 - 2014 Argentina Brazil Chile Colombia Costa Rica Ecuador Peru Venezuela Population sizea 43 132 000 205 620 000 18 006 000 48 584 000 4 832 000 16 205 000 30 814 175 30 620 000 Urban populationb 92% 85% 89% 76% 76% 64% 76% 89% GDPc 12 509 11 384 14 528 7 903 10 415 6 345 6 541 12 771f GINI indexd (22) 42,3 52,9 50,5 53,5 49,2 47,3 44,7 46,9g Obesity in adultse 26,3% 20% 27,8% 21% 24.3% 18,7% 21,1% 24,8% Source: Prepared by the authors based on the study literature. a Census reports from the country’s national statistical offices. b Percent of the country’s total population (22). c Gross Domestic Product per capita in current US$ (23). d Measure of statistical dispersion, intended to measure the distribution of wealth in a country (21). e Body Mass Index ≥ 30 (24). f World Bank 2012; all others, 2014. g 2006; all others, 2013. 2 Rev Panam Salud Publica 41, 2017 Fisberg et al. • Lessons from a multicenter study in Latin America Special report TABLE 2. Fieldwork characteristics in the Latin American Study of Nutrition and Health, 2014 - 2016 Argentina Brazil Chile Colombia Costa Rica Ecuador Peru Venezuela Study name Estudio Estudo Estudio Latino Estudio Latino Análisis del El estudio del Estudio Latino Estudio Latino Argentino Brasileiro Americano de Americano de balance balance enérgico Americano de Americano de de Nutrición de Nutriçao Nutrición y Nutrición y energetico y de una muestra Nutrición y Nutrición y y Salud e Saude Salud- Chile Salud- factores de poblacional, Salud- Perú Salud- Colombia riesgo de Ecuador Venezuela obesidad en la población costarricense Type of institution Research Private children’s University University University University Research Research institution and hospital and institution institution and university federal university university Number of researchers 7 7 10 5 3 3 12 12 on team Number of cities 12 23 9 11 7 9 10 13 evaluated n participants 1 200; 319 2 000; 569 870; 308 1 230; 405 790; 282 800; 268 1 100; 360 1 134; 368 (total; accelerometer) Administration mode Tablet Tablet Tablet Tablet Tablet Paper Tablet Paper Fieldwork period October October September October November September November March 2014 – 2014 – 2014 – 2014 – 2014 – 2014 – 2014 – 2015 – July 2015 July 2015 July 2015 July 2015 July 2015 July 2015 May 2015 February 2016 Number of fieldwork interviewers 45 40 25 42 13 22 36 36 Source: Prepared by the authors based on the project survey. FIGURE 1. Organizational structure of Latin American Study of Nutrition and Health ELANS gained greater detail and depth. Project, 2016 Additionally, an external advisory board experienced with similar epidemiologi­ Chairs/coordinating cal studies provided decision­making center guidance required to continue moving the project forward. Lessons learned: Advisory board (a) A collaborative, multidisciplinary team of experts with extensive scientific International project Academic/research experience and effective communication Co-chaira managersa Technical support teams per country tools is essential.(eight teams) (b) Supervision by an international team is crucial. The appointment of an experienced and proactive coordinating Contract Research Principal investigator to mentor and counsel other Organization (CRO) investigator sites ensures consistency and support where it is needed. (c) Designating/assigning an interna­ Intake assesment Local project tional project manager who provides external team manager overarching communication and can e nsure issues are quickly detected and measures are continuously improved. Physical activity external team Statistics team CHALLENGE 2. DEVELOPING AND IMPLEMENTING A SINGLE STANDARDIZED PROTOCOL Statistics team Local team At the start, when the decision to col­ lect common data across the sites was a International team composed by chairs/coordinating center, co-chair and international project managers. made, it became apparent that a single Source: Prepared by the authors based on the project survey. protocol and a uniform set of data Rev Panam Salud Publica 41, 2017 3 Special report Fisberg et al. • Lessons from a multicenter study in Latin America collection tools would be required. The centers. The Multiple Pass Method (10) It is noteworthy that the PIs faced many process of creating a single protocol in­ was unanimously chosen. All PIs agreed problems with the purchase of imported volved 15 months of collaborative work to use the Nutrition Data System for Re­ equipment (e.g., accelerometers, scales, with in­p erson meetings and conference search software (Minnesota University, stadiometers). Due to customs and mail calls with the PIs. All decisions were Minneapolis, Minnesota, United States; delivery problems, all sites had to delay guided by scientific evidence and field­ NDSR) to allow ELANS inter­country the start of the fieldwork. work logistics. The advice from experi­ comparisons. An extensive process of The socioeconomic questions used in enced external advisors with differing harmonization was initiated, as de­ ELANS were designed based on ques­ expertise was important. scribed elsewhere (11). tionnaires used by each national statis­ Considering the large number of sites, The International Physical Activity tics office, or otherwise used most eight in this case, a tools standardization Questionnaire has been validated in frequently by each country (15 – 20). process was developed to ensure equal countries of Latin America; however, the Due to differing variables used across standards for data collection (8) at every Mexican (Spanish) version, (12) adapted countries to determine Socioeconomic site. The choice of dietary method used by the International Study of Physical Ac­ Levels (SEL), a rule was developed to in the ELANS was based on existing sur­ tivity and Built Environments was se­ equate the different classification sys­ vey studies, such as NHANES (9) and lected for use after cultural adaptations tems of the eight countries (Table 3). HELENA (3), as well as the experience for wording and examples (13, 14). Based on this, three levels of classifica­ of some of the PIs. Repeated 24­hour re­ The objective measure of physical ac­ tion were established for SEL and in­ calls (24hr) are considered useful tools in tivity was a crucial aspect of the ELANS cluded equivalent characteristics for all providing national­ and group­level esti­ that provided accurate estimates of phys­ countries: high, medium, and low. The mates of usual dietary intakes of individ­ ical activity and energy expenditure in same procedure was performed to estab­ uals, as well as in describing usual intake this Latin American population. Due to lish equalization across levels of educa­ distributions of populations, using ap­ logistic and financial matters, efforts tional in the eight countries, and also propriate statistical approaches and con­ were made to ensure that a range of resulted in three classifications: primary trolling for intra­individual variability. 25% – 40% of each sample would wear to incomplete secondary schooling; The standardization of procedures for the accelerometer for 7 days. complete secondary to incomplete 24hr administration and analysis was A major concern was that all sites higher/tertiary education (technical/ necessary to ensure equivalence of the should use the same or equally­reliable university); and complete higher educa­ dietary outcomes across participating equipment validated by previous studies. tion (technical/university). TABLE 3. Socioeconomic and educational level equalization between eight countries participating in the Latin American Study of Nutrition and Health, 2016 Argentina Brazil Chile Colombia Costa Rica Ecuador Peru Venezuela Socioeconomic levela High A, B, C1 A1, A2, B1 A, B, C1 5 / 6 A, B, C1 A, B A A, B, C+ Medium C2, C3 B2, C1 C2, C3 3 / 4 C2, C3 C+ B, C C Low D1, D2 C2, D, E D, E 1 / 2 D1, D2 C-, D D, E D, E Educational level Primary – – Illiterate – I lliterate – Illiterate – I lliterate – Illiterate – I lliterate – I lliterate – I lliterate incomplete – C omplete or – Complete or – Complete or incomplete – Incomplete – C omplete or – Complete or – Complete or – C omplete or secondary incomplete incomplete basic primary or incomplete incomplete incomplete incomplete schooling primary fundamental – Incomplete humanistic baccalaureate primary primary primary basic – I ncomplete – Incomplete – Incomplete middle – I ncomplete – Incomplete – Incomplete – I ncomplete secondary secondary secondary secondary secondary secondary – Incomplete baccalaureate C omplete – C omplete – C omplete – Complete humanistic – Complete – C omplete – Complete – C omplete – Complete secondary – secondary secondary – Complete middle baccalaureate secondary or secondary or secondary middle and incomplete – I ncomplete – Incomplete – Incomplete – I ncomplete vocational baccalaureate – Incomplete diversified higher/tertiary tertiary higher professional-technical technical – Incomplete – I ncomplete university education – I ncomplete education – I ncomplete bachelor – Incomplete university university university university Complete – Complete – C omplete – C omplete professional- – Complete – Complete – More years – Complete – C omplete higher tertiary higher technical technical or university of superior technical technical education – Complete education – C omplete bachelor technological – M asters or education – Complete – Complete university – C omplete school – Complete doctorate (without university university – Post-graduate of non-commissioned university post-graduate) – P ost-graduate – Doctorate officers – Specialization, – Complete – Post-graduate masters or university doctorate – Post-graduate Source: Prepared by the authors based on the project survey. a Classifications according to those provided by National Institutes of Statistics of each country. 4 Rev Panam Salud Publica 41, 2017 Fisberg et al. • Lessons from a multicenter study in Latin America Special report Lessons learned: data collection simultaneously at differ­ identification badge to prospective partic­ ing sites—it became apparent that a stan­ ipants and wore an identifying apron; and (a) Input from local teams is essential dardized and uniform application of during the first visit, an informative letter, during protocol development and plan­ questionnaires and measurements was including the researchers’ contact infor­ ning of field work to ensure achievable needed. These led to the decision to use a mation, was provided to each participant. goals. Contract Research Organization (CRO) Other issues identified during the pilot (b) Developing a feasible and realistic with offices in all eight countries in­ study were discussed and corrected, as protocol may be more difficult than first volved in ELANS. This CRO was respon­ needed. expected. Standardization and harmoni­ sible for all participant sampling and zation of foods in software require exten­ recruitment and data collection and en­ Lessons learned: sive and innovative work. The use of try, except for dietary recalls, which were accelerometers requires caution, as com­ managed by each local institution. All (a) A pilot study is important to ensur­ pliance is difficult to achieve. Measure­ procedures were continuously moni­ ing feasibility, efficiency, and adherence ment and definitions of socioeconomic tored by the international team, and a to protocols and procedures. Although it status and educational level across coun­ weekly meeting of PIs was held through­ delayed the study start, the pilot proved tries require close attention to ensure out the fieldwork. to be important in preventing measure­ comparability. Frequent contact with field supervi­ ment errors and improving participant (c) Equipment availability and pro­ sors was necessary to ensure quality data compliance. curement in different countries requires collection. At the start, there were many (b) Thorough training is fundamental prior planning and ample time. Possible mistakes due to the interviewers’ lack of for interviewers and ample time should delivery delays and customs tariffs experience; later, there seemed to be less be allotted for this purpose. Interviewers should be included in the schedule and attention to detail. To accomplish the with experience or training in health sci­ budget. data collection on time, local interview­ ences are preferable. ers were hired and trained to work in (c) Continuous supervision of inter­ CHALLENGE 3. OBTAINING different parts of each country simulta­ viewers’ tasks is critical for maintaining ETHICS APPROVALS neously. Interviewers with previous the data quality of the fieldwork. health research experience produced (d) A good relationship and open The process for ethical reviews and b etter quality interviews. Faster field­ c ommunication between the academic/ timeframes for obtaining approvals work was experienced in some countries, research team and the CRO are key to vary greatly among countries. To ad­ such as Chile, where interviewers were conducting efficient supervision and dress this challenge and to reduce the health science students. monitoring procedures. burden of conducting multiple local Standardized training and operating ethics reviews, an external institutional manuals were designed for use in all CHALLENGE 5. ENSURING review board (IRB) was selected to per­ countries; however, revisions by the DATA QUALITY form a centralized review of the proto­ i nternational team were made on an col, the informed consent template, and as­needed basis to ensure equivalent Data reliability and credibility are es­ other study documents. In this case, measures, described elsewhere (7). sential to the success of ELANS. Threats Western IRB (WIRB, Puyallup, Washing­ Three operating manuals provided to data quality were identified during ton, United States) was chosen as an in­ well­structured guidance for interview­ the protocol design process, the pilot dependent i nternational organization. ers on anthropometric measurements, study, and the first weeks of the field­ However, to obtain WIRB approval, dietary intake recalls, and accelerometer work. Thus, the coordinating investiga­ each site also had to submit the protocol use and preparation. A fourth operating tors/chairs proposed adopting procedures and informed consent through a local manual was designed to ensure proper in all sites to ensure high­quality data IRB. After approval of local IRBs, use of the NDSR software by researchers and reliable information. These included ELANS received WIRB approval. working with the dietary data. The inter­ preparatory meetings; detailed operat­ viewers’ work was continuously super­ ing manuals; site visits; technical visits to Lesson learned: vised during the data collection period. participating centers; interviewer train­ To ensure functionality of data collec­ ing; close monitoring of data collection (a) Obtaining ethics approvals for a tion procedures, a pilot study was per­ and data entry; retraining of interview­ multicenter study must be carefully formed before fieldwork began. It tested ers when needed; concurrent query man­ planned and, preferably, centralized, the tools and accompanying procedures agement and fieldwork supervision; with extra time allotted for reviews at at all the sites and identified performance partial database generation; inconsis­ multiple levels. differences across them (8). One of the tency checks of anthropometric, physical main difficulties during the pilot study activity, and food intake data; and when CHALLENGE 4. was a refusal to participate, due either possible, a return visit to the participant’s ACCOMPLISHING to time constraints, lack of interest, or household to correct unclear, incomplete, SIMULTAENOUS FIELDWORK fear of strangers. To overcome this, sev­ or questionable responses. eral strategies were used: leaflets on Monthly, each center sent informa­ During the common protocol design, the ELANS project were distributed in the tion on the status of interviews, entry the complexity of the logistics—a large neighborhoods prior to selection of house­ of 24hr food records in NDSR, and the number of interviewers performing field holds; interviewers showed an official accelerometer files, plus reported any Rev Panam Salud Publica 41, 2017 5 Special report Fisberg et al. • Lessons from a multicenter study in Latin America difficulties or impediments to the coor­ CHALLENGE 6. EXTRACTING The types of inconsistencies that were dinating center. It is worth noting DATA AND DATABASE found led the team to develop a second that considerable effort went into CONSISTENCY phase of consistency­checking post data food­matching between local foods/ collection. Each academic/research site recipes and the foods available in the In all countries, except Ecuador and team was responsible for this second NDSR database, as described by Venezuela, the CRO had computer tab­ phase. A specified procedure was used Kovalskys and colleagues (11). lets available for data collection. When to detect possible errors of data consis­ the electronic form was created, internal tency before the generation of each final Lessons learned: consistency checks were already pro­ database (Figure 2). All data collected grammed into the software to alert the were simultaneously reviewed by the (a) Maintaining regular, study­wide interviewers to outliers and other issues CRO, the accelerometer center and its communication among PIs, the CRO, and during real­time data entry. Physical Activity consultant team (PA), the accelerometer, nutritional, and statis­ Throughout the fieldwork, data qual­ and each academic/research site team. tical centers is critical to data quality. ity­control was carried out periodically. The PA consultants and each academic/ FIGURE 2. Flowchart for fieldwork closure in the Latin American Study of Nutrition and Health, 2016 PAa consultants b Academic team CRO team 1 2 Data analysis Data cleaning 3 Uploading of intake data Valid Compositions of Anthropometric accelerometers final sample data (IDs)0 (IDs)c (IDs)c Anthropometric outliers Valid intake data Dietaryc definitions (IDs) recomendations No Match1 No(IDs)c 2 Anthropometric No 2 consistency Intake YES No 3 YES consistency No General No YES 3 consistency 2 1 2 3 : Connectors YES : Entity : Process/Research Activity : Document PAa consultants Academic team CRO b : Decision team : Storage Shading represents the area responsible for each quality control process: White: accelerometer center (Physical Activity consultant team) Accelerometer CRO Intake d d d Light gray: database center DB DB DB (Control Research Organization) Dark gray: each academic/research site team Source: Prepared by the authors based on the project survey. a Physical activity. b Control Research Organization. c Subject identification. d Database. 6 Rev Panam Salud Publica 41, 2017 Fisberg et al. • Lessons from a multicenter study in Latin America Special report research team performed participant and guidance were required by all the various strategies detailed, this project identification (ID)­matching between research teams. Special consideration overcame various challenges to ensure their databases (accelerometer database was given to the variations in fortified the integrity of its cross­site data and food intake database, respectively) local foods and processed food prod­ collection. and the CRO database. Additionally, ucts, and to correcting micronutrient The entire research team at each par­ each academic/research team con­ contents of foods according to local ticipating center and the coordinating ducted anthropometric and dietary food composition tables. All of the re­ center—the PIs, local investigators, con­ i ntake consistency checks. Upon suc­ searchers provided input for this pro­ sultants, coordinators, research assis­ cessful completion, a general consis­ cess and approved the final version of tants, data managers, system analysts, tency check was then performed by the resulting document. statisticians, network managers, and each academic/research team. Thereaf­ accountants—proved to be essential to ter, final versions of the datasets were Lessons learned: study development and performance. generated by the PA consultants, CRO, Ideas and solutions, as well as planning and academic/research team. If any (a) Extensive expertise in dietary as­ for analysis and interpretation of re­ step was not successfully concluded sessment among the researchers is im­ sults, arose naturally when coming during this process, it was restarted, as portant because this sensitive part of the from a team thinking and working to­ shown in Figure 2 by connectors 1, 2, data collection required the greatest gether. The institutions involved all and 3. If any participant needed to be number of corrections to inconsistencies. benefited from the exchange of knowl­ excluded from the sample, the academic (b) Data cleaning demands close atten­ edge among researchers, the equipment team informed the CRO, and the case tion to detail and collaboration of multi­ acquisitions, the use of internationally­ was replaced with a matched case— ple stakeholders. recognized methodology, and wider same gender, age group, SEL, and geo­ (c) Correct use of accelerometers relies study dissemination. graphic area—from the oversample. on thorough interviewer training so that Multicenter cooperative studies while Common sources of errors included they can, in turn, effectively instruct par­ challenging, offer great potential for simple typographical errors; consis­ ticipants. Timely delivery of data to the building a scientific base for studies on tency errors, such as birth date and age PA team by the CRO is needed for correc­ nutrition and health. Considering the did not concur; differing participant ID tions to be made. A 70% accelerometer methods and experience of other multi­ and/or demographic information be­ compliance rate should be taken into ac­ center studies while developing the tween first and second visit; insufficient count, with fewer valid versus measured study design of the ELANS was impor­ accelerometer wear time (hours/day cases in the final sample. tant for surmounting possible obstacles and/or number of days); programming (d) A data analysis and publication and allowing faster progress. To this end, or accelerometer malfunctions; or ab­ plan that ensures appropriate data the lessons learned during ELANS pro­ sence of PA logs. After identifying the management and dissemination must vide new perspectives for better plan­ source of each error, most could be be developed with input from all team ning of financial, staff, and technological r esolved by communication between members and be followed meticulously resources for similar future studies. the centers. If the entry was unclear, through the end of the study. missing, or otherwise suspicious, field Acknowledgements. The authors staff were contacted for correction or FINAL CONSIDERATIONS would like to thank the following indi­ verification at participating house­ viduals at each of the participating sites holds. Accelerometer data errors were The development of a multicenter who made substantial contributions to corrected in the office when possible; household cross­sectional survey of the the ELANS: Luis Costa, Regina Fisberg, when not, the individuals were moved nutritional and physical activity status of Alejandra Guidi, Mariela Jauregui, Beate to the group without accelerometer use adolescents and adults in eight countries Lloyd, Brenda Lynch, and Bruno Zoca de (60% of the sample). When this group of Latin America was an innovative and Oliveira. was full, the rest were assigned to the fundamental step toward better under­ oversample group. As a considerable standing behaviors and their relation­ The following are members of ELANS number of cases were found to be in­ ship to health in the region. This survey Study Group: valid due to insufficient wear time, pro­ contributed to several lessons learned Chairpersons: Mauro Fisberg and Irina gramming error, or device malfunction, regarding the organization of a multi­ Kovalskys. the final percentage was reduced to center study. The feasibility of perform­ Co-chair: Georgina Gómez Salas. 25% of the sample. ing such a study depends on many Core Group members: Lilia Yadira Cortés Each site was responsible for verify­ considerations, including standardiza­ Sanabria, Mauro Fisberg, Georgina ing the quality of the data registered in tion of data collection, maintenance of Gómez Salas, Marianella Herrera­ its 24hr food records, and as needed, high data quality, and collaboration Cuenca, Irina Kovalskys, Rossina field staff were contacted for correction across sites. Among the greatest barriers G abriella Pareja Torres, Attilio Rigotti, or verification at participants’ house­ to conducting studies such as this one are and Martha Cecilia Yépez García. holds. Several steps were developed for the inherent organizational complexity, External advisory board: Berthold analysis of dietary intake data consis­ recruitment of collaborators and research Koletzko, Luis A. Moreno, Michael Pratt, tent with NDSR. This was a difficult staff, institutional cooperation, develop­ and Katherine L. Tucker. phase that required validation from ment of infrastructure, and identifica­ Project Managers: Viviana Guajardo and several researchers. Technical support tion of resources. By implementing the Ioná Zalcman Zimberg. Rev Panam Salud Publica 41, 2017 7 Special report Fisberg et al. • Lessons from a multicenter study in Latin America International Life Sciences Institute, Instituto de Investigación Nutricional, Peru: Costa Rica, Pontificia Universidad Argentina: María Paz Amigo, Fernando Mellisa Abad, Maria Reyna Liria, Krysty Católica de Chile, Pontificia Uni­ Cardini, Viviana Guajardo, Ximena Meza, Rossina Pareja Torres, and Mary versidad Javeriana, Universidad Cen­ Janezic, and Irina Kovalskys. Penny. tral de Venezuela/Fundación Bengoa, Instituto Pensi–Hospital Infantil Sabara, Universidad Central de Venezuela: Pablo Universidad San Francisco de Quito, Brazil: Natasha Aparecida Grande de Hernández, Marianella Herrera­Cuenca, and Instituto de Investigación Nutricio­ França, Mauro Fisberg, Agatha Nogueira Maritza Landaeta, Betty Méndez, nal de Perú. The funders had no role in Previdelli, and Ioná Zalcman Zimberg. Guillermo Ramírez, and Maura Vasquez. study design, data collection, analysis, Pontificia Universidad Catolica de Chile: Accelerometry analysis: Claudia Alberico the decision to publish, or the prepara­ Ó scar Castillo, Guadalupe Echeverría, and Priscila Bezerra Gonçalves. tion of this manuscript. MF is a member Leslie Landaeta, and Attilio Rigotti. Physical activity advisor: Gerson Luis de of the Board of Directors of Danone In­ Pontificia Universidad Javeriana, Colombia: Moraes Ferrari. stitute International (Paris, France). Yuri Milena Castillo, Lilia Yadira Cortés Dietary intake advisor: Ágatha Nogueira KLT received consulting fees from the Sanabria, Luisa Fernanda Tobar, and Luz Previdelli. Coca Cola Company to participate. Nayibe Vargas. Universidad de Costa Rica, Costa Rica: Funding. The ELANS is supported Conflict of interests: None declared. Anne Chinnock, Georgina Gómez Salas, by a scientific grant from the Coca Cola and Rafael Monge Rojas. Company (Atlanta, Georgia, United Disclaimer. Authors hold sole respon­ Universidad San Francisco de Quito, States) and by grants and/or support sibility for the views expressed in the E cuador: Lucia Eguiguren, Mónica Villar from the Instituto Pensi/Hospital In­ manuscript, which may not necessarily Cáceres, and Martha Cecilia Yépez fantil Sabara, International Life Science reflect the opinion or policy of the RPSP/ García. Institute of Argentina, Universidad de PAJPH and/or PAHO. REFERENCES 1. Bangdiwala SI, De Paula CS, Ramiro LS, 8. Rog DJ. Designing, managing and analyz­ 15. Associação Brasileira de Empresas de Muñoz SR. Coordination of international ing multisite evaluations. In: Newcomer Pesquisa. Critério padrão de classificação multicenter studies: Governance and ad­ KE, Hatrt HP, Wholey JS, eds. Handbook econômica Brasil. Available from: www. ministrative structure. Salud Publica Mex. of practical program evaluation. 4th ed. abep.org/criterioBrasil.aspx Accessed on 2003;45(1):58–66. New Jersey: Jossey­Bass; 2015. 10 December 2015. 2. Sprague S, Matta JM, Bhandari M, Dodgin 9. United States Centers for Disease 16. Comisión de Enlace Institucional AAM­ D, Clark CR, Kregor P, et al. Multicenter Control and Prevention. National Health SAIMO­CEIM. Nivel socioeconómico. collaboration in observational research: and Nutrition Examination Survey Antecedentes, marco conceptual, enfoque improving generalizability and efficiency. (NHANES). Survey methods and analytic metodológico y fortalezas. Buenos Aires: J Bone Joint Surg Am. 2009;91(suppl guidelines. 2014. Available from: www. Comisión de Enlace Institucional AAM­ 3):80–6. cdc.gov/nchs/nhanes/survey_methods. SAIMO­CEIM; 2006. 3. Moreno LA, De Henauw S, González­ htm Accessed on 21 October 2015. 17. Departamento Administrativo Nacional Gross M, Kersting M, Molnár D, Gottrand 10. Moshfegh AJ, Rhodes DG, Baer DJ, Murayi de Estadisticas de Colombia. Proyecciones F, et al. Design and implementation of the T, Clemens JC, Rumpler WV, et al. The US nacionales y departamentales de pobla­ Healthy Lifestyle in Europe by Nutrition Department of Agriculture Automated cion 2005­2020. Estudios Postcensales No. in Adolescence Cross­Sectional Study. Int J Multiple­Pass Method reduces bias in the 7. 2010. Available from: www.dane.gov. Obes. 2008;32(suppl 5):S4–11. collection of energy intakes. Am J Clin co/files/investigaciones/poblacion/con­ 4. Katzmarzyk PT, Barreira T V, Broyles ST, Nutr. 2008;88:324–32. ciliacenso/7_Proyecciones_poblacion.pdf Champagne CM, Chaput J­P, Fogelholm 11. Kovalskys I, Fisberg M, Gómez G, Rigotti Accessed on 10 December 2015. M, et al. The International Study of A, Cortés LY, Yépez MC, et al. 18. Instituto Nacional de Estadística y Censos Childhood Obesity, Lifestyle and the Standardization of the food composition de Ecuador. Encuesta de estratificación de Environment (ISCOLE): design and meth­ database used in the Latin American nivel socioeconómico, 2011. Available ods. BMC Public Health. 2013;13:900. Nutrition and Health Study (ELANS). from: www.inec.gob.ec/estadisticas/?op­ 5. van Stralen MM, te Velde SJ, Singh AS, De Nutrients. 2015;7:7914–24. tion=com_content&view=article&i­ Bourdeaudhuij I, Martens MK, van der 12. Salvo D, Reis RS, Sarmiento OL, d=112&Itemid=90& Accessed on 10 Sluis M, et al. European Energy balance Pratt M. Overcoming the challenges of December 2015. Research to prevent excessive weight Gain conducting physical activity and built 19. Asociacion Investigadores de Mercado. among Youth (ENERGY) project: Design environment research in Latin America: Grupos socioeconómicos Chile, 2012. and methodology of the ENERGY cross­ IPEN Latin America. Prev Med. 2014;69: Available from: http://comunicaciones. sectional survey. BMC Public Health. S86–92. udd.cl/mcom/files/2013/06/Informe­ 2011;11:65. 13. Kerr J, Sallis JF, Owen N, De Bourdeaudhuij Actualización­GSE­2012.pdf Accessed on 6. Bautista LE, Casas JP, Herrera VM, I, Cerin E, Sugiyama T, Reis R, et al. 10 December 2015. Miranda JJ, Perel P, Pichardo R, et al. Advancing science and policy through a 20. Instituto Nacional de Estadística. The Latin American Consortium of coordinated international study of physi­ República Bolivariana de Venezuela. Studies in Obesity (LASO). Obes Rev. cal activity and built environments: IPEN Síntesis estadística de pobreza e indicado­ 2009;10:364–70. adult methods. J Phys Act Health. res de desigualdad, 2011. Available from: 7. Fisberg M, Kovalskys I, Gómez Salas G, 2013;10:581–601. www.ine.gov.ve/documentos/Boletines_ Rigotti A, Cortés Sanabria L, Herrera­ 14. Sallis JF, Cerin E, Conway TL, Adams Electronicos/Estadisticas_Sociales_y_ Cuenca M, et al. Latin American Study of MA, Frank LD, Pratt M, et al. Urban envi­ Ambientales/Sintesis_Estadistica_ de_ Nutrition and Health (ELANS): rationale ronments in 14 cities worldwide are Pobreza_e_Indicadores_de_Desigualdad/ and study design. BMC Public Health. r elated to physical activity. Lancet. pdf/BoletinPobreza.pdf Accessed on 10 2016;16:93. 2016;387:1–10. December 2015. 8 Rev Panam Salud Publica 41, 2017 Fisberg et al. • Lessons from a multicenter study in Latin America Special report 21. World Bank. GINI index (World Bank esti­ ZS/countries Accessed on 7 January 2014. Available from: http://apps.who. mate). World Development Indicators, 2016. int/gho/data/view.main.2450A?lang=en 2013. Available from: http://data.world­ 23. World Bank. GDP per capita (current US$). Accessed on 7 January 2016. b a n k . o rg / i n d i c a t o r / S I . P O V. G I N I World Development Indicators, 2014. Accessed on 7 January 2016. Available from: http://data.worldbank. 22. World Bank. Urban population (% of org/indicator/NY.GDP.PCAP.CD/coun­ total). World Development Indicators, 2015. tries Accessed on 7 January 2016. Available from: http://data.worldbank. 24. World Health Organization. Global Health Manuscript received on 29 August 2016. Accepted o r g / i n d i c a t o r / S P. U R B . T O T L . I N . Observatory data repository. Obesity, for publication on 9 January 2017. RESUMEN En este informe se examinan los retos de llevar a cabo un estudio transversal multicén­ trico en países con culturas diversas y se transmiten las enseñanzas extraídas. Se usó Elaboración de un estudio el Estudio Latinoamericano de Nutrición y Salud (ELANS) como un estudio de facti­bilidad realizado en el período 2014–2015, que incluyó las ciudades más populosas de colaborativo multicéntrico ocho países de América Latina (Argentina, Brasil, Chile, Colombia, Costa Rica, en América Latina: Ecuador, Perú y Venezuela) y abarcó a cerca de 40% de la población de la Región de las enseñanzas extraídas del Américas. La muestra establecida incluyó a 9 000 personas de 15 a 65 años de edad, y fue estratificada según la ubicación geográfica (solo zonas urbanas) y por sexo, edad y Estudio Latinoamericano situación socioeconómica. de Nutrición y Salud Se encontraron seis retos principales: estructurar los equipos y seleccionar los sitios; preparar un protocolo único; obtener las aprobaciones éticas; terminar simultánea­ mente el trabajo sobre el terreno; velar por la calidad de los datos y extraer datos y mantener la uniformidad en todas las bases de datos. Las enseñanzas extraídas mues­ tran que la armonización, los estudios piloto, la uniformidad de los procedimientos, el riguroso control de la calidad de los datos y la comunicación y colaboración entre todos los sitios son imperativos. Los obstáculos incluyeron la complejidad de la orga­ nización, el reclutamiento de colaboradores y personal de investigación, la coope­ ración institucional, el desarrollo de infraestructura y la definición de los recursos. Antes de comenzar el trabajo sobre el terreno, se debe llegar a un consenso acerca de mediciones y resultados uniformes y la metodología de recopilación de datos, así como un plan para la gestión y el análisis de los datos y la comunicación, publicación y difusión de los resultados del estudio. A pesar de que estos estudios constituyen un desafío, hacen posible establecer una base científica para los estudios sobre la nutri­ ción, la actividad física y otros temas de salud, al facilitar las comparaciones entre los países de América Latina. Palabras clave Estudio multicéntrico; vigilancia nutricional; encuestas nutricionales; América Latina. Rev Panam Salud Publica 41, 2017 9 Special report Fisberg et al. • Lessons from a multicenter study in Latin America RESUMO Neste relato se examinam os desafios de realizar um estudo transversal multicêntrico em países com culturas diversas e os ensinamentos tirados. O Estudo Latino­americano Elaboração de estudo de Nutrição e Saúde (ELANS) serviu de estudo de viabilidade, englobando as cidades mais populosas de oito países da América Latina (Argentina, Brasil, Chile, Colômbia, cooperativo multicêntrico Costa Rica, Equador, Peru e Venezuela) em 2014–2015, representando cerca de 40% da na América Latina: população das Américas. A amostra pretendida compreendeu 9.000 indivíduos com ensinamentos do Projeto 15 a 65 anos de idade e foi estratificada por localização geográfica (exclusivamente áreas urbanas), sexo, idade e nível socioeconômico. do Estudo Latino-americano Os seis principais desafios identificados foram: estruturar as equipes e selecionar as de Nutrição e Saúde sedes, elaborar um único protocolo, obter as aprovações dos comitês de ética, realizar trabalho de campo simultâneo, assegurar a qualidade dos dados e extrair os dados e manter a consistência em todas as bases de dados. Os ensinamentos tirados demons­ tram serem imprescindíveis harmonização, estudo­piloto, uniformidade dos procedi­ mentos, ótimo controle da qualidade dos dados e comunicação e colaboração entre as sedes. Entre as barreiras estão a complexidade organizacional, recrutamento de colaboradores e pessoal de pesquisa, cooperação institucional, desenvolvimento de infrae strutura e identificação dos recursos. Antes do início do trabalho de campo, deve­se chegar a um consenso sobre medidas e resultados uniformes e metodologia de coleta de dados assim como um plano para o gerenciamento e análise dos dados, comunicação, publicação e disseminação dos resultados dos estudos. Apesar da sua complexidade, tais estudos têm um grande potencial de fundar uma base científica para estudos de nutrição, atividade física e outros tópicos relacionados à saúde, ao mesmo tempo que facilitam comparações entre os países da América Latina. Palavras-chave Estudo multicêntrico; vigilância nutricional; inquéritos nutricionais; América Latina. 10 Rev Panam Salud Publica 41, 2017