Cohort Profile: The Epidemiology of Chronic Diseases Cohort (EpiDoC)

Non-communicable chronic diseases are the leading cause of death and the main contributor to disease burden worldwide, accounting for 86% of all deaths in Portugal.1 Several modifiable behavioural risk factors, such as unhealthy dietary habits, physical inactivity, tobacco use and harmful use of alcohol, are the main risk factors for these diseases. Thus, the existence of epidemiological data on chronic diseases and their determinants (i.e. socioeconomic and demographic factors), associated factors and consequences are important public health tools for designing and developing strategies to tackle the burden of non-communicable diseases. In 2011, a prospective cohort study called Epidemiology of Chronic Diseases (EpiDoC) aimed to create a large population database for medical and health-related research in Portugal. To our knowledge, the EpiDoC study constitutes one of the first Portuguese prospective large cohort studies, including a representative sample of the Portuguese population, with the primary aim of examining the health determinants and outcomes of chronic non-communicable diseases and their impact on health care resource consumption. The EpiDoC study was designed by researchers from NOVA Medical School in Lisbon with close collaboration between social and biomedical scientists, ensuring a thorough multidisciplinary approach. The first wave of this cohort study, named EpiDoC 1 (EpiReumaPt), occurred between September 2011 and December 2013. Its primary aim was to assess rheumatic and musculoskeletal disease (RMD) prevalence and its burden in Portugal. This wave had two phases: the first consisted of a face-to-face interview, and the second included a detailed clinical evaluation of RMD performed by rheumatologists. This baseline assessment also enabled the creation of a population-based biobank (i.e. DNA, serum and total blood samples) for identifying genetic predictors and serum risk factors for chronic diseases. Musculoskeletal imaging data were also collected, in particular peripheral dual energy X-ray (DXA) in all second phase participants and X-ray of the affected joint(s). Similar to other cohort studies,1,2 the scope of the EpiDoC study has expanded over time. So far, two subsequent waves have been completed: EpiDoC 2 (March 2013–July 2015) and EpiDoC 3 (September 2015–July 2016). In both waves, data were collected through a phone interview. EpiDoC 2 (CoReumaPt) focused on lifestyle behaviours and their determinants, with a secondary goal of identifying innovative patient solutions for coping with disability. EpiDoC 3 (Saúde.Come) assessed inequalities in access to healthy food and health services, with a focus on food insecurity and its determinants and health consequences. The EpiDoC study was performed according to the principles established by the Declaration of Helsinki and revised in 2013 in Fortaleza. Ethical approval was obtained from the National Committee for Data Protection (Comissão Nacional de Proteção de Dados) and NOVA Medical School Ethics Committee. Ethical committees of regional health authorities also approved the study. EpiDoC is a prospective closed cohort study including a nationally representative sample of adults (≥18 years old) who were non-institutionalized and living in private households in Portugal Mainland and Islands (Azores and Madeira).3 Portugal is a south-western European country with a resident population of 10 562 178, of whom 8 million are adults (4 072 122 men and 4 585 118 women).4 During the past two decades, life expectancy in Portugal has been increasing. Data from the World Health Organization indicate that life expectancy in Portugal was 83.9 years for women and 78.2 years for men in 2015. In addition, as in other European countries, the Portuguese population has been undergoing demographic changes. The Portuguese population pyramid shows an increasing number of individuals at the top and a decreasing number at the bottom, indicating a new structure of the Portuguese population with fewer young people and more elderly. In 2015, the old-age dependency ratio was 31.1 per 100 persons of working age, which is the ratio between the number of persons aged ≥65 years (i.e. when individuals are generally economically inactive) and those aged 15–64 years.5 Portugal is divided into seven regions according to the Nomenclature of Territorial Units for Statistics II (NUTS II): Norte, Centro, Lisboa e Vale do Tejo, Alentejo, Algarve, Região Autónoma dos Açores (the Azores) and Região Autónoma da Madeira (Madeira). At the NUTS II level, the Norte region has the largest population density (34.7%), followed by Lisboa e Vale do Tejo (26.6%) and Centro (22.4%) (Figure 1). The other NUTS II regions (Alentejo, Algarve, the Azores, and Madeira) encompass small towns and villages with lower population densities and higher desertification rates. Portuguese population density distribution according to the 7 NUTS II. Considering the primary aim of EpiDoC 1, the sample size was calculated based on the estimated prevalence of rheumatic diseases with a 95% confidence interval (CI), and standardized for age and sex according to the total adult population of the studied areas. Assuming that the expected prevalence of rheumatic diseases was between 0.5% and 1%, and expecting a drop-out rate of 50%, it was estimated that a total of 9000 individuals should be recruited. To obtain regional representativeness, the sample size was stratified according to dimensions and characteristics of the seven Portuguese regions. Population recruitment was conducted by Centro de Estudos e Sondagens de Opinião da Universidade Católica Portuguesa (CESOP-UCP), and multistage random sampling was used for participant selection. In EpiDoC 1, candidates for participation were visited at their homes by a team of trained interviewers. Locations were selected as the primary unit of sampling according to the Census 2001. Selected households and their addresses were identified using a random selection of points in the map of each location, where the interviewer began a systematic step count (defined for each locality based on its size). Each selected household was visited, with no previous contact, up to three times (including evenings and weekends) if no candidate participant was present during the first visit. In each household, an individual ≥18 years old with permanent residence and the most recently completed birthday was selected to be a participant in the EpiDoC study. Before participant interviews, the EpiDoC team gave information about study details and aims at local churches, primary care centres and municipalities. Local priests, health providers and municipality employers helped us to spread the information and motivate participation. EpiDoC 1 enrolled 10 661 participants and was primarily designed to estimate the prevalence of RMDs. To provide a comprehensive understanding of the burden of RMDs, this wave had the secondary aim of evaluating quality of life, physical function, mental health, work status and health care resource consumption, with the purpose of identifying differences in these and other outcomes between individuals with and without RMDs.3 EpiDoC 1 data collection consisted of two phases. Phase 1 involved face-to-face interviews conducted by a team of trained interviewers (non-physicians) through door-to-door visits. Phase 2 involved clinical observations with physical examination performed by rheumatologists, for participants identified as potentially having an RMD (using a screening questionnaire applied at Phase 1) and 20% of asymptomatic individuals. All procedures occurred between September 2011 and December 2013. Of the 10 661 participants selected in Phase 1, 7451 had a positive RMD screening and 3210 had a negative RMD screening. A total of 8152 participants were contacted in Phase 2: 7451 with a positive RMD and 701 (∼20%) without an RMD as previously defined in the study protocol. Of these, 4275 did not attend a clinical observation by a rheumatologist. Therefore, at the end of Phase 2, there were 3877 clinical observations with physical examination performed by rheumatologists; 3198 participants received validation of an RMD diagnosis and 679 did not have an RMD diagnosis. In Phase 1, a structured questionnaire using a computer-assisted personal interview (CAPI) system was used to collect data. Questions on rheumatic symptoms were asked, and an algorithm for screening each RMD was applied. An individual was considered to have a positive screening: if he/she mentioned a previously known RMD; if any of the specific disease algorithms in the screening questionnaires were positive; or if the participant reported muscle, vertebral or peripheral joint pain in the previous 4 weeks.3 Phase 2 was performed by rheumatologists at the local primary care centre for all participants who were identified as having a positive RMD screening. All clinical laboratory and imaging data were verified by a team of three experienced rheumatologists, and diagnoses were confirmed according to validated criteria.3 All participants enrolled in EpiDoC 1 (10 661 participants) were invited to participate in a follow-up study, of whom 10 153 (95.2%) signed consent forms and agreed to participate. For follow-up waves (EpiDoC 2 and 3), data were collected using a structured questionnaire administered by phone call interviews using a CAPI system. A core questionnaire was used in each EpiDoC wave, with additional questions added according to the focus of each wave. In EpiDoC 2 and 3, when a participant was not available, additional attempts were made at different times up to a maximum of six attempts. The last contact attempt had to follow the previous contact by least 1 month; only then was the contact attempt abandoned. EpiDoC 2 was the first follow-up wave, with data collected between March 2013 and July 2015. EpiDoC 2 included 7591 participants (out of 10 153 eligible participants) representative of the adult Portuguese population, resulting in a response rate of 71.2% from EpiDoC 1. Considering that the main risk factors for non-communicable diseases are unhealthy lifestyle behaviours, EpiDoC 2 employed the core structured questionnaire but included more detailed questions on lifestyle behaviours, such as physical activity, dietary habits, tobacco and alcohol use and sleeping habits. Questions regarding innovative patient solutions for coping with disability were also included. EpiDoC 3 occurred between September 2015 and July 2016 and included 5653 participants, resulting in a response rate of 55.7% from EpiDoC 1. This wave continued to employ the core structured questionnaire but included questions on food insecurity, its determinants and its health consequences. This particular interest in food insecurity was based on a growing awareness of social inequalities in health and modifiable risk factors for chronic diseases, such as dietary patterns, as well as the economic crisis faced by Portugal in previous years. The participation rate declined from EpiDoC 1 to EpiDoC 3, similar to most other population-based studies.2,6Table 1 presents the characteristics of participants in the cohort. There were no significant differences in any categories of variables between the three waves. Characteristics of the participants in the cohort SD, standard deviation. Characteristics of the participants in the cohort SD, standard deviation. The EpiDoC study employed cross-sectional and longitudinal study designs (Figure 2). As it used a closed cohort, no new participants were added in any wave. Table 2 presents the attrition rates between EpiDoC 1 and 2, EpiDoC 1 and 3, and EpiDoC 2 and 3. Characteristics of the participants in the cohort (attrition rate) Characteristics of the participants in the cohort (attrition rate) Flowchart of EpiDoC study. The participation rate in EpiDoC 3 was of 53.03%. The attrition was most pronounced in younger adults (18–29 years old). Of the 10 661 participants in EpiDoC 1,509 (4.8%) refused to sign the consent form for follow-up. Of the resulting 10 153 eligible participants for EpiDoC 2, 79 (0.8%) had died, 179 (1.8%) wished to leave the study and 917 (9.0%) had an invalid contact. Thus, a total of 1639 participants were lost to follow-up; these subjects had a mean age of 55 years, and 962 (58.7%) were women. Between EpiDoC 2 and 3, 51 (0.6%) participants had died, 232 (2.6%) wished to leave the study and 721 (8.0%) had an invalid contact. Thus, a total of 1004 participants were lost to follow-up; these individuals had a mean age of 56 years, and 620 (61.8%) were women. Figure 1 shows the flowchart of the EpiDoC study. Data collection included measures for five domains that were central to the longitudinal study: sociodemographic characteristics, lifestyle characteristics, health and clinical characteristics, health care resource consumption, a population-based biobank (total blood, serum and DNA) and imaging data (peripheral DXA and X-ray of the affected joint) (Table 3). For reasons of longitudinal comparison, most measurement tools were used consistently across waves. However, some measurement tools were updated or improved, new measurement tools were added and old measurement tools were removed as needed. Data collected over EpiDoC study Data collected over EpiDoC study Of the measurements obtained across all three waves, lifestyle variables included smoking habits, alcohol intake and physical exercise. Health variables included anthropometric measures, self-reported chronic diseases, rheumatic diseases, a health assessment questionnaire (HAQ)7 and the European Quality of Life Survey with five dimensions and three levels (EQ-5D-3L).8,9 Employment variables included employment status, retirement due to disease, retirement due to RMD, work absenteeism due to disease, work disability due to RMD, unemployment due to disease, unemployment due to RMD, number of working hours/week and changed employment status due to RMD. Health care resource variables included hospitalization events (in previous 12 months since last contact), their reason and their duration. Concerning falls and bone fractures, variables included any falls or bone fractures and the number and location of bone fractures. Sociodemographic data, including sex, age, ethnicity, years of education and education level, and marital status, were collected only in EpiDoC 1, based on the assumption that these characteristics would not change over time. Other information obtained only in EpiDoC 1 were household income, household composition, coffee intake and health information from the 36-item Short Form Survey (SF-36).10 Information obtained only in EpiDoC 1 and 2 were the Hospital Anxiety and Depression Scale (HADS)11 and home care assistance (in previous 12 months or since last contact), its provider and its payer. Information obtained only in EpiDoC 2 and 3 included: sleep habits; frequency of watching TV, using computer/videogames/tablets and using the internet; number of meals per day; frequency of soup, vegetable, fruit, meat, fish, milk/dairy and water consumption; numbers of medical appointments (in previous 12 months or since last contact), private versus public medical appointments, private medical appointments with or without insurance, public medical appointments in a hospital/health care centre and private or public medical appointments by specialty. Information obtained only in EpiDoC 3 were frequency of olive oil, wine, beans, fat and sugar consumption; attitudes toward food; a food insecurity scale; and characteristics of food acquisition and preparation. Population-based biobank and imaging data were collected in EpiDoC 1 during medical appointments at the local primary care centre. Blood samples were collected from 3608 participants (DNA, serum and whole blood). Taking into consideration the imaging reservoir, there were a total of 3342 participants who had a forearm bone mineral density evaluation through peripheral DXA. Also, bone mineral assessment (BMA) using a high-resolution digital X-ray machine (D3A, France) was collected from 2422 wrists and 2228 calcaneus bones. Simple X-rays were performed to examine 438 hands, 122 hips, 479 knees, 1265 lumbar spines, 691 thoracic spines and 206 cervical spines, according to participants’ musculoskeletal complaints. All data collected, including biobank and imaging data, are detailed in Table 3. Over 24 peer-reviewed journal publications based on EpiDoC data have been published to date, covering a wide range of scientific domains. A full list of publications can be found on our website[http://cedoc.unl.pt/epidoc-unit/]. Sample overviews of study data are in 1, 2 and and 95% of confidence interval of reported chronic diseases and lifestyle was not included at and 95% of confidence interval of reported chronic diseases and lifestyle was not included at In EpiDoC 1, socioeconomic of the Portuguese adult a social and health of an was that of the adult Portuguese population had a of data from EpiDoC 2 that and a education are associated with an unhealthy lifestyle and higher prevalence of chronic in health is a public health, with food insecurity one of its main insecurity is defined as a in a healthy due to economic and is a of EpiDoC 3 a prevalence of food insecurity and its and unhealthy dietary insecurity was associated with non-communicable diseases, lower quality of life and higher health care resource using EpiDoC data have questions and about the to food insecurity, not only to individual health status but also to public health Considering health and health-related characteristics, blood level, and were self-reported the Portuguese adult The prevalence of in Portugal is similar to that reported in other close was the of individuals of one or more RMDs, who did not have a previous diagnosis of 3877 This be by the number of rheumatologists in Portugal 100 and the of awareness the population about these diseases, as RMD symptoms are as of the helped a new for of by the National of Health in collaboration with the EpiDoC research The RMD with the prevalence in Portugal was pain 95% which was more in women in men pain with age, and its prevalence was in the age 95% the impact of on health-related quality of life, physical and mental health the Portuguese population, EpiDoC data that with have more health care resource consumption, were more and had more in the previous 12 with participants with no In EpiDoC 1, a number of people to have due to This to years of working life lost and potentially due to self-reported are also to at least 0.5% of the the burden of and the to RMD which is a to the of to the of rheumatic EpiDoC 1 also a prevalence of other chronic diseases Portuguese adults such as and The are a population for chronic diseases, whom the of two or more chronic diseases is leading to quality of life and The most chronic diseases in the were rheumatic disease and were and were Similar were found for Portuguese of whom more did not and more were EpiDoC 2 estimated a prevalence of and Portuguese of and with or were more to higher levels of physical disability and lower quality of and clinical data have been used in of adult the impact of falls and fractures, level, and health, as well as on and and bone in In EpiDoC publications have our understanding of socioeconomic and health inequalities Portuguese the elderly. that unhealthy are more the most and are associated with a higher prevalence of chronic non-communicable diseases and higher health care resource consumption. The EpiDoC study has also on the burden of rheumatic diseases in Portugal. shows a to the and to provide care to rheumatic EpiDoC work is aimed at the determinants and burden of and other chronic non-communicable diseases, mental and diseases. be at understanding health The main of the EpiDoC study are its population and sample of measures and biobank blood is its research with a team laboratory and The different of the three waves are also a as have expanded the scope of the EpiDoC study to a more cohort study. The EpiDoC study also has some such as its attrition which is similar to that of other and was not different between the three waves. In EpiDoC 2 and EpiDoC 3, data were collected by phone to attrition by using for and and to all is that diseases were a detailed and comprehensive questionnaire included a screening for RMD All measurement tools and were validated and the screening of was validated by an algorithm by Each wave was of a structured comprehensive questionnaire which was for participant and The EpiDoC research and Data from our cohort and are for researchers who a research to the scientific details about questionnaire and clinical measurements can be found on our A research form can be and to An EpiDoC all for and to access data and use of EpiDoC is a prospective population-based closed cohort study that health The study primarily aimed to rheumatic diseases, but its scope has to other chronic diseases, and and diseases. health of the adult population ≥18 in Portugal were completed: EpiDoC 1 (September EpiDoC 2 (March 2013–July 2015) and EpiDoC 3 (September 2015–July 2016). EpiDoC have a total of years, with an attrition rate of EpiDoC 1, 2 and 3 had 10 7591 and participants, The EpiDoC sample is representative of the Portuguese In EpiDoC 1, participants were and most were 10 and In EpiDoC 2, participants were and the mean age of all participants was years. In EpiDoC 3, participants were and the mean age of all participants was years. EpiDoC data are to researchers who research to the scientific details can be found on our The EpiDoC study has had of EpiDoC 1 and EpiDoC 2 were by from the Portuguese of Health and from Medical Centro de de and individual rheumatologists. EpiDoC 3 was by the Health by of