Cohort Profile: The International Mobility In Aging Study (IMIAS)

Why was the cohort set up?

Mobility, the ability to move in one’s environment, is an essential feature of human functioning.¹ Among older adults, mobility disability is more frequent in women than in men and differences are largely unexplained. Results from the Established Populations Epidemiologic Study of the Elderly (EPESE) showed that the prevalence of mobility disability in women increased from 22% at age 70 to 62% at age 80, whereas in men disability reached only 15% and 38% at comparable ages.² This gender gap in mobility disability in old age was narrower and had decreased in recent decades in Sweden.³ In less industrialized countries, gender mobility disability differences are larger. Using data from the SABE surveys (SAlud BienEstar, or Health and Wellbeing) on older people in seven Latin America and Caribbean (LAC) cities,⁴ we found a mobility disability prevalence odds ratio for women compared with men of 2.4, when controlling for life course exposures and health factors.⁵

The International Mobility In Aging Study (IMIAS) study was designed by researchers from the Université de Montréal and Queen’s University in Canada, Universidade Federal do Rio Grande do Norte in Natal, Brazil, Universidad de Caldas in Manizales, Colombia, and researchers affiliated with the Albanian School of Public Health in Tirana, in a close collaboration between social and biomedical scientists.

The main objective of IMIAS is to increase knowledge oof the effects of gender differences on mobility disability (prevalence, incidence and recovery) in ageing and on the impact of potentially modifiable risk factors, with detailed hypotheses on the pathways linking life course exposure to violence, poverty, social isolation and reproductive history with mobility disability in old age. We use an interdisciplinary approach to conduct research on gender differences in mobility across societies at different stages of the epidemiological transition. Herein, the term ‘gender’ emcompasses the biological and social differences between men and women.

Who is in the cohort?

IMIAS is a population-based, longitudinal, prospective cohort, multicentre, multidisciplinary study conducted in five cities: Tirana (Albania), Natal (Brazil), Manizales (Colombia), Kingston (Ontario, Canada) and Saint-Hyacinthe (Quebec, Canada). It examines the prevalence, incidence of and recovery from mobility disability and associated risk factors in n = 2002 community-dwelling men and women aged 65–74 years at baseline. Participants were followed to establish the baseline prevalence and the 4-year incidence of and recovery from mobility disability, allowing for the study of both cross-sectional and longitudinal relationships between risk factors and mobility disability. Inclusion of five research sites allowed analyses of the impact of the diversity of environments in terms of life course past and current living social and economic conditions, gender roles and current neighborhood environments on mobility disability. IMIAS included communities that differed in their social, cultural, socioeconomic and geographical environments and in their health care systems (Table 1).

Table 1.

Country profiles: demographic, health and economic indicators

	Albania			Brazil			Canada			Colombia
	Total	Men	Women	Total	Men	Women	Total	Men	Women	Total	Men	Women
Demographic indicators
Total population (millions), 2015a	2.9	1.5	1.4	206	101	107	35.9	17.8	18.1	48.2	23.7	24.5
Population aged over 65 (millions), 2015a	0.4	0.2	0.2	16.4	7.0	9.0	5.8	2.6	3.2	3.4	1.5	1.9
% population over 65, 2015	12	12	13	8	7	9	16	15	17	7.0	6.3	7.8
Life expectancy at birthb	73	71	74	73	70	77	81	79	83	75	72	79
Health indicators
Healthy life expectancyb	64	64	64	64	62	66	73	71	75	66	64	69
% life expectancy in good health	88%	90%	86%	88%	89%	86%	86%	90%	90%	88%	89%	87%
Obesity (aged over 18)c	21.1	20.9	21.3	21.6	18	24.9	28.8	28.8	28.8	21.9	17.0	26.1
Economic indicators
Gini coefficienta	29			51.5			33.7			53.5
Gross Domestic Product/per capitaa	10 397			14 445			42 891			12 988
Gender Inequality Index Ranke	51			92			18			89
Health services
Health expenditures ($/per capita)b	244			606			4409			284
Universal Health Care indexd	62			77			>80			76
Health as % GNPb	7.0%			8.4%			10.1%			6.1%

GNP, gross national product.

^aUnited Nations Development Program. Human Development Reports 2015. [https://esa.un.org/unpd/wpp/DataQuery/].

^bWHO Health Statistics 2010. [http://www.who.int/whosis/whostat/2010/en/].

^cWHO 2015. [http://www.who.int/gho/ncd/risk_factors/overweight_obesity/obesity_adults/en/].

^dWHO 2015. Universal Health Coverage Portal. [http://apps.who.int/gho/cabinet/uhc-service-coverage.jsp].

^eUNDP 2016 Human Development Report. Table 5. Gender Inequality Index. [http://hdr.undp.org/en/composite/GII].

The sample was stratified by sex, with the aim to recruit 200 men and 200 women registered at community health clinics and through public health insurance databases. The sample in Natal was randomly drawn from the registers of the Family Health programme at three low-income and two middle-income areas of the city. The Manizales sample was randomly drawn from the Public Health Insurance database. At Tirana, participants were randomly selected from the population registered at two neighbourhood health centres in one middle class area of the city. In the Canadian research sites, the ethics committees did not allow direct contact with the potential participants. Thus in Kingston and Saint-Hyacinthe, potential participants were invited by letter from their primary care physicians to contact our field coordinator in order to participate in the study. In Saint-Hyacinthe, the sample was stratified by neighbourhoods. In Kingston this stratification was not possible because the Ethics Board requested that invitation letters be sent directly by the participating clinics to avoid knowledge of the addresses of the potential participants among the research team.

Exclusion criteria

At baseline, participants with scores lower than 4 (severe cognitive impairment) in the orientation scale of the Leganes Cognitive Test were excluded from the study because they were considered unable to answer the questionnaire and to complete the physical function tests, as well as freely consent to the study.⁶ The numbers of excluded people were zero in Kingston, one in Saint-Hyacinthe and Tirana, two in Manizales and five in Natal.

Ethics

Ethical approval for this project was obtained from the ethics review committees of the research centres of the University of Montreal Hospitals (CR-CHUM), Queen’s University (Kingston), the Albanian Institute of Public Health, the Federal University of Rio Grande do Norte (Brazil) and the University of Caldas (Colombia).

How often have they been followed up and how was the field work conducted?

The first data collection (baseline) took place in 2012. Before fieldwork, research and field staff were trained in all aspects of data collection, questionnaire administration and physical measurements. Cohort members underwent biennual home follow-up visits. The first follow-up assessments were conducted in 2014 and the second in 2016. IMIAS has expanded its collaborative efforts with nested sub-studies.

At all sites, study procedures were carried out by a trained interviewer at the participant’s home unless the participant requested otherwise. Interviewers at each site received standard training based on protocol instructions and data entry forms. Assessments in Tirana were done by public health professionals and graduate students, in Natal by physiotherapists, in Manizales by local nurses, in Saint-Hyacinthe by retired high school teachers and in Kingston by lay people with university degrees. The principal investigators and the study coordinator trained all interviewers at each site in the 5 days preceding fieldwork, using the study manual of procedures and role playing. Retraining took place a few weeks later during data collection. Training and retraining sessions were repeated in 2014 and 2016.

The questionnaires, all data collection documents and manuals of procedures were available in the local languages: Albanian, Spanish, Portuguese, English and French. Questionnaires and manuals of procedures are available upon request. The main scales were validated in pilot studies conducted in Brazil, Colombia and Quebec.^6,7–12 In Tirana, translation into Albanian and adaptation to the local population were also carried out, although no pilot study was conducted at Tirana.

What has been measured?

Box 1 presents the variables included in the interview questionnaire, functional tests and biomarkers from blood and saliva analysis. The questionnaire included sections on demographic and socioeconomic variables, self-reported chronic conditions, depressive symptoms, medications currently used, falls and fear of falling, disability, life course history of socioeconomic conditions, exposure to domestic violence, reproductive history, physical activity, smoking and alcohol and illegal drug use, quality of life, neighbourhood social, economic and physical characteristics, social networks and support, social participation, gender roles, decision making power, financial autonomy and life space assessment. New variables were included in the second wave interviews: housing quality, driving experience, resilience,¹³ personal mastery,¹⁴ pain, urinary incontinence and an additional disability scale: the Late Life Disability Index.¹⁵

Box 1.

Summary of variables collected in IMIAS

Demographic data Household membership Living arrangements Marital status Education Income and assets Earnings Sources of income Occupation Employment situation Economic hardship Migration history Physical health Self-rated health Chronic conditions (SABE) Medication use Vision test (ETDRS Tumbling E chart) Falls/balance Falls Efficacy Scale International (FES I) Functional limitations (Nagi questionnaire) Disability, mobility and activities of daily living Mobility Assessment Tool (MAT) videos Reproductive history (only for women) Access and use of health care services Behavioural health Smoking/smoking history Alcohol consumption Physical activity: IPAQ

Mental health Cognitive function: MoCA/Leganes Cognitive Test (LCT) CES-D depression scale Social and civic participation Social activities Neighbourhood physical and social environment Friends-family networks and support Life-space assessment Psychosocial factors Gender roles: Bem Sex roles inventory Decision autonomy Victimization and fear of victimization Early life circumstances (first 15 years) and childhood adversity. Violence Hurt, insulted, threaten and scream (HITS) Two questions from Canadian Health Questionnaire Physical examination and performance Height and weight; waist circumference Blood pressure Grip strength Short Physical Performance Battery  (SPPB): balance, gait speed and chair stands Blood assays Haemoglobin and haematocrit Glycated haemoglobin Albumin Total, LDL, HDL cholesterol and triglycerides High sensitivity C-reactive protein (HS-CRP) Interleukin 6 (IL-6)

HDL, high-density lipoprotein; LDL, low-density lipoprotein; ETDRS, Early Treatment Diabetic Retinopathy Study; FES I, Falls Self-Efficacy Scale; IPAQ, International Physical Activity Questionnaire; CES, Center for Epidemiologic Studies Scale-Depression;

Two tests of cognitive function were used. The Leganes Cognitive Test (LCT), originally developed to screen for dementia in populations with low education and used previously in cognitive ageing research,⁶ was administered to all IMIAS participants. The Montreal Cognitive Assessment (MoCA) was used only in the Canadian sample, since during pilot testing of MoCA we documented its lack of validity in Manizales.⁸

Vision assessment, blood pressure, grip strength, physical performance based on assessments of gait speed, balance and chair stands [the Short Physical Performance Battery (SPPB)], see: [http://www.grc.nia.nih.gov/branches/leps/sppb/index.htm], were also carried out.¹⁶

Participants from all five sites underwent fasting blood tests. A total of 1728 participants out of the 2002 total sample provided blood (86%). Blood samples were taken by nurses or a trained phlebotomist. Conventional blood analyses were done at the local hospitals. Inflammatory markers were analysed as follows: serum from Saint-Hyacinthe and Kingston was analysed at Kingston General Hospital, affiliated with Queen’s University, serum from Manizales at the University Hospital of the Caldas and serum from Natal at a certified local commercial laboratory. Inflammatory markers at Tirana were not determined. Salivary cortisol was collected from a subsample of 309 participants, instructed to collect saliva on two consecutive days at awakening, 30 min and 60 min after awakening, at 15.00  h and before bedtime. All information was collected on laptops, using WILLIAM (William^MD MultiSpectra Inc.), resulting in more efficient interviews and greater ease in producing databases.

Baseline response rates and attrition

Baseline response rates in Manizales, Natal and Tirana were higher than 90%. Response rates in Canada were low; these low rates were induced by the recruitment procedures required by the ethics review boards at the Canadian research sites. Only 30% of people who received the invitation letter from their primary care doctor called our research field team, and 95% of them agreed to participate in the study, giving an overall response rate of 28%. The final samples were 398 at Kingston, 401 at Saint-Hyacinthe, 394 at Tirana, 407 at Manizales and 402 at Natal. The small variations around the planned sample size of 400 were due to the logistics of fieldwork.

Comparisons with the 2006 Canadian census data on education show that participants in Kingston were more highly educated than the general population of that age group in that city, whereas participants in Saint-Hyacinthe had levels of education close to the census data for that age group in Saint-Hyacinthe. In particular, whereas the proportion of participants with more than high school education was 79% in our Kingston sample, this proportion was 55% in the 2006 Canadian census for the Kingston area. In Saint-Hyacinthe, these proportions were 48% and 46%, respectively.

Attrition between the 2012 baseline data collection and the latest follow-up in 2016 was relatively small. Mortality was 2.3% in Kingston, 3.2% in Saint-Hyacinthe, 6.6% in Tirana, 7.4% in Manizales and 8.5% in Natal. At 4 years, of the 1891 subjects not known to have died, 1527 participated in the follow-up assessment (81%). The proportion of completed interviews in 2016, of those recruited in 2012 and not known to have died, was 75% at Kingston and at Saint-Hyacinthe, 82% at Tirana, 83% at Manizales and 67% at Natal.

What has been found? Key findings and publications

Quantification of the gender gap in mobility disability

The first IMIAS objective was to measure the magnitude of the gender gap in mobility disability and to increase understanding of gender differences in life course exposures related to mobility. Prevalence of self-reported mobility disability (difficulty walking 400 m without help or difficulty climbing a flight of stairs without assistance) and activities of daily living (ADL) disability [any difficulty in performing six basic ADL (walking across a room, bathing, getting dressed, getting up, eating and going to the bathroom)] were significantly higher in women than in men in all sites except for Kingston. Poor physical performance (assessed by an SPPB score <8) was more frequent in women than in men in Manizales, Natal and Tirana, the three non-Canadian cities. Among men, few differences in physical performance, mobility or ADL disability were observed across cities. Among women, poor SPPB mobility and ADL disability prevalences were significantly elevated for Manizales, Natal and Tirana compared with Kingston. These results provide preliminary support for our hypothesis that observed gender physical performance and mobility and ADL disability gap is related to gender inequalities prevailing in less gender-egalitarian societies.¹⁷

Life course adversity, reproductive history, gender roles and domestic violence as gender-related risk factors for physical performance and mobility disability in old age

IMIAS showed evidence of the early origins of social and economic inequalities in physical performance during old age: low physical performance was associated with childhood social and economic adversity, semi-skilled occupations, living alone and insufficient income in later life.¹⁸ Early social adversity was associated with high levels of inflammation in old age¹⁹ and inflammation in old age was associated with low physical performance.²⁰ All these associations between life course adversity, inflammation and physical performance were similar in men and women.

IMIAS provided evidence supporting the hypothesis that adolescent childbirth increases the risk of developing chronic diseases and physical limitations in older age: after adjustment for study site, age, education, childhood economic adversity and lifetime births, women who gave birth at a young age had 1.75 [95% confidence interval (CI): 1.17–2.64] odds ratio for poor physical function compared with women who gave birth after 18 years of age.²¹

Gender roles have been related to mortality, mental health and health behaviours. Masculinity and femininity traits could be shaped by societal institutions, family, school, work environment and the society. Previous research has not examined the associations between gender roles and physical function. Upon validation of a brief 12-item version of the Bem Sex Roles Inventory (BSRI),²² we classified subjects into four types (masculine, feminine, androgynous and undifferentiated), unveiled cross-sectional and longitudinal associations between these gender role types and physical performance and identified mediating pathways between BSRI and physical performance. Feminine and undifferentiated gender roles are independent risk factors for mobility disability and low physical performance.²³ Masculinity predicts higher maintenance of good physical function. Gender roles influence health behaviours which in turn contribute to chronic conditions and faster decline of lower extremity physical function.²⁴

Domestic violence was frequent in the life of older adults in all five cities, with excess in women compared with men.^25,26 Although reported physical violence in the last 6 months was rare, psychological violence both from the partner and from family members was reported frequently and was associated with low support from family and friends. Family violence was more frequent in multigenerational households.²⁷ In the Canadian subsample of IMIAS, risk factors associated with both current and lifetime violence included: having a previous experience of violence involving an intimate partner; witnessing violence between family members in childhood; and having poor quality relationships with family and friends.²⁶

In IMIAS, life course violence was demonstrated to be a risk factor for both self-reported mobility and measured physical performance.^18,25 Multiple chronic conditions, inflammation, depression and lack of physical activity were shown to constitute pathways between domestic violence, and mobility disability and physical performance.²⁸

Chronic conditions as risk factors for mobility disability, and clinical management of chronic conditions

Since chronic conditions are well-known factors for mobility disability and physical decline, and our work demonstrates they are important mediators between socioeconomic adversity, adolescent childbirth, gender roles and domestic violence, and mobility disability and low physical performance, we have devoted part of our research to estimate the prevalence and clinical management of chronic conditions in our research settings. We focused our research on hypertension, falls, depression and visual impairment, since these are preventable conditions with clear diagnoses and usually available and relatively inexpensive treatment.

Hypertension prevalence ranged from 53% in Saint-Hyacinthe to 83% in Tirana. More than two-thirds of hypertensive participants were aware of their condition (from 67% in Saint-Hyacinthe to 85% in Tirana); women were more aware than men. Though most of those aware of their hypertensive condition were being treated pharmacologically, associations between awareness, and physical activity and refraining from smoking, were weak. Control among treated hypertensive participants was low, especially in Tirana and Natal. Diabetes and physical inactivity were associated with poor hypertension control.²⁹

The prevalence of falls in the previous year was 27% with significant site differences, ranging from 39% in Kingston to 18% in Tirana.³⁰ Risk of mobility disability was associated with fear of falling³¹ and associated with reduction in life space in all sites of the study.³² Furthermore, using classification and regression tree analysis, we proposed a simple algorithm to predict falls in primary care based on three simple indicators: fear of falling, number of previous falls and time to complete five chair rises.³³ We explored social capital, neighbourhood-level socioeconomic status and built environment factors as potential factors in the aetiology of the occurrence of falls among older people.³⁴

Prevalence of clinically relevant depression varied widely (Brazil 6.3% to Albania 46.6%), and was higher in women than in men. Low education, insufficient income, living alone, multiple chronic conditions and poor physical performance were all significantly associated with depression prevalence in all sites,³⁵ and the androgynous gender role was associated with low prevalence of depression independent of biological sex.³⁶ We reported socioeconomic inequalities in psychotropic drug use among IMIAS participants. An inverse association was observed between socioeconomic standing and psychotropic drug use in Canada, whereas in Latin America greater antidepressant use was associated with high education level and elevated income.³⁷

The prevalence rates of visual impairment in Natal, Manizales and Tirana were 10%, 22% and 15%, respectively. The percentages of those who saw an eye care provider in the past year in Natal, Manizales and Tirana were 54%, 37% and 26%, respectively. Risk factors for visual impairment were older age, being a woman, having a low income and having experienced physical violence from a family member. Factors associated with seeing an eye care provider in the past year included being a woman, more years of formal education, being in a higher income category and history of physical violence from a family member.³⁸

For the interested reader, IMIAS has provided opportunities for research on active ageing,³⁹ resilience,⁴⁰ diurnal cortisol profiles and physical function,⁴¹ weakness,⁴² inflammatory markers,²⁰ social support⁴³ and self-rated health.^44,45

What are the main strengths and weaknesses?

The main strengths of the IMIAS include:

1. IMIAS gives new results on the gender-physical function gap by examining ageing populations living in diverse societies from high-, middle- and low-income countries, using identical and rigorous study designs and data collection methodology and widely used physical and cognitive function measurement tools, which have been validated in the studied populations.

2. IMIAS comprises community-based samples from five distinctive research sites in four countries, maximizing variability in life course exposures and health and disability outcomes.

3. The sample size of 2002 men and women is large enough to examine social and reproductive history influences on mobility, allowing for relatively precise confidence intervals around estimates.

4. The broad spectrum of simultaneous measures across multiple domains provides a comprehensive picture of physical and mental health and functioning in this older age group.

5. All instruments and questionnaires were pilot tested in Canada, Colombia and Brazil in preceding studies, and are available in the five study languages.

6. The study involves researchers from a variety of disciplines (epidemiology, geriatric and family medicine, public health, occupational therapy, physiotherapy, psychology and statistics), across a number of academic institutions in different countries. This broad investigative scope has allowed the cohort to be well characterized, collecting an extensive array of factors (biomedical, behavioural, demographic, economic, social and psychological) that contribute to the health status of participants.

Early feedback suggests that the study could have been enhanced by improving:

1. Response rates in Canada. The baseline response rates were low (30%) in Canada, but these low rates are mostly caused by the indirect recruitment procedure imposed by the human subjects committees at the Canadian research sites. Comparisons of 2006 census data on education show that participants in Kingston are more highly educated than the general population in that city, whereas participants in Saint Hyacinthe have levels of education that are consistent with census data. Mobility disability in the Canadian sites are close to those in the 2012 Canadian Survey on Disability, which used a comparable question (22% for women and 18% of men older than 65, and 14% in the 65–74 age group, with no differences between sexes for that age group).⁴⁶

2. The sample was limited to participants aged 65–74 years. The participants were young-older adults, and do not represent a broad age spectrum of older adults. This age selection was justified by our interest in assessing mobility in populations at an age when mobility is very dynamic, with expected high rates of recovery. In addition, we expected to minimize selective survival by restricting the study to a relatively young group.

3. Because of the sampling, focusing on urban dwellers of middle-sized cities, we acknowledge that our samples do not represent all community-dwelling older adults in Albania, Brazil, Canada and Colombia.

Can I get hold of the data? Where can I find out more?

All collected source data are maintained and stored at the IMIAS Office in the Universidade Federal do Rio Grande do Norte, Brazil. Contact information can be found at the study website: [http://www.imias.ufrn.br]. Researchers interested in collaborative work are invited to contact the IMIAS Publication Committee as indicated at the website.

Profile in a nutshell

• IMIAS is a population-based, longitudinal, prospective cohort, multicentre, multidisciplinary study conducted in five cities: Tirana (Albania), Natal (Brazil), Manizales (Colombia), Kingston (Ontario, Canada) and Saint-Hyacinthe (Quebec, Canada).

• It examines the prevalence, incidence of and recovery from mobility disability and associated risk factors in 2002 community-dwelling men and women aged 65-74 years at baseline. Participants were followed to establish the baseline prevalence and the 4-year incidence of and recovery from mobility disability, allowing for the study of both cross-sectional and longitudinal relationships between risk factors and mobility disability.

• The first data collection (baseline) took place in 2012. The first follow-up assessments were conducted in 2014 and the second in 2016. Attrition between the 2012 baseline data collection and the latest follow-up in 2016 was relatively small. Mortality was 2.3% in Kingston, 3.2% in Saint-Hyacinthe, 6.6% in Tirana, 7.4% in Manizales and 8.5% in Natal. At 4 years, of the 1891 subjects not known to have died, 1527 participated in the follow-up assessment (81%).

• A wide range of information was collected via questionnaires, physical examinations and biological specimens in the five cities.

• Specific proposals for collaboration are welcome, and interested researchers are invited to contact at the study website [http://www.imias.ufrn.br].

Funding

IMIAS was funded by the Canadian Institutes of Health Research (CIHR), Institute of Aging, Mobility in Aging Initiative ‘New Emerging Team: Gender differences in mobility: what we can learn about improving mobility in old age’ (Grant number 108751).