An environmental scan was conducted to identify Canadian data sources across all life stages that potentially include measures of intrinsic capacity, functional ability and the environment. An environmental scan is a methodological approach used to examine a range of practices, policies, programs, and data sources in order to describe the current landscape and inform decision-making [6]. Although no established guidelines for environmental scans exist [7], we consulted five phases of the RADAR-ES framework [6], and have adapted reporting items from PRISMA where applicable.

A search was first conducted on Maelstrom (https://www.maelstrom-research.org/), which is an international database used to facilitate collaborative epidemiological research through rigorous data documentation, harmonisation, integration, and co-analysis, globally, and catalogues study metadata for over 393 studies. A second search was conducted on the Atlas of Longitudinal Datasets (https://atlaslongitudinaldatasets.ac.uk/), which is a list of longitudinal datasets from across the world. Since both are curated catalogues rather than bibliographic databases, keyword search strategies could not be applied. For each, Canadian studies were identified using the country filter. The remaining eligibility criteria were applied at the screening stage. The catalogues were scanned in 2023 and rechecked in January 2025. As both are curated registries of longitudinal studies, hand-searching and citation checking were not undertaken.

Eligible studies needed to have the following criteria: 1) a general population sample, including gender specific cohorts; 2) were not a COVID-19-specific cohort or study; 3) had ongoing data collection; and 4) utilised a longitudinal, cohort or repeat cross-sectional design. Studies that did not have a clear data access policy for a broader research community were excluded. These eligibility criteria were deemed important if future data harmonisation were to take place.

Records were screened by a single reviewer. Information from eligible studies were extracted including study name, location, sample size, location in Canada, start year, study design, availability of data support documentation, availability of data, number of waves of data collection and if studies were linked to administrative data. Extraction was performed by a single reviewer using a standardised extraction form. Where linkage to administrative data could not be verified from study documentation or study websites, this was recorded as “Unsure” rather than inferred; these entries therefore reflect the absence of publicly available information.

Studies were mapped to life stages and domains of intrinsic capacity, functional ability, and the environment as proposed by the Framework of Healthy Ageing, using data support documentation. To explore comparability of data in included studies, measures of intrinsic capacity, functional ability and the environment were selected. Selected measures are only an exemplar, acting as an exploratory exercise and does not mean they are the only measure of each domain, or the only measure available within studies. For some domains, two measures were selected, if a single measure for all life stages was not suitable. Selection of relevant measures was made using the Baseline report on Healthy Ageing [2], and other relevant frameworks (e.g., Nurturing Care Framework [8], Integrated Management of Childhood Illness [9] and the Adolescent Well-being Framework [10]). Candidate measures were identified from these frameworks by the research team, and expert consultation on the basis of relevance to the sub-domain, applicability across the greatest number of life stages, and availability in at least one included study.

A total of 262 Canadian studies were identified from both Maelstrom and the Atlas of Longitudinal Datasets. After removing duplicates, 238 studies remained. After restricting studies to the general population sample and those that used a cohort or repeat cross-sectional design, 68 studies were eligible. Studies were excluded if they were a COVID-19-specific study (n = 17), did not have ongoing data collection (n = 17), did not have data support documentation or a data dictionary publicly available (n = 9), did not have possible access to data or access was restricted (n = 8), and did not collect individual-level data (n = 2). A final 14 studies were identified to be eligible for inclusion in a possible life course data harmonisation study. Figure 1 outlines the process of identification of relevant studies.

Among the 14 included studies, 10 used a cohort design, and the remaining four used a repeat cross-sectional design. One study was classified as a clinical trial on Maelstrom, however, the intervention was embedded within an observational cohort design. A total of 9,424,056 participants are included across the eligible studies, with three studies having no limit on participant number as they used a repeat cross-sectional design, thus do not have a fixed sample size. There were six studies that included participants across Canada, not limited to a single province or territory. A complete description of the studies included can be found in Table 1.

The studies included were mapped to life stages (Table 2). Life stages include Childhood (0–9 years); Adolescents (10–19 years); Young adults (20–29 years); Adults (30–44 years); Middle aged adults (45–59); Older adults (60–84 years); Oldest adults (85+ years). No study followed participants from birth across the life course, spanning all life stages. The Manitoba Multigenerational Cohort included participants of all life stages, linking participants across generations within a family. Three studies the Canadian Healthy Infant Longitudinal Development (CHILD) study, Canadian Health Survey on Children and Youth (CHSCY) and Healthy Life Trajectories Initiative (HeLTI) included data on children, however, participants have not yet been followed into subsequent life stages. The remaining studies recruited participants from other life stages. For example, the Canadian Community Health Survey (CCHS) included participants from all life stages except for childhood. Most of the other studies recruited participants at various adulthood life stages, with only six studies including adults in the oldest adult age range.

Almost all studies included at least one measure of intrinsic capacity, functional ability and the environment (Table 2). However, the Labour Force Survey only included measures of functional ability. Tables 3–5 map eligible studies to specific measures of intrinsic capacity, functional ability and the environment as a preliminary assessment of comparability of measures across studies which could inform potential harmonisation.

Intrinsic capacity. For each selected measure of vitality, psychological capacity, sensory and locomotion, at least one study had data for each life stage (Table 3). Executive functioning was selected as the sub-domain of cognition; however, it was difficult to select a measure that was relevant and applicable to all life stages. The selected measure (verbal fluency) was not measured in many studies, aside from the Canadian Longitudinal Study on Aging (CLSA). Beyond this measure and sub-domain, many studies did assess other aspects of cognition using different measures.

Functional ability. It was difficult to identify one single measure of each domain of functional ability that was relevant to each life stage. For be mobile, learn, grow, make decisions and meeting basic needs, two measures were selected. For the sub-domain, Contribution to society, the selected measure (Participating in volunteering) was available for almost all life stages except for Children (Table 4).

Environment. Selected measures for the environment at both community- and individual-levels were available in some data sources for most life stages, except for childhood where no studies evaluated the selected individual-level measure (Table 5).

It was beyond the scope of this paper to explore methodological consistencies across the identified studies. Although harmonisation may be possible, a more nuanced consideration of the methodological complexities involved in harmonisation must be made. The studies identified vary in terms of study design, and although we only included general population samples, the recruitment and sampling strategies across the studies vary. For instance, studies that employed non-probability sampling techniques may result in an underrepresentation of a certain group of the population, like those who may not be inclined to participate in studies (e.g., those who are sick). Whereas, other studies may have employed probability sampling techniques, like using random-digit dialling, meaning every individual within a target population has an equal chance of being recruited into the study. This creates selection bias, which can lead to biased estimates when harmonising data, particularly, if the exposure or outcome of interest is related to the likelihood of being sampled. This may result in an underestimation of the true effect. If there is substantial heterogeneity in study design and sampling methods across the identified studies, researchers may consider using meta-analysis techniques rather than pooling data directly or conducting stratified analyses by sampling strategy or study design. In addition, given that some identified studies used a longitudinal design, there is a risk of attrition bias, as participants who remain in the study may differ systematically from those lost to follow-up. If those who do not remain in the study share similar characteristics (e.g., age group, racial background, income level), they then are underrepresented in the study, leading to biased estimates when pooling data. It is also important for descriptive analyses of attrition over time to be explored, and where possible statistical approaches like multiple imputation or weighting could be used prior to harmonising data to address differential dropout or missingness across studies.

Measures must consider the whole person assessing both health and well-being. At the time of this review, we found that some domains of the Healthy Ageing framework may not exactly capture relevant aspects of health at other life stages. As proposed in the recent WHO Life Course framework [11], these domains were extended and modified which were informed from other frameworks including the Nurturing Care Framework, Integrated Management of Childhood Illness and the Adolescent Well-being Framework to capture developmental processes. For example, vitality, which measures metabolic and energy imbalances, was broadened to include thriving, which aligns with health among children and adolescents, and sexual and reproductive capacity was introduced for relevant life stages. These domains now better evaluate health at each life stage and across the life course. Building on these domains, further exploration is needed to determine available and relevant measures at different stages of life. Having this understanding can allow for trajectories of different capacities or abilities to be modelled, using measures that are appropriate at different life stages. For example, executive functioning is an important sub-domain of cognition across the life course. However, it was evident selecting one measure for all life stages was not suitable. Verbal fluency, the exemplar measure selected, was available only in the CLSA among adults aged 45 years and older. This does not indicate that cognition is unmeasured in the other included studies, but rather reflects how cognition is assessed. Verbal fluency, like most validated cognitive measures, requires in-person, trained administration, and among the included studies only those with a face-to-face assessment component were positioned to collect it. Studies relying on self-report or interview-administered questionnaires can capture self-reported cognitive difficulties or diagnoses, but not performance-based cognition. The same constraint operates in childhood, where cognitive assessment requires age-appropriate measures. Harmonisation of cognition across the life course will consequently depend not only on identifying measures of the same underlying construct, but on establishing whether that construct is consistent across life stages. Mapping measures across the life course must recognise that what is typically assessed within cognition shifts with age, with childhood measures focused on learning and development and later life measures on decline and impairment [12]. A life course perspective requires cognition to be captured at all life stages, so that whether these measures align, or reflect similar underlying domains, can be determined. A similar issue arises for the environment in childhood, where only one selected measure for neighbourhood crime was available in the CHILD study. It is possible this is a result as these measures are typically asked of adult respondents about their own circumstances, and their absence for children may reflect the exemplars chosen rather than an absence of environmental measurement. Other environmental measures relevant to childhood are likely to be available within the included studies, and identifying measures that are both relevant and available at each life stage is a necessary step before the extent of any gap can be established. Future work should therefore examine the full range of environmental measures held across studies, and consider how child-level environmental exposures could be derived from data already collected, for example by assigning household-level measures reported by adults to children in the same household, or by linking geocoded residential histories to census, air quality, and built environment data. Harmonisation should extend beyond choosing a single measure across the life course and instead focus on identifying key constructs relevant to each life stage, followed by selecting valid and reliable measures appropriate to those stages and transforming these measures so they are on the same scale.

To further understand harmonisation potential, it will be necessary to establish the appropriate infrastructure. The Health Data Research Network Canada has developed a strategic plan to support innovation in the use of data, focusing specifically on multi-regional data use [13]. Strategic goals include developing and improving services and supports for data access, diversifying data use, establishing partnerships, strengthening data use to improve equity and supporting Indigenous-led data sovereignty, and ensuring purposeful and ongoing public and community involvement. Building on this initiative, and modelling similar research collaborations, such as MINDMAP [14], which is an integrated database infrastructure for research in ageing, a platform must be developed to determine how datasets could be integrated, considering availability of data and associated costs. Previous work has been undertaken to combine data from different cohorts to examine life course trajectories. For instance, harmonisation of five prospective cohort studies from Belarus and the United Kingdom was completed to examine associations of parental education, sex, and race/ethnicity with children’s weight trajectories [15]. This example illustrates the possibility of combining cohorts to model trajectories; however, it is noted caution should be taken in interpreting findings given the heterogeneity across cohorts, and that a single cohort could drive a particular finding.

This review only focused on studies that included general population samples. Future work should consider and identify how data on special or vulnerable populations, including Indigenous and ethnically diverse samples, can be included to ensure a database/harmonised dataset is representative of the entire Canadian population. It is also important to consider principles that have been put in place to protect and respect data sovereignty. For instance, the Ownership, Control, Access and Possession (OCAP) principles have been created to respect Indigenous data sovereignty [16, 17]. If harmonisation were to take place, further exploration of key constructs and measures would be needed to determine appropriate and relevant measures for different populations. To do this, stakeholders and representatives from these groups must be involved in all aspects of the research process.

This environmental scan is the first to comprehensively review Canadian data sources that may be harmonised to create a pooled life course dataset. This is particularly important given the growing need to leverage pre-existing data to overcome the time, cost, and feasibility constraints of collecting longitudinal data across all life stages, while still capturing continuity across life stages to inform life course-oriented research and policy. Using a comprehensive approach that drew on Maelstrom and the Atlas of Longitudinal Datasets, we identified a wide range of Canadian data sources spanning multiple life stages, many of which can be linked to existing administrative data. However, it is possible studies may have been missed, or data on these platforms were not complete. For most studies, individual websites and documentation were reviewed to ensure the most accurate data were included in the current review. In a potential harmonisation of data, restrictions accessing data may still exist, including fees or delays in accessing data, which may create added challenges. Whenever harmonisation is used, there is the potential for a loss of information as data from different studies may have to be categorised to ensure comparability. Although this is a limitation of harmonisation, evidence generated could provide an indication of where continued or future research should be focused. Given screening and data extraction were conducted by a single reviewer rather than in duplicate, the possibility of error in study selection or extraction cannot be excluded. As an exploratory exercise, a single exemplar measure was selected for each sub-domain of intrinsic capacity, functional ability, and the environment. These measures were chosen to test comparability across studies and life stages, and were not intended to be exhaustive. Where a study is not listed against a selected measure, this indicates that the specific exemplar was not available, and not that the domain is unmeasured in that study. Many of the included studies collect other measures within the same domains, including alternative assessments of cognition, and a fuller mapping of all available measures within each domain was beyond the scope of this scan. Additionally, this review did not examine the demographic, socioeconomic, or lifestyle variables held within the included studies. Our mapping was restricted to the domains of intrinsic capacity, functional ability, and the environment, and characterising how variables such as race and ethnicity, immigration status, sex and gender, income, and education are defined and measured across studies was not an objective of this current scan. This is an important gap, as these variables are central to intersectional analysis, and a harmonised dataset that pools individuals across studies can only support equity-oriented research if they are captured comparably.