Assistive technology access in longitudinal datasets: a global review

Functional limitations become more prevalent as populations age, emphasising an increasingly urgent need for assistive technology (AT). Critical to meeting this need trajectory is understanding AT access in older ages. Yet few publications examine this from a longitudinal perspective. This review aims to identify and collate what data exist globally, seeking all population-based cohorts and repeated cross-sectional surveys through the Maelstrom Research Catalogue (searched May 10, 2022) and the Disability Data Report (published 2022), respectively. Datasets incorporating functional limitations modules and question(s) dedicated to AT, with a wave of data collection since 2009, were included. Of 81 cohorts and 202 surveys identified, 47 and 62 meet inclusion criteria, respectively. Over 40% of cohorts were drawn from high-income countries which have already experienced significant population ageing. Cohorts often exclude participants based on pre-existing support needs. For surveys, Africa is the most represented region (40%). Globally, 73% of waves were conducted since 2016. ’Use’ is the most collected AT access indicator (69% of cohorts and 85% of surveys). Glasses (78%) and hearing aids (77%) are the most represented AT. While gaps in data coverage and representation are significant, collating existing datasets highlights current opportunities for analyses and methods for improving data collection across the sector.


Introduction
The World Health Organization (WHO) estimates that the global average life expectancy has increased from 66.8 to 73.4 years, between 2000-2019 [1]. This increase in longevity comes with unique and urgent concerns, as populations (and the individuals who comprise them) experience ageing differently. The 2015 World Report on Ageing and Health stresses a need to understand drivers of variation in healthy ageing [2,3] or the development and maintenance of functional ability that enables wellbeing in older age [4]. An increase in the average life expectancy is often driven by better survival at all ages and does not always indicate improved health at older ages, emphasising the importance of distinguishing years lived in good health from all years lived [5,6]. Ageing populations tend to demonstrate an increase in the prevalence of functional limitations, both physical and cognitive, which impact physical and emotional health and societal participation [7][8][9]. There is an increasing need for assistance and support in these populations, which emphasises the growing role of assistive technology (AT) access in an ageing/aged world [10][11][12][13]. Yet limited data and low political prioritisation of AT have prevented a thorough understanding of the trajectory of this need. This review aims to collate all datasets currently or explicitly planned to be longitudinal and include dedicated questions on AT indicators to establish a global minimum dataset to facilitate research on AT access.
Assistive technology is defined by the WHO as "an umbrella term covering the systems and services related to the delivery of assistive products and services. Assistive products maintain or improve an individual's functioning and independence, thereby promoting their well-being," [14] and their global relevance was highlighted with the 2022 publication of the Global Report on AT [13]. However, which specific products and services are available and accessible through national provision infrastructure vary substantially between (and often within) countries [15][16][17][18]. The WHO promotes the development of national priority assistive product lists (APLs) which clearly define the most relevant AT for a specific country context, based on the unique needs of its population [19,20]. Measuring and monitoring access to AT in a population is critical to support healthy longevity.
Systematic [21] and scoping reviews [18,22] have identified that most research publications on AT access are singlewave cross-sectional studies, which lack follow-up and inhibit the exploration of trends in this sector. Only ten studies were identified in a large systematic review that had any longitudinal/follow-up consideration, comprising <5% of all 207 included studies [21]. However, a wealth of standardised longitudinal data do exist in the forms of repeated crosssectional administrative surveys and population cohort studies, where questions on AT are incorporated. Yet the vast majority of these data have not been utilised in longitudinal analyses.
Forecasting AT access with longitudinal data can describe inequities between and within countries, defined by key demographic, geographic, and health variables [23,24]. AT access serves as an aggregated health indicator and describes a population's capacity to initially and repeatedly access products and services needed to support different areas of individual functioning. A longitudinal perspective allows the identification of disparities in healthy ageing and bottlenecks in provision at multiple positions on the continuum of care. Despite this critical relevance to healthy ageing and potential to produce nuanced forecasts that are valuable for provision planning, these data have not been used to study AT access in this capacity. An essential first step to improve data use in this sector is to identify what data are available, for which countries, and what characterises their strengths and limitations for producing evidence needed to expand AT access. An additional output of this research will be a dedicated section of the AT2030 data portal [25], constituting a baseline repository of longitudinal datasets that include AT, which can be updated as new datasets emerge or are identified through methods outside this review.

Methods
Two types of population-based datasets were sought for this review: repeated cross-sectional surveys and cohorts. To support the comparability of datasets generated by these studies, only those approaching disability through functional assessments were used. As there is no systematic way of searching for all potential data sources, this paper maintains the remit of a scoping review.
For the purposes of this review, assistive products are focused on and defined using the WHO's list of 50 priority assistive products [20]. These data may include unspecified assistive products that are grouped, products defined inseparably with their essential services, or a single product alone, as long as it is clear that either of these three options includes one product found on the priority list. However, PDAs (e.g. smartphones) or services (e.g. physical therapy) are not included if represented in isolation.
We further assume AT access can be indicated in parts by needing, having, using, or being dis/satisfied with an assistive product. Self-reported need for an AP approximates a measure of demand that, when considered in line with the other outcomes, can indicate where access is limited (e.g. reporting need for an AP while also reporting not having it indicates unmet need for the AP [26]. These indicators of access are in line with those used by the WHO's rapid assistive technology assessment survey, launched in 2020, that maps populationlevel access to AT in terms of need, demand, supply, and satisfaction with AT [27].

Eligibility
For inclusion in this study, cohorts and surveys had to meet the following criteria: 1. Incorporate a functional limitation module, characterised as one or more questions asking a participant to estimate their level of difficulty in a certain functional domain or corresponding activity (e.g. seeing, hearing, mobility, self-care, communication, and remembering).

Cohort studies
The search was conducted on 10 th May, 2022, and yielded 81 cohorts, of which 47 meet inclusion criteria. Two are based on specific disease registers (see Table 1). To report results by country, cohorts are double-counted where population-based samples were taken in multiple countries (n = 58). Ten samples (10/58, 17%) are from the United States, with eight (14%) from Canada, five (9%) each from the United Kingdom, Australia, and Sweden, and four (7%) from Japan. No cohorts include populations in South America or the Middle East.
Of 47 cohorts, 44 either provide a min/max age for enrolment or state there is no min/max requirement. The minimum age is under 18 for five studies (5/44, 11%), and between 18-30 for nine studies (20%). Thirty-eight (84%) have a maximum age of 65 or older; 22 (49%) have no age maximum. Five are women-only and one is men-only. Five studies also mention a language requirement for enrolment. Of 15 cohorts with inclusion/exclusion criteria beyond age and location, 12 (12/15, 80%) exclude persons due to a pre-existing health limitation or diagnoses, and four (27%) exclude institutionalised persons, meaning those who are incarcerated or residing in long-term care facilities.
Thirty-three cohorts (69%) only collect data on 'Use' of APs, making it the most represented indicator. Four (8%) collect data on over 20 specific APs. Six (13%) ask which functions/activities of daily living an individual uses APs to support, rather than what types of APs are used, meaning individual APs are not disaggregable. Thirty-three surveys (69%) collect indicators specifically for glasses, 32 (67%) for hearing aids, and 21 (44%) for wheelchairs, and nine (19%) for prosthetics, meaning the rest of the collected data can be disaggregated for these individual APs, compared to three (6%) that only collect indicators for grouped APs. Fifteen (31%) only ask about glasses, hearing aids, and/or wheelchairs, but no other APs.

Cross-sectional surveys
Two hundred and eighty questionnaires, representing waves from 202 unique surveys, were reviewed. Eighty-five of these questionnaires (corresponding to 62 surveys) met inclusion criteria (see Table 2).
Twenty-five surveys (25/62, 40%) are set in the WHO African Region, most commonly South Africa (n = 4). Eleven (18%) are set in the Region of the Americas and the Western Pacific Region, with two each in Brazil, Mongolia, and the Philippines. Eight (13%) are set in the European Region, with no countries represented more than once. In the Southeast Asian Region, four surveys were identified with two from Bangladesh. Three are set in the Eastern Mediterranean Region, with two from Palestine.
Overall, 23 waves (23/85, 27%) were conducted from 2009-2015, and 62 (73%) from 2016-2021. All data collected in these surveys was self-reported, given by the participant or a proxy respondent in the same household, and include individuals at nearly all ages.
Fifty-three surveys (53/62, 85%) collect data on 'Use' of APs. Four surveys (6%) specifically ask participants what APs they think they need, all of which are surveys dedicated to disability and functioning, and three of which collect these data for over 20 individual APs. Fifty-three surveys (53/62, 85%) collect data that can be disaggregated for glasses, 52 (84%) for hearing aids, 15 (24%) for wheelchairs, and eight (13%) for prosthetics. Eight also collect data for grouped APs that cannot be separated out by type, though these groupings can be disaggregated by functional domain for two surveys. Continued.

Discussion
Collating these datasets and comparing study designs and world regions highlights critical data gaps, opportunities for analyses that are possible with existing data, and key opportunities to improve data collection across the sector. These steps are critical to generate evidence, demonstrate need, mobilise political will, and ultimately expand access to AT. A significant limitation is the representativeness of population cohorts, especially when seeking to understand AT access on a global scale. Most cohort data are from high-income, English-speaking countries that have already aged considerably (27/58, 47%), which restricts their international relevance and highlights the need for cohorts from diverse contexts when studying inequities in healthy ageing trajectories. Almost 75% (34/47) focused all or in part on participants over 65, which restricts investigations of how health at younger ages affects healthy longevity. Utilising these cohorts to represent disabilities in a population is also complex, given how often participants are specifically excluded for pre-existing conditions, use of assistance, or residence in long-term health facilities. Further, only three cohorts and no surveys collected data on the timing of use (when a person began using AT in their lifetime). These gaps prevent the exploration of disparities in AT access between people ageing with a disability and people acquiring disability as a result of ageing. Unlike most surveys which include functioning modules for all ages over five, the Multiple Indicator Cluster Surveys (MICS) focus on women aged 18-50, affecting their relevance to ageing studies. In both cohort studies and cross-sectional surveys, vision and hearing are proportionately overrepresented compared to other functional domains. Subsequently, less is known about APs that are useful for functional limitations with lower prevalence. This limited awareness can obfuscate the true demand for these APs in a population, which hinders efforts to expand access to them. Though national-level surveys are routinely conducted, waves of data collection are often separated by multi-year gaps. Further, 73% of surveys have also only incorporated AT modules in waves conducted in the last six years. Though more waves are explicitly planned, very few presently exist that can be used to establish trends or understand how AT access at younger ages in the past has affected the current population.
Despite these fundamental gaps, these datasets hold great potential to inform on AT access in a population. Current AT data collection methods allow the identification of clusters of need, and with capacity for longitudinal analyses, past trends can be studied, and forecasts can be developed to support policy planning and better understand the significance of AT access to healthy ageing trajectories. For example, where the single AT indicator 'Use' is collected, we can learn the number of people with a functional limitation who have had access to an AP across a range of ages, and whether they still experience difficulty even when using the AP communicates who is experiencing under-met need. This information is also useful to AT innovators, who need to identify and learn about the populations for whom certain APs are not reaching their potential. Unmet or under-met needs may be more common at certain age groups, geographies, functional domains, or associated with pre-existing health conditions. Adding data on satisfaction is another opportunity to differentiate between people who use the AP because it works for them, or who use it only because it is better than nothing. Not having AT despite experiencing functional difficulties indicates to policymakers that there is a bottleneck in provision, which could be due to market-level factors, like the cost or availability of the AP, or challenges for individuals who need it when accessing essential health care services. With a longitudinal approach, changing states of AT access can also be examined and factors influencing when an individual transitions from having an unmet to a met need for AT can be identified. Where these data can be disaggregated by key demographic and health variables, a more comprehensive and dynamic understanding of an ageing/aged country's national-level AT needs and opportunities can take shape.
As evidenced by this review, these analyses are possible in many cases but not for all populations that have been surveyed or included in cohorts, demonstrating where and how data collection can be improved. Many studies only collect data that can be disaggregated for glasses and/or hearing aids, limiting our capacity to study other APs and their relationship to functional difficulties outside of vision and hearing. Many countries of all income contexts are also missing data on all but one or two APs. Data in this sector could be improved by adding or expanding existing AT modules in existing multi-wave surveys or setting up population-based cohorts where data are particularly scarce or non-existent. Collection methods can be bolstered by incorporating questions on multiple specific APs when asking about functioning and activities of daily living. Indeed, cross-sectional data meeting our criteria will be expanded if the MICS program and broader demographic and health surveys (DHS) continue to include the functioning modules with dedicated questions on APs. Improving data collection and data sharing across this sector overall supports the development of global reports, such as those on Assistive Technology [13] and Ageing [31] which inform policy and operational planning for countries and regions as they prepare for the dynamic support needs of a growing aged population. To support these efforts, we will incorporate the findings of this review into the AT2030 data portal, which can be updated with new waves and sources to provide an up-to-date view of available data across this sector.

Limitations
This scoping review contributes new data sources to a global minimum dataset on AT access. Though both sources of AT data are comprehensive, some relevant cohorts and surveys may have been missed.
We also limited our inclusion criteria to studies utilising functioning modules. It is possible AT questions may also be included in surveys that assess disability in binary or alternative terms, or even surveys that do not consider disability at all. However, based on the infrequent inclusion of AT-specific questions even in dedicated functioning modules, we understand these cases to be exceptions.
Commercial and health-record data providing insight on AT access may also be available, though these sources were not included in the remit of this review due to difficulty searching and accessing them systematically.

Conclusion
Longitudinal data on AT access are a critical component for planning policy and provision that will support aged/ageing populations. By collating a global dataset indicating what kind of data are available and where, we have identified current data gaps, opportunities to learn from existing data, and recommendations for improving data collection going forward.

Contribution statement
Authors

Data sharing
All data collected for this study have been included and will also be made available at at2030.org by April 2023.

Conflict of interests
Authors have no conflicts of interest to declare.

Ethics statement
Ethical approval was not required for this study, as all data is publicly available.

Funding statement
This project was led by Jamie Danemayer in her role as a PhD student and researcher under the AT2030 programme. The AT2030 programme is funded by UK Aid from the UK Government and is led by the Global Disability Innovation Hub. However, no specific funding for this paper exists and the funder had no role in its production.