Harmonising assistive technology access data for evidence-based decision-making: guidance and implications

Executive summary

Population ageing demands that governments plan proactively for a growing number of older people and an increasing demand for products and services that enable their independence and participation. Assistive technology (AT) plays a central role in supporting functioning across the life course, and demand for AT rises substantially with age. Preparing for this demographic shift ideally requires forward-looking policies informed by all available evidence; understanding of changes in AT demand and access over time; and improved estimates regarding under-represented groups.

While new data collection is still critical for policy development, the effective planning that needs to happen now will necessarily have to rely on (and therefore make better use of) existing evidence and data, especially in lower resourced settings. Population-level AT data are typically characterised by several common limitations: AT is not consistently included in large surveys, and where AT access is measured, coverage varies across products, definitions, and outcomes. Further, groups at risk of exclusion are often underrepresented even in large datasets, and factors like stigma and the accessibility of survey enumeration can disproportionately affect their participation in the survey, leading to small sample sizes after disaggregation and reduced statistical confidence in survey estimates. Despite these limitations, historical data can still be used with consideration and careful interpretation to inform policy development and future research.

Harmonisation is a data cleaning process that offers a potential way to make use of fragmented datasets and derive policy insight. This statistical method aligns similar but non-identical variables across surveys conducted within the same population, allowing previously incomparable datasets to be analysed together (i.e. producing ‘joint analyses’). When surveys occur at similar time points, harmonised data can be pooled to produce more robust estimates, improving representation of groups often under-sampled in large surveys, such as very old adults and people with disabilities. Or, when surveys span multiple years, harmonised datasets can reveal trends in AT outcomes over time.

Harmonisation of AT data can therefore enable policymakers to better understand disparities in AT access across population groups defined by age, gender, and other key characteristics collected in each dataset, as well as differences across types of assistive products or functional needs. Trend analyses can also help assess how population, policy, or system changes relate to patterns of access, supporting more targeted and equitable planning for service expansion. Harmonised evidence can also inform improvements in future data collection by highlighting the value of consistent AT measurement across surveys and functional domains.

Harmonised data therefore provide an important opportunity to strengthen evidence for future AT provision, with implications for research, innovation, data collection, and policy decision-making. Though this method has been used in wider public health applications to combine datasets and monitor temporal trends, harmonisation of AT data has rarely been undertaken due to the historical scarcity of comparable surveys and variation in how AT outcomes are defined and measured. This article proposes a practical framework and method to address these challenges and demonstrate how existing data can be leveraged to support forward-thinking policy development.

The statistical method detailed in this article is a harmonisation logic that aims to enable the direct comparison of nonidentical AT assessment questions, which are each commonly used in global health surveys, namely, a ‘direct’ question set (e.g., “Without using a hearing aid, do you have difficulty hearing?”) and an ‘aggregate’ question set (e.g., “Do you have difficulty hearing, even if using hearing aids?”). The logic proposed in the present paper can be operationalised to calculate and directly compare descriptive statistics across multiple surveys and examine trends over time, creating a longitudinal dataset of AT access indicators and demographic variables. Augmenting datasets in such ways can enable governments in lower resourced, data-poor settings to make better use of the data that are available; with these improvements, these datasets can forecast AT demand, delineate access disparities between demographically-defined groups, and inform equitable resource allocation at scale. Validation of this logic in diverse contexts will be needed to fully explore its efficacy. 

This article is structured as follows: first, the authors introduce the importance of harmonisation in the AT data space and the overall aims of this proposal. The Methods describe how commonly used, nonidentical AT assessment modules may be harmonised in practice, using the proposed logic. Implications of the broad applications of this logic are described, in terms of research and innovation, new data collection, and policy decision-making. Finally, a reflection on the limitations of this logic and the potential for harmonisation to improve evidence-based decision-making in the AT sector conclude this.