The proportion of the world's population over 60 years of age is estimated to increase to approximately 2 billion individuals by 2050, nearly doubling from 12% of the world population to 22% (World Health Organization, 2018a). With such a drastic shift in global demographics, the incidence of age-related chronic health conditions is also expected to increase. Indeed, the prevalence of audition and vision degradation increases with age and can have global impacts on cognition (Salthouse et al., 1996; Baltes and Lindenberger, 1997; Porto et al., 2016) and temporal perception (Gordon-Salant and Fitzgibbons, 1999; Grose and Mamo, 2010; de Boer-Schellekens and Vroomen, 2014; Brooks et al., 2018). Combining information across the senses can improve localization, discrimination, and speed of responses to objects, however, the central nervous system (CNS) must bind together the appropriate signals (Calvert et al., 2004). One cue that the CNS can use to determine whether or not stimuli should be bound together into a single percept (multisensory integration) is the temporal relation, how close in time two or more signals are to one another; research has revealed, using not only non-human animal models but also through studies conducted with humans, that signals that appear closer in time are more likely to be integrated [Vroomen and Keetels, 2010; see also King (2005) for a review of strategies used by the CNS to bind appropriate cues]. As temporal perception is affected by changes in unisensory processing, changes in auditory and visual acuities can act as indicators that may provide insight into changes associated with the multisensory integration processes within the aging population. Within the auditory domain, an estimated 466 million people worldwide have disabling hearing loss (World Health Organization, 2021a) and it is estimated that between 25 and 40% of older adults aged 65 and over, can be classified as having hearing impairment (Yueh et al., 2003). It has been found that the prevalence of hearing loss rises with age, ranging from 40 to 66% in adults over the age of 75 years and more than 80% in those older than 85 years of age (Cooper and Gates, 1991; Yueh et al., 2003; Walling and Dickson, 2012). Further it has been found that after the age of 60, hearing typically declines by about 1 dB annually and that men usually experience greater hearing loss and earlier onset compared to women (Lee et al., 2005). More than 90% of older individuals with hearing loss have age-related sensorineural hearing loss, which is a gradual symmetric loss of hearing—predominately of higher frequencies—that is worse in noisy environments (Yueh et al., 2003). Note however, that in an epidemiology study conducted by Lin et al. (2011) where data related to hearing abilities of older adults aged 70 and over was used from the 2005–2006 cycle of the National Health and Nutritional Examination Survey, it was found that the prevalence of hearing loss varied depending on the tonal frequencies, the audiometric thresholds used to define hearing loss, and whether hearing loss was considered in the better or worse hearing ear. They reported hearing loss prevalence rates from 16.5% when hearing loss was defined as using 0.5, 1, and 2 kHz (standard pure tone averages; PTA) with a 40 dB threshold in the better ear to 99.7% when hearing loss was defined as using 3, 4, 6, and 8 kHz (high-frequency PTAs) with a 15 dB threshold in the worse ear. Although they found that most reports of hearing loss prevalence used a 25 dB threshold, standard PTA (0.5, 1, and 2 kHz) or speech frequency PTA (0.5, 1, 2, and 4 kHz), and obtained measures in either the worse or better ear, there was still a high degree of variability of hearing loss reported. The range was narrower but spanned 44.8% when using the standard PTA in the better ear to 75.1% when using speech frequency PTA in the worse ear. Thus, the definition used when measuring hearing loss is crucial especially when some researchers may be utilizing a more rigid inclusion criteria as compared to others.

Shifting our focus toward the visual domain, worldwide, approximately 185 million people over the age of 50 years are visually impaired (World Health Organization, 2017), with cataracts, age-related macular disease, and refractive errors being the most common causes of visual impairment in older adults (Buch et al., 2001; World Health Organization, 2017). Although cost effective interventions such as cataract surgery and corrective glasses have shown to be effective, only 22 and 37% of individuals living in upper-middle and high-income countries respectively have reported having an eye exam during the preceding year (World Health Organization, 2017). The World Health Organization (WHO) defines visual impairment based on the International Classification of Diseases 11 (2018b) classification in the following categories for acuities measured at a distance of 2–4 m: mild visual impairment is defined as acuity worse than 6/12–6/18, moderate visual impairment is defined as acuity worse than 6/18–6/60, and severe visual impairment is defined as visual acuity that is worse than 6/60–3/60 in both eyes. In other words, the WHO defines visual impairment as best corrected visual acuity of <20/40 but ≥20/400, while many researchers (especially in the US) commonly define visual impairment as best corrected visual acuity that is worse than 20/40 but better than 20/200 in the better eye (Buch et al., 2001; World Health Organization, 2018; also see Supplementary Table 1 for details regarding the WHO's definition of visual impairment. This table contains Snellen and LogMAR values). In a study conducted by Buch et al. (2001) comparing the prevalence of visual impairment in 944 individuals as defined by the WHO and the criteria most commonly used in the US, it was found that 2.6% of those aged 70–74 years and 4.8% of those aged 75–80 years had visual impairments according to the WHO's definition (worse than 20/60–20/400 in the better eye; World Health Organization, 2004). However, these values differed based on the criteria used in most US studies where 3.1% of those aged 70–74 years and 8.0% of those between 75 and 80 years of age had visual impairment. Here, once again, we are reminded that the definition of impairment used by researchers is crucial and that some researchers may be excluding more participants than others.

Given the changes in sensory acuities associated with aging, accounting for such changes in hearing and vision is crucial as it may increase the quality and validity of the data obtained. The integration of auditory and visual cues into a unified percept is a fundamental process with an evolutionary benefit as it allows the observer to respond to external events more quickly and accurately relative to unisensory information alone (Stein and Stanford, 2008). Such an ability to integrate auditory and visual cues into a coherent percept has been thought to be beneficial for everyday function, for example in improving perception of speech in noise (Sumby and Pollack, 1954) and in improving driving performance (Ramkhalawansingh et al., 2016), both of which are especially relevant for the aging population. Research has time and again revealed that there are three principles that underlie multisensory processing, the first two principles suggest that the more temporally and spatially coincident (Meredith et al., 1987; Stein et al., 2014; Baum and Stevenson, 2017) two sensory cues are, the more likely they are to be bound together and result in a unified percept. The third principle states that unisensory signals that are weakly effective on their own are more likely to benefit from integration (Stein et al., 2009, 2014; Baum and Stevenson, 2017). This third principle of inverse effectiveness however does not hold true when the unisensory component that would be bound into a multisensory percept becomes unreliable and can result in a reduction in multisensory benefits as observed through models of optimal integration (Ross et al., 2007; Ma et al., 2009; Baum and Stevenson, 2017). Indeed, effective multisensory integration is dependent on both peripheral sensory organs as well as higher cognitive processes. As significant changes in sensory systems (e.g., decrease in visual acuity and an increase in auditory acuity thresholds) and cognitive function (e.g., decline in executive function and memory) are associated with healthy aging, it is not surprising that multisensory integration also changes with age (Rapp and Heindel, 1994; Kalina, 1997; Liu and Yan, 2007; Mozolic et al., 2012; Baum and Stevenson, 2017; Fjell et al., 2017). Indeed, older adults have been found to have longer response times in audiovisual detection tasks (†Laurienti et al., 2006; †Mahoney et al., 2011; †Couth et al., 2018; †Basharat et al., 2019), exhibit wider temporal binding windows (TBWs; the window of time within which information from different modalities is integrated and perceived as simultaneous; †Setti et al., 2011b; †Bedard and Barnett-Cowan, 2016; †Basharat et al., 2019), are more likely to be distracted by irrelevant stimuli within and across modalities [Poliakoff et al., 2006; see de Dieuleveult et al. (2017) for a detailed review regarding the effects of aging on multisensory integration], but they are also more likely to exhibit greater multisensory enhancement [see Mozolic et al. (2012) and de Dieuleveult et al. (2017) for detailed reviews] compared to younger adults. Further, it has been found that such changes in multisensory integration are exacerbated in those living with mild cognitive impairment and dementia. Research has revealed that those living with MCI and dementia tend to have longer response times, exhibit wider temporal binding windows, are more likely to experience attention impairment, and are less likely to benefit from multisensory enhancement compared to healthy controls (Wu et al., 2012; Chan et al., 2015; Murray et al., 2018). These results suggest that both cognitive function and sensory abilities must be accounted for when conducting multisensory integration related research with the aging population.

A decline in sensory abilities can affect the reliability, or the precision of a sensory estimate, with which the central nervous system integrates cues from auditory and visual modalities and can thus reduce the benefits typically gained through the multisensory process (Ernst and Bülthoff, 2004; Odegaard and Shams, 2016). Note however, that reduced acuity may also help to explain the increased benefits of multisensory integration in the aging population through the lens of the principle of inverse effectiveness (Mozolic et al., 2012). With a decline in auditory and visual acuity, the unisensory cues from these modalities would be presented just above threshold levels, thus the principal of inverse effectiveness would predict that integration of these weakly effective cues would produce gains much larger than the sum of their parts, suggesting that individuals with reduced sensitivity or acuity (i.e., older adults) may experience enhanced sensory integration. Thus, accounting for age-related sensory loss is essential in multisensory literature as it impacts the reliability of the incoming information and thus the likelihood of integration. It should however be noted that in a recent review conducted by de Dieuleveult et al. (2017) where the performance of older adults was compared to younger adults on unisensory and multisensory stimuli, it was found that although older adults did not always exhibit slower response times on the unisensory stimuli, they continued to show multisensory facilitation, indicating that inverse effectiveness may be one of many processes involved in the enhancement observed for multisensory cues in older populations (†Peiffer et al., 2007; Guerreiro et al., 2014, 2015).

Regardless of the underlying mechanisms, some research shows that changes in audition and vision impact temporal perception, not just within each modality, but also between these sensory modalities. Within the auditory domain, older age impairs temporal order judgments (Gordon-Salant and Fitzgibbons, 1999), duration discrimination (Fitzgibbons and Gordon-Salant, 1994, 1995; Gordon-Salant and Fitzgibbons, 1999), and reduces sensitivity to temporal fine structure (Grose and Mamo, 2010). Within the visual modality, age also impairs visual temporal judgments (de Boer-Schellekens and Vroomen, 2014), reduces flicker sensitivity (Mayer et al., 1988; Kim and Mayer, 1994), and reduces critical flicker frequency (Lachenmayr et al., 1994). When assessing age-related changes to audiovisual temporal perception, researchers find that older adults are more susceptible to the sound-induced flash illusion (†Setti et al., 2011a; McGovern et al., 2014), are more susceptible to the temporal ventriloquist effect (de Boer-Schellekens and Vroomen, 2014), and have wider temporal binding windows (†Bedard and Barnett-Cowan, 2016; †Basharat et al., 2018, 2019). Further, as aging increases the prevalence of ocular disease and hearing loss, this can lead to impairment in temporal perception (Phipps et al., 2004; Gin et al., 2011; Gallun et al., 2014). Although many, but not all, audiovisual multisensory paradigms include auditory- and visual-only conditions to gain insight into the workings of auditory and visual systems, we believe that accounting and screening for age-abnormal changes in the auditory and visual modalities will allow researchers to draw more reliable conclusions related to how audiovisual integration (the binding of auditory and visual cues into a unified percept) changes with age without being confounded by uncorrected vison and hearing. Our preliminary search revealed that researchers are not employing as much scientific rigor as would be necessary to account for auditory and visual acuity changes within the multisensory integration literature. While some researchers rely on self-reported measures obtained from participants, and others measure acuities in the lab or research centers to determine eligibility, some researchers however do not collect or account for visual and/or auditory acuities whatsoever. Further, a standardized criterion for what constitutes “normal” vision and hearing does not seem to be used and does not exist within the multisensory integration literature (Brooks et al., 2018).

Here, we aimed to determine the scope of the problem and collected information regarding what practices researchers are following in the literature to screen for vision and hearing impairment. We collected descriptive statistics regarding the number of researchers who screened for auditory and visual acuities (and how they reported it), those who used self-reported measures, and finally those who did not utilize any form of acuity measurements. We also aimed to determine what cut-off scores are being used when researchers do measure the acuities within a research or laboratory setting and what types of questions are asked to obtain self-reported perceptions of auditory and visual acuities. In addition to visual and auditory acuity measures, we also assessed how researchers define healthy aging (e.g., if cognitive impairment is accounted for and if so, how it is being measured). This scoping study will help provide a map of the methods researchers are utilizing and will help determine whether or not a meta-analysis can be conducted to further understand the scope of the issue with the current dataset.

The methods of the current study have been registered with the Open Science Framework (Basharat and Barnett-Cowan, 2020). The scoping review was conducted according to the framework proposed by Arksey and O'Malley (2005) and the suggestions that have been developed by Levac et al. (2010).

Our review demonstrates that only 63.8 and 51.4% of studies examining audiovisual integration in aging, measure auditory thresholds and visual acuities respectively and that less than half of the studies (45.8%) that measure acuities screen both sensory modalities for age-abnormal changes. Further, a key finding is that a consistent definition of what constitutes normal hearing and vision is not employed within studies that screen for audiometric thresholds and visual acuities. Additionally, we found that although 41 and 39 studies use self-reported measures to screen for normal hearing and vision respectively, only nine studies reported the questions that were presented to participants for auditory screening, while only six studies reported the questions used to screen for self-reported visual impairment (see Tables 4, 5 below and Supplementary Table 6 for further information). In addition, as one may expect, a variety of tasks and behavioral outcomes of interest (e.g., discrimination or detection, mean response time, susceptibility to the sound induced flash illusion, etc.; see Supplementary Tables 3, 4 and Figure 2 above for details) were used in the studies selected for this scoping review; thus, the variability present in the data, from the screening measures to the multiple different tasks used makes it difficult to recommend a meta-analysis at the moment. It should however be noted that of the 2,462 articles, the 72 that were selected based on the inclusion and exclusion criteria specified above in the methods section and in the protocol for this review (https://osf.io/v3snz/; Basharat and Barnett-Cowan, 2021) were more focused on the aging process rather than on specific behavioral outcomes related to any specific task. Thus, future researchers whose research questions can be addressed using a meta-analysis can use either a rigid criterion (e.g., include studies that tested discrimination or detection response time only) to look for studies that use the same tasks or a more lenient criteria (e.g., compare the impact of aging on additional sensory modalities including somatosensation in a given task such as for detection or discrimination tasks) to capture a larger set of studies.

Our results indicate that more studies measured auditory thresholds compared to visual acuity (46 vs. 37 respectively; see Table 4 for further information). This is not surprising given that the prevalence of hearing loss rises with age ranging from 46 to 60% to more than 80% in adults aged 75 and 85 years respectively (Cooper and Gates, 1991; Yueh et al., 2003; Walling and Dickson, 2012) and is much higher than the prevalence for visual disorders that range from 2.6 to 8.0% in adults aged 70–74 and 75–80, respectively (Buch et al., 2001). However, we recommend testing both sensory modalities to ensure that stimuli presented to all participants are perceived at the appropriate thresholds (e.g., suprathreshold) required for accurate results. Note however that depending on the study design and the types of stimuli used, additional control conditions may be required. Further, and as alluded to above, the studies of audiovisual integration included in this review have adopted inconsistent screening definitions, especially for the auditory acuity, making it difficult to compare results between studies. A potential solution for this lack of standardization, is using the definitions of hearing loss and visual impairment that are recommended by the WHO; hearing loss is defined as “a speech-frequency pure-tone average at 0.5, 1, 2, and 4 kHz frequencies of >20 dB HL in both ears” and visual impairment is defined as “ <20/40 (no visual impairment) but greater than or equal to 20/400 (severe visual impairment) in both eyes” (Mathers et al., 2000; World Health Organization, 2018b, 2021b). We also found that more females than males were tested, both in the younger (62.8%) and older (54.2%) adult populations. Surprisingly, 29.2 and 34.7% of the studies failed to report gender for older and younger adults respectively, which may impact the ratio of men to women currently seen in this review. Note here that five studies were included that only tested older adults and if those studies were removed, we would be left with a comparable sample to the younger population of 461 older males and 657 older females (compared to 406 younger males and 688 younger females). However, we decided to keep these articles in the scoping review as they provide useful information regarding the screening procedures for inclusion of older adults, used in the literature.

Although a large number of studies (76.39%) used cognitive reporting to ensure that the participants included were cognitively intact (MMSE, MoCA, and DemTect, a dementia screening test, self-reported lack of cognitive impairment), many different scores were used to include or exclude individuals (see Table 3 and Supplementary Table 5 for further information regarding the inclusion criteria used for cognitive impairment and for various other inclusion criteria used by the studies included in this review). Although the variability in scores was somewhat expected as various cut off scores have been used for the detection of mild cognitive impairment (MCI) for both the MoCA (26, 25, 24, and 23; Nasreddine et al., 2005; Luis et al., 2009; Davis et al., 2015; Ciesielska et al., 2016; Milani et al., 2018) and the MMSE (28, 27, 26, 24; Anderson et al., 2007; Markwick et al., 2012; Creavin et al., 2016; Kvitting et al., 2019; Erdodi et al., 2020), we were surprised by the preferred use of the MMSE over the MoCA. As the MMSE was designed to screen for dementia at a time where the concept of MCI did not exist, the MoCA has been found to be a more sensitive test for detection and screening for early cognitive impairment compared to the MMSE (Markwick et al., 2012; Ciesielska et al., 2016). Further, research reveals that performance on the MMSE is affected by race, education, language, and gender (Tombaugh and McIntyre, 1992; Grigoletto et al., 1999; Wood et al., 2006), while the MoCA was designed as an alternative method of cognitive screening and is thought to account for the limitations that affect the MMSE [Nasreddine et al., 2005; Ciesielska et al., 2016; however please see a review by Siqueira et al. (2019) which indicates that both the MoCA and the MMSE are impacted by educational level]. Moving forward, we recommend that researchers use the MoCA to detect MCI as it was specifically designed to screen for mild cognitive impairment and it accounts for educational level differences through the addition of a point to the final score for those with <12 years of formal schooling (Nasreddine et al., 2005; Siqueira et al., 2019).

Further, we found that the 72 studies included in this scoping review used different tasks with various methodology, aims, and varying behavioral outcomes of interest (refer to Supplementary Tables 3, 4, 6, 8, 9 as well as Figure 2 above for details). Overall, the most common behavioral outcomes were thus “accuracy or proportion correct or percent correct” and “mean and median response time” measures. Given that the articles included used 28 different outcomes of interest to assess multisensory integration, it is difficult to suggest conducting a meta-analysis with the specific articles that we have used in this scoping review. However, we strongly believe that there is a sufficient amount of data available in the 1,500+ articles that were screened for this scoping review and thus suggest utilizing either a more rigid inclusion criteria (e.g., utilizing only speech recognition or response time tasks) or a broader inclusion criteria (e.g., including studies that do not mention the aging process) for those interested in conducting a meta-analysis.

This scoping review is not without its limitations. An inherent limitation of any given scoping review is that it provides breadth rather than depth on a topic (Arksey and O'Malley, 2005; Levac et al., 2010). While this scoping review provides a broad view of how studies are screening for age-abnormal sensory changes through the use of auditory and visual acuities, we are unable to determine the effectiveness of accounting for unisensory changes in multisensory integration research within this scoping review. As such, future research using meta-analyses is necessary to determine whether the results obtained from studies that screen for auditory and visual acuities differ from those that only use self-reported measures. We do however believe that providing a breath of knowledge will prove to be useful for researchers in understanding and further investigating multisensory integration within the aging population. Another limitation is that the majority of the literature in this review stems from developed countries (Economic Analysis Policy Division, 2020) and therefore it is not clear whether these findings extend to developing countries. However, it is also not clear whether the recommendations to correct the limitations associated with accounting for sensory acuities would not be applicable to the research conducted in developing nations, thus, we would extend our recommendations to developing nations unless future research indicates otherwise. Additional research with the inclusion of studies from developing nations is necessary to elucidate this matter. An additional limitation of the current study is that only studies published in English were included, limiting the review to articles that were either published in English-speaking countries, which may explain the predominance of the literature stemming from developed countries, or to those that had the funds for translation services. Finally, we conducted this scoping review using behavioral studies as we were concerned that behavioral studies may be conducted with less rigor as compared to neuroimaging studies, however, further research investigating the use of auditory and visual acuity screening methods with neuroimaging studies will not only provide insight, but is necessary to ensure standardized methods are used throughout the literature.

In conclusion, we found that only approximately 64 and 51% of studies measure for age-abnormal hearing and vision respectively and that within these studies a consistent definition of what constitutes normal hearing and vision is not found. Further, we found that many studies screen for one sensory modality (audition) more than the other modality. Here, we recommend screening for both age-abnormal hearing and vision and using the World Health Organization's definitions of hearing loss and visual impairment. Further, we find that many researchers use the MMSE for MCI screening instead of the MoCA and we recommend the utilization of the latter cognitive assessment as it has been found to be more sensitive toward the detection of MCI. We found that many different tasks were used to assess audiovisual integration in younger and older adults ranging from speech recognition to the stream bounce task, thus various behavioral outcomes were obtained ranging from accuracy to stream bounce susceptibility, making it difficult to suggest conducting a meta-analysis with this particular dataset. We do however believe that a meta-analysis can be conducted with the abundant data that exists within audiovisual literature; if you wish to conduct a meta-analysis, we recommend using either a more strict or a less strict inclusion criteria depending on your research question of interest.

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author/s.

AB drafted the manuscript with input from AT and MB-C. AB and AT read the abstracts and titles of all 1,546 articles and 105 full-text articles to screen the studies for inclusion. Any disagreements were discussed between the two reviewers until a consensus was reached or by arbitration of a third reviewer MB-C. All authors contributed to study conception, development, manuscript revision, read, and approved the submitted version.

This research was supported by an Ontario Research Fund grant and a Natural Sciences and Engineering Research Council of Canada Discovery Grant RGPIN-03977-2020 to MB-C. AB was supported by the Ontario Graduate Scholarship. University of Waterloo Interdisciplinary Trailblazer Fund to MB-C.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.