Introduction
Disability means having difficulty or being unable to perform a range of activities (Rantakokko, Portegijs, Viljanen, Iwarsson, & Rantanen, Reference Rantakokko, Portegijs, Viljanen, Iwarsson and Rantanen2017; Verbrugge & Jette, Reference Verbrugge and Jette1994). It is not an inevitable aspect of aging (Jehn & Zajacova, Reference Jehn and Zajacova2019); however, the rate of disability increases as people age (Government of Canada, 2011; Okoro, Hollis, Cyrus, & Griffin-Blake, Reference Okoro, Hollis, Cyrus and Griffin-Blake2018). Similar trends are found in mobility disability, the most common type of disability experienced by older adults (Government of Canada, 2011; Okoro et al., Reference Okoro, Hollis, Cyrus and Griffin-Blake2018). According to the 2011 Federal Disability Report (FDR), one in four Canadians 65–74 years of age experienced mobility disability. This increases to two in five Canadians 75–84 years of age and three in five for those 85 years of age and older (Government of Canada, 2011). Furthermore, middle-aged Canadians (40–64 years) have been experiencing flat or increasing disability trends across different demographics, such as level of education and province of residence (Jehn & Zajacova, Reference Jehn and Zajacova2019), which is particularly problematic, as these individuals represent our future older population.
Mobility refers to the ability to move oneself (independently or by using assistive devices or transportation) from one place to another (Webber, Porter, & Menec, Reference Webber, Porter and Menec2010; World Health Organization, 2001) and to perform different activities in their environments (Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020). Mobility is considered a prerequisite for maintaining a good quality of life and is an important health outcome indicator in older adults (Abbott, Reference Abbott2009; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020). Mobility disability has been associated with increased mortality risk and health care utilization, as well as decreased functional independence and quality of life (Freiberger, Sieber, & Kob, Reference Freiberger, Sieber and Kob2020; Hardy, Kang, Studenski, & Degenholtz, Reference Hardy, Kang, Studenski and Degenholtz2011; Ni et al., 2017; Pruitt et al., Reference Pruitt, Glynn, King, Guralnik, Aiken and Miller2008). Mobility loss with aging is commonly described as a downward trajectory with a steeper decline shown late in life (Ferrucci et al., Reference Ferrucci, Cooper, Shardell, Simonsick, Schrack and Kuh2016). However, age-related impairments in mobility-associated physiological systems can be compensated for by changing the manner or the frequency of performing a task requiring mobility (Ferrucci et al., Reference Ferrucci, Cooper, Shardell, Simonsick, Schrack and Kuh2016). Overt mobility disability only manifests when the severity of the mobility deficit becomes too severe to be compensated for (Ferrucci et al., Reference Ferrucci, Cooper, Shardell, Simonsick, Schrack and Kuh2016). The intermediary phase, occurring with progressive loss of function and preceding the onset of disability, is referred to as “pre-clinical disability” (Fried, Herdman, Kuhn, Rubin, & Turano, Reference Fried, Herdman, Kuhn, Rubin and Turano1991), “pre-clinical mobility disability” (Fried et al., Reference Fried, Bandeen-Roche, Williamson, Prasada-Rao, Chee and Tepper1996), or “pre-clinical mobility limitation”(Mänty et al., Reference Mänty, Heinonen, Leinonen, Törmäkangas, Sakari-Rantala and Hirvensalo2007). According to a recent study (Richardson, Beauchamp, Bean, et al., Reference Richardson, Beauchamp, Bean, Brach, Chaves and Guralnik2023), a consensus among a group of international experts in pre-clinical disability highlights that the concept of pre-clinical disability should encompass task modification, particularly routine mobility tasks, so that these tasks can be performed without visible disability. As a result, the study introduced Pre-clinical Mobility Limitation (PCML) as the preferred term, which was utilized for this review.
PCML was first conceptualized by Fried et al. (Reference Fried, Herdman, Kuhn, Rubin and Turano1991). They defined it as a functional stage at which people can compensate for functional decline by modifying their task performance and thereby maintaining their function without reporting outright difficulty (Fried et al., Reference Fried, Herdman, Kuhn, Rubin and Turano1991). Evidence demonstrates that people reporting PCML are at an elevated risk of experiencing future overt disability in physical performance (Fried et al., Reference Fried, Bandeen-Roche, Williamson, Prasada-Rao, Chee and Tepper1996; Reference Fried, Young, Rubin and Bandeen-Roche2001; Mänty et al., Reference Mänty, Heinonen, Leinonen, Törmäkangas, Sakari-Rantala and Hirvensalo2007). Additionally, it is suggested that PCML is associated with the onset of multiple adverse health outcomes such as falls, depressed mood, and obesity in older adults (Clough-Gorr et al., Reference Clough-Gorr, Erpen, Gillmann, Von Renteln-Kruse, Iliffe and Beck2008; Hirvensalo et al., Reference Hirvensalo, Sakari-Rantala, Kallinen, Leinonen, Lintunen and Rantanen2007; Mänty et al., Reference Mänty, Heinonen, Viljanen, Pajala, Koskenvuo and Kaprio2010; Naugle, Higgins, & Manini, Reference Naugle, Higgins and Manini2012; Wolinsky et al., Reference Wolinsky, Miller, Andresen, Malmstrom, Miller and Miller2007). PCML is proposed as “an early warning system” for altered risk factors (Fried, Bandeen-Roche, Chaves, & Johnson, Reference Fried, Bandeen-Roche, Chaves and Johnson2000; Wolinsky et al., Reference Wolinsky, Miller, Andresen, Malmstrom, Miller and Miller2007). Individuals in this transition stage may represent an optimal group to receive interventions on prevention rather than recovery when one or multiple domains of disability is established (Clough-Gorr et al., Reference Clough-Gorr, Erpen, Gillmann, Von Renteln-Kruse, Iliffe and Beck2008; Gregory et al., Reference Gregory, Szanton, Xue, Tian, Thorpe and Fried2011; Higgins et al., Reference Higgins, Janelle, Naugle, Knaggs, Hoover and Marsiske2012). In the primary prevention of disability among older adults, intervening in the PCML period may be a uniquely effective strategy to reduce the burden of disability in this population (Weiss, Hoenig, & Fried, Reference Weiss, Hoenig and Fried2007). For older adults experiencing overt disabilities in performing some tasks, identifying other tasks for which pre-clinical modification has been reported has also been found to be an effective prevention strategy (Young, Boyd, Guralnik, & Fried, Reference Young, Boyd, Guralnik and Fried2010).
Although empirical findings provide sufficient evidence to move toward interventional studies targeting PCML, studies in this field have provided inconclusive results. For example, in a 12-week single-blinded randomized controlled trial (RCT) involving older adults with pre-clinical gait dysfunction, it was shown that a motor learning program, when compared with a standard treadmill walking program, resulted in improvements in mobility performance parameters, such as gait speed and motor skill (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013). Furthermore, in a 12-month quasi-experimental trial, it was discovered that older adults with PCML participating in a physical therapy program augmented with mobile tele-health technology exhibited significant differences in gait speed compared to controls (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019). However, other RCT studies reported little changes in impairments or functional limitations for older adults with PCML engaged in a 24-week Tai Chi or seated flexibility exercise program (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012). Many factors may contribute to these conflicting findings, such as tools used to identify PCML stage, participants’ eligibility criteria, and interventions selected. A preliminary search also showed that the outcomes measuring changes in PCML vary among studies (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008). Some used self-reported measures, such as the pre-clinical disability screening tool by Fried et al. (Reference Fried, Young, Rubin and Bandeen-Roche2001) or the Pre-clinical Mobility Disability Scale by Mänty et al. (Reference Mänty, Heinonen, Leinonen, Törmäkangas, Sakari-Rantala and Hirvensalo2007), whereas others argued that performance-based measures can identify limitations in physical function better than self-report measures (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013). In addition, the heterogeneity of the recruited samples’ characteristics (e.g., age, gender, cognitive level) and the variation in the intervention used (e.g., type, frequency, length, intensity, and delivery professions) may also contribute to the variation in results.
Therefore, it would be productive to understand the types of interventions used to address issues of PCML, the measures employed to assess PCML changes, and the characteristics of the participants with PCML recruited for interventional studies. To our knowledge, there are no reviews that have synthesized PCML interventions. This scoping review aims to provide a comprehensive understanding of PCML intervention studies in middle-aged and older adults that have been tested or planned, describe how they have been conducted and reported, identify the knowledge gaps in current literature, and make recommendations about future research directions regarding interventions for PCML. Specifically, the review questions are:
-
• What types of rehabilitation interventions are used to address PCML in middle-aged and older adults?
-
• What measures are used to assess PCML changes in intervention studies? What measures are used to assess other outcomes in these intervention studies?
-
• What are the characteristics of the baseline samples included in the PCML intervention studies (e.g., participants’ baseline characteristics, eligibility criteria, PCML stage assessment)?
Methods
This review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews (PRISMA-ScR) checklist (Tricco et al., Reference Tricco, Lillie, Zarin, O’Brien, Colquhoun and Levac2018) and the JBI approach developed by JBI and the JBI Collaboration (JBIC) group (Peters et al., Reference Peters, Godfrey, McInerney, Munn, Tricco and Khalil2020). The JBI approach was underpinned by the methodological frameworks initially proposed by Arksey and O’Malley (Reference Arksey and O’Malley2005) and further enhanced by Levac, Colquhoun, and O’Brien (Reference Levac, Colquhoun and O’Brien2010). This scoping review was registered with Open Science Framework (https://osf.io/b6zr4). A detailed review protocol was posted on medRxiv (https://medrxiv.org/cgi/content/short/2022.10.22.22280644v1). The PRISMA-ScR Checklist was also included (Appendix III).
Eligibility Criteria
For inclusion in this review, articles had to meet the following criteria: reporting rehabilitation interventional studies on middle-aged individuals (45–64 years) or older adults (≥ 65 years of age) with PCML outcomes, or reporting functional change outcomes consistent with the PCML stage. Measurements used to determine participants’ PCML stage and assess PCML changes could include either self-reported measures or physical performance measures, as long as the term PCML or its synonym (e.g., pre-clinical disability, subclinical disability, perceived disability) was indicated explicitly in the studies. Additionally, any non-surgical or non-pharmacological intervention could be considered a rehabilitation intervention, following the definition provided by McGlinchey et al. (Reference McGlinchey, James, McKevitt, Douiri, McLachlan and Sackley2018). Studies using a single-component rehabilitation intervention or as a part of a multifaceted intervention were included, regardless of frequency, intensity, length, and who delivered them. Because of limited evidence in this field, various study designs were considered, including RCTs, non-randomized controlled trials (NRS), controlled clinical trials, pre-post studies, interrupted time series studies, uncontrolled longitudinal studies, case studies, registered trials, and protocols. Abstract-only publications were not considered for inclusion. Studies published in languages other than English were excluded because of limited resources for translation.
Data Sources and Search Strategies
Seven electronic databases (MEDLINE®, Embase, Allied and Complementary Medicine Database [AMED], PsycInfo, Cumulative Index to Nursing and Allied Health Literature [CINAHL], Web of Science, and Cochrane CENTRAL) were used to locate relevant intervention studies (English evidence from inception to March 19, 2022). The search strategy was generated using the Population, Concept, Context (PCC) framework and refined after consulting with a McMaster research librarian. The text words included in the titles and abstracts of relevant articles ([preclinical* OR pre-clinical or subclinical* OR sub-clinical or incipient* OR perceived OR early] adj3 [mobility OR function* OR physical] adj3 [disab* OR declin* OR limit* OR deficit* OR dysfunction* OR impair* OR difficult* OR deteriorat* or modif*]) and the index terms used to describe the articles (“middle aged” or “aged” or “aged, 80 and over”) were used to develop a complete search strategy for Ovid MEDLINE (see Appendix I). The search strategy was tailored to the specific requirements of each included database, including all identified keywords and index terms. Google Scholar was utilized to identify any other primary sources within grey literature. In addition, a hand search of the reference list of retrieved articles and review articles was conducted to identify any further studies not yet captured. A follow-up search was implemented over 3 months, ending on July 2, 2022, to identify additional articles after the initial search on March 19, 2022.
Study Selection
Titles and abstracts of all identified citations were imported into Covidence (Veritas Health Innovation, Melbourne, Australia), and duplicated citations were removed. A calibration exercise was undertaken by two reviewers (A.L. and A.M.) independently screening 20 citations to evaluate reviewer agreement and identify any revisions required in the screening process. Following the calibration period, the two reviewers (A.L. and A.M.) independently conducted the title and abstract screening and full-text screening. Attempts were made to contact the source author for articles with missing or unclear data. Any disagreements on study selection were resolved by discussions or consensus involving a third reviewer (J.R.). Reviewers were not blinded to author or journal information.
Data Extraction
A data extraction form was developed to chart characteristics of the selected studies such as study design, sample size, participants’ characteristics (age, gender, setting) by group (intervention vs. comparator), and specific details about the interventions (e.g., content length, frequency, single or multifaceted, outcomes reported) (see Appendix II). The form was reviewed, pilot-tested, and modified by the reviewers (A. L., A.M., J.R.) before the full-text review process. Two reviewers (A. L. and A.M.) conducted the data extraction independently and discussed any conflicts that arose.
Data Synthesis
The extracted data were presented in tabular form that aligns with the research questions of this scoping review. The type, content, length, intensity, and frequency of interventions were presented, as well as whether they were delivered as a single method or part of a multifaceted approach. Participants’ characteristics were presented by listing age, gender (% of female), per cent of participants identified as being at the stage of PCML, and the measurements used. Self-reported or physical performance measured PCML outcomes and other outcomes reported in the studies were also presented. A narrative summary accompanied the tabulated results and described how the results relate to the review questions.
Results
Selection of PCML Interventional Studies
Using the key search descriptors, 2,291 potentially relevant studies were yielded after 2,061 duplicates were removed (Figure 1). After abstract screening and full-text reviewing, a total of 15 studies met the eligibility criteria and 14 of them were selected for this review (see Appendix IV), with one being excluded because of missing critical data. A large number of studies (n = 2,245) were excluded upon screening at the title and abstract stage because the key terms used in the search strategy also corresponded to other study designs (e.g. cross-sectional study) or the pre-clinical phase of other diseases (e.g., Alzheimer’s disease, hypothyroidism). Thirty-two studies not meeting the eligibility criteria were excluded during the data extraction stage. The reasons for exclusion were: “inappropriate patient population (n = 17)”, “inappropriate study design (n = 5)”, “abstract only (n = 4)”, “duplicate (n = 4)”, “non-English publication (n = 1)”, and “missing data (n = 1)” (Figure 1).
Figure 1. PRISMA flow diagram of study selection.
Characteristics of Included Studies
All included studies were published from 2008 onwards, over the 15 years since PCML was conceptualized by Fried et al. (Reference Fried, Herdman, Kuhn, Rubin and Turano1991). Two protocols (Ni et al., 2017; Rantanen et al., Reference Rantanen, Pynnönen, Saajanaho, Siltanen, Karavirta, Kokko, Karvonen, Kauppinen, Rantalainen, Rantakokko, Portegijs and Hassandra2019) and their respective published articles (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020) were grouped, resulting in a total of 12 studies being reported. Of those, 11 were RCTs, and one was an NRS (quasi-experimental). Five were conducted in the United States (42%), three in Canada (25%), two in Finland (17%), and two in Australia (17%). Most studies employed exercise as an intervention type, making up 58.3% (7/12) of trials. Other than one study with no clear statement (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013), studies indicated that they recruited community-based samples.
Results of Individual Sources of Evidence
The findings were organized based on the scoping review questions: (1) sample characteristics, (2) intervention type, (3) PCML measures, (4) measures to assess other outcomes, and (5) changes in PCML outcomes.
Sample characteristics
Of the 12 studies, most (n = 8) indicated a specific PCML stage assessment as one of the inclusion criteria, including subclinical gait dysfunction and impaired motor skill in walking (n = 1) (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013); Fried Self Report Task Modification and Disability Scale (Fried scale) (n = 4) (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Bennett & Hackney, Reference Bennett and Hackney2018; Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012, Reference Day, Hill, Stathakis, Flicker, Segal, Cicuttini and Jolley2015); Manty Pre-clinical Mobility Disability Scale (Manty scale) (n = 2) (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Richardson et al., Reference Richardson2020); and Task Modification Scale (MOD) (n = 1) (Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007). There was only one study that considered middle-aged (≥ 40 years) participants (Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020); all others had an inclusion criteria of being at least 55 years of age, which is often the youngest age for being considered a “Senior Citizen” (https://www.seniorliving.org/life/senior-citizen/). Other than age and PCML stage measures, inclusion criteria related to cognitive level (n = 6) and medical condition (n = 8) were used. The Mini-Mental State Examination (MMSE) was the most frequently used measure in these studies (4/6, 66.7%), but the eligibility cut-off scores varied across studies, with the lowest score being 18 (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019) and the highest being 25 (Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020). Among the 12 studies, eight explicitly specified that participants needed to be medically stable (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013), have no major illness (Richardson et al., Reference Richardson2020), receive clearance from physicians (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012, Reference Day, Hill, Stathakis, Flicker, Segal, Cicuttini and Jolley2015; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008), or achieve a designated cut-off score on functional tests such as the Short Physical Performance Battery (SPPB) with a baseline score of 4–12 and < 20% in the range of 11–12 for the intervention group (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019); or the Life Space Assessment (LSA) with a baseline score of 52.3–90 (Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020), or demonstrate the ability to walk 500 m unaided in a walking test (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019).
Although four studies did not require the PCML stage measure as an eligibility criterion (Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020), all studies included a PCML measurement. All or a portion of the participants were identified as individuals at the PCML stage, except for in two protocols that had not completed participant recruitment at the time of submission (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Richardson et al., Reference Richardson2020). Participants in the 10 completed studies were all older adults ≥ 55 years of age, including both genders. The sample size in the studies varied depending on the intervention type. Studies using exercise as an intervention either had small recruitment (n = 20–40) or high drop-out rates and low attendance, which resulted in a relatively small sample size. However, studies using other types of interventions had larger sample sizes. For example, 237 participants were included in an educational intervention study (Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008) and 510 participants were recruited using a knowledge translation intervention (Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020) (see Table 1).
Table 1. Main findings of included studies
Note.
a Refer to the sample size analyzed unless otherwise noted.
b Defined as near-normal gait speed (≥1.0 m/sec walked at their usual, self-selected speed on a 4-m instrumented walkway) and impaired motor skill in walking (figure of 8 walk test (≥8 sec)
c Participants were initially tested (pre-control) and asked to continue their normal daily activities for an 8–10-week control period, tested again (post-control) and randomly assigned to intervention groups.
d Only those who participated in the pre-trial monitoring (n = 139, 68% of the total sample) were invited to participate in the post-trial monitoring.
RCT = randomized controlled trial; NRS = non-randomized controlled trials; PCML = pre-clinical mobility limitation; IG = intervention group; CG = control group; MMSE = Mini-Mental State Examination; SPPB = Short Physical Performance Battery; MoCA = Montreal Cognitive Assessment; RT = resistance training; FRT = functional + resistance training; FT = functional training; LLFDI = Late-Life Function and Disability Instrument; CS-PFP10 = Continuous Scale Physical Functional Performance 10 item test; SF36PF = Medical Outcome Survey- SF36 Physical Function; PPT = Physical Performance Test; PAC = peak aerobic capacity; SF-36 = Short Form-36; SAILS = Structured Assessment of Independent Living Skills; NPPS = Nagi Physical Performance Scale; ED = emergency department; WOMAC = Western Ontario and McMaster Universities Arthritis Index Osteoarthritis Index; BP = blood pressure; HR = heart rate; 6 MWT = 6-Minute Walk Test; BBS = Berg Balance Scale; BDI = Beck Depression Inventory; PA = physical activity; TUG = Timed Up and Go Test; RAPA = Rapid Assessment of Physical Activity; KT = knowledge translation; LSA = Life-Space Mobility Assessment; HQoL = health-related quality of life; WHOQOL = WHO Quality of Life; ABC = Activities-Specific Balance Confidence Scale; RPE = Borg Rating of Perceived Exertion scale; FRAX = WHO Fracture Risk Assessment Tool; SM = self-management; SPMSQ = Short Portable Mental Status Questionnaire; HUI3 = Health Utilities Index- Mark III; B/A = before and after; UAB LSA = University of Alabama at Birmingham Study of Aging Life-Space Assessment; YPAS = Yale Physical Activity Survey; GDS = Geriatric Depression Scale
Intervention type
Only one study used a multifaceted intervention (exercise combined with a mobility self-management [SM] program, Richardson et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020), while all others used a single intervention. Exercise was the most frequently used intervention type in the studies (7/12, 58.3%) (Bennett & Hackney, Reference Bennett and Hackney2018; Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012, Reference Day, Hill, Stathakis, Flicker, Segal, Cicuttini and Jolley2015; Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Richardson et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020). Of these, four studies used other exercise types as controls, one compared exercise to usual care (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013), one compared it to a nutrition intervention (Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008), and one compared to participants’ initial status, which was 8–10 weeks before the intervention. During the control period, all participants continued their usual activity for 8–10 weeks and were then randomly assigned to intervention groups (Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007). Other types included an educational intervention (Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008), tele-physical therapy (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019), knowledge translation (Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020), and counselling (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020). The duration of exercise interventions in the studies was shorter (8–48 weeks) with an average of 22 weeks compared with that of other intervention types, which mainly lasted 12 months or longer (Table 1).
PCML measures
As outlined in Table 1, not all studies assessed PCML outcomes, and there were various measures of PCML in the studies where this was assessed. The Manty scale was the most frequently used measure in the studies, including two RCTs (Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020) and two RCT protocols (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Richardson et al., Reference Richardson2020). However, the questions selected from the Manty scale also varied among these four studies: some asked about the difficulties in 2 km walking (Richardson et al., Reference Richardson2020; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020), and others added questions about difficulties in 0.5 km walking and climbing up one flight of stairs (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020). Additionally, in one RCT (Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020), there was no explicit indication of whether the question regarding task modification had been further asked for those responding “able to manage without difficulty” to 2 km walking. The study also only categorized two groups, rather than three or four groups, as suggested in previous studies (Mänty et al., Reference Mänty, Heinonen, Leinonen, Törmäkangas, Sakari-Rantala and Hirvensalo2007, Reference Mänty, Heinonen, Viljanen, Pajala, Koskenvuo and Kaprio2010). The Fried scale (Fried et al., Reference Fried, Bandeen-Roche, Williamson, Prasada-Rao, Chee and Tepper1996) was used to assess PCML outcomes in two studies (Bennett & Hackney, Reference Bennett and Hackney2018; Richardson et al., Reference Richardson2008). However, one of them (Bennett & Hackney, Reference Bennett and Hackney2018) only used seven items of activities instead of the 27 items recommended in the Fried scale (Fried et al., Reference Fried, Bandeen-Roche, Williamson, Prasada-Rao, Chee and Tepper1996). Continuous Scale Physical Functional Performance 10 Item Test (CS-PFP10), MOD and timed performance on eight tasks of daily life, and gait speed and impaired motor skill in walking were used in one study each (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008). Gait speed over 4 m was also assessed in three studies as a component of the SPPB (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Bennett & Hackney, Reference Bennett and Hackney2018) or as an independent test (Richardson et al., Reference Richardson2020). However, these studies did not specify it as a PCML outcome measure.
Measures to assess other outcomes
There were outcomes other than those for the PCML measured in these interventional studies. These measures were grouped into eight major categories according to the functional dimension/domain they evaluated.
-
1. Physical function and disability outcomes: self-report Late-Life Function and Disability Instrument (LLFDI) (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012), Nagi Physical Performance Scale (NPPS) (Richardson et al., Reference Richardson2008), and complete SPPB (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Bennett & Hackney, Reference Bennett and Hackney2018; Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020) or part of SPPB such as chair stands (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008).
-
2. Balance outcomes: number of falls (Day et al., Reference Day, Hill, Stathakis, Flicker, Segal, Cicuttini and Jolley2015; Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008), balance tests such as single-leg balance time or Berg Balance Scale (BBS) or Timed Up & Go (TUG) (Bennett & Hackney, Reference Bennett and Hackney2018; Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012; Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008), balance confidence assessed by Activities-specific Balance Confidence Scale (ABC Scale) (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019), and balance self-efficacy (SE) (Richardson et al., Reference Richardson2020).
-
3. Lower extremity strength (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008, Reference Richardson2020).
-
4. Endurance outcomes: measured by 400 m walk (Bennett & Hackney, Reference Bennett and Hackney2018) or 6-minute walk test (6MWT) (Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012).
-
5. Physical activity (PA): measured by self-reporting (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Richardson et al., Reference Richardson2020; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020) or performance tests (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012; Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020).
-
6. Mobility outcomes: assessed using LSA (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019; Siltanen et al., Reference Siltanen, Portegijs, Pynnönen, Hassandra, Rantalainen, Karavirta, Saajanaho and Rantanen2020) or self-reported (Richardson et al., Reference Richardson2020).
-
7. Quality of life outcomes: measured by WHO Quality of Life (WHOQOL) (Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019), 12-Item Short-Form Health Survey (version 2) (SF-12v2) (Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007), or 36-Item Short-Form Health Survey (SF-36) (Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Richardson et al., Reference Richardson, Chan, Risdon, Giles, Mulveney and Cripps2008, Reference Richardson2020).
-
8. Psychological outcomes include SE (Richardson et al., Reference Richardson2020) and depression (Day et al., Reference Day, Hill, Jolley, Cicuttini, Flicker and Segal2012, Reference Day, Hill, Stathakis, Flicker, Segal, Cicuttini and Jolley2015; Edgren et al., Reference Edgren, Karinkanta, Rantanen, Daly, Kujala, Törmäkangas, Sievänen, Kannus, Heinonen, Sipilä, Kannas, Rantalainen, Teittinen and Nikander2019).
Changes in PCML outcomes
Changes in PCML outcomes as a result of the intervention were evaluated by whether there was an improved score in PCML measures and/or a decreased proportion of participants with PCML (Table 1). Although only eight studies reported the PCML results after the intervention, the findings varied regarding the outcome measures and intervention effectiveness.
Out of the six studies reporting changes in the scores of PCML measures, five studies revealed significant improvements in the intervention group’s PCML scores compared to the controls (Bean et al., Reference Bean, Brown, Deangelis, Ellis, Senthil Kumar, Latham, Lawler, Ni and Perloff2019; Brach et al., Reference Brach, Van Swearingen, Perera, Wert and Studenski2013; Manini et al., Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007; Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Richardson et al., Reference Richardson2008). For example, Brach et al. (Reference Brach, Van Swearingen, Perera, Wert and Studenski2013) reported that compared with the controls, participants with pre-clinical gait dysfunction receiving motor learning exercise intervention improved significantly more in gait speed (0.13 vs. 0.05 m/sec, p = 0.008) and motor skill (-2.2 vs. -0.89 sec, p < 0.001). Richardson et al. (Reference Richardson2008) compared participants receiving detailed feedback from physiotherapists (PT)/ occupational therapists (OT) about physical function assessment results over 18 months with those who did not receive feedback in the control group. They found statistically significant group/time interactions for Fried scale (frequency subscale, F = 3.78, p ≤ 0.05; modification subscale, F = 4.58, p ≤ 0.05). Another study reported that changes in PCML outcomes in the intervention group compared with controls differed by measures. There were significant improvements in gait speed and self-reported difficulty climbing a flight of stairs, but not in the difficulty of walking 0.4 km (Bennett & Hackney, Reference Bennett and Hackney2018).
Three studies assessed the impact of interventions on the percentage of participants with PCML. However, there is variation in the results across these studies (Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008; Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020; Rantanen et al., Reference Rantanen, Pynnönen, Saajanaho, Siltanen, Karavirta, Kokko, Karvonen, Kauppinen, Rantalainen, Rantakokko, Portegijs and Hassandra2019). One study reported that the number of participants with PCML in the intervention group (16-week supervised walking interventions) decreased significantly from 66 to 25% (Moore-Harrison et al., Reference Moore-Harrison, Speer, Johnson and Cress2008), while another study reported that the number of participants with PCML decreased but not significantly (Rantanen et al., Reference Rantanen, Pynnönen, Saajanaho, Siltanen, Karavirta, Kokko, Karvonen, Kauppinen, Rantalainen, Rantakokko, Portegijs and Hassandra2019). A study used knowledge translation as an intervention for 12 weeks and reported that the number of participants with PCML decreased in both the intervention and control groups (3.9% vs. 5.2% at the end of the study, respectively; 4.6% vs. 5% at the follow-up 3 months, respectively), but that there were no significant between-group differences at the end of the study (p = 0.59) or at follow-up (p = 0.19) (Neil-Sztramko et al., Reference Neil-Sztramko, Smith-Turchyn, Richardson and Dobbins2020).
Discussion
It is more than 20 years since Fried proposed that people in the PCML stage may particularly benefit from interventions preventing disability onset (Fried et al., Reference Fried, Herdman, Kuhn, Rubin and Turano1991). However, this review was only able to identify 14 articles available in the PCML interventions field. Two of them are protocols, and the respective published articles were also included. The majority of the studies (9/12, 75%) were published after 2011, 10 years after the PCML concept was introduced (Fried et al., Reference Fried, Herdman, Kuhn, Rubin and Turano1991), 67% of which were published over the past 5 years. This may reflect the recent increase in research interest and growth in this field. However, all studies originated from the United States, Canada, Australia, and Finland. Skewed distributions in the PCML publication could be partly the result of the research group’s monolingualism, which prevented review of non-English publications. However, it also suggests that the concept of PCML needs to be better recognized and explored worldwide, as it will benefit both health and longevity (Richardson, Reference Richardson2020).
This review included intervention studies with a full or partial percentage of participants at PCML stages. However, the measurements used to assess PCML vary from self-report measures (e.g., Fried scale) to physical performance measures (PPMs) (e.g., CS-PFP10). These two types of measures have been shown to assess different characteristics of function in older adults, and have their own respective advantages (Babić, Bertić, Križanec, & Babić, Reference Babić, Bertić, Križanec and Babić2020; Cavanaugh, Richardson, McCallum, & Wilhelm, Reference Cavanaugh, Richardson, McCallum and Wilhelm2018): PPMs are objective and tend to be reproducible, less influenced by language and education, and more reliable in assessing change over time; self-report measures are accessible, easier to apply, and sensitive to capturing very early changes in function. Furthermore, self-reported measures were associated closely with objective PPMs (Young et al., Reference Young, Boyd, Guralnik and Fried2010). Self-reported measures can be a valid benchmark in detecting PCML in some tasks in older adults who were already disabled in other tasks (Young et al., Reference Young, Boyd, Guralnik and Fried2010). However, the variations between PPMs and self-reported measures, with respect to administration mode and outcome reporting, limit the comparison across studies in this review. In addition, validated measurements were modified by selecting part of the tasks or adding more tasks in some studies. For example, Manini et al. (Reference Manini, Marko, VanArnam, Cook, Fernhall, Burke and Ploutz-Snyder2007) added the activity of “laundry basket lift and carry” on the top of the seven activities previously used to calculate the MOD score (Manini, Cook, VanArnam, Marko, & Ploutz-Snyder, Reference Manini, Cook, VanArnam, Marko and Ploutz-Snyder2006); while Bennett and Hackney (Reference Bennett and Hackney2018) selected seven out of the 27 activities originally used in Fried scale. This raises the question of what task(s) should be selected to ask about for early detection of disability progression.
In this review, eight categories of outcomes were summarized as associated with PCML outcomes in the studies: physical function and disability outcomes, balance, lower extremity strength, endurance, PA, mobility, quality of life, and psychological outcomes. The physical decline in muscle strength, endurance, mobility, and balance collectively contribute to the limitation of functional capacity, which can be influenced by PA (Tomás, Galán-Mercant, Carnero, & Fernandes, Reference Tomás, Galán-Mercant, Carnero and Fernandes2018). The prediction of PCML on falls, adverse health outcomes, diminished quality of life, and the onset of difficulties in various tasks, not limited to activities of daily living (ADLs)/ instrumental activities of daily living (IADLs) but encompassing a broader range of activities, has been reported in both cross-sectional and longitudinal studies (Clough-Gorr et al., Reference Clough-Gorr, Erpen, Gillmann, Von Renteln-Kruse, Iliffe and Beck2008; Fried et al., Reference Fried, Bandeen-Roche, Chaves and Johnson2000; Katz, Morris, & Yelin, Reference Katz, Morris and Yelin2008; Wolinsky et al., Reference Wolinsky, Miller, Andresen, Malmstrom, Miller and Miller2007). PCML presents both as a risk marker associated with impairment or limitation and a mediating factor for disability progression (Weiss et al., Reference Weiss, Hoenig and Fried2007). This suggests that the rationale for PCML interventional studies extends beyond focusing solely on PCML outcomes to explore how underlying etiological factors can be modified to prevent the progression of the disease to mobility disability (Fried et al., Reference Fried, Young, Rubin and Bandeen-Roche2001). Additionally, it emphasizes the importance of assessing and reporting associated outcomes to identify if any of the multiplicity of risk factors can be minimized through the intervention.
This review does not seek to gauge the effectiveness of rehabilitation interventions on PCML-associated outcomes, but rather to describe what associated outcomes were assessed in currently available PCML interventional studies. However, also of importance is what outcomes should be utilized in combination with PCML outcomes when exploring interventions in the pre-clinical stage, for preventing or delaying overt disability progression (Fried et al., Reference Fried, Young, Rubin and Bandeen-Roche2001). It is noteworthy that 33 per cent (4/12) of studies in this review reported psychological outcomes. Psychological constructs/outcomes such as lower SE, higher anxiety, and depressed mood were associated with PCML (Higgins et al., Reference Higgins, Janelle, Naugle, Knaggs, Hoover and Marsiske2012; Hirvensalo et al., Reference Hirvensalo, Sakari-Rantala, Kallinen, Leinonen, Lintunen and Rantanen2007). This can be further investigated in future studies, as such an approach may advance the use of compensatory strategies and serve as a focus for future interventions aimed at disability prevention (Higgins et al., Reference Higgins, Janelle, Naugle, Knaggs, Hoover and Marsiske2012).
This review identified the types of available evidence in the PCML rehabilitation intervention literature, including exercise, education, tele-physical therapy, knowledge translation, and counselling. The effectiveness of interventions was evaluated by measuring changes in PCML scores or the percentage of participants with PCML. All five studies that used PCML scores as an outcome measure demonstrated effectiveness, showing an improved score. However, discrepancies arose in studies that assessed the percentage of participants with PCML, with only three studies available for analysis. In the study in which exercise was used as an intervention and PPM as a measurement, a significant decrease in PCML participants was observed in the intervention group. This change differed significantly when compared with the control groups. Conversely, when using a self-reported scale (i.e., Manty scale) as an outcome measurement, the decreased percentage of PCML participants in the intervention groups (counselling or Knowledge Translation intervention) either showed no significant change over time or did not differ significantly from the control groups. These discrepancies suggest the benefits of using exercise as an intervention and highlight the sensitivity of using PPM as an outcome measure. Yet, this finding must be interpreted with caution, given the heterogeneous nature of the sample, diversity of the eligibility criteria, difference in the length of the intervention and follow-up, and various PCML measures across the studies. The results need to be confirmed by a systematic review with a larger number of studies. Together these findings may inform future intervention strategies so that they can be clinically effective in disability prevention.
Future Research Directions
The mobility concepts are multifactorial and complex (Freiberger et al., Reference Freiberger, Sieber and Kob2020). Correspondingly, mobility disability may result by multiple functional declines across multiple systems (Fried et al., Reference Fried, Young, Rubin and Bandeen-Roche2001). Therefore, a multifaceted intervention may be the most effective approach to jointly benefit older adults in the PCML stage. An example of this approach is the combination of exercise and a mobility SM program (Richardson et al., Reference Richardson2020). Future research should employ multifaceted interventions with different intervention components to address questions on maintaining functional competence, preventing adverse outcomes, and reducing the burden of disability in this population.
The measurements that assess PCML need to have established sensitivity to change in this population to capture the pre-clinical signs of functional decline. Performance-based measures are objective, quick, inexpensive, and more reliable in assessing change over time, whereas self-reported measures are sensitive to early changes in function. The use of performance-based ADL/IADL assessment independent of or in conjunction with self-report as a method of identifying pre-clinical disability in older adults was reported in a previous study (Toto, Terhorst, Rogers, & Holm, Reference Toto, Terhorst, Rogers and Holm2017). By combining both, researchers can have greater specificity in detecting incident mobility difficulty than by using either alone. In addition, self-report of functional limitation is also associated with education (Gregory et al., Reference Gregory, Szanton, Xue, Tian, Thorpe and Fried2011), culture or socio-environmental differences associated with culture (Spencer, 2008), gender (Lorenz, 2009), body sensation, and coping practice that allows bodily limitations to remain unnoticed or conceals limitations from others (Lorenz, 2007, 2010). Future research using self-reported measures should also consider these factors when implementing measures.
Conclusion
This scoping review identified 14 articles focusing on PCML interventions, including exercise, education, tele-physical therapy, knowledge translation, and counselling. The review demonstrated limited published work on interventions to address PCML, especially on middle-aged populations and older adults not considered to be community dwelling. It also revealed the complexity and varieties of currently available PCML measures, which leads to the difficulty in applying a unique measure that is valid and reliable to use, and compares findings across PCML studies. This review provides preliminary evidence that rehabilitation interventions with persons with PCML may help to delay and prevent disability progression. However, because the quality of the studies was not evaluated, the findings should be interpreted cautiously.
Acknowledgments
We gratefully acknowledge the help of Sadaf Ullah, clinical outreach librarian at McMaster Health Science Library, for sharing her expertise and providing guidance on refining search strategies for this review.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0714980823000685.
Open access
