Web-Based Software Applications for Frailty Assessment in Older Adults: Current Status and Insights Into Future Development

Background: A crucial aspect of continued senior care is the early detection and management of frailty. Developing reliable and secure electronic frailty assessment tools can benet virtual appointments, a need especially apparent since the COVID-19 pandemic. An emerging effort has targeted web-based software applications to improve accessibility and usage. Methods: We conducted an environmental scan through MEDLINE and Google searches (last updated on June 1st, 2021) to identify currently available web applications, each of which was evaluated and assigned a rating score based on eight featured categories. Results: Twelve web-based frailty assessment applications were found, chiey provided by the USA (50%) or European countries (42%) and focused on frailty grading and outcome prediction for specic patient groups (58%). The categories that scored well among the applications included the User Interface (2.67/3) and the Cost (2.75/3). Other categories had a mean score of 1.5 or lower. The least developed features in the existing web applications included Data Saving. Conclusions: This is the rst study that has compiled a comprehensive list of frailty assessments available online, described their usage and evaluated their advantages and limitations. The study emphasized several essential features with future web application development to support early detection and management of frailty with virtual care.


Background
Frailty is a state of increased vulnerability, re ected by the accumulation of de cits in health with age due to declines in resilience [1,2]. Frailty can lead to increased risks of adverse outcomes, including hospitalization and mortality [3][4][5]. The Frailty Index (FI) has been used to assess the level of frailty, calculated as the proportion of health de cits that a person has over the total number of health de cits under consideration [1,2].
Emerging data have highlighted the relationship between frailty and the COVID-19 deaths and other adverse effects in older adults, who are more likely to have accumulated more de cits of multiple organ systems [6][7][8]. Following the physical distancing guidelines, virtualized approaches are being increasingly adopted to continue senior care and frailty management while reducing in-person appointments [9,10]. Besides the pandemic, the older adult demographic is already beset by transportation related issues that prevent easy access to primary care (e.g. mobility problems, lack of caregiver ability to drive, no drivers license, remote area) [11,12], increasing the need for reliable virtual care.
Web-based health assessments have potential to aid in virtual healthcare. In general, web-based assessments are easy to access, available for free or at a minimal cost, more time e cient, and can allow automatic calculations leading to immediate results [13][14][15][16]. With a pressing demand for maintaining physical distancing guidelines, reliable web-based assessments may be especially helpful for the future of frailty assessments.
Frailty assessments have traditionally relied on manual data entry and processing, which can be time-consuming and error-prone [16,17]. Recent research has enabled an electronic FI calculation in the Electronic Medical Records (EMR) system, which has allowed for frailty measures in primary care for millions of older patients in the UK and several other European countries [18,19].
More recently, an eFI was constructed based on the electronic Comprehensive Geriatric Assessment (eCGA, a multidisciplinary diagnostic assessment that evaluates many domains of older adults' health and care needs) [20,21]. Being available in EMR systems and as standalone software that runs on personal computers [20], the eFI-CGA is time-e cient and cost-effective. Even so, accessing the EMR from home can be infeasible, and downloading/installing the standalone version upgrades can be inconvenient.
Motivated to develop a web-based software tool, we conducted a thorough search and evaluation of the currently available web applications for frailty assessment. The purpose of this study is to provide insights to guide future web-based frailty assessment software development, including the web-based eFI-CGA. Our speci c objectives were to: 1) understand what web applications exist for frailty assessment; 2) describe the usability of these applications; and 3) evaluate the challenges and opportunities of future web-based frailty assessments.

Search Terms
We conducted an environmental scan to collect and organize information [22,23] pertaining to web-based frailty assessments.
Google search was chosen as our rst source of information, given the focus of the search and the known coverage of Google on online resources. An independent MEDLINE database search was also employed to extracted information from peer-reviewed research literature. Two independent researchers (RC and HL) conducted each search (Fig. 1), and last updated the information on June 1st 2021.
The initial searches were performed utilizing three sets of keywords: Set 1: "online" or "web-based" or "website-based"; Set 2: "geriatric" or "frailty"; and Set 3: "assessment" or "software tool" or "application" or "calculator". For the Google search, each individual term in sets 1 through 3 were combined, creating 24 unique search inputs (Appendix 1). For the MEDLINE search, the three sets of search terms were combined to create the search string: ("online" OR "web-based" OR "website-based") AND ("geriatric" OR "frailty") AND ("assessment" OR " software tool" OR "application" or "calculator"). A "*" sign was used to represent different su xes of the same word in the search inputs/string where applicable.

Search Methods
Google search: All personal Google accounts were logged out of, and the history, cookies, and cache were reset before each search to minimize the effects of Google search customization. The rst 50 results (which was found to be su cient to cover all relevant material) for each Google search using a unique input term (excluding advertisements) were retrieved. This yielded a total of 1200 initial results (Fig. 1). All the resulted website titles and previews were sequentially screened to include websites related to frailty assessment. The selected websites were then fully scanned to ensure they either hosted a web-based frailty assessment or included a direct external link to a web-based frailty assessment meeting our criteria (Fig. 1).
MEDLINE search: Article titles and abstracts were screened to include the publications discussing frailty assessment. Names of speci c frailty related web-based assessments were then extracted from these relevant articles. Using this information, an additional Google search of each of the individual web application names was conducted to ensure the assessment remains available online and meets the selection criteria described below (Fig. 1).
Finally, the results from the Google and MEDLINE searches were combined, yielding a nal list of web applications for further evaluation and analysis (Fig. 1).

Selection Criteria
Web-based frailty assessments were considered if they met the following inclusion criteria: (1) related to frailty, (2) had a llable form for data collection, (3) hosted on a web domain, and (4) interfaced in English. The exclusion criteria were: (1) not available online, (2) only showed a non-llable PDF version, and (3) not accessible for research or use.

Evaluation and Analysis
Considering the common features of web applications, each application was evaluated applying IEEE recommendations for scoring that consisted of a comprehensive list of criteria [24][25][26][27][28][29][30]. The criteria of the evaluation included a total of 13 categories of software features ( Table 2). These included user friendly interface, effective data saving, completeness of health domain inclusion, completeness of health item inclusion, completion the cost of usage, results interpretation availability, instructions and training availability, remote access and conductance possibility (Table 2, top panel); and time e ciency of assessment, algorithm e ciency, security, environmental requirements, and browser requirements ( Table 2, bottom panel). The categories were relevant to frailty and widespread application potential (e.g. promoting digital health and bene ting virtual care).
For each web application identi ed, an assessment score was assigned using a rating grade ranging from 0 to 3 for a given category, with higher scores meaning better (Table 2). For a category that could not be adequately evaluated (i.e. missed being reported most websites), scoring of the category was not included in the sum up score, but was still discussed. The evaluation and scorings were conducted independently by two researchers (RC and HL). Any discrepancies were resolved through discussions between the researchers and consensuses were achieved upon the majority vote involving additional researchers (e.g., AM, XS).

Results
The nal set of results consisted of twelve web-based frailty assessment applications, all of which were developed over the past ve years (2015 -2020). The majority of the web applications were provided by the USA (50% and UK or European countries (42%), with a focus on frailty grading and outcome prediction for speci c patient groups (58%). The basic information for each of the applications is summarized below and described in Table 1.
Released in 2015, the Edmonton Frail Scale consists of 11 assessment items. It provides a simple way to assess frailty in older adults and can be completed in 5 minutes on average.
Also released in 2015, the Myeloma Frailty Score Calculator aids in the prognosis of elderly myeloma patients. It assesses 31 items, which can be saved in a PDF document along with the calculated score.
In the following year, the Johns Hopkins Frailty Assessment Calculator was released to assess the ve-item frailty phenotype [31]. A free trial is limited to 5 calculations; unlimited calculations and guidebook and database access can be obtained with an annual subscription.
The Geriatric 8 (G8) Health Status Screening Tool was recently updated in 2017, for used to identify older cancer patients who may bene t from a CGA. Free accounts can download a PDF assessment; a subscription leads to unlimited downloads and the ability to add notes to the PDF.
The Liver Frailty Index was also released in 2017 to assess physical frailty in patients with chronic liver disease and/or cirrhosis. It assesses three performance-based items: grip strength, chair stands, and balance, with detailed instructions including diagrams for use.
Also released in 2017, the QFrailty Risk Calculator assesses older adults' risk of developing frailty involving over 40 items of 10 health domains. The software estimates the frailty degrees (e.g "mild" or "severe"), and the two-year hospitalization and death risks.
The Frailty Risk Calculator was released in 2018 and estimates the probability of hospitalization or death within the next year.
Also released in 2018, the Senior Health Calculator uses the CGA items to produce a FI. Fifty items on medical history, functional status, performance tests, and nutritional status are assessed, and FI calculation can be based only on the rst two domains. The input data, FI, and summary may be saved or printed as PDF.
The Modi ed Frailty Index was recently updated in 2019 and assesses the morbidity and mortality risks in older general surgery patients. Free accounts can download a PDF of the assessment; a subscription leads to unlimited downloads and adding notes to the PDF.
The Frailty Index Calculator, a simple calculator developed in 2020, allows users to enter the number of health de cits present in a patient and the total number of de cits measured, and gets an FI score using the formula FI = No. de cits present / No. de cits measured. Users need to have the input numbers ready somewhere else rst.
The Myelodysplastic syndromes (MDS) Speci c Frailty Index was released in 2020 and evaluates frailty in patients with myelodysplastic syndrome. Seventeen items are included to calculate a frailty scale ratio, composite score, and estimated survival outcome. Users may manually copy/paste the input and output to the local computer.
The Cumulative Illness Rating Scale-Geriatric (CIRS-G) is a brief assessment that quanti es the disease burden in older adults. The Assessment uses 14 multiple-choice items to produce a frailty score. With a free account, users can copy assessment inputs and results to the local computer.
Eight feature categories could be applied to scoring these web applications (i.e., assigning a 0 through 3 to each category), making 24 the highest possible sum-up score that a web application could receive ( Table 2). Figure 2 shows the sum-up scores of these web applications. The two categories that scored with high values among the applications were User Interface and Cost (2.67/3 and 2.75/3, respectively), whereas the other categories had a mean score of 1.5 or lower ( Figure 2). The category with the lowest score was Data Saving, with only 4 assessments not scoring a 0, as most of the web applications permitted no or very primitive data saving.

Discussion
We conducted this study to understand what frailty assessment tools are available online and to evaluate their usability considering a large number of feature categories. To the best of our knowledge, this is the rst attempt to systematically identify and summarize web-based frailty assessment applications. This is also the rst known study that comprehensively evaluates the applicability and limitations of the web applications. The research has allowed us to generate important insights into successful future development of online software tools in support of early detection and management of frailty.
The assessments under evaluation showed several essential merits and have multiple advantages. For example, most applications have developed a highly friendly user interface. One crucial bene t of having a web-based assessment is the convenience and ease of completing the assessment with simple selections and mouse clicks [15,18]. Assessments realized the functionality through employing radio button selection for "yes or no" questions where appropriate, rather than using the more time-consuming drop-down menus or text boxes. Most applications also appeared to be highly cost e cient. Making assessments available for free will maximize the potential for adoption and impact. Having accessible frailty assessments widely available will encourage use and support for frailty management.
Despite these bene cial qualities, several important areas may be improved in future development of web-based frailty tools. For example, the assessment pages often lacked clear and comprehensive instructions or training materials. Including these is helpful for potential users to accurately and effectively complete the assessment, especially regarding the speci cs of some performance items (e.g. speci c version of the sit to stand test). Similar arguments can apply to including helpful materials for results interpretation. Even though an assessment produced a score or frailty rating, the number often lacked context for clear meaning without interpretation. Adding this can help make appropriate patient care planning.
It is also worth noting that the existing web applications commonly included limited options for data saving and retrieval.
Although copying of the completed assessment might be allowed by some, working with the data or even reloading the saved assessments for processing/editing were typically unmanageable. This feature is needed in scenarios where an assessment must be completed over more than one session due to time restrictions, interrupted internet connections, and other disruptions that patients may experience with virtual care [32].
Further, a majority of the assessments considered only a limited number of health domains and/or total items. Because frailty is a multidimensional syndrome characterized by the loss of physiological reserve in multiple health systems [1,2] frailty assessments can have improved precision relating the outcomes when a wide range of health domains and items are considered, for a comprehensive overview of patients' health. In fact, it has been recommended that de cit accumulation based frailty index to include no less than 30 individual items when possible [33,34].
Our study has several limitations. Our search was limited to Google and MEDLINE. While this is a widely accepted approach, it is still possible that some existing tools only available from less prominent resources may have been overlooked. Additional, focusing on the functionality and usability, the content validity of the web applications is beyond the scope of the current investigation, as each assessment warrants a focused separate study [35]. Also, several categories in the evaluation criteria (time e ciency, security, algorithm e ciency, environmental requirements, and browser requirements) were not assessable due to the lack of data provided by the web applications under evaluation. Information on these aspects can be fundamental in software appraisal [20,24,[27][28][29][30]; future developments are encouraged to take into account these into the implement and report how these aspects are addressed.
Despite these limitations, our study is contributable to the research eld and is meaningful to advancing early detection and management of frailty. We have applied the established software evaluation approaches [24][25][26][27][28][29][30] for insightful understanding of the applications, highlighting the important features for inclusion in the future development, including our ongoing effort in advancing a fully functional web-based eFI-CGA (https://e -cga.ca/). It is seen that having reliable and effective web-based health assessments will enable virtual care via digital health as a common practice in the future.
This study is also relevant to the COVID-19 pandemic and its prevailing consequences and the upcoming new norms, as it has been established that healthcare providers need exible health assessments to accommodate physical distancing restrictions [10]. This has been re ected by the rapidly increase number of requests, since the beginning of the pandemic, for obtaining a copy of the standalone eFI-CGA software tool [20]. A more widespread uptake of web-based frailty assessments is anticipated, as they can allow for frailty management to continue throughout times of limited contact, as they are not limited by the need to access and download software. These remarks are amongst the situation that transportation-related issues that prevent easy access to primary care already disadvantage older adults. Given the many reasons why virtual care is useful for seniors and their general positive perspectives towards virtual care and tele-technologies [36], this timely study highlights the currently available web-based assessments and provides insights on developing web applications for use in virtual assessments.

Conclusions
The Consent for publication: Not applicable.
Availability of data and materials: All data (i.e., the websites) analysed during this study are included in this published article (Table 1).An additional data le listing the original data (i.e., the websites) and the associated scores assigned for the analysis is also provided (Appendix 2).
Competing interests: The authors declare that they have no competing interests.  https://www.mdcalc.com/cumulative-illness-rating-scalegeriatric-cirs-g#use-cases 1 Year the tool was created, unless otherwise indicated below 2 Organization/company provided the web application, regardless the original assessment. 3 As described in the tool 4 The webpage under the present evaluation * Publication date of the corresponding article ** Date of last revision  (2) Compatible with more browsers. (3)(4) Compatible with most major browsers (≥5)