Scoping review findings are as follows.²⁵⁾ First, commonly assessed domains were physical health, psychological/cognition, functional, and mobility/balance. Second, mode of administration includes healthcare professionals, trained volunteers, and self-administration. Third, duration of BGA ranged from less than 5 minutes to 20 minutes. Fourth, except for improved identification of older adults with higher or unexpressed needs, no other significant changes were observed.

These findings guided the initial draft of Delphi statements. For example, the use of trained volunteers to administer BGA prompted a statement regarding its feasibility in Singapore. This was followed by a consultation with three study advisors, who are lead or adjunct scientists with a national research institute focusing on ageing research. The advisors are also senior consultants or the head of Geriatric Medicine departments of public hospitals in Singapore and hold professorships with local universities. They were chosen due to their involvement and extensive experience in practice, education, and research on geriatric medicine. The draft statements were edited based on their feedback and questions were also anchored on the need for BGA to be within 15 minutes to ensure consistency among participants. The statements were also piloted with two additional senior consultants, resulting in minor revisions. The finalized version contained 72 statements.

A condensed set of experts is recommended for specialized clinical topics,³³⁾ with 12 participants deemed sufficient for reliable consensus.³⁴⁾ We aimed to complete the Delphi with a minimum of 12 participants. Considering a 25% attrition rate from a previous study,³⁵⁾ a purposive sample of 16 doctors were invited via email. The invitees are consultants, senior consultants, or heads of service in Geriatric Medicine departments of public hospitals in Singapore, or doctors in family medicine focusing on geriatric issues at public clinics in Singapore.

Participants rated statements on a scale of 1 (strongly disagree) to 7 (strongly agree) and could also provide comments in text boxes. Consensus was defined a priori, combining the common definition of level of agreement, and dispersion of responses based on interquartile range for additional robustness.^33,36) Consensus is achieved if 75% of respondents disagreed (rating of 1 and 2) or agreed (rating of 6 and 7) with a statement, and if the interquartile range of the response is 1 or less. Based on participants’ feedback, non-consensus statements in Round 1 were edited, and new statements were added for Round 2. Participants had to respond to both rounds. Non-respondents in Round 1 were not invited to Round 2. The study was conducted electronically via email.

Participants’ quantitative responses were entered into an Excel spreadsheet to calculate the percentage of agreement/disagreement and interquartile range for each statement. Qualitative responses in text boxes were read and discussed among the study team to understand the quantitative results. Findings were also discussed with the study advisors before the results synthesis.

Participants agreed on using BGA to detect geriatric syndromes for older adults in community and primary care, specifically targeting those showing geriatric syndromes or with mild to moderate frailty. There was no consensus on targeting all older adults aged 75 and above in these settings. Participants questioned the efficiency and the healthcare system’s capacity to cope with the influx of referrals from these broad screenings. There was no consensus on targeting older adults receiving regular primary care services or living alone.

Participants agreed that BGA in community settings should assess physical health, psychological/mental health, functional abilities, mobility/balance, and social situation. Participants also agreed that these domains, together with medication use, should be part of BGA in primary care settings. There was also consensus on the sub-domains for each domain (e.g., visual & hearing impairment for the domain of physical health). Information on domains and sub-domains is in Supplementary Table S3. There was no consensus on assessing comorbidity for the physical health domain in community-based settings. One participant felt that comorbidity assessments would prolong the BGA without adding significant value, while another expressed concerns about the accuracy of older adults’ self-reports.

Administration of BGA by healthcare professionals was agreed as feasible. There was no consensus on administration by trained non-healthcare professionals or self-administration by older adults. One participant highlighted potential difficulties in maintaining competency standards for trained non-healthcare professionals, while another expressed concerns on the validity of self-administration.

For mode of data collection, participants agreed that self-reporting based on older adults’ own subjective perception, such as how an older adult is feeling, should be accompanied by objective assessments. Conversely, self-administered tools collecting objective (e.g., older adults’ height or number of times they can perform an action) and subjective measurements should be accompanied by an informant report. Validity issues with self-reporting and self-administering were highlighted by participants.

Participants agreed that a positive BGA result (i.e., geriatric syndromes detected) should lead to a CGA referral. There was no consensus on recommending follow-up interventions without CGA for cases with a single identified geriatric syndrome. One participant questioned the likelihood of identifying only a single geriatric syndrome. Two participants commented that intervening without CGA depends on the geriatric syndrome identified. Another suggested that assessments and interventions can be conducted as needed, instead of expecting all assessments to be completed before intervening. The need for a clear care pathway and adequate resources to support older adults in follow-up interventions without a CGA was emphasized by a participant.

Older adults showing potential geriatric syndromes should be targeted for a BGA to detect geriatric syndromes. For example, an older adult who keeps bumping into things and does not always respond when addressed could be seen as showing signs of hearing and visual impairment. A BGA could confirm the impairment or determine if the issues are caused by other geriatric syndromes such as cognitive impairment. Statements targeting broader groups such as those: (1) aged 75 and above, (2) receiving regular primary care services, or (3) living alone did not achieve consensus. The cost and consequences of broad screening underlined participants’ preference for a targeted approach, which aligns with the targeted frailty^37,38) and cognitive screening³⁸⁾ recommendations by national organizations in other countries.

Participants agreed that BGA in community settings should assess physical health, psychological/mental health, functional abilities, mobility/balance, and social situation, while BGA in primary care should also include medication use. This difference arose from a deliberate omission of a statement on medication assessment in the community, as suggested by all three study advisors as community assessors may not have the prerequisite background. Participants’ consensuses almost mirrored the scoping review.²⁵⁾ Most BGA identified assessed physical health, psychological/mental health, functional abilities, and mobility/balance. However, social situation was the least assessed domain.²⁵⁾ This may reflect a recent emphasis on social connection in Singapore’s Ministry of Health action plan for older adults,³⁹⁾ prompting participants to include social situation assessment.

For each domain, participants agreed on a comprehensive list of sub-domains to assess, including frailty, incontinence, visual and hearing impairment, weight loss, cognitive function, mood, activities of daily living, fall history, and caregiver presence. This list differed from the BGA identified in the scoping review. For example, the Rapid Geriatric Assessment assessed frailty, sarcopenia, anorexia of aging, and mild cognitive impairment or dementia,⁴⁰⁾ while the Brief Risk Identification of Geriatric Health Tool assessed activities of daily living, cognitive impairment, and fall history.⁴¹⁾ Even the more comprehensive BGA were missing sub-domains participants agreed on, such as behavioural problems, gait disturbances, and pain.⁴²⁾ This discrepancy may stem from participants underestimating the time required to assess all suggested sub-domains.

Participants agreed on the feasibility of healthcare professionals administering BGA. This is consistent with the scoping review, where majority of BGA were administered by healthcare professionals.²⁵⁾ Conversely, the statement on BGA administration by non-healthcare professionals only came close to consensus, even though several BGA identified in the scoping review did so.⁴²⁾ One participant’s concern on maintaining trained volunteers’ competency standards to a similar level to healthcare professionals could explain the non-consensus. Previous research showed mixed results regarding assessments by laypeople versus professionals. While excellent inter-rater reliability was observed for Timed Up and Go tests conducted by caregivers and student physiotherapists,⁴³⁾ low inter-rater reliability was observed for nutritional screening conducted by trained volunteers and a dietician.⁴⁴⁾ Differences were also noted in assessments of neuropsychiatric symptoms between clinicians and caregivers.⁴⁵⁾ Timed Up and Go test measures performance while assessments of neuropsychiatric symptoms and nutrition involve subjective judgements whereby domain expertise is likely to play a bigger role. Delphi consensus are informative, but does not mean that a “correct” answer has been found.³²⁾ The use of trained non-healthcare professionals for BGA remains a possibility. One suggestion is that they could handle performance assessments while healthcare professionals manage more complex subjective evaluations.

Participants agreed on either a combination of self-administered tools and informant reports, or a combination of self-reporting questions and objective assessments for data collection. Concerns on the validity of self-reporting BGA led researchers to add objective measures.⁴⁶⁾ Other studies on self-administered or self-reporting tools cite the cost of using a healthcare professional as a reason for their research and have reported promising results.^47,48) A comparison of self-administered and healthcare professional administered BGA found that the self-administered version performed well on questions about falls, fatigue, weight loss, and multimorbidity.⁴⁷⁾ Areas for improvement, such as clarifying ambiguous wording, were also identified. Further research could explore improved versions of these tools to assess whether healthcare professional administration or supplementary assessments are still necessary.

Participants agreed that a positive BGA should lead to a CGA referral. There was no consensus on recommending follow-up interventions without a CGA for cases with only one identified geriatric syndrome. Participants expressed the need for a CGA referral as interpreting a positive BGA result can be challenging without further investigation. This aligns with the British Geriatrics Society’s recommendations for a short frailty screening, followed by a CGA for older adults identified with frailty.³⁷⁾ However, referring all positive BGA cases for CGA may create additional burden on healthcare systems. In this study, we were unable to determine the precise conditions where a CGA is unnecessary. Nonetheless, other studies have explored situations where a CGA is warranted, such as the identification of “red flags” based on the Geriatric 5Ms during a brief assessment for older adults visiting an emergency department.⁴⁹⁾ There is also potential to combine a BGA with management pathways for specific conditions when a CGA could be clinically unnecessary or for cases where older adults are not inclined to proceed with a CGA. For instance, the Rapid Geriatric Assessment was linked with assisted management pathways to screen and manage frailty, sarcopenia, anorexia of aging, and cognitive decline.⁴⁷⁾ Under the assisted management pathways, an exercise intervention could be recommended to frail or sarcopenic older adults, while cognitive stimulus therapy would be offered to older adults identified with cognitive issues.⁵⁰⁾ Preliminary research showed positive physical outcomes for sarcopenic older adults and positive cognitive outcomes for older adults with cognitive issues.⁵⁰⁾

Findings indicated that BGA should be targeted at older adults showing potential geriatric syndromes. Healthcare professionals should administer the BGA to assess older adults’ physical health, psychological/mental health, functional abilities, mobility/balance, and social situation in a community setting, while also including medication use in primary care. Detection of geriatric syndromes should lead to a CGA referral. This is a preliminary care process based on the consensuses achieved, and further research is needed to clarify BGA’s role as well as the support needed in the care process.

Participants’ consensuses suggested selective administration of BGA to identify older adults needing CGA. This raises the question on how the target group of older adults (i.e., show signs of geriatric syndromes or mild to moderate frailty) should be identified. Perhaps an even briefer screening tool is needed. Alternatively, greater public education on geriatric syndromes and frailty could prompt family members of older adults to look for the signs and bring their older adults for a BGA.

Strict adherence to the consensus of BGA administration by healthcare professionals may strain the healthcare system. Therefore, using trained personnel or self-assessments should be considered through careful selection of domain-specific instruments that are least susceptible to biased interpretation. Using trained personnel for performance assessments could allow healthcare professionals to work on other tasks requiring their clinical expertise. Alternatively, regular refresher training could address the issue of maintaining standards among non-healthcare professionals.

First, a limitation of the Delphi technique is the arbitrariness of the consensus criteria. We used dispersion of responses on top of level of agreement to add robustness to the consensus criteria. Second, a two-round Delphi meant that we were unable to further explore some issues, especially statements close to consensus. Third, we were unable to get participants to prioritize the essential elements of a BGA. This could be due to the use of a Likert scale instead of a ranking system. Alternatively, the 15 minutes time constraint may not have been sufficiently emphasized. It was only indicated at the start of the section and thus could have slipped participants’ mind as they progressed to other pages. Efforts to mitigate the first and second limitations were done via discussions between the study team and study advisors, such as considering statements that came close to consensus.³⁶⁾

This Delphi study aimed to build expert consensus on domains that are essential for a BGA, and appropriateness, feasibility and implementability of BGA in community and primary care settings in Singapore. Aside from the domains and assessments for a BGA, participants also agreed that (1) administration of BGA should be targeted, (2) BGA should be used to identify those requiring CGA, (3) interventions should only be recommended after a CGA, and (4) BGA should be administered by a healthcare professional. Implications of such recommendations and potential areas for further research were also discussed.