The global prevalence of dementia is projected to increase substantially, from 57.4 million cases in 2019 to 152.8 million in 2050, primarily due to population growth and an aging population, with a greater proportion of cases occurring among women than men [1]. International bodies such as Alzheimer’s Disease International [2] and the International Association of Gerontology and Geriatrics [3] have advocated for active case-finding of dementia among high-risk older adults. This involves extending beyond tertiary hospitals to proactively identify high-risk individuals through routine annual cognitive evaluations in community-dwelling older adults [3]. However, the potential benefits of these cognitive evaluations in the community are diminished when individuals do not pursue timely medical follow-up after receiving a positive case-finding result for cognitive impairment (ie, mild cognitive impairment [MCI] or dementia). Timely medical follow-up is essential, not only for initiating appropriate medical treatments (eg, cognitive enhancers), but also for accessing comprehensive care management (eg, behavioral, financial, legal, and end-of-life issues), and social support (eg, support for caregivers) [4]. Delayed follow-up can have far-reaching consequences, including compromised well-being [5,6] and increased risk of early institutionalization in long-term care facilities [7-9].

Despite these risks, little is known about the frequency of actual medical help–seeking following a diagnosis of cognitive impairment. A large national survey conducted in China between 2015 and 2018 revealed that 71.4% (1974/2766) of individuals with dementia and 97.2% (6926/7125) of individuals with MCI did not seek medical treatment [10]. In Singapore, unpublished data from a pilot case-finding study [11] by the corresponding author (TML) as part of a larger implementation project to train and implement PENSIEVE-AI, an artificial intelligence (AI) tool for early cognitive impairment detection in the community, further illustrate this challenge. Among community-dwelling older adults diagnosed with cognitive impairment and referred directly to the Singapore General Hospital (SGH) memory clinics, 20.8% (10/48) defaulted on their appointments. Of these individuals, 8 were diagnosed with MCI and 2 with dementia. These missed follow-ups occurred despite deliberate efforts to facilitate medical follow-up, including direct referrals to SGH memory clinics for individuals living near SGH, text message appointment reminders, and proactive coordination by a clinic case manager before follow-up appointments. Although these data were derived from a single, relatively well-resourced acute tertiary hospital in Singapore, they likely underestimate the extent of the default follow-up rates in the broader population. Together, these findings highlight a persistent and multifaceted challenge of timely medical follow-up after receiving a diagnosis of cognitive impairment. Addressing this gap requires a better understanding of the determinants of medical help–seeking behavior in this context.

A systematic review [12] identified 5 key illness perception factors associated with delayed help-seeking among individuals with dementia and their caregivers, including cultural beliefs of caregiving duties, acceptance of diagnosis, lack of knowledge about dementia, complex health care systems (including both the positive and negative experiences from prior health care professionals and services), and perceived threat to personal independence. However, the review focused on help-seeking intentions rather than actual help-seeking behaviors. This distinction is important due to the well-established intention-action or intention-behavior gap in health behaviors [13-15]. For clarity, throughout this paper, help-seeking without the qualifier medical refers to both formal (ie, health care professionals) and informal (ie, family and friends) sources of help-seeking support. In contrast, medical help–seeking behavior refers specifically to the actual actions taken to seek and obtain care from formal health care providers, such as neurologists, psychiatrists, or general practitioners.

Building on this, a recent qualitative study [16] extended medical help–seeking intentions in individuals with dementia and identified 8 key determinants of medical help–seeking behavior among individuals with MCI. These determinants can be conceptually clustered into 3 broad categories, including psychological processes, comprising disease knowledge, perceived disease threat, symptom attribution, negative valence emotions, and reliance on cognitive strategies; socioeconomic and structural conditions, that is, financial stability and accessible health care services; and social support. However, participants in this study were recruited from a memory clinic and were already on medical follow-up. As such, the determinants influencing individuals with MCI who did not seek medical help remain underexplored, leaving an important unaddressed gap in the context of medical help–seeking following positive case-finding in the community.

To address these methodological and practical knowledge gaps, this qualitative study aims to explore the barriers and facilitators influencing medical help–seeking behavior among individuals diagnosed with MCI or dementia following positive community-based case-finding results in Singapore. Specifically, what are the barriers and facilitators to medical help–seeking behavior among community-dwelling individuals diagnosed with cognitive impairment following a positive case-finding result? This study forms part of a larger implementation effort involving the deployment of PENSIEVE-AI but is methodologically independent from the PENSIEVE-AI tool development study [11]. In other words, although the participants in this qualitative study were recruited from the PENSIEVE-AI pilot study [11], this study focuses specifically on exploring the determinants of medical help–seeking after MCI or dementia diagnosis. Given the complex, multifactorial nature of medical help–seeking behavior, a theory-informed approach was adopted to guide data collection and analysis. This approach enables the systematic identification of key determinants grounded in an established theoretical framework to inform the future development of targeted interventions and implementation strategies to improve medical help–seeking follow-up [17].

This study applies the capability, opportunity, motivation-behavior (COM-B) framework [18,19], a widely used systematic framework for understanding behavior, designing targeted interventions, and achieving health behavior change across diverse settings, including dementia screening [20]. Within this framework, capability refers to the physical and psychological ability to perform the behavior; opportunity pertains to the physical and social environments; and motivation encompasses both automatic processes, such as emotions and impulses, and reflective processes, such as evaluation and plans [18,19]. Its utility lies not only in identifying behavioral determinants but also in linking the COM-B diagnosis with the 9 intervention functions that encircle the COM-B framework of the behavior change wheel (BCW): education, persuasion, enablement, incentivization, coercion, training, restrictions, environmental restructuring, and modeling [18,19,21]. This allows for the systematic development of tailored strategies specific to the behavioral determinants [18,19].

This study arose from a recent pilot case-finding study in Singapore that screened approximately 1800 community-dwelling older adults for early cognitive impairment using the PENSIEVE-AI tool [11]. Multimedia Appendix 1 outlines the steps of the PENSIEVE-AI case-finding pilot study. Community-dwelling older adults were recruited and underwent comprehensive cognitive assessments. Assessments of MCI, dementia, or normal cognition were made by 3 dementia specialists using an internationally recognized gold-standard consensus approach based on all the information from the comprehensive assessment. Community-dwelling older adults living in the SGH catchment area with positive case-finding results for MCI or dementia received direct medical appointments for follow-up at SGH memory clinics, whereas those living outside the SGH catchment area were strongly advised to seek medical attention from primary care doctors for referral to memory clinics within their home vicinity.

Forty-one eligible PENSIEVE-AI participants diagnosed with MCI or dementia and their corresponding caregivers from the pilot case-finding study [11] were contacted for the semistructured interviews over the phone. Participants who declined to participate did not preclude their caregivers from participating in the study. Interviewees were selected using maximum variation sampling to capture diverse experiences of medical help–seeking following positive case-finding. Sampling variation was based on participants’ age, gender, ethnicity, living arrangement, relationship with caregiver, prior memory concerns, and whether participants had sought medical follow-up after their PENSIEVE cognitive impairment diagnosis. These characteristics were selected because they are expected to influence the cognitive impairment appraisal, access to formal and informal support for medical help–seeking, and caregiver involvement in decision-making regarding medical help–seeking. Figure 1 details the recruitment and selection process.

For participants diagnosed with dementia, only their caregivers were interviewed due to practical considerations such as impaired decision-making capacity, reliance on interviewer’s skill for consistent self-reports, and risk of distress during lengthy interviews. For participants diagnosed with MCI, interviews were conducted with participants alone, with participants and caregivers together, or with caregivers only, depending on participant preferences. This strategy aimed to improve interview participation rates. Interviews were conducted in either English or Mandarin Chinese. All interviews were audio recorded, anonymized, transcribed verbatim, and verified by another author (SRL or XX). Mandarin Chinese interviews were transcribed verbatim in the source language, coded in English, and only the extracts were translated into English by SRL and verified by XX. All authors are bilingual and proficient in English and Mandarin. All interviews were conducted by SRL, with XX accompanying one interview when a male participant was home alone. Interviews continued until data saturation was reached. An interview guide based on COM-B [18] was developed and used for the semistructured interviews (Multimedia Appendix 2). The COREQ (Consolidated Criteria for Reporting Qualitative Research) checklist was used as a guide for reporting our study [22]. Informed consent was taken at the start of the interview.

The data were analyzed using framework analysis [23] involving the following steps: (1) familiarizing with the interviews and transcripts; (2) independent inductive coding of the transcripts (ie, open coding) using Microsoft Word’s comments function and the coding macro [24] to export the codes and the associated text to Microsoft Excel; (3) grouping codes into defined facilitator and barrier categories; (4) indexing the transcripts according to the existing codes and categories; (5) charting the categories into the COM-B matrix, including quotation references; and (6) performing data interpretation.

The identified COM-B components were used to guide the selection of relevant intervention functions from the BCW. The coding, indexing, charting, and selection were performed by SRL and XX, who have respective backgrounds in psychology and sociology. The categories were developed together through discussion with TML, who has expertise in geriatric psychiatry and public health, and subsequently reviewed by the remaining authors with expertise in geriatric medicine, medicine, health services research, and public health. Disagreements in the categorization of barriers and facilitators, mapping to COM-B components, and selection of intervention functions were resolved through discussion. We maintained an audit trail of coding comments, categorization and selection decisions, and engaged in reflexive discussions throughout the analysis.

This study was approved by the SingHealth Centralized Institutional Review Board (reference number 2021/2590). Informed consent procedures for participants with cognitive impairment included a simplified capacity assessment in accordance with the Mental Capacity Act. Participants were required to demonstrate their ability to understand, retain, and use the study information to make an informed decision, and to communicate their decision to participate by paraphrasing key information, including the voluntary nature of participation and the right to withdraw at any time. The interviewer remained attentive to verbal and nonverbal cues of distress and paused interviews when necessary to ensure participants’ and caregivers’ well-being and to accommodate caregiving responsibilities. Participants and caregivers were reminded that they could terminate the interview at any time, and breaks were offered when needed. Data were kept confidential, with access restricted to the study team. Interview data were anonymized during transcription, with identifiable information removed and pseudonyms assigned where appropriate. Participants received Singapore $40 (US $1=1.26 Singapore Dollars) as a token of appreciation.

Between May 2023 and January 2024, 26 individuals were interviewed across 21 unique households, comprising 12 households with a family member diagnosed with MCI and 9 households with a family member diagnosed with dementia. The mean interview duration was 53 minutes, ranging from 25 to 71 minutes. Of the 21 household interviews, 17 were conducted in person, and 4 by telephone. The median age group for participants with MCI was 70‐79 years, while the median age group for the interviewed caregivers was 60‐69 years. Most interviewees were Chinese, living with family members in a 4‐5 room public housing apartment. Notably, 23.8% of households did not take further medical follow-up action after the positive findings from the pilot case-finding program, with 25.0% for MCI households and 22.2% for dementia households. Detailed sociodemographic characteristics of the interviewees are presented in Table 1. Five interviews were conducted in Mandarin Chinese.

Based on the in-depth interviews, we identified 15 determinants encompassing both facilitators and barriers of medical help–seeking behavior. These determinants were mapped onto the COM-B domains. Detailed descriptions of each determinant, along with illustrative verbatim quotes, are detailed in the sections below. Figure 2 provides an overview of these determinants as mapped onto the COM-B domains and subdomains.

This qualitative study identified 15 barriers and facilitators to medical help–seeking behavior following a diagnosis of MCI or dementia, categorized using the COM-B model. The discussion first contextualizes the findings within the existing literature, followed by integrated evidence- and practice-based recommendations.

Our analyses on inaccurate knowledge about MCI and dementia, inability to recognize the symptoms of MCI and dementia, complicated health care processes, and low health literacy to navigate the health care processes aligned with commonly reported themes in the literature, including normalization of problems, lack of knowledge about dementia, and the complexity and inaccessibility of health care services [12,16,29,30]. Lay beliefs about seeking medical care when sick mirrored the finding of the review by Hill et al [30] on the perceived lack of benefit in medical help–seeking. Both reflect mismatches between clinical recommendations and the lay understandings that can undermine the perceived value of medical treatment, highlighting the need to increase public awareness of the benefits of timely medical intervention and care.

Strong affective states triggered by positive case-finding results were consistent with the findings of Jiao et al [16], who noted that individuals diagnosed with MCI perceived the disease as inherently threatening, characterized by negative valence emotions, such as anxiety and fear regarding disease progression. While Devoy and Simpson [29] identified fear as a barrier to medical help–seeking, our findings suggest a more nuanced, dual role, in which fear, conceptualized as a strong affective state triggered by the case-finding results, functions as both a facilitator and a barrier (eg, Extract 32). This duality reflects the multidimensional nature of fear, encompassing affective, cognitive, and psychobiological dimensions [31], with each influencing medical help–seeking behavior in different ways [32]. In the context of cancer screening, for instance, the cognitive and affective dimensions of fear facilitated cancer screening participation, while the psychobiological dimension hindered it [32].

A unique barrier identified in this study was the perceived inaccuracy of the case-finding tool. As participants were recruited from the larger case-finding pilot program using an AI tool, concerns about the tool’s accuracy arose despite participants receiving diagnoses through the gold-standard consensus method. Future implementation of the PENSIEVE-AI tool [11] will need to explicitly address accuracy concerns and clarify the role of the AI tool in the cognitive impairment diagnosis process, particularly given that the consensus approach by memory specialists is not typical in routine clinical practice.

Medical help–seeking was further complicated by anosognosia in individuals with dementia and symptom fluctuation among individuals with MCI. Anosognosia, the lack of insight into one’s own cognitive impairment, is a characteristic of dementia [33] that may contribute to denial and resistance to medical follow-up. This is evident in Extracts 10 and 41, where a caregiver described how her mother was in denial and misattributed cognitive symptoms to normal aging, features potentially indicative of the presenting symptoms of anosognosia. Similarly, among individuals with MCI, a clinically and neuropathologically heterogeneous group [34], the progression to dementia varies; while approximately 46% of individuals with MCI aged 65 years and older progress to dementia within 3 years [35], rising to 65% for those aged 75 years and older [36], others reported symptom improvement [37,38]. This symptom fluctuation makes early signs of cognitive impairment harder to recognize, contributing to uncertainty and delays in seeking medical help, as reflected in Extracts 7, 11, and 29.

Despite the identified barriers, several facilitators supported medical help–seeking behaviors. These included the ability to recognize cognitive impairment symptoms and the adoption of practical strategies to navigate symptom-related challenges, such as tracking medical appointments. Strong social capital, particularly reliance on caregivers and friends to reschedule medical appointments, provide transportation, and accompany individuals to medical appointments, facilitated medical follow-up behavior. Additionally, caregivers with profamily workplace policies and the case-finding result letter served as a physical cue to prompt medical follow-up action. For some individuals diagnosed with cognitive impairment, the strong affective states triggered by the case-finding results also prompted medical follow-up, particularly among those who valued proactive disease management through medical professional monitoring of disease progression. These facilitators align with the literature highlighting the importance of positive social support, disease knowledge, and the value of early detection in promoting medical help–seeking [16,29,30].

Our findings contrast with the paradoxical finding of Jiao et al [16] that individuals with MCI were less likely to seek medical help when using cognitive strategies. Our study found that individuals with MCI actively adopted strategies to track medical appointments, reflecting problem-solving efforts by the diagnosed individuals and their families to manage cognitive impairment symptoms. Notably, those with strong social capital and support networks were able to overcome some of the current barriers identified in this study, including physical impairments, low literacy to navigate health care processes, inadequate physical infrastructure, complicated health care processes, and lay beliefs about seeking medical care when sick. Facilitators unique to this study included caregivers with profamily workplace policies and the case-finding results letter as a cue for action. The former strengthened social capital, while the latter is a distinctive finding in this qualitative study on determinants of medical help–seeking behavior following a cognitive impairment diagnosis via a community-based AI case-finding tool.

Our findings extend the Alzheimer disease (AD) system framework of Ball et al [39] by specifying the barriers and facilitators influencing medical help–seeking behavior following a cognitive impairment diagnosis. These determinants can be situated within the AD system framework, between “health care system patient flow” and “access to support, care, and delivery,” specifically following the “timely and accurate diagnosis” node [39]. Notably, current results demonstrate that the pathway from cognitive impairment case-finding or diagnosis to medical help–seeking is nonlinear. Instead, it is a process punctuated by interrelated determinants at multiple junctures, through a series of interactions between individual capacities and motivations, physical and social dynamics and opportunities, and broader structural conditions.

Although COM-B provides a systematic structure for identifying these determinants, several exhibit cross-domain interdependence. For instance, while physical impairments hindering medical follow-up are categorized under the physical capability subdomain, this barrier is closely linked to the social opportunity subdomain; many participants often relied on caregivers for transportation and company, a form of social capital that is both an enabler and a barrier. Such overlaps highlight the complexity and dynamic nature of medical help–seeking behavior following cognitive impairment case-finding results. Recognizing these interplays is essential for a nuanced understanding of medical help–seeking behavior and for designing interventions that address barriers within their broader systemic and contextual realities rather than in isolation.

In view of this complexity, we developed actionable recommendations by mapping BCW intervention functions onto our analytic findings. In addition to barriers, facilitators across the COM-B domains were integrated to bolster support for timely medical follow-up among individuals with cognitive impairment. Each determinant was mapped to one or more BCW intervention functions, as detailed in the matrix in Table 2. Training, coercion, and restrictions of the behavior change intervention functions were excluded; training addresses knowledge and skill gaps that are already covered by the education and enablement intervention functions, while coercion and restrictions are not suitable in this context. Further details of the recommendations are provided in Textbox 1.

The education behavior change intervention function addresses barriers related to psychological capability, social opportunity, and both automatic and reflective motivation. Consistent with Ball et al [39], educational content should be tailored to different audiences (ie, the general public, consumers, and health care professionals). Points 1-3 outlined in Textbox 1 apply across all audiences, whereas Point 4 targets the general public and health care professionals. Building on evidence showing improved public dementia knowledge and attitude [40], education efforts, public campaigns, and media should evolve beyond depicting individuals with severe dementia to include individuals living with MCI and early stages of dementia. In doing so, they can illustrate how individuals adapt and continue to live positively and meaningfully despite cognitive decline. This shift will enable the public to gain a more accurate understanding of early cognitive impairment, emphasizing that dementia is more than just memory loss; it also involves language difficulties, problem-solving challenges, and changes in mood and behavior. By providing a broader perspective, it can increase psychological capability and reflective motivation, thereby encouraging timely medical help–seeking behavior even in the early stages of the disease.

In tandem with education, persuasion strategies can activate both automatic and reflective motivation processes. For individuals reluctant to seek medical help after receiving positive case-finding results, carefully designed result letters may serve as a behavioral nudge, drawing on psychological and behavioral science principles, such as nudges [41-43] and loss-framed messaging that emphasizes the negative consequences of delaying medical follow-up [44]. Point 2, engaging the public through expert figures and community leaders to offer insights and perspectives on early cognitive impairment, is a powerful persuasion strategy that draws on the principle of authority to engage the public and encourage medical help–seeking behavior by deferring to the experts (eg, memory specialists) [45,46], and positioning these experts as catalysts for behavior change.

The enablement intervention function is necessary to reduce structural and procedural barriers. Our study supports simplifying the referral pathways; participants receiving direct referrals to SGH memory clinics achieved a 100% attendance rate (ie, for those living in the SGH vicinity, n=10), compared to 54.6% for those with indirect referrals to memory clinics within their home vicinity (ie, participants living outside of the SGH vicinity, n=6). The differences point to the value of simplifying the referral process to reduce cognitive load and barriers in navigating the health care system. At the organizational and policy level, establishing clear practice guidelines for referrals from primary care to memory clinics would reduce practice variations across the health care system and improve the opportunity to timely access medical care.

Finally, environmental restructuring is essential to address physical opportunity barriers, particularly for individuals living with mobility challenges. Limitations in Singapore’s public transport can hinder attendance at follow-up medical appointments, particularly during peak commuting hours when public transport is crowded and less accessible. For instance, the Singapore public buses are only able to accommodate one or 2 wheelchair users at a time [47], requiring additional wheelchair users to wait for the next available bus. Environmental restructuring and structural modifications, such as increasing wheelchair capacity on buses and improving clinic accessibility, would not only ease travel burdens but also signal a societal commitment to inclusion by proactively supporting the diverse needs of people living with cognitive and physical impairments.

This study has several limitations. First, the sample was drawn from participants in the PENSIEVE-AI study [11], raising the possibility of self-selection bias whereby participants may have been more inclined to follow up on their cognitive impairment. To mitigate this, we purposively included individuals who had not sought medical follow-up and expressed no intention to do so unless their symptoms became more severe (eg, difficulty returning home using a familiar route). Second, the sample lacked ethnic diversity, with limited representation from Malay and Indian minority groups relative to the Chinese majority group. Third, the long intervals between memory clinic follow-ups, typically 6-12 months, introduce uncertainty regarding continued long-term medical follow-up, particularly for individuals with MCI whose symptoms are typically less severe and may be perceived as less disruptive than those associated with dementia.

A further limitation concerns the exclusion of individuals with dementia from direct interviews. Although based on ethical and practical considerations regarding potential distress and communication challenges, this warrants critical reflection on the representation gap inherent in relying on proxy accounts. This reliance placed emphasis on caregivers’ experiences and perspectives in identifying barriers and facilitators of the medical help–seeking journey following dementia diagnosis, inadvertently glossing over the affective experiences and personal autonomy of the individuals with dementia themselves. In particular, their personal decision-making in medical help–seeking, sense-making of the dementia diagnosis, and challenges in navigating health care services, as well as the interactions with caregivers and health care professionals may be underrepresented, despite being highlighted in the broader dementia literature [48-50].

Prior research emphasizes the importance of including individuals with dementia to address epistemic injustice and support person-centered approaches in research [51-56]. While caregiver accounts were sufficiently informative for understanding the barriers and facilitators to medical help–seeking in this study, we acknowledge the importance and value of including the perspectives of individuals with dementia. Future studies should adopt more inclusive and adapted approaches, such as process consent, to ethically and effectively engage individuals with dementia. Finally, evaluating the feasibility and effectiveness of our recommendations was beyond the scope of this qualitative study. Future research should apply the acceptability, practicability, effectiveness, affordability, side-effects, and equity (APEASE) criteria [19] in collaboration with stakeholders, including policymakers and community representatives (eg, patient advocates, representatives from caregiver organizations, community health care workers, or volunteers), to refine and evaluate these recommendations.

Despite these limitations, this study makes several important contributions. First, while prior research has primarily focused on medical help–seeking intentions [12,29], this study explored determinants of actual medical help–seeking behavior following a cognitive impairment diagnosis. Second, by including individuals with MCI and the caregivers of individuals with MCI or dementia who had not sought medical follow-up, we addressed a key sampling gap in earlier qualitative work [16]. This provided a more nuanced understanding of the determinants of medical help–seeking behavior, particularly among those not yet engaged in follow-up care. In doing so, this study extends the AD system framework [39] by highlighting the nonlinear processes involved in seeking medical treatment.

This study identified key barriers and facilitators to medical help–seeking behavior following a cognitive impairment diagnosis, demonstrating the dynamic and nonlinear processes in accessing medical treatment. Building on these findings, we developed actionable recommendations grounded in the COM-B framework and the BCW intervention functions. We advocate for a whole-of-society approach, with implications extending to policymakers, primary care providers, tertiary hospitals, and the transport and media authorities. To maximize the impact of these recommendations, ongoing collaboration with stakeholders and careful evaluation using the APEASE criteria are needed to address practical considerations and promote equity in facilitating timely medical follow-up after positive case-finding results or diagnosis.