A systematic review was conducted according to the PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines in Checklist 1 [23] and registered in the PROSPERO (International Prospective Register of Systematic Reviews; protocol number CRD42024521363). A member of the public who was a carer of a person with dementia was involved during the conceptualization of the study protocol. Amendments made to the original protocol can be found in the revision notes of the protocol in the PROSPERO.

A comprehensive search with the assistance of an experienced medical librarian was conducted using 4 databases, such as MEDLINE (Ovid), Embase (Ovid), PsycInfo (Ovid), and SCOPUS. Both keywords and medical subject headings were included to ensure that the search was exhaustive.

We based the main categories of the search strategy on the PICOS (Population, Intervention, Comparison, Outcomes, and Study Design) framework, using the following main terms: (1) dementia, (2) BPSD, and (3) telephone interventions. The former included all types of dementia, and BPSD comprised terms specific to the individual symptoms. The latter incorporated mobile health apps, electronic consultation, or telehealth delivered through a telephone. The search strategy was inspired by 2 other previous systematic reviews with a similar methodology [12,24]. The full search strategy for each database can be found in the Multimedia Appendix 1. The studies’ extraction from the databases was performed on March 14, 2024. There was no limit on the date range applied to the publications extracted. No gray literature was sought.

Five reviewers (AA, ACS, ERS, JU, and MK) performed the title and abstract screening of each of the 4355 papers (2 reviewers per paper) in Covidence (Veritas Health Innovation Ltd) independently. Conflicts between reviewers were resolved through discussion with a further researcher (PN). Three independent reviewers (ACS, JU, and RA) performed the full-text review of each study (2 reviewers per study), and the conflicts were discussed between these reviewers. Studies that did not have a full-text version available were excluded. Those studies that did not explicitly mention the type of technology used were also removed to reduce the uncertainty of the results for this review. Neither time nor language restrictions were applied. When papers were written in a language other than English, the authors searched for an available English version. Table 1 shows the criteria used and the rationale behind each of them. Multimedia Appendix 2 has further details on the reasoning behind each selection criterion.

Risk of bias was assessed using the study design–specific tools available from the National Heart, Lung, and Blood Institute (controlled trials or pre-post study intervention scales). This tool provides a set of questions to evaluate the internal validity of studies in systematic reviews, including randomization of participants, dropout rates, and outcome measurement. Studies were classified as poor, fair, or good, depending on the number of criteria they fulfilled according to the authors’ judgments and their study design [25]. Studies were not excluded based on the outcome of the risk of bias assessment. ACS conducted these assessments, which were reviewed independently by 5 coauthors (ALN, AEO, BH, GG, and PN). This risk of bias assessment was subsequently incorporated into the GRADE (Grading of Recommendations Assessment, Development and Evaluation) certainty of evidence evaluation.

The data were extracted by ACS. Results were collected using a data extraction table in Excel (Microsoft Corporation). The final results were reviewed by the other authors independently (ALN, AEO, BH, GG, and PN). Unresolved discrepancies were arbitrated by ACS.

A narrative synthesis of the results was conducted, with the relevant study characteristics and intervention details and outcomes presented in the results section. The effect measures reported were those of each study, as well as the results of any statistical analyses. Secondary outcomes reporting usability, side effects, and acceptance of interventions by the participants were also extracted. The significance of the results was established when P≤.05. The most relevant follow-up measurements were retrieved from studies to facilitate a more concise synthesis of the results. A meta-analysis was not performed due to the heterogeneity of the study designs and lack of sufficient standardized effect measurements.

As a systematic review examining publicly accessible secondary data, ethical approval was not necessary for this study. The primary studies considered in this review carried out this process individually.

The PRISMA flowchart in Figure 1 summarizes the decision pathway for the final inclusion of the studies in this review. The electronic search of the 4 databases yielded a total of 7921 study records. After removing duplicates, 4355 papers were screened for title and abstract, of which 70 papers were included for full-text review (Figure 1). A total of 13 papers [26-38] were selected for the final analysis.

In Figure 1, other interventions refer to when the type of technology, or application of the technology in the papers did not meet the inclusion criteria of this review. Other blended interventions refer to studies that used telephones alongside other digital technologies, or included an in-person component.

Table 2 shows the characteristics of the studies included. All were conducted in high-income countries: 7 studies [26,27,29,30,36-38] in the United States, 2 studies [31,33] in Germany, 2 [32,34] in Italy, 1 study [35] in Korea, and 1 study [28] in the United Kingdom. Overall, 2 studies [26,27] used the same sample and intervention; therefore, they were merged into one for analysis. The studies were conducted between 2014 and 2024. In total, 8 [26-31,33,34,38] out of the 12 studies [26-38] were randomized controlled trials (RCTs), 2 [32,35] were nonrandomized controlled trials (ie, controlled trials without the process of randomization), and 2 [36,37] were pre-post study interventions with no control group. The sample size ranged from 20 to 440, with a total of 1685 participants.

This systematic review identified 12 studies [26-38] from 5 high-income countries that examined the effectiveness of different types of telephone interventions. Overall, 3 categories of telephone interventions were demonstrated to improve BPSD and BPSD-related informal caregiver burden. Only 1 [27] study reported on BPSD-related hospitalizations, indicating a statistically significant decrease in ED and hospital admissions. In total, 4 [28,36-38] of the 12 studies also reported on secondary outcomes, revealing good usability, acceptability, and no specific adverse effects reported by participants.

To the best of our knowledge, this is the first systematic review summarizing the effectiveness of telephone interventions in the management of BPSD by informal caregivers and patients at home. It is also the first to assess the impact of these technologies on key stakeholders—individuals living with dementia and informal caregivers—on health care resource usage. In the same way, the cost and relative simplicity of the interventions, compared to other more technologically complex approaches, could further increase the feasibility of implementation in low-resource settings and more constrained health care systems [14,48].

However, it must be considered that BPSD and BPSD-related caregiver burden are highly complex and multifactorial. While the NPI-Q and NPI-Q-D are standardized and widely used scales, they do not entirely capture the impact and change of individual symptoms or the complexity of factors that affect caregiver burden, respectively. For instance, each symptom can respond to treatment differently [49]. Similarly, some symptoms can cause more burden for caregivers than others; for example, aggressions may cause more caregiver distress than others, such as appetite changes. In the same way, there are a variety of factors that affect caregivers' burden, such as cognitive symptoms or level of social support available [1,50]

Therefore, while the NPI-Q-D provides valuable insight, it captures just some part of this experienced burden.

This review focused on quantitative measurements; therefore, a qualitative review was not feasible. However, understanding and interpreting the extent to which the different types of interventions are effective would be key to ensure that they have a clinically significant impact on complex phenomena, such as BPSD and related burdens [51,52]. This is especially important with scales such as NPI-Q or NPI-Q-D, for which currently there is no established threshold to define a clinically meaningful change [53].

Given the complexity of technology use in dementia care, qualitative findings could help enhance the interpretation of quantitative outcomes and support the development of more personalized interventions. It can also aid in identifying specific barriers or difficulties dyads may experience when using these technologies, especially those with limited health and digital literacy levels, disabilities, or language difficulties [54].

The improvements that telephone interventions demonstrated in BPSD-related caregiver burden and BPSD in care recipients, as well as the positive usability and acceptability scores, support the potential use of telephone interventions as a tool for managing these symptoms in home settings. In addition, most studies included in this review are RCTs, reducing the likelihood of confounding factors impacting the effectiveness of the various interventions, such as the dyads' ages or the relationship of caregivers with patients living with dementia.[13,55]

These interventions also have the potential to enable continuity of care for many families, given the flexibility and accessibility these treatments provide for patients with dementia, informal caregivers, and health care professionals. These services may also complement existing in-person support and contribute to a more holistic and equitable model of dementia care [56].

However, this review also highlights important gaps that require further exploration.

First, all studies from this review emanated from high-income countries. As such, it is crucial to interpret these findings according to the context of the health care system within each place, as factors, such as health and social care professionals’ knowledge about dementia, staff and resource availability, and culture can play a key role in the effectiveness of these interventions. [2,57]

More studies focusing on LMICs are needed to ensure the applicability of findings in diverse health care settings. One way in which this could be achieved is through intercountry collaborations, which could support research centers in these regions in having necessary funding and resources to conduct these trials. These efforts could be grounded in strong community engagement, involving dyads, local professionals, and institutions from LMICs. [58,59] Developing telephone co-interventions with stakeholders in these communities will be key to ensuring they are accessible, acceptable, and effective in different contexts from those in these studies [60] In addition, future systematic reviews could incorporate studies conducted in LMICs by expanding searches or the inclusion criteria of the interventions, to include supplementary databases.

The long-term effectiveness of these interventions warrants further research, particularly given the importance of continuity of care in dementia and the current fragmented nature of dementia care.[61,62] More RCTs with longer follow-up periods will be key to evaluating the sustainability and feasibility of these interventions at scale. Similarly, cost-effectiveness analyses are needed to support future policy recommendations. This could be achieved by tracking changes in hospital admission rates and associated costs when dyads use these tools and transparently illustrating the costs of these interventions.[63]

Finally, telephone interventions to manage BPSD in the community have the potential to help achieve the "Global Action Plan on the Public Health Response to Dementia" [21], which aims for at least 75% of families worldwide to receive dementia training and support. They can also become an essential component of national digital transformation programs, such as the NHS long-term plan [22]. However, it is important to tailor the interventions according to the needs, resources, and health care staff available in each region. For example, in lower-income settings, the telephone platforms could prove more helpful because they are asynchronous. [64] Patients and caregivers would not require internet access or the immediate availability of a health care professional in order to benefit from these interventions. Investing in large-scale research on these interventions could be highly efficient in leveraging these mainstream technologies within health care systems to achieve more equitable dementia care both within and across countries.[65]

Telephone interventions delivered through psychosocial and educational calls and online platforms are promising tools for reducing BPSD-related caregiver burden. Personalized telephone interventions, including patients’ and informal caregivers’ input in the treatment plan, may improve community-dwelling patients’ BPSD severity and frequency. However, the certainty of evidence for both outcomes was low; therefore, these findings should be interpreted with caution. Further high-quality research of the interventions in low- and middle-income countries with longer follow-up periods, including cost-effectiveness analyses and greater consistency in intervention design and outcome measurement, is required to establish the global generalizability of these interventions and inform future practice.