Effectiveness of Digital Health Interventions on Body Weight and Dietary Intake Outcomes Among Culturally and Linguistically Diverse (CALD) and Indigenous Populations: A Systematic Review

ABSTRACT

Introduction

Digital health interventions are effective for weight management and improving dietary intake, but studies in culturally and linguistically diverse (CALD) and Indigenous populations are limited. The aim of this systematic review is to evaluate the effectiveness of digital health interventions on body weight and dietary intake outcomes in CALD and Indigenous populations.

Methods

MEDLINE, Embase, Scopus, and Cochrane databases were searched on December 28, 2022 (PROSPERO: CRD42023394058). Inclusion criteria were randomized controlled trials (RCTs) conducted in high‐income English‐speaking countries with free‐living adults ≥ 18 years. Trials had to report both weight and dietary outcomes, with ≥ 50% participants from CALD/Indigenous backgrounds or outcomes reported by race/ethnicity. Two reviewers independently screened records. Risk of bias was assessed using the Cochrane RoB 2 tool. Results were synthesized descriptively and presented in graphs and tables.

Results

From the 1984 records identified, nine RCTs were included, which involved a total of 2716 participants. Eight trials were conducted in the United States, and only one trial included Indigenous participants. Significant body weight changes occurred in three trials. Significant diet quality changes occurred in three trials. Most trials had high retention rates (≥ 80%) but low intervention adherence (< 50%). Risk of bias was low for most trials.

Conclusion

Limited evidence supports the effectiveness of digital health interventions for improving body weight and dietary intake outcomes in CALD and Indigenous populations. The predominance of US‐based trials, female‐dominated samples, and hybrid intervention designs limits generalizability. Future research should prioritize inclusive practices and standalone digital designs to establish effectiveness in these populations.

Abbreviations

BMI

body mass index

CALD

culturally and linguistically diverse

DASH

Dietary Approaches to Stop Hypertension

HDS

Healthy Diet Score

HEI

Healthy Eating Index

IDVD

interactive digital video disc

IVR

interactive voice response

PICOS

Population, Intervention, Comparison, Outcomes, Study

PRISMA

Preferred Reporting Items for Systematic Reviews and Meta‐Analyses

PROSPERO

Prospective Register for Systematic Reviews

RCT

randomized controlled trial

SD

standard deviation

SMS

short message service

95% CI

95% confidence intervals

1. Introduction

The increasing prevalence of diet‐related chronic diseases, including cardiovascular disease and type 2 diabetes mellitus, is a significant global public health concern [1]. Weight management and diet quality play a crucial role in managing and preventing chronic diseases [2, 3, 4]. Due to social and economic disparities, priority populations are at greater risk of poorer diet quality and obesity, making them more likely to be affected by multiple chronic conditions throughout their lifetime [5, 6]. Among the most recognized priority population groups impacted are culturally and linguistically diverse (CALD) and Indigenous populations in high‐income countries where the main language spoken is English [5, 7]. Digital health interventions, which leverage digital technologies, such as mobile health apps, telephone support, SMS, and the Internet, to create, store, process, and communicate health‐related information [8, 9, 10], hold significant potential to promote health equity for these populations [11, 12]. This potential stems from the abilities of digital technologies to address barriers related to health system factors and care delivery, such as geographic isolation and transportation barriers through remote delivery and language challenges through multilingual interfaces [11, 12, 13]. The use of digital health interventions has been previously shown to improve body weight and dietary intake outcomes; however, this has been primarily demonstrated in non‐CALD and non‐Indigenous populations [14, 15]. Considering this demonstrated effectiveness, digital health interventions may prove similarly helpful in addressing the heightened diet and weight‐related health challenges facing CALD and Indigenous populations.

Digital health intervention trials conducted in CALD and/or Indigenous populations are limited [16], with very few identified in reviews of the literature [17, 18, 19, 20]. A scoping review of Indigenous studies [17] and three systematic reviews, which either exclusively focused on CALD populations [19] or included studies where most participants were from CALD populations [18, 20], showed inconsistent results for weight loss, dietary intake, and/or physical activity outcomes [17, 18, 19, 20]. The varied outcome measures investigated in these reviews also make it challenging to interpret and compare dietary intake and weight change findings. For example, Bennett et al. [19] only assessed weight changes, Myers‐Ingram et al. [18] evaluated changes in weight and physical activity, and Karimi et al. [20] focused exclusively on dietary behaviors and related health outcomes. Given the role of both body weight and dietary intake in chronic disease development, particularly among CALD and Indigenous populations, evaluating the impact of digital health interventions on diet and weight outcomes concurrently in these population groups is needed. The aim of this systematic review is to evaluate the effectiveness of digital health randomized controlled trials on both body weight and dietary intake outcomes among CALD and Indigenous population groups.

2. Materials and Methods

The protocol for this systematic review was registered with the International Prospective Register for Systematic Reviews and is publicly available (PROSPERO, registration number: CRD42023394058). This review was conducted using the guidelines outlined in the Preferred Reporting Items for Systematic Reviews and Meta‐Analysis (PRISMA) [21]. The completed PRISMA checklist is available in the Supplementary Information.

2.1. Eligibility Criteria

This review included trials on digital interventions in adults from CALD or Indigenous populations published in English. Eligibility criteria were structured using the PICOS framework: Population, Intervention, Comparison, Outcomes, and Study design [21].

2.2. Population

Trials were required to be conducted in high‐income countries where the main language is English (Australia, United States, United Kingdom, Ireland, New Zealand, and Canada). Trials were included if participants were free‐living adults over the age of 18 years, and a minimum of 50% of the participants were from a CALD and/or Indigenous background. If less than 50% of the sample was from a CALD and/or Indigenous background, trials that reported on desired outcome measures by race or ethnicity were eligible for inclusion. No restrictions on health status were applied; however, trials were excluded if the participants were pregnant, 6‐months postpartum, or had undergone bariatric surgery.

Both CALD and Indigenous populations were included in this review as they both commonly experience healthcare and service disparities. However, the terms “CALD” and “Indigenous” were kept separate as each group faces distinct challenges and has unique needs related to their cultural, linguistic, and ethnic backgrounds [22]. Definitions of CALD populations vary across different contexts and sources [22, 23]. For the purposes of this review, the term CALD populations refers to groups and individuals who “differ from the English‐speaking majority in terms of culture, language, race, religion, values, or beliefs.” [23] Despite this term being more popularized in Australia [22, 23], it was selected to avoid the hierarchy and deficit‐based perspective often associated with other terms [24]. The term Indigenous populations refers to groups and individuals who have “a historical continuity with pre‐invasion and pre‐colonial societies that developed on their territories, consider themselves distinct from other sectors of the societies now prevailing on those territories, or parts of them.” [25]

2.3. Intervention Types

Eligible trials used digital health technologies, including digital platforms, apps, devices, telecommunications, or other related electronic tools, to deliver healthy lifestyle interventions. These technologies were required to be interactive to facilitate the active exchange of information and enable active participant involvement in targeted lifestyle behaviors. Trials were excluded if the intervention did not use interactive digital health technologies or provided food, financial support, or supplements/medications that could impact dietary behaviors and body weight.

2.4. Comparator

Trials were required to include a control group that received either no intervention or standard care that involved no use or minimal use of digital technologies.

2.5. Outcomes

The outcomes of interest were changes in both dietary intake (energy, nutrients, food groups, or diet quality scores) and body weight (weight or BMI). Outcomes were required to be assessed at baseline and a minimum of one additional time point post‐intervention. Additionally, given associations with improved health outcomes, this review evaluated whether trials reported clinically significant weight loss among participants, defined as a 5% or more loss in body weight [26]. Intervention adherence, referring to the extent that participants follow prescribed procedures outlined by the intervention [27], and participant retention were also assessed as reported by trials.

2.6. Study Design

Only randomized controlled trials (RCTs) written in English with the aim of investigating the efficacy of digital healthy lifestyle interventions were included. Review and observational articles were excluded.

2.7. Search Strategy

An initial search was conducted following electronic multidisciplinary databases and trial registries MEDLINE, Scopus, Embase, and Cochrane on December 28, 2022, with no publication date restriction imposed. The search was rerun on August 5, 2024, with publication date restrictions applied to include only trials published since the initial search. No additional trials were identified. Trial registries were included among the databases searched to capture any relevant unpublished RCTs and to be made aware of any relevant RCTs that have been completed with results and are awaiting publication.

Databases were searched using a combination of six search term groups. Relevant keywords and subject headings were used to identify the search term groups. Searches were completed using Boolean search techniques, such as “AND” and “OR,” based on the PICOS framework (see Supplementary Information). The six search term groups were: (1) technology, (2) dietary intake, (3) body weight, (4) adult, (5) cultural population/s, and (6) randomized controlled trial. A Health Sciences librarian was consulted during the search strategy development to verify the search terms, strategies, and results. The detailed search strategy for each database, including all keywords and filters, is provided in the Supplementary Information.

2.8. Data Collection and Analysis

2.8.1. Study Selection

Records were imported from the selected databases into Covidence (Veritas Health Innovation Ltd., Melbourne, Australia), a Web‐based software platform for systematic review management [28]. Once imported, automatic and manual deduplication was undertaken, with two independent reviewers screening the remaining records. Records were initially screened by title and abstract and then by full text. The reasons for exclusion were collated, and any conflicts between the reviewers were resolved through discussion and re‐examination of the study. If consensus could not be achieved, a third author was available to resolve conflicts. The references of the included trials were manually screened for further relevant trials. A PRISMA flow diagram was used to report the screening process [21] (Figure 1).

FIGURE 1.

PRISMA flowchart.

2.8.2. Data Extraction

Extraction was completed using Covidence software. The data extracted included study details (author, year of publication, country of study, duration), participant characteristics (sample size, baseline characteristics, retention, CALD, and Indigenous population), intervention characteristics for all intervention arms, and all dietary intake and body weight outcomes at baseline and at further assessed time points. The same two reviewers independently extracted the data. A third reviewer was available to resolve any discrepancies.

Participant and intervention characteristics were extracted and presented in summary tables for all included trials. Due to the limited number of trials included and substantial heterogeneity in intervention design, outcome measures, and follow‐up periods, meta‐analysis, subgroup analyses, and sensitivity analyses were not feasible. To help interpret results and simplify comparisons between trials, forest plots were produced in Microsoft Excel [29] to visually present the effect sizes and 95% CIs of dietary intake and body weight data across trials. Established methods were used to calculate mean differences and 95% CIs in cases where different statistical summary measures were provided for dietary intake and body weight outcomes [30, 31]. Average changes observed in outcome variables within each group were used to calculate mean differences between groups, and the standard error of difference was calculated for the determination of 95% CIs of difference.

Standardized outcomes were applied to heterogeneous dietary intake and body weight measures across trials to enhance the interpretability of assessed outcomes further. Diet quality outcomes evaluated using different scoring systems were standardized based on their respective scales and presented as percentage differences to reflect relative changes in diet quality between intervention and control groups. Fruit and vegetable intake in cups, or cup equivalents per day, was converted to standard daily servings. As most trials were conducted in the USA, this unit conversion was guided by the US dietary guidelines where the quantity equivalent of one cup/cup equivalent equates to two servings of fruit and vegetables [32, 33]. Body weight outcomes evaluated in pounds were converted to kilograms (kg).

2.8.3. Quality

The same two reviewers independently assessed the risk of bias in eligible trials. Risk of bias was assessed using the Cochrane Risk of Bias (RoB 2) tool for randomized trials [34]. Assessments were presented using traffic light plots showing domain‐level judgments for each study. Assessments of publication bias and formal certainty of evidence were not conducted due to the limited number of trials included.

3. Results

3.1. Study Selection

In total, 3081 studies were identified in the initial search. After 1097 duplicates were removed, 1984 remained for abstract and title review. A total of 1772 studies did not meet the inclusion criteria and were excluded. The full texts of 212 studies were then assessed, and 11 studies were determined to be eligible. Three of the 11 studies indicated that they had evaluated the outcomes of interest but had not published related data. The authors were contacted to obtain outcome data; however, the data were retrieved for only one study [35]. Consequently, the remaining two studies were excluded [36, 37]. One of the 11 studies was a post hoc analysis of collected dietary data from a digital health intervention trial [38]. Upon reviewing the reference list for the study, it was found that the authors had reported on body weight data for the same trial in a prior publication [39]. The trial was, therefore, included. During the second database and trial registry search, screening and eligibility assessment determined that no new trials met the criteria for inclusion in this review. Therefore, nine RCTs that explored the impact of a digital health intervention on body weight and dietary intake outcomes were included in this review, with outcomes reported in 10 publications (Figure 1) [35, 38, 39, 40, 41, 42, 43, 44, 45, 46].

Two reviewers with postgraduate qualifications and experience conducting systematic and curriculum reviews conducted the screening process. Reviewers discussed the inclusion and exclusion criteria prior to screening, with further discussions held after completing approximately 10% of screening to review decisions and resolve any conflicts. For title and abstract screening, proportionate agreement was 0.90 (Cohen's κ = 0.53). For full‐text screening, proportionate agreement was 0.87 (Cohen's κ = 0.45). These calculated kappa values indicate moderate inter‐rater agreement [47].

3.2. Study Characteristics

All trials were conducted between 2004 and 2022. There were 2716 participants in total across all the nine trials (see Table 1). Eight of the nine trials were conducted in the USA [39, 40, 41, 42, 43, 44, 45, 46], with only one conducted in New Zealand [35]. Seven trials involved participants from various ethnic backgrounds [35, 39, 41, 42, 44, 45, 46], and all participants in the remaining two trials were African‐American [40] and Latino [43]. Only one trial included participants from an Indigenous background [35]. Seven trials were conducted on individual adult participants [35, 39, 40, 41, 42, 43, 44]. The remaining two trials were conducted on parent–child dyads [46] and families [45]. This review only reports on results from the individual adult participants in these trials.

TABLE 1.

Characteristics of the randomized controlled trials included in the systematic review.

Author (year), study name, country	Aims	Duration (months) a	Sample size, n (retention) b	Mean age (SD) (years)	Female (%)	Race and ethnicity	Outcomes reported	Assessment method	Control arm	Intervention arm/s
Brewer et al. [40] (2022), FAITH!, United States	Improve cardiovascular health among African Americans in faith communities	8.5	85 (80.0%)	54.2 (12.3)	70.6	African American	Diet: HDS Body wt: BMI	HDS: self‐administered FFQ (short Delta NIRI FFQ) Body wt and ht. measured	Digital: i) Activity monitor	Digital: i) Activity monitor ii) Mobile app: 10 heart health education modules and weekly quiz; self‐tracking FV intake; moderated discussion platform and weekly health posts; 2 x weekly in‐app personalized messages tailored to stage of change; church leaders and past participants' video testimonials
										Non‐digital: i) Cluster level social incentives ii) Heart health booklet (at end)
Chang et al. [41] (2010) Mothers In Motion, United States	Prevent wt gain in low‐income mothers	10.5	129 (40.7%)	25.4 (4.0)	100	African American, White	Diet: FV intake Body wt: wt	FV: telephone‐administered short assessment form (NCI FV Short Assessment Form) Body wt measured	Non‐digital: i) Bi‐annual nutrition education during WIC recertification	Digital: i) 5 × 10–15 min interactive DVD chapters (1/fortnight): prompted evaluations of health statements; culturally sensitive testimonials on adopting healthy lifestyle behaviors; behavior change goal‐setting exercises and quizzes ii) 5 × 30 min moderated PSGTs (1/alternate fortnights): interactive DVD chapter discussions; goal sharing; behavior change challenge discussions; encouragement.
										Non‐digital: i) 5 x interactive DVD chapter summary pamphlets, ii) bi‐annual nutrition education during WIC recertification.
Kandula et al. [42] (2015), SAHELI, United States	Reduce ASCVD risk among South Asians	6	63 (100%)	50.0 (7.5)	63	Asian Indian, Pakistani	Diet: FV intake Body wt: wt	FV: interviewer‐administered F2F/ telephone 24HR Body wt measured	Non‐digital: i) Mailed explanation of clinical results and recommended PC visit, ii) monthly translated education mailings on ASCVD, diet, exercise, and wt loss	Digital: i) 6 × 15 min telephone counseling calls: motivational interviewing; behavior change goals; problem‐solving
										Non‐digital: i) Mailed explanation of clinical results and recommended PC visit, ii) 6 × 60–90 min weekly F2F group classes on heart‐disease related topics, iii) 4 × optional cultural heart health events.
Ni Mhurchu et al. [35] (2019), OL@‐OR@, New Zealand	Improve adherence to health‐related behavior guidelines among Māori and Pasifika communities	3	1451 (84.4%)	37.6 (12.8)	69.7	Māori, Pasifika	Diet: FV intake Body wt: wt and BMI	FV: self‐administered survey (New Zealand Health Survey) Body wt self‐reported	Digital: i) Control version of web and smartphone app: weekly thank you msg (app notification); countdown till access to active app version; data collection questionnaires (e.g., health behaviors; diet, PA)	Digital: i) Active version of web and smartphone app: setting customizable goals; self‐tracking dietary intake and PA, virtual rewards; 4–5 weekly culturally tailored reminders (e.g., goal review) and motivational messages sent as app notifications; access to recipes, PA information, and links to local support services; data collection questionnaires.
Rosas et al. [43] (2020), Vida Sana, United States	Promote wt loss among Latino adults in primary care	24	191 (96.9%)	50.2 (12.2)	61.8	Latino	Diet: DASH and FV intake Body wt: wt and BMI	DASH and FV intake: interviewer‐administered multiple‐pass 24HR Body wt and ht measured	Digital: i) Activity monitor Non‐digital: i) Usual PC	Digital: i) Diet and PA self‐monitoring via tracking app and activity monitor, ii) Weekly individualized feedback via smartphone app (1st 6 months), iii) Support as needed via smartphone app (2nd 6 months), iv) 12 × monthly emails (during 12‐month maintenance)
										Non‐digital: i) Usual PC ii) 16 × F2F sessions targeting wt loss, PA, and caloric intake goals (12 × wkly, then 4 × bimonthly in 1st 6 months), iii) 6 × monthly F2F sessions targeting maintenance (2nd 6 months).
Staten et al. [44] (2004), Arizona WISEWOMAN Project, United States	Increase PA and FV consumption in uninsured women, primarily Hispanic, over age 50	12	326 (67.0%)	57.2 (4.8)	100	Hispanic, African American, white (non‐Hispanic)	Diet: FV intake Body wt: wt and BMI	FV: interviewer‐administered F2F/ telephone 24HR Body wt and ht measured	Non‐digital: 3 × HCP counseling: i) Health education brochures ii) PA and FV intake discussed (benefits, barriers, and goals)	Intervention arm 1 Non‐digital: i) Same as control arm ii) 2 × health education classes: nutrition and PA, iii) 12 × monthly newslettersc
										Intervention arm 2 Non‐digital: i) Same as intervention arm 1 Digital: i) 26 × telephone support calls (1 × fortnight): PA and FV intake discussion, health behavior modification tips; knowledge assessments; bi‐monthly walk invite
Steinberg et al. [38] (2019) / Bennett et al. [39] (2018), Track, United States	Promote wt loss among low‐income adults with elevated cardiovascular risk	12	351 (92.0%)	50.7 (8.9)	68	Hispanic, non‐Hispanic black, non‐Hispanic other, non‐Hispanic white	Diet: DASH and DASH‐nutrient Body wt: wt and BMI	DASH and DASH‐nutrient: self‐administered FFQ (110‐item Block FFQ) Body wt and ht measured	Non‐digital: i) Usual PC, ii) Quarterly newsletters: health; financial; safety information, iii) Self‐help materials: wt loss and maintenance iv) Community resource list: healthy eating; PA; wt management	Digital: i) 4 × computer generated and ranked goals cycled every 8 weeks (3 × personalized goals and 1 × common intervention goal), ii) Daily weighing on network‐connected scales with feedback delivered weekly via SMS, iii) Weekly SMS prompt/IVR calls (2–3 min/call) to monitor progress, with feedback delivered weekly via SMS/IVR, v) 18 × 20–30 min wt loss coaching calls (4 × weekly, then 6 × bi‐weekly, then 8 × monthly): progress review; barrier reduction strategies; skills training; community resources; wt maintenance plans
										Non‐digital: i) Pedometers, behavior change tip materials (binder and DVDs), ii) 1–4 brief wt loss counseling delivered by PCP (encourage behavior change and participation), iii) Worksheet with current and goal wt
Wieland et al. [45] (2018), Healthy Immigrant Families, United States	Improve dietary quality and PA among immigrant and refugee families	24	44 families; 70 adults (68.2%)	39.1 (10.9)	71.4	Hispanic, Somali, Sudanese	Diet: HEI and FV intake Body wt: BMI	HEI and FV: self‐administered 24HR (ASA24: Web‐based) Body wt measured	i) Contacted up to 4 times for participation appreciationii) To receive intervention post‐trial completion	Non‐digital (1st 6 months): i) 12 × 30–90 min home visits to deliver intervention modules (6 × health eating, 4 × PA, 2 × reinforce content) and knowledge/health behavior assessments.
										Digital (2nd 6 months): i) 12 × 15 min fortnightly telephone calls to adult participants for goal progress updates and delivery of module content summary.
Wright et al. [46] (2013), HEAT, United States	Improve health behaviors in families from underserved populations	3	50 parent–child dyads; 50 adults (86.0%)	40.0 (9.1)	96	African American, White, Other	Diet: FV intake Body wt: wt and BMI	FV: interviewer‐administered screener (Block 2007 Fat, Sugar, Fruit, and Vegetable screener) Body wt and ht measured	To receive intervention post‐trial completion	Digital (parent): i) 2 × weekly IVR calls: 1st call: goal review; progress feedback; education; goal setting, and 2nd call: self‐report progress update.
										Non‐digital (parent): i) IVR call guidebook.

Abbreviations: 24HR, 24‐h dietary recall; ASCVD, atherosclerotic cardiovascular disease; BMI, body mass index; DASH, Dietary Approaches to Stop Hypertension; F2F, face‐to‐face; FFQ, food frequency questionnaire; FV, fruit and vegetable; HEI, Healthy Eating Index; HDS, Healthy Diet Score; ht, height; IVR, interactive voice response; PA, physical activity; PC, primary care; PCP, primary care provider; PSGT, peer support group teleconference; WIC, Women, Infants, and Children; wt, weight.

^a Trial duration is inclusive of intervention duration and maintenance/follow‐up phases, with durations converted into months if reported in weeks.

^b Retention rate evaluated at the final time point assessed.

^c Assumed non‐digital delivery method as intervention delivery occurred between 1998 and 2000.

Eight of the nine trials were two‐arm RCTs [35, 39, 40, 41, 42, 43, 45, 46]. The trial by Staten et al. [44] was a three‐arm RCT with two intervention groups where only one group received a digital intervention combined with a non‐digital intervention. Participants in two trials were all female [41, 44] and were predominantly female in the remaining seven [35, 39, 40, 42, 43, 45, 46]. Health status restrictions were placed on the participants in four of the nine trials [39, 41, 42, 43]. They were required to be living with overweight or obesity [39, 41, 43], a chronic disease [39, 42], or at least one chronic disease risk factor [42, 43]. Trial duration, inclusive of intervention delivery, follow‐up, and maintenance phases, ranged from three months [35, 46] to twenty‐four months [43, 45]. Across trials, participant retention varied from 41% [41] to 100% [42] at final assessed time points.

3.3. Intervention Components

The aim of three trials was weight management‐related: two at achieving weight loss [39, 43] and one at preventing weight gain (see Table 1) [41]. Two trials specifically targeted diet and physical activity [44, 45], two trials focused on cardiovascular‐related behaviors [40, 42], and two trials addressed generic health behaviors [35, 46]. Cultural tailoring and adaptations were not used in three trials’ intervention design [39, 44, 46].

Three trials used a mobile app to deliver the digital intervention featuring data tracking and individualized feedback delivery [35, 40, 43]. Resource access was also available through an app for two trials [35, 40] but was delivered via email for the other [43]. Chang et al. [41] used an interactive digital video disc (IDVD) to deliver the intervention. The IDVD involved interactive chapters requiring participants to engage with content using an input device and was supported by fortnightly moderated teleconferences with other participants.

Five trials used telecommunication support to deliver digital interventions [39, 42, 44, 45, 46]. Telephone calls primarily provided health education and counseled participants through behavior change [39, 42, 44, 45, 46], with the frequency ranging from 6 [42] to 26 calls [44]. Telephone support in four of the trials was delivered by a human [39, 42, 44, 45]. Wieland et al. [46] used an interactive voice response (IVR) to deliver tailored health education. Bennet et al. [39] used IVR to collect self‐monitoring data and delivered personalized feedback in response via SMS.

Eight of the nine trials delivered digital interventions complemented by the use of non‐digital resources [39, 40, 41, 42, 43, 44, 45, 46], with the remaining trial only using a Web and mobile app as the intervention [35]. Digital interventions were supported by the following: home visits [45]; soft and hard copies of healthy lifestyle education resources [39, 40, 41, 44, 46]; face‐to‐face education or coaching delivered by healthcare or medical professionals [39, 41, 44]; face‐to‐face events [42]; and face‐to‐face classes with other participants [42, 43, 44].

3.4. Effectiveness of Digital Health Intervention on Body Weight Outcomes

Five trials reported BMI and weight (kg) together [35, 39, 43, 44, 46], two reported BMI [40, 45], and two reported body weight (kg and pounds) [41, 42]. Mean differences in weight (see Figure 2) and BMI (see Figure 3) have been visually presented on forest plots and grouped by trial duration: 3 months, 6–12 months, and ≥12 months. Three trials reported significant changes in weight between the intervention and control group; −1.5 kg at 6 months [42], –4.4 kg at 6 months and −3.8 kg at 12 months [39], and −2.1 kg at 12 months [43]. Two of the three trials reporting BMI demonstrated a statistically significant difference in BMI between the intervention and control group; −0.8 kg/m² at 12 months [43], and −1.6 kg/m² at 6 months and −1.4 kg/m² at 12 months [39]. These trials also reported that intervention participants had a 23.7% (p < 0.001) [39] and 16.7% (p = 0.003) [43] higher likelihood of achieving at least a 5 % weight loss compared to control participants at 12 months.

FIGURE 2.

Mean differences and associated 95% confidence interval of difference of weight across grouped assessment time points.

FIGURE 3.

Mean differences and associated 95% confidence interval of difference of BMI across grouped assessment time points.

3.5. Effectiveness of Digital Health Intervention on Dietary Intake Outcomes

The methods used to capture dietary intake varied and included 24‐h dietary recalls (24HR) (interviewer‐administered [42, 43, 44] and self‐administered web‐based ASA24 [45]), self‐administered food frequency questionnaires (FFQs) (short Delta NIRI FFQ [40] and 110‐item Block FFQ [38]), and brief instruments (self‐administered [35, 46] and telephone‐administered [41] pre‐existing screeners and surveys).

Diet quality indices were used in four trials to evaluate dietary intake; the Healthy Diet Score (HDS) [40], Dietary Approaches to Stop Hypertension (DASH) nutrient score [38], DASH food score [38, 43], and Healthy Eating Index (HEI) score [45]. Given the use of multiple diet quality scores, for the purposes of this review, diet quality changes have been visually presented on a forest plot as percentage differences between intervention and control groups' scores (see Figure 4). The forest plot illustrates that intervention group participants generally had higher diet quality scores than those in the control group. Significant improvements in diet quality scores in intervention participants compared to control were reported in three of the four trials [38, 40, 45], ranging from 12% [38] to 17% [40]. Steinberg et al. [38] reported an increase of 1 point in the DASH nutrient score that was associated with a significant decrease of 1.37 kg in weight (p = 0.002). Wieland et al. [45] reported a significant change in the HEI score for the control group after receiving the intervention at 12 to 24 months (p = 0.01).

FIGURE 4.

Mean percentage differences and associated 95% confidence interval of difference of standardized diet quality scores across assessed time points.

Seven trials evaluated fruit and vegetable intake [35, 41, 42, 43, 44, 45, 46], using varied measures. Fruit and vegetable intake was measured using cups [41, 46], serving sizes [35, 42, 43, 44], or cup equivalents per day [45]. For the purposes of this review, units for fruit and vegetable intake were standardized to the number of servings per day. Changes in fruit and vegetable intake across trials and assessed time points are visually presented on a forest plot and grouped by trial duration: 3 months, >3 months to < 12 months, and ≥12 months (see Figure 5). Changes in fruit and vegetable intake ranged from −3.4 [41] to +3.2 [41] servings per day. Differences observed were however not significant or statistical significance was not evaluated.

FIGURE 5.

Mean differences and associated 95% confidence interval of difference of fruit and vegetable intake across grouped assessment time points.

Six trials assessed other dietary components including fat [41, 42, 43, 46], energy [42, 43, 46], alcohol [35], and sugar [45]. These outcomes were not consistently evaluated across trials and no significant changes were observed. Therefore, these dietary intake outcomes were not included in this review.

3.6. Intervention Adherence

Seven trials reported on intervention adherence [35, 39, 40, 41, 42, 43, 46]. Brewer et al. [40] reported that all intervention participants successfully logged into the mobile app but only 40% completed at least half of the 10 required modules. Chang et al. [41] reported that 66% of the intervention group indicated they had viewed one or more chapters, and 48% participated in one or more teleconferences. Kandula et al. [42] reported that 16% of intervention participants completed at least half of the six counseling calls with no participants completing all six. Ni Mhurchu et al. [35] reported that only 7% of 23,233 notifications were recorded as read, 61% of intervention participants did not open any, and 26% of participants set at least one behavior change goal. Two trials did not comment on intervention adherence [44, 45].

Three trials that assessed intervention adherence examined their effects on dietary intake and body weight outcomes. Rosas et al. [43] found that greater session attendance and diet, weight, and physical activity monitoring were associated with greater weight loss. Bennett et al. [39] reported significantly greater weight loss in intervention participants who completed over 80% of the required daily weighing (p = 0.0004), counseling calls (p = 0.01), and self‐monitoring (p = 0.004). Participants completed 54–100% of weekly self‐monitoring (median 93.2%) and 50–100% of coaching calls (median 89%). Participants completed 42.9% of required daily self‐weighing on 1.2–4.5 days/week (median 2.8 days). Wright et al. [46] reported that 76% of the adults called the IVR once or more of the 24 required times. The mean number of total completed IVR calls was 9.1/24; education and behavior calls (mean 5.2/12) and tracking calls (mean 3.9/12). Participants who completed four or more IVR calls ate significantly fewer calories than those who completed less than four (−454.0 kcal; p = 0.02).

3.7. Risk of Bias

A summary of the risk of bias is shown in Figure 6. All trials had some risk of bias in the measurement of the outcome domain due to the use of self‐reported dietary intake. Six trials had a low risk of bias in the remaining four domains [34]. One trial was considered to have a high risk of bias in the deviations from the intended intervention domain due to limited information being provided to determine if intention‐to‐treat analysis was used [45]. Three trials were considered to have some concerns of bias in the selection of the reported results domain, as there was insufficient information available to determine if data were analyzed in accordance with a pre‐specified plan [42, 44, 46]. Consequently, eight trials had some concerns of bias for the overall risk of bias, with one trial considered high‐risk [45].

FIGURE 6.

Cochrane risk‐of‐bias for randomized trials (RoB 2) summary for all trials.

4. Discussion

This review aimed to evaluate the efficacy of digital health interventions on body weight and dietary intake in CALD and Indigenous populations. Five of the nine digital health intervention trials included in this review demonstrated statistically significant changes in body weight or dietary intake outcomes between intervention and control participants, with only one trial demonstrating significant improvement in both. There is therefore minimal evidence to support the effectiveness of digital health intervention trials for improving body weight and dietary intake outcomes in CALD and Indigenous populations. The small number of identified trials included also confirms the limited representation of CALD and Indigenous populations in digital health intervention trials [16, 17].

This review found that only two trials achieved clinically significant weight loss outcomes greater than 5% of body weight [39, 43]. These results are similar to those demonstrated in systematic reviews evaluating digital health interventions on weight outcomes in CALD populations [18, 19]. A suggested intervention approach to promote clinically significant weight loss is through the delivery of personalized care [48]. This approach is suggested to encourage successful outcomes by making intervention content more relevant to participants, reducing the mental effort required to understand, engage with, and adhere to the content [49, 50]. This is particularly important when designing interventions for priority populations due to the significant impact of genetic, dietary, cultural, and socioeconomic factors on obesity prevalence and outcomes. CALD and Indigenous populations consistently express the need for personalized and culturally sensitive care to enhance the relevance and thus effectiveness of healthcare interventions [51, 52, 53, 54]. Nearly all included trials used personalized feedback as part of their intervention, with most also using formative research [41, 42, 43, 44, 45] or co‐design [35, 40, 43, 45] to ensure intervention content met participants' cultural needs. This, however, did not consistently translate into clinically significant weight loss among included trials. This may be attributed to several factors, such as statistical power and intervention design [55, 56]. Seven of the nine trials included in the review did not report on statistical power [40, 41, 42, 43, 44, 45, 46]. Four of these trials identified themselves as pilot studies [40, 41, 42, 46], where the lack of formal power calculations and statistical significance testing is methodologically justified [57]. All four trials acknowledged that their findings warranted larger RCTs, with one trial explicitly outlining intentions to conduct a larger trial in future [42]. However, the overall lack of power reporting across trials makes it challenging to determine if they were adequately powered to detect significant effects. Furthermore, the intervention design of included trials, which often involved multi‐component approaches, might have influenced the outcomes. While multi‐component interventions are generally viewed as beneficial for weight loss [58, 59, 60], their complexity can present challenges in implementation and adherence, particularly within culturally diverse adult populations. Disparities in access to resources can result in reduced capacity among participants to engage in interventions [61, 62], affecting the equitable uptake of multicomponent lifestyle interventions. This is reinforced by a qualitative process evaluation [63] that was conducted for the included SAHELI trial [42]. The multicomponent nature of the intervention notably contributed to participant burden, which was identified as a key barrier to intervention adherence [63]. The challenge of intervention adherence was not specific to the SAHELI trial alone.

Across identified trials in this review that evaluated adherence [35, 39, 40, 41, 42, 43, 46], less than half of participants adhered to the required digital intervention components. Adherence has been noted to play a role in improving dietary intake and body weight outcomes in digital health interventions [64, 65]. This is supported by three of the identified trials which demonstrated that greater intervention adherence led to significantly greater changes in weight loss [39, 43] and energy intake [46]. Consequently, the low intervention adherence across the identified trials may have contributed to the inconsistent and minimal improvements observed in dietary intake and body weight outcomes. This challenge of low adherence is commonly credited to activities and digital features of the intervention as well as participant characteristics that may serve as barriers to intervention engagement [66, 67, 68, 69]. The clinical heterogeneity of participant demographics and intervention design, such as intervention content and delivery method, across included trials makes it difficult to determine which factors contributed to the observed low adherence. Variations in digital health intervention adherence and engagement by ethnicity or race may exist [70, 71]; however, this was not reported on in identified trials. Future research should examine differences in adherence to digital health intervention components by conducting detailed subgroup analyses to examine how participant characteristics, such as ethnicity and race, interact with intervention components to help determine which digital intervention components are most appropriate for use in specific cultural populations.

Differences in dietary intake varied across assessed outcomes. Of the four trials reporting diet quality scores in this review, three demonstrated significant changes. A previous systematic review and meta‐analysis reported that digital health interventions may lead to improvements in diet quality outcomes; however, due to the lack of statistical significance in the observed changes, improvements were not guaranteed [14]. Of the seven trials in this review that evaluated fruit and vegetable intake, none exhibited a significant change. A similar result was observed in a previous review and meta‐analysis [14], which demonstrated no changes in pooled studies reporting on combined fruit and vegetable intake. The contrasting finding between significant changes observed in diet quality outcomes and fruit and vegetable intake may be due to diet quality scoring systems integrating multiple dietary factors into a single metric, providing a comprehensive assessment of overall dietary patterns [72]. This approach is sensitive to variations in dietary patterns and can reflect broader shifts in dietary intake, even if changes in food group servings may not be as pronounced when assessed in isolation. The use of a diet quality metric is also a preferred dietary intake evaluation approach as it may more effectively predict health outcomes when compared to isolated food groups or nutrients [73]. Furthermore, despite most identified trials having multiethnic cohorts, dietary intake data were reported collectively for all participants rather than by ethnicity subgroups. Variations in food group and nutrient intake exist across CALD and Indigenous populations [74, 75]. Consequently, focusing on the intake of individual macronutrients or food groups rather than overall diet quality, as was done in most trials, may not have been sufficient to capture total dietary intake improvements. Future trials should consider approaching dietary intake evaluations using diet quality scoring systems. Such scoring systems should also consider assessing at‐risk dietary components that are prevalent within cultural dietary practices [76] to better reflect the unique dietary patterns of specific cultural groups and predict the impact of interventions on relevant health outcomes.

Dietary assessment methods varied across the included trials. Trials tended to use repeated measures of low‐cost, non‐burdensome, self‐report dietary assessments, such as 24HRs, FFQs, and dietary screeners [77], to capture changes in dietary intake over the trial period. This is consistent with the findings of a previous review that reported on dietary assessment methods used in digital health interventions [14]. Measurement inaccuracies, such as systematic and random error, exist in all dietary intake data captured using self‐reported dietary assessment methods [77]. A 24HR is, however, notably more accurate at capturing intake than an FFQ or brief instrument due to its detailed approach to probing and low cognitive load [77]. However, it is not impervious to errors, which are primarily due to daily fluctuations in intake. While it is possible to counteract errors within 24HRs by conducting recalls on multiple days at each dietary intake assessment time point [78, 79], only one trial did this [44]. This trial, however, did not report any significant changes in dietary intake among participants. All included trials in this review also used repeated self‐report dietary assessment methods, which, in addition to systematic and random error, may be prone to differential measurement error. This error refers to differences in how accurately and consistently participants report dietary intake at different timepoints and under different study arms within an intervention [80]. The cumulative effect of differential and measurement error produces greater variability in dietary intake data, which can reduce statistical power and misconstrue the true impact of delivered interventions on dietary intake outcomes [80], particularly in digital health interventions where small effect sizes are often seen [81]. No single dietary assessment method is likely to be suitable for all populations and study contexts; however, future trials may consider exploring digital dietary assessment tools, such as image‐based methods. The nature of image‐based dietary assessment, which captures entire meals and eating occasions rather than isolated food items [82], makes it particularly suitable for assessing overall dietary patterns and diet quality. While these approaches remain subject to self‐report limitations, such as social desirability bias [83], validation studies have demonstrated good agreement for energy assessment in both controlled feeding and community‐dwelling settings [84, 85]. Evidence also suggests these methods may reduce some of the measurement challenges inherent in traditional dietary assessment approaches, such as portion size estimation errors and recall bias [83]. These tools may therefore offer a promising approach for improving dietary assessment in digital health interventions.

Long‐term evaluation monitoring was notably limited across trials. Two of the identified trials included in this review sought to evaluate dietary intake and body weight outcomes beyond 12 months [43, 45], but only one evaluated differences between the control and intervention arms [43]. This trial by Rosas et al. [43] showed a predominant downward trend for dietary intake outcomes and demonstrated that significant differences in weight loss outcomes were not maintained. This is a common result of prolonged interventions aimed at improving dietary intake and weight outcomes in the broader literature. Weight regain [86] and reduced compliance with dietary recommendations [87] are expected. However, given the paucity of literature surrounding digital health interventions evaluating dietary intake and body weight outcomes in CALD and Indigenous populations, future trials should aim to observe longer follow‐up durations to determine the long‐term intervention effectiveness in this population group.

4.1. Strengths and Limitations

To our knowledge, this is the first systematic review to evaluate the effectiveness of digital health interventions on body weight and dietary intake in CALD and Indigenous populations. This systematic review identified a small number of eligible trials, as many trials did not specify the ethnicity of participants and were therefore excluded. Among the included trials, only two provided details of sample size calculations [35, 39]. This raises concerns that many of the trials did not have sufficient power to detect the true effect of interventions [88], warranting prudence when interpreting results. Few trials were captured during screening that evaluated digital health interventions in Indigenous populations. This may have been a consequence of a non‐exhaustive search strategy due to a lack of standardized identification criteria [89] and the low representation of Indigenous populations in the broader digital health literature [17]. There was a predominance of US‐based trials and female participants, limiting the applicability of findings to the broader CALD and Indigenous populations. Most of the included digital trial interventions were supported by non‐digital components. Therefore, it would be remiss not to acknowledge the potential role that non‐digital components may have contributed to the observed results. These limitations collectively highlight significant gaps in the current evidence base for digital health interventions in CALD and Indigenous populations, demonstrating the need for more targeted and well‐designed research approaches in these populations.

It is also important to consider our search approach. Our search strategy required both weight and diet outcomes to be mentioned in identified trials. Acknowledging that interventions might report weight and diet outcomes across multiple publications, efforts were made during title and abstract screening to include digitally delivered interventions mentioning either outcome alone for full‐text assessment. Although this approach aimed to be comprehensive, some relevant trials where outcomes were published separately may not have been captured. Future reviews could explore using broader search terms (e.g., weight OR diet) followed by full‐text screening of relevant studies to potentially capture additional RCT interventions that assess both outcomes across separate publications.

5. Conclusions

This review reveals limited evidence for digital health intervention effectiveness in CALD and Indigenous populations for weight management and dietary intake improvement, with poor adherence rates suggestive of engagement challenges. The heterogeneity in intervention design and participant demographics across trials additionally makes it difficult to identify which intervention components are most effective for these priority populations.

To address these evidence challenges, funding bodies should prioritize digital health research in priority populations. Future research should emphasize inclusive practices to better represent and engage CALD and Indigenous populations in digital health interventions. Such research can facilitate understanding of how participant characteristics interact with intervention components and ultimately support more effective interventions for sustainable weight management and dietary behavior change in these population groups. Given rapid advancements in digital health technologies, ongoing evaluation of their effectiveness in CALD and Indigenous populations is warranted to build the evidence base needed for informed clinical practice and policy development.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Appendix S1: Completed PRISMA Abstract Checklist and PRISMA 2020 Checklist.

Appendix S2: Search strategy (MEDLINE, EMBASE, COCHRANE, SCOPUS).

Hassan A., Breare H., Rollo M., et al., “Effectiveness of Digital Health Interventions on Body Weight and Dietary Intake Outcomes Among Culturally and Linguistically Diverse (CALD) and Indigenous Populations: A Systematic Review,” Obesity Reviews 27, no. 3 (2026): e70035, 10.1111/obr.70035.

Data Availability Statement

The data that supports the findings of this study are available in the supplementary material of this article.